240 thesis methods

Scope

240 manually assigned study-unit/estimand/endpoint/model scaffolds, with 12 expanded prospectus refinements taking precedence

Notice

Research-design mathematics only. No fresh medical-source verification, demonstrated novelty, validated protocol, clinical result, asserted effect size, powered recruitment target or data access is claimed. Formulas are candidate estimators/identification and planning scaffolds; outcome, decision, assay, death and access choices remain unresolved.

Source precedence

The 12 expanded prospectuses refine the corresponding original briefs. All other records use full original question, aims, design, outcomes and analysis. Scientific source metadata remain inherited/unverified.

Mathematical conventions

Potential-outcome notation marks a causal target, not automatic identification. Prediction/assay/cohort models retain their narrower interpretation.
All outcome links, thresholds, contrasts, time windows, scales, covariate sets and intercurrent-event rules are candidates until preregistered. Mean, risk, rate, ordinal and independent-draw superiority effects answer different questions.
Randomized assignment effects include subsequent clinician-directed treatment and incomplete attendance under the relevant strategy; ordinary adjustment for postassignment variables can change the target or induce bias.
An experimental unit can differ from the independent biological/source/transport unit. Repeated samples improve measurement precision but cannot manufacture independent donors, participants, lots, trials, laboratories or policy events.
Linear-Gaussian contrast variance is a planning approximation for that model only. Nonlinear/marginal, clustered, censored, survival, adaptive and interaction designs require estimator-specific or simulation planning.
Death, missing living outcomes, valid assay nondetects and technical assay failures are distinct states. No conventional living score is imputed after death and no arbitrary positive potency replaces a nondetect.
Independent-draw superiority is a contrast of outcome distributions, not the probability a particular person benefits. Individual counterfactual effects are not observed ground truth.
Verification gates concern design/provenance/calculation/statistics separately; arithmetic correctness cannot validate biomedical inputs, model assumptions or causal efficacy.

E01 — Which personal responders remain responders?

Rank: 7
Cluster: Quantified-self, measurement, and N-of-1 causal methods
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: REPLICATE; allocation or exposure unit: eligible workday within participant, separately in each randomized phase; independent sampling unit: participant; repeated measurement unit: workday vigilance test within phase; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Freeze phase-one predictions before phase-two outcomes; estimate prediction uncertainty and lagged prompts; never condition on taking the break.
Primary estimand: phase-two interaction of assignment with frozen phase-one personal prediction
Candidate primary outcome: mean reciprocal reaction time in the standardized afternoon test
Primary finalized: No
Mathematical model: type: effect-modification/replication calibration simulation with prediction uncertainty, repeated randomization and independent validation units; formulas: E(Y_it | A_it,P_i,X_it)=alpha_i+tau A_it+kappa A_it P_i+gamma^T X_it+f_time(t). With unrestricted fixed alpha_i the time-invariant prognostic main effect of P_i is absorbed, not separately estimable.
Primary predictive-effect parameter is kappa in independent randomized evaluation; assess calibrated average effects and proper held-out loss separately.
Phase one: random slopes are shrunk; freeze P_i before phase two. Phase two alone estimates kappa, with participant/day dependence and phase-one uncertainty sensitivity.; variables: Y: mean reciprocal reaction time in the standardized afternoon test; i: participant; t: workday vigilance test within phase; the registered visit/time-window index, not another independent unit; A: eligible workday within participant, separately in each randomized phase; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: phase-two interaction of assignment with frozen phase-one personal prediction; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; P i: frozen pre-evaluation personal/phenotype effect prediction (or pretrial acute-response contrast for P19), with scale fixed before evaluation; eta: prognostic main effect of prediction; kappa: treatment-by-frozen-prediction coefficient, distinguishing predictive modification from prognosis; X it: registered predecision or baseline covariates; no future measurements; f time(t): predeclared practice/calendar trend; family code: REPLICATE
Sample size design: family: effect-modification/replication calibration simulation with prediction uncertainty, repeated randomization and independent validation units; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure
distribution/reliability of frozen predictions
true effect heterogeneity scenarios
within-person randomization schedule
independent validation sample/cluster coverage; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Freeze phase-one predictions before phase-two outcomes; estimate prediction uncertainty and lagged prompts; never condition on taking the break.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Predictions are frozen before evaluation; noisy person-specific estimates are not observed counterfactual truth; training/evaluation dependence must be propagated.
A nonzero interaction does not by itself establish useful calibration or superiority over a population-only policy.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: A sufficiently precise absence of phase-two treatment-effect moderation would undermine reproducible personalization. Even a positive interaction would not establish useful calibration, accurate individual counterfactual effects or superiority to the population-average prediction.; preanalysis gate: Freeze phase-one predictions before phase-two outcomes; estimate prediction uncertainty and lagged prompts; never condition on taking the break.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Freeze phase-one predictions before phase-two outcomes; estimate prediction uncertainty and lagged prompts; never condition on taking the break.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Freeze phase-one predictions before phase-two outcomes; estimate prediction uncertainty and lagged prompts; never condition on taking the break.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/6; question: Does a personal treatment-effect prediction frozen after phase one predict between-person differences in the randomized phase-two afternoon-vigilance effect, tested by the treatment-by-prediction interaction? Prediction calibration and superiority are separate secondary questions.; aims: Estimate measurement reliability before experimentation.
Run randomized individual crossover trials of activity-break prompts versus neutral prompts.
Repeat the experiment months later and test whether initial response predicts replication.; design verbatim planning only: Two eight-week phases separated by approximately four months use repeated within-person workday randomization to accessible activity-break offers or attention-matched neutral prompts. Fix eligibility, weekday/schedule strata and carryover rules. Freeze phase-one shrinkage-based personal predictions before phase two; vigilance measurement and research contact follow the same schedule in both conditions.; outcomes verbatim: Primary reproducibility test: the phase-two treatment-by-frozen-personal-prediction coefficient for standardized afternoon vigilance response. The independently measured outcome is prespecified mean reciprocal reaction time; average assignment effects, fatigue, discomfort and practical policy value are secondary.; analysis verbatim inherited not authoritative if refined: Fit the preregistered phase-two treatment-by-frozen-prediction interaction with assignment and participant terms, propagating phase-one prediction uncertainty and repeated-day dependence. Compare with population-average prediction using untouched phase-two data. Do not treat noisy person-specific contrasts as observed causal ground truth; prespecify carryover, practice, missingness and multiplicity sensitivity analyses.; falsification verbatim: A sufficiently precise absence of phase-two treatment-effect moderation would undermine reproducible personalization. Even a positive interaction would not establish useful calibration, accurate individual counterfactual effects or superiority to the population-average prediction.; source ids unverified leads: n-of-1-methods-review-2025
digital-n-of-1-physiology-2026; prospectus precedence: Yes; prospectus file: thesis-prospectuses.json; prospectus design and estimand: Each participant completes two eight-week randomized phases separated by approximately four months. Randomization occurs at the eligible workday within person, balancing weekdays and predefined work-schedule strata; neutral prompts have comparable notification contact. The intervention is an offer to take ordinary accessible movement breaks, not enforced exercise. The prompt-assignment effect is the causal daily estimand. The primary reproducibility parameter is the phase-two treatment-by-frozen-personal-prediction coefficient, with average treatment effect and participant intercepts included.; prospectus analysis plan: Fit a phase-one hierarchical model with participant intercepts and treatment slopes, day-of-week effects, predefined time trends, and within-person autocorrelation. Freeze code and predictions before unlocking phase two. In phase two, estimate the treatment-by-prediction interaction and compare predictive scores with a population-only model using a prespecified scoring rule. Include day and participant dependence in uncertainty estimates. Examine lagged assignment for carryover. Avoid conditioning the primary analysis on taking the prompted breaks, because adherence is post-randomization.; precedence note: Expanded prospectus refinements govern this scaffold; verbatim earlier brief material is retained for traceability, not as a contradictory analysis instruction. F15 is explicitly two arms; F30 has no unjustified symptom noninferiority claim.

E02 — The false-discovery cost of checking every wearable metric

Rank: 44
Cluster: Quantified-self, measurement, and N-of-1 causal methods
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: SIM; allocation or exposure unit: permissible null-label assignment or simulated study dataset; independent sampling unit: independent participant trajectory; independently generated evaluation replicate; repeated measurement unit: days and metric families within trajectory; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: operating characteristics under explicit null/data-generating scenarios; no clinical treatment effect; specific constraint: Autocorrelation, stopping, missingness and selected outcomes must be reproduced; resampling days as independent would understate false discovery.
Primary estimand: probability that an outcome/stopping rule makes any false confirmatory discovery under the null
Candidate primary outcome: indicator of at least one false rejection per complete simulated or randomized-null study
Primary finalized: No
Mathematical model: type: Monte Carlo operating-characteristic precision plus independent-participant transport robustness; formulas: FWER_hat_r=(1/B) sum_b 1{at least one false confirmatory rejection under rule r in null replicate b}.
Monte Carlo SE ~= sqrt[FWER_hat_r(1-FWER_hat_r)/B] for independent simulation replicates; if replicates share resampled data, also quantify original-participant resampling uncertainty.; variables: Y: indicator of at least one false rejection per complete simulated or randomized-null study; i: independent participant trajectory; independently generated evaluation replicate; t: days and metric families within trajectory; the registered visit/time-window index, not another independent unit; A: permissible null-label assignment or simulated study dataset; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: probability that an outcome/stopping rule makes any false confirmatory discovery under the null; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; B: number of independently generated null evaluation replicates, not number of observed days; b: simulation/randomized-null replicate; r: fixed stopping/outcome-selection rule; FWER hat r: empirical familywise false-rejection frequency; family code: SIM
Sample size design: family: Monte Carlo operating-characteristic precision plus independent-participant transport robustness; formula: For independent replicates only, MCSE(p_hat)=sqrt[p_hat(1-p_hat)/B]; choose B for prespecified Monte Carlo precision, separately from human-data precision.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Autocorrelation, stopping, missingness and selected outcomes must be reproduced; resampling days as independent would understate false discovery.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Null labels must be exchangeable under the actual allocation design; simulate serial correlation, optional stopping and missingness.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: The proposed rule fails if nominal error control breaks under realistic missingness or its apparent gains disappear with reference-grade outcomes.; preanalysis gate: Autocorrelation, stopping, missingness and selected outcomes must be reproduced; resampling days as independent would understate false discovery.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Autocorrelation, stopping, missingness and selected outcomes must be reproduced; resampling days as independent would understate false discovery.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Autocorrelation, stopping, missingness and selected outcomes must be reproduced; resampling days as independent would understate false discovery.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/43; question: Which stopping and outcome-selection rules prevent false personal discoveries without making self-experiments impractically long?; aims: Build empirically calibrated null streams from repeated reference measurements.
Compare informal checking with preregistered sequential and fixed-horizon rules in simulation.
Prospectively test decision quality in low-risk randomized self-experiments.; design verbatim planning only: Assume 180 volunteers contribute twelve weeks of activity and symptom data; 80 subsequently randomize neutral versus movement reminders. Simulations preserve observed autocorrelation and missingness rather than inventing independent daily observations.; outcomes verbatim: Primary: empirical false-positive rate under randomized null labels. Secondary: decision delay, false negatives, participant understanding, and independent replication.; analysis verbatim inherited not authoritative if refined: Separate algorithm development and evaluation cohorts. Control familywise error for confirmatory outcomes; examine false-discovery procedures only for labeled exploration and quantify sensor uncertainty.; falsification verbatim: The proposed rule fails if nominal error control breaks under realistic missingness or its apparent gains disappear with reference-grade outcomes.; source ids unverified leads: n-of-1-methods-review-2025
digital-n-of-1-physiology-2026
wearables-living-review; prospectus precedence: No

E03 — Do uncertain dashboards improve decisions or merely reduce confidence?

Rank: 185
Cluster: Quantified-self, measurement, and N-of-1 causal methods
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT_COUNT; allocation or exposure unit: participant randomized to one display format; independent sampling unit: participant; repeated measurement unit: daily activity decisions; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Fix the primary display comparison and independent cancellation rubric; nonresponse is not cancellation or inactivity.
Primary estimand: uncertainty-plus-guidance versus point-score offer effect on unnecessary cancellations
Candidate primary outcome: count of unnecessary cancellations of planned ordinary activities during the prespecified follow-up
Primary finalized: No
Mathematical model: type: overdispersed count contrast planning by simulation, with follow-up/cluster/zero inflation assumptions; formulas: Y_it ~ NegativeBinomial(mu_it,phi); log(mu_it)=alpha_t+z(A_i)^T theta_t+beta Y_i0+gamma^T X_i+b_cluster(i)+u_i.
Delta = standardized E[Y | do(A=1)] - E[Y | do(A=0)] under randomization; report ratio exp(c^T theta) only when the log-mean model supports it.; variables: Y: count of unnecessary cancellations of planned ordinary activities during the prespecified follow-up; i: participant; t: daily activity decisions; the registered visit/time-window index, not another independent unit; A: participant randomized to one display format; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: uncertainty-plus-guidance versus point-score offer effect on unnecessary cancellations; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; mu it: conditional mean of the actual per-thesis count; phi: overdispersion parameter, with Var(Y)=mu+mu^2/phi under this parameterization; z(A): prespecified indicators for the actual arms or randomized policy; family code: RCT_COUNT
Sample size design: family: overdispersed count contrast planning by simulation, with follow-up/cluster/zero inflation assumptions; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Fix the primary display comparison and independent cancellation rubric; nonresponse is not cancellation or inactivity.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: A reduction in reported anxiety accompanied by worse function or unchanged avoidance would refute the proposed practical benefit.; preanalysis gate: Fix the primary display comparison and independent cancellation rubric; nonresponse is not cancellation or inactivity.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Fix the primary display comparison and independent cancellation rubric; nonresponse is not cancellation or inactivity.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Fix the primary display comparison and independent cancellation rubric; nonresponse is not cancellation or inactivity.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/184; question: Does showing calibrated uncertainty around recovery scores change behavior, anxiety, and functional performance?; aims: Co-design interpretable uncertainty displays with diverse users.
Randomize display format while holding underlying sensor data constant.
Test whether effects persist after feedback is withdrawn.; design verbatim planning only: Planning assumption: 240 existing wearable users receive eight weeks of point scores, uncertainty intervals, or intervals plus decision guidance, followed by four weeks without scores. Assignment occurs at participant level to avoid format carryover.; outcomes verbatim: Primary: unnecessary cancellation of planned ordinary activities. Secondary: health anxiety, score comprehension, daily function, and independent vigilance performance.; analysis verbatim inherited not authoritative if refined: Adjust for baseline anxiety and prior tracking intensity; randomization identifies total display effects. Prespecify one primary comparison, model repeated measures, and distinguish missing reports from genuine inactivity.; falsification verbatim: A reduction in reported anxiety accompanied by worse function or unchanged avoidance would refute the proposed practical benefit.; source ids unverified leads: belief-effects-health-review-2026
nocebo-communication-review-2026
wearable-stress-measurement-review-2026; prospectus precedence: No

E04 — Firmware drift as an invisible self-experiment confounder

Rank: 189
Cluster: Quantified-self, measurement, and N-of-1 causal methods
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: DID; allocation or exposure unit: observed device-version update, not research randomization; independent sampling unit: independent update/version event with participants nested or crossed by device; repeated measurement unit: paired device-reference visits over calendar time; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: quasi-experimental policy/update association with conditional causal interpretation; observational timing is not randomization; specific constraint: Choose one confirmatory error endpoint; participants do not create additional independent firmware changes; require unaffected comparison and update dates.
Primary estimand: update-associated change in heart-rate/activity measurement error relative to unaffected versions
Candidate primary outcome: device-minus-reference error in a prespecified heart-rate or activity endpoint
Primary finalized: No
Mathematical model: type: staggered adoption/event-study precision with independent adoption units and bias/parallel-trend sensitivity; formulas: ATT(g,t)=E[Y_t-Y_(g-1) | G=g] - E_w[Y_t-Y_(g-1) | G>t or G=never], with predeclared eligibility/comparator weights.
Aggregate only identified group-time effects with explicit nonnegative target weights. Estimate pre-adoption leads and placebo dates; no favorable pretrend test proves parallel untreated trends.; variables: Y: device-minus-reference error in a prespecified heart-rate or activity endpoint; i: independent update/version event with participants nested or crossed by device; t: paired device-reference visits over calendar time; the registered visit/time-window index, not another independent unit; A: observed device-version update, not research randomization; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: update-associated change in heart-rate/activity measurement error relative to unaffected versions; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; G: first policy-adoption cohort (M40) or documented device/version update cohort (E04); g,t: adoption cohort and outcome calendar period; E w: comparison-group expectation after justified pre-exposure covariate weighting; ATT(g,t): group-time average effect, causally interpreted only under conditional parallel untreated trends/no anticipation/appropriate interference assumptions; family code: DID
Sample size design: family: staggered adoption/event-study precision with independent adoption units and bias/parallel-trend sensitivity; formula: Use design-specific simulation or independent-unit resampling for the stated estimand and target confidence width/power; preserve all allocation, censoring, clustering and measurement-error structure.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Choose one confirmatory error endpoint; participants do not create additional independent firmware changes; require unaffected comparison and update dates.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Common eligibility/time zero, no anticipation, credible untreated-trend comparability, stable composition and no unaddressed simultaneous policies/updates.
Use independent adoption/update units for uncertainty; people and readings do not create new independent adoption events.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: If discontinuities track reference physiology equally or occur before updates, firmware attribution is unsupported.; preanalysis gate: Choose one confirmatory error endpoint; participants do not create additional independent firmware changes; require unaffected comparison and update dates.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Choose one confirmatory error endpoint; participants do not create additional independent firmware changes; require unaffected comparison and update dates.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Vary never-treated versus not-yet-treated comparisons, admissible pretrends, event windows, concurrent-policy exclusions and missing-laboratory observation weights.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Choose one confirmatory error endpoint; participants do not create additional independent firmware changes; require unaffected comparison and update dates.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/188; question: How often do device and algorithm updates create changes mistaken for physiological improvement?; aims: Create a version-aware provenance system for wearable exports.
Measure paired consumer and reference signals across naturally occurring updates.
Evaluate correction and warning methods in historical self-experiments.; design verbatim planning only: Assume 150 adults wear two devices for nine months, with scheduled reference ECG and activity assessments and event-triggered visits after updates. Device models are staggered; updates are observed rather than forced.; outcomes verbatim: Primary: update-associated shifts in heart-rate and activity measurement error. Secondary: altered responder classification and between-device transportability.; analysis verbatim inherited not authoritative if refined: Use interrupted time-series and difference-in-differences with unaffected device versions where available. Adjust season, illness, and behavior; prespecify a narrow outcome family and model reference-instrument error.; falsification verbatim: If discontinuities track reference physiology equally or occur before updates, firmware attribution is unsupported.; source ids unverified leads: wearables-living-review
wearables-remote-monitoring-2026
rwd-digital-platform-cohort-2026; prospectus precedence: No

E05 — When disengagement masquerades as successful optimization

Rank: 111
Cluster: Quantified-self, measurement, and N-of-1 causal methods
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: MISSING; allocation or exposure unit: participant logging-support randomization; observation process otherwise natural; independent sampling unit: participant; repeated measurement unit: device-off interval and auxiliary follow-up; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: missing-data bias sensitivity and randomized logging-support effect separately; MNAR is not empirically solved; specific constraint: The discrepancy is a selection-bias target; randomized logging support does not identify all unobserved outcomes under MNAR.
Primary estimand: difference between complete-case and intensive-follow-up estimates of one prespecified symptom/function mean
Candidate primary outcome: same symptom/function score measured by routine logs and independent follow-up
Primary finalized: No
Mathematical model: type: two-phase follow-up/selection precision with MNAR pattern-mixture scenario simulation; formulas: Within a fixed baseline/arm stratum, mu(delta)=Pr(R=1)E[Y|R=1]+Pr(R=0){E[Y|R=1]+delta}.
Primary discrepancy = mu_complete_case - mu_intensive_followup; the latter is a better-observed estimate, not guaranteed truth. Sweep delta over an externally justified range and preserve auxiliary outcomes.; variables: Y: same symptom/function score measured by routine logs and independent follow-up; i: participant; t: device-off interval and auxiliary follow-up; the registered visit/time-window index, not another independent unit; A: participant logging-support randomization; observation process otherwise natural; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: difference between complete-case and intensive-follow-up estimates of one prespecified symptom/function mean; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; R: routine outcome observed indicator; delta: unidentified pattern-mixture mean shift for missing versus observed outcomes in the same stratum; mu complete case,mu intensive followup: same endpoint/population estimators under distinct follow-up procedures; family code: MISSING
Sample size design: family: two-phase follow-up/selection precision with MNAR pattern-mixture scenario simulation; formula: Use design-specific simulation or independent-unit resampling for the stated estimand and target confidence width/power; preserve all allocation, censoring, clustering and measurement-error structure.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: The discrepancy is a selection-bias target; randomized logging support does not identify all unobserved outcomes under MNAR.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Auxiliary intensive follow-up can remain missing/selective; never equate device removal with wellness.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: If intensive follow-up reveals no outcome differences during missing periods, the hypothesized benefit inflation is not supported.; preanalysis gate: The discrepancy is a selection-bias target; randomized logging support does not identify all unobserved outcomes under MNAR.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: The discrepancy is a selection-bias target; randomized logging support does not identify all unobserved outcomes under MNAR.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
The discrepancy is a selection-bias target; randomized logging support does not identify all unobserved outcomes under MNAR.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/110; question: How much do mood-related device removal and selective symptom logging bias personalized health conclusions?; aims: Classify technical, intentional, and illness-related missingness using brief participant interviews.
Estimate bias in common before-after optimization analyses.
Randomize low-burden retention support to test whether inclusion improves inference.; design verbatim planning only: Planning assumption: 350 heterogeneous self-trackers followed for twelve months; device-off periods trigger optional, privacy-preserving check-ins. A nested randomized support intervention offers flexible logging or usual reminders, without financial penalties.; outcomes verbatim: Primary: discrepancy between complete-case and intensive-follow-up estimates. Secondary: retention, symptom trajectories, participation burden, and subgroup representation.; analysis verbatim inherited not authoritative if refined: Combine inverse-probability weighting, pattern-mixture sensitivity analyses, and auxiliary telephone outcomes. Separate missingness mechanisms; avoid claiming unverifiable identification when missing-not-at-random assumptions dominate.; falsification verbatim: If intensive follow-up reveals no outcome differences during missing periods, the hypothesized benefit inflation is not supported.; source ids unverified leads: wearables-remote-monitoring-2026
digital-phenotyping-relapse-review-2026
rwd-digital-platform-cohort-2026; prospectus precedence: No

E06 — A digital twin must predict an intervention, not just tomorrow

Rank: 156
Cluster: Quantified-self, measurement, and N-of-1 causal methods
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: CATE; allocation or exposure unit: eligible day randomized within participant in development and independent cohort; independent sampling unit: participant, including independent external-cohort participants; repeated measurement unit: randomized day with afternoon fatigue; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: No observed individual counterfactual ground truth; repeated assignment supports noisy personal contrasts, not exact individual effects.
Primary estimand: calibration slope and held-out loss for frozen model predictions of conditional prompt-assignment effects
Candidate primary outcome: afternoon fatigue at the prespecified post-prompt time
Primary finalized: No
Mathematical model: type: effect-modification/replication calibration simulation with prediction uncertainty, repeated randomization and independent validation units; formulas: E(Y_it | A_it,P_i,X_it)=alpha_i+tau A_it+kappa A_it P_i+gamma^T X_it+f_time(t). With unrestricted fixed alpha_i the time-invariant prognostic main effect of P_i is absorbed, not separately estimable.
Primary predictive-effect parameter is kappa in independent randomized evaluation; assess calibrated average effects and proper held-out loss separately.
A randomized pseudo-outcome U_it=(A_it-p_it)Y_it/[p_it(1-p_it)] has E(U_it | H_it)=conditional assignment effect under the randomized design; compare frozen predictions with aggregated U using noise-aware scores.; variables: Y: afternoon fatigue at the prespecified post-prompt time; i: participant, including independent external-cohort participants; t: randomized day with afternoon fatigue; the registered visit/time-window index, not another independent unit; A: eligible day randomized within participant in development and independent cohort; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: calibration slope and held-out loss for frozen model predictions of conditional prompt-assignment effects; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; P i: frozen pre-evaluation personal/phenotype effect prediction (or pretrial acute-response contrast for P19), with scale fixed before evaluation; eta: prognostic main effect of prediction; kappa: treatment-by-frozen-prediction coefficient, distinguishing predictive modification from prognosis; X it: registered predecision or baseline covariates; no future measurements; f time(t): predeclared practice/calendar trend; p it: known eligible-day randomization probability; H it: predecision observed history; U it: unbiased noisy randomized-effect pseudo-outcome, not an individual true effect; family code: CATE
Sample size design: family: effect-modification/replication calibration simulation with prediction uncertainty, repeated randomization and independent validation units; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure
distribution/reliability of frozen predictions
true effect heterogeneity scenarios
within-person randomization schedule
independent validation sample/cluster coverage; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: No observed individual counterfactual ground truth; repeated assignment supports noisy personal contrasts, not exact individual effects.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Predictions are frozen before evaluation; noisy person-specific estimates are not observed counterfactual truth; training/evaluation dependence must be propagated.
A nonzero interaction does not by itself establish useful calibration or superiority over a population-only policy.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: If the twin predicts outcomes well but fails randomized treatment contrasts or external calibration, its personalization claim is falsified.; preanalysis gate: No observed individual counterfactual ground truth; repeated assignment supports noisy personal contrasts, not exact individual effects.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: No observed individual counterfactual ground truth; repeated assignment supports noisy personal contrasts, not exact individual effects.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
No observed individual counterfactual ground truth; repeated assignment supports noisy personal contrasts, not exact individual effects.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/155; question: Can a personalized physiology model predict randomized intervention effects better than simpler population or within-person models?; aims: Benchmark observational forecasting using identical training data.
Freeze models before randomized activity-break perturbations.
Test intervention-effect calibration in a second recruitment site.; design verbatim planning only: Assume 200 sedentary adults contribute a six-week baseline and an eight-week experiment with repeated within-person randomization of eligible days to activity-break or neutral prompts; an independent 100-person cohort evaluates transportability using the same randomized design. Collect repeated afternoon fatigue outcomes, prespecify eligibility and carryover, and freeze baseline-trained models before randomized evaluation. The intervention uses ordinary movement adapted to ability.; outcomes verbatim: Primary: calibration of predicted individual intervention effects on afternoon fatigue. Secondary: vigilance, aggregate treatment effects, forecast error, and decision regret.; analysis verbatim inherited not authoritative if refined: Compare prespecified physiological twins, causal statistical models and simple averages using nested training splits. Evaluate population and prespecified conditional-effect calibration against randomized contrasts. For individual average effects, use repeated within-person assignment with a hierarchical measurement-error model and report estimation uncertainty rather than treating estimated person-specific contrasts as ground truth. Assess multiple-model selection and external calibration on untouched data.; falsification verbatim: If the twin predicts outcomes well but fails randomized treatment contrasts or external calibration, its personalization claim is falsified.; source ids unverified leads: digital-twins-healthcare-review-2026
digital-twin-causal-trials-2026
digital-n-of-1-physiology-2026; prospectus precedence: No

E07 — Separating sweat physiology from sensor performance

Rank: 219
Cluster: Quantified-self, measurement, and N-of-1 causal methods
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: AGREE; allocation or exposure unit: measurement condition; any randomized condition order is separate from sensor comparison; independent sampling unit: participant; repeated measurement unit: paired analyte-specific sensor/reference sample; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: measurement agreement/reliability; condition contrasts are causal only where the condition itself was randomized; specific constraint: Reference uncertainty, sensor lag and sweat rate require analyte-specific models; sweat concentration is not automatically a systemic concentration.
Primary estimand: out-of-condition bias and agreement of sweat-analyte estimates against validated references
Candidate primary outcome: sensor-minus-reference analyte concentration in its physical unit
Primary finalized: No
Mathematical model: type: paired hierarchical agreement/bias/limits precision with independent participants/donors/labs and replicate allocation; formulas: D_ij = S_ij - R_ij = beta_0 + beta_context^T C_ij + u_i + b_lab + epsilon_ij.
Context-specific agreement limits use beta_0 + beta_context^T C +/- z_(1-alpha/2) sqrt(Var(D_new | target context)); distinguish a new person from a repeat reading.
If reference error is material, S_ij = T_ij + b_S + e_S,ij and R_ij = T_ij + b_R + e_R,ij; identifiability needs reference validation/replicates or stated constraints.; variables: Y: sensor-minus-reference analyte concentration in its physical unit; i: participant; t: paired analyte-specific sensor/reference sample; the registered visit/time-window index, not another independent unit; A: measurement condition; any randomized condition order is separate from sensor comparison; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: out-of-condition bias and agreement of sweat-analyte estimates against validated references; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; S ij: sensor/test result in the stated endpoint unit; R ij: paired validated reference result in the same unit; D ij: paired test-minus-reference error, matching the per-thesis outcome; C ij: prespecified context, firmware/device and subgroup features; T ij: latent underlying quantity; not automatically identifiable; b S,b R: test and reference biases; e S,e R: test and reference measurement errors; b lab: laboratory effect when multiple laboratories actually sampled; D new: prediction error for the prespecified future measurement context; family code: AGREE
Sample size design: family: paired hierarchical agreement/bias/limits precision with independent participants/donors/labs and replicate allocation; formula: Plan confidence width for bias and agreement limits by hierarchical simulation; n readings is not n independent people. Acceptance limits in physical units must be independently justified.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Reference uncertainty, sensor lag and sweat rate require analyte-specific models; sweat concentration is not automatically a systemic concentration.
Reference calibration, matched time/unit and appropriately modeled reference error; correlation is not agreement.
Future-context prediction assumptions must match held-out persons/labs/conditions; missing/failing readings are outcomes of the measurement system.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: If added physiology variables fail held-out calibration or blood-sweat relationships are unstable, the proposed systemic-monitoring interpretation fails.; preanalysis gate: Reference uncertainty, sensor lag and sweat rate require analyte-specific models; sweat concentration is not automatically a systemic concentration.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Reference uncertainty, sensor lag and sweat rate require analyte-specific models; sweat concentration is not automatically a systemic concentration.
Assess proportional/heteroscedastic error, reference uncertainty and subgroup-specific failure, not just a global correlation.
Repeat estimates with and without questionable quality strata under predeclared rules; report excluded/failing measurements explicitly.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Reference uncertainty, sensor lag and sweat rate require analyte-specific models; sweat concentration is not automatically a systemic concentration.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/218; question: Can a biochemical sweat sensor remain calibrated across changes in sweat rate, skin condition, and ambient temperature?; aims: Partition analytic error using controlled reference samples.
Estimate within-person physiological lag and dilution effects.
Prospectively validate frozen calibration across ordinary environmental conditions.; design verbatim planning only: Planning assumption: 80 adults undergo repeated supervised assessments and a two-week ambulatory phase. Compare a multianalyte sweat platform with laboratory sweat assays and clinically collected reference samples where ethically justified.; outcomes verbatim: Primary: out-of-condition agreement and calibration stability. Secondary: sensor failure, lag, skin irritation, and differential error by skin characteristics.; analysis verbatim inherited not authoritative if refined: Use hierarchical errors-in-variables models; separate repeated samples from independent participants. Prespecify analyte-specific acceptance limits and correct the confirmatory analyte family for multiplicity.; falsification verbatim: If added physiology variables fail held-out calibration or blood-sweat relationships are unstable, the proposed systemic-monitoring interpretation fails.; source ids unverified leads: sweat-biosensors-review-2025
sweat-glucose-passive-platform-2024
sweat-multianalyte-microfluidic-2025; prospectus precedence: No

E08 — Cuffless blood pressure outside the calibration chair

Rank: 18
Cluster: Quantified-self, measurement, and N-of-1 causal methods
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: AGREE; allocation or exposure unit: device measurement and calibration context, without treatment allocation; independent sampling unit: participant; repeated measurement unit: paired cuffless/reference reading at visit; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: measurement agreement/reliability; condition contrasts are causal only where the condition itself was randomized; specific constraint: Agreement includes calibration decay and failed readings; repeated readings are not independent people or a diagnosis.
Primary estimand: person-level bias and clinically meaningful misclassification across real-world conditions
Candidate primary outcome: cuffless-minus-reference blood pressure in mmHg
Primary finalized: No
Mathematical model: type: paired hierarchical agreement/bias/limits precision with independent participants/donors/labs and replicate allocation; formulas: D_ij = S_ij - R_ij = beta_0 + beta_context^T C_ij + u_i + b_lab + epsilon_ij.
Context-specific agreement limits use beta_0 + beta_context^T C +/- z_(1-alpha/2) sqrt(Var(D_new | target context)); distinguish a new person from a repeat reading.
If reference error is material, S_ij = T_ij + b_S + e_S,ij and R_ij = T_ij + b_R + e_R,ij; identifiability needs reference validation/replicates or stated constraints.; variables: Y: cuffless-minus-reference blood pressure in mmHg; i: participant; t: paired cuffless/reference reading at visit; the registered visit/time-window index, not another independent unit; A: device measurement and calibration context, without treatment allocation; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: person-level bias and clinically meaningful misclassification across real-world conditions; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; S ij: sensor/test result in the stated endpoint unit; R ij: paired validated reference result in the same unit; D ij: paired test-minus-reference error, matching the per-thesis outcome; C ij: prespecified context, firmware/device and subgroup features; T ij: latent underlying quantity; not automatically identifiable; b S,b R: test and reference biases; e S,e R: test and reference measurement errors; b lab: laboratory effect when multiple laboratories actually sampled; D new: prediction error for the prespecified future measurement context; family code: AGREE
Sample size design: family: paired hierarchical agreement/bias/limits precision with independent participants/donors/labs and replicate allocation; formula: Plan confidence width for bias and agreement limits by hierarchical simulation; n readings is not n independent people. Acceptance limits in physical units must be independently justified.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Agreement includes calibration decay and failed readings; repeated readings are not independent people or a diagnosis.
Reference calibration, matched time/unit and appropriately modeled reference error; correlation is not agreement.
Future-context prediction assumptions must match held-out persons/labs/conditions; missing/failing readings are outcomes of the measurement system.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: If quality flags improve mean agreement by disproportionately discarding specific groups, the proposed equitable reliability solution fails.; preanalysis gate: Agreement includes calibration decay and failed readings; repeated readings are not independent people or a diagnosis.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Agreement includes calibration decay and failed readings; repeated readings are not independent people or a diagnosis.
Assess proportional/heteroscedastic error, reference uncertainty and subgroup-specific failure, not just a global correlation.
Repeat estimates with and without questionable quality strata under predeclared rules; report excluded/failing measurements explicitly.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Agreement includes calibration decay and failed readings; repeated readings are not independent people or a diagnosis.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/17; question: Which real-world conditions cause clinically consequential error in cuffless blood-pressure estimates?; aims: Map error sources against validated reference measurements.
Develop quality flags without using outcome information.
Validate locked flags and subgroup performance in a separate cohort.; design verbatim planning only: Assume 220 adults spanning blood-pressure ranges and skin pigmentation complete repeated clinic assessments and ambulatory monitoring over six months. Consumer cuffless readings are compared with validated cuff measurements and specialist reference procedures.; outcomes verbatim: Primary: person-level agreement and frequency of clinically meaningful misclassification. Secondary: calibration decay, failure rates, and equity of usable measurements.; analysis verbatim inherited not authoritative if refined: Use repeated-measures agreement methods and errors-in-variables models, not correlation alone. Prespecify subgroup interactions and multiplicity correction; measure pigmentation directly rather than treating race as optical physiology.; falsification verbatim: If quality flags improve mean agreement by disproportionately discarding specific groups, the proposed equitable reliability solution fails.; source ids unverified leads: cuffless-bp-aha-statement-2026
wearables-living-review; prospectus precedence: No

E09 — When meal, symptom, and wearable clocks disagree

Rank: 170
Cluster: Quantified-self, measurement, and N-of-1 causal methods
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: TIMING; allocation or exposure unit: eligible movement-prompt event randomized within participant; independent sampling unit: participant; repeated measurement unit: timestamped event and heart-rate response window; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: timing-error effects on a defined randomized movement contrast; unrelated meal/symptom associations remain observational; specific constraint: Do not infer temporal ordering from noisy logged clocks; identify timing-error distribution using independent timestamps.
Primary estimand: bias and interval coverage for the randomized movement-assignment effect when timestamps are uncertain
Candidate primary outcome: heart-rate response over a fixed independently timestamped event window
Primary finalized: No
Mathematical model: type: latent-event-time measurement-error and randomized-contrast coverage simulation; formulas: T_logged,ie=T_true,ie+U_ie; estimate the timing-error distribution from independently timestamped reference events.
Y_it=alpha_i+tau A_ie K(t-T_true,ie)+f_time(t)+epsilon_it; integrate over latent T_true or propagate reference-calibrated timing draws, then assess bias and interval coverage for tau under known randomized assignment.; variables: Y: heart-rate response over a fixed independently timestamped event window; i: participant; t: timestamped event and heart-rate response window; the registered visit/time-window index, not another independent unit; A: eligible movement-prompt event randomized within participant; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: bias and interval coverage for the randomized movement-assignment effect when timestamps are uncertain; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; e: randomized prompt/event; T logged,T true: logged and independently referenced/latent true event times in the same clock unit; U ie: timing error, potentially person/device dependent; K: prespecified response/lag-window function, not chosen to maximize the effect; tau: registered randomized activity-assignment response parameter; family code: TIMING
Sample size design: family: latent-event-time measurement-error and randomized-contrast coverage simulation; formula: Use design-specific simulation or independent-unit resampling for the stated estimand and target confidence width/power; preserve all allocation, censoring, clustering and measurement-error structure.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Do not infer temporal ordering from noisy logged clocks; identify timing-error distribution using independent timestamps.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Timing error and reference uncertainty need independent calibration; variable clock drift cannot be recovered solely from the response correlation.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: If alignment fails held-out effect recovery or requires unavailable reference information, its practical causal-inference advantage fails.; preanalysis gate: Do not infer temporal ordering from noisy logged clocks; identify timing-error distribution using independent timestamps.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Do not infer temporal ordering from noisy logged clocks; identify timing-error distribution using independent timestamps.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Do not infer temporal ordering from noisy logged clocks; identify timing-error distribution using independent timestamps.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/169; question: How does uncertainty in event timing distort causal conclusions from combined wearable, meal, and symptom streams?; aims: Measure clock offsets and retrospective label error against an independent reference.
Develop probabilistic alignment separating recording delay from physiological latency.
Validate frozen methods using randomized ordinary activity events with known assignment times.; design verbatim planning only: Planning assumption: 160 adults complete reference sessions and four weeks of tracking. Independently timestamped meal and movement events provide labels; randomized accessible movement versus neutral prompts supply causal contrasts without changing clinical care.; outcomes verbatim: Primary: bias and interval coverage for randomized activity effects on heart-rate responses. Secondary: false temporal ordering, spurious meal-symptom associations, logging burden, and transfer across devices.; analysis verbatim inherited not authoritative if refined: Use latent-event-time models with person-level validation splits. Account for activity, reporting propensity, autocorrelation, and reference uncertainty; preregister lag windows and control exploratory trigger multiplicity.; falsification verbatim: If alignment fails held-out effect recovery or requires unavailable reference information, its practical causal-inference advantage fails.; source ids unverified leads: n-of-1-methods-review-2025
digital-n-of-1-physiology-2026
wearables-living-review; prospectus precedence: No

E10 — The courier as a hidden component of a home blood test

Rank: 94
Cluster: Quantified-self, measurement, and N-of-1 causal methods
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: LAB_FACTOR; allocation or exposure unit: split specimen assigned to ex-vivo transport condition; independent sampling unit: participant source specimen; independent transport runs for transport-level inference; repeated measurement unit: aliquot, assay replicate and analyte; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: controlled in-vitro/ex-vivo allocation contrast when culture allocation is randomized; no direct human therapeutic-efficacy claim; specific constraint: Transport effects occur ex vivo; patient physiology is not randomized; separate matrix validation, assay censoring and specimen rejection.
Primary estimand: matrix/temperature/delay effects on deviation from promptly processed paired reference
Candidate primary outcome: home-kit analyte minus prompt-reference analyte concentration
Primary finalized: No
Mathematical model: type: multilevel donor-by-preparation/condition experimental contrast precision; independent-donor and independent-batch allocation simulation; formulas: Y_djbrt = alpha + z(A_djb)^T theta + gamma^T X_d + u_d + v_b + q_dj + epsilon_djbrt.
Delta_model = c^T theta standardized to sampled donors/conditions; r is a technical repeat, not another donor or manufacturing batch.
For two controlled factors B,C, add beta_B B + beta_C C + beta_BC B*C; estimate interaction as mu_11-mu_10-mu_01+mu_00 on the chosen additive scale. Donor age/disease attributes themselves are not randomized.; variables: Y: home-kit analyte minus prompt-reference analyte concentration; i: participant source specimen; independent transport runs for transport-level inference; t: aliquot, assay replicate and analyte; the registered visit/time-window index, not another independent unit; A: split specimen assigned to ex-vivo transport condition; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: matrix/temperature/delay effects on deviation from promptly processed paired reference; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; d: independent tissue/cell donor; j: independently allocated culture/explant/construct unit within donor; b: independent preparation/manufacturing/differentiation batch as appropriate; r: technical assay repeat or cell observation within allocated culture; z(A): indicators for the named controlled preparation/state/reference conditions; u d,v b,q dj: donor, batch and allocated-culture effects; crossed donor/preparation structure must follow actual allocation; X d: pretreatment donor/model attributes and baseline function only; B,C: the actual two condition/context factors named in this thesis; fix levels and identify which factors are randomized; beta BC: interaction parameter; if context is observational it is not a causal effect of changing donor age/disease; family code: LAB_FACTOR
Sample size design: family: multilevel donor-by-preparation/condition experimental contrast precision; independent-donor and independent-batch allocation simulation; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Transport effects occur ex vivo; patient physiology is not randomized; separate matrix validation, assay censoring and specimen rejection.
Randomize independently allocated cultures/explants within blocking factors; donor and preparation/batch are the biological replication units for transport.
Technical replicates/cells/fields reduce measurement error but do not increase independent donor count.
Culture viability, identity loss, assay failure and lineage selection are outcomes/selection processes; do not silently condition treatment comparisons on successful posttreatment models.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: If flags fail independent agreement or disproportionately reject remote households without alternatives, the proposed testing improvement fails.; preanalysis gate: Transport effects occur ex vivo; patient physiology is not randomized; separate matrix validation, assay censoring and specimen rejection.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Transport effects occur ex vivo; patient physiology is not randomized; separate matrix validation, assay censoring and specimen rejection.
Leave donors and manufacturing batches out; compare crossed/nested covariance structures supported by the design.
Report viable/failed model coverage, censored assays, donor-context heterogeneity and robustness to contamination/handling controls.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Confirmed donor/preparation access and allocated culture design; validated functional assay, dynamic range, technical error, detection limits and identity/viability gates.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Transport effects occur ex vivo; patient physiology is not randomized; separate matrix validation, assay censoring and specimen rejection.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/93; question: Which collection and transport conditions make home-sampling results misleading measures of physiological change?; aims: Partition collection-method and transport effects using paired reference samples.
Randomize ex-vivo shipping conditions to quantify analyte-specific degradation.
Validate rejection flags during ordinary seasonal courier delivery.; design verbatim planning only: Planning assumption: 200 adults provide professionally supervised paired home-kit and central-laboratory samples across two seasons. Split samples undergo randomized ex-vivo transport conditions; participants experience no experimental exposure. Prespecified nutritional and metabolic markers include B12 where matrix validation permits.; outcomes verbatim: Primary: analyte-specific deviation from promptly processed references. Secondary: clinically consequential classification changes, hemolysis, rejection frequency, repeat-testing burden, and transport cost.; analysis verbatim inherited not authoritative if refined: Use paired mixed models separating matrix, temperature, delay, and participant effects. Randomize laboratory batches, propagate reference uncertainty, correct analyte-family multiplicity, and validate thresholds in held-out samples.; falsification verbatim: If flags fail independent agreement or disproportionately reject remote households without alternatives, the proposed testing improvement fails.; source ids unverified leads: rwd-digital-platform-cohort-2026
nih-ods-b12-2025; prospectus precedence: No

E11 — Does cleaner home air produce measurable personal benefit?

Rank: 54
Cluster: Environmental exposures, air, noise, water, and workplaces
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: CROSS_INT; allocation or exposure unit: participant-season randomized active/sham air-cleaner sequence; independent sampling unit: participant; repeated measurement unit: blood-pressure reading within crossover period and season; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Baseline exposure modifier frozen using season one; period/season/carryover and rescue plans define estimand; postassignment exposure is a mediator.
Primary estimand: season-two replication of baseline-exposure modification of the randomized systolic-pressure contrast
Candidate primary outcome: validated home systolic blood pressure in mmHg
Primary finalized: No
Mathematical model: type: paired/crossover contrast precision with period, carryover and within-unit covariance; simulation for actual design; formulas: g(E[Y_ipt | A_ip,X_i]) = alpha_i + tau A_ip + pi_p + rho A_i,p-1 + gamma^T X_i + f_time(t) + b_cluster(i,p). Add tau_S A_ip S + kappa A_ip B_i with season S and independently frozen baseline exposure B_i; report the second-season interaction.
Target the registered within-unit assignment contrast, standardized over period/history; direct and lag effects are distinct parameters.; variables: Y: validated home systolic blood pressure in mmHg; i: participant; t: blood-pressure reading within crossover period and season; the registered visit/time-window index, not another independent unit; A: participant-season randomized active/sham air-cleaner sequence; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: season-two replication of baseline-exposure modification of the randomized systolic-pressure contrast; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; p: randomized crossover period/session/episode, not the outcome probability; pi p: period effect; include calendar/season/sequence terms supported by the design; rho: prespecified lagged-assignment/carryover parameter; cannot in general identify arbitrary persistent carryover in a short two-period trial; f time(t): prespecified within-period recovery/day trend; g: identity for mean outcomes, log for positive/count means, or logit for binary/proportion outcomes as specified below; S: season; explicitly distinguish development season from prospective replication season; B i: baseline exposure measured before the relevant allocation; tau S,kappa: treatment-by-season and treatment-by-baseline-exposure coefficients; family code: CROSS_INT
Sample size design: family: paired/crossover contrast precision with period, carryover and within-unit covariance; simulation for actual design; formula: For a simple two-condition paired mean only: Var(Delta_hat)=sigma_D^2/n_independent; actual cluster/period/lag models require design-specific simulation or c^T(D^T V^(-1)D)^(-1)c.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; sigma D^2: variance of independent-unit paired contrasts including measurement error; n independent: number of independent participants/teams/donors, not repeated samples; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Baseline exposure modifier frozen using season one; period/season/carryover and rescue plans define estimand; postassignment exposure is a mediator.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Concealed randomized sequence, a scientifically justified recovery interval and no anticipation; period/carryover cannot be repaired solely by a nonsignificant test.
If carryover is plausible, report lag/period and first-period or otherwise valid design sensitivity; do not drop periods based only on a post hoc carryover p-value.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: With verified particle reduction and adequate precision, failure of the prespecified baseline-exposure relationship to predict second-season blood-pressure contrasts would undermine reproducible personalization. A precise absence of benefit in the prespecified high-exposure group would challenge the vascular hypothesis; a null overall mean alone would not exclude subgroup benefit.; preanalysis gate: Baseline exposure modifier frozen using season one; period/season/carryover and rescue plans define estimand; postassignment exposure is a mediator.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Baseline exposure modifier frozen using season one; period/season/carryover and rescue plans define estimand; postassignment exposure is a mediator.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Compare planned lag/carryover structures and a valid first-period sensitivity where its sample supports inference; assess informative absent periods.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Baseline exposure modifier frozen using season one; period/season/carryover and rescue plans define estimand; postassignment exposure is a mediator.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/53; question: Which adults experience reproducible functional or vascular benefit when indoor particle exposure is reduced?; aims: Validate personal particle estimates against reference monitors.
Estimate blinded filtration effects on exposure and blood pressure.
Test whether baseline exposure predicts independently replicated benefit.; design verbatim planning only: Planning assumption: 160 urban adults complete randomized active-versus-sham air-cleaner crossover periods in each of two seasons, with repeated reference blood-pressure measurements in every period. Use the first season to estimate a prespecified baseline-exposure response relationship and the second as prospective replication, accounting for season. Do not introduce pollution; maintain clinically necessary protection and care, with a predefined severe-air-quality rescue plan.; outcomes verbatim: Primary: validated home systolic blood pressure. Secondary: personal particulate exposure, respiratory symptoms, attention, noise annoyance, and electricity cost.; analysis verbatim inherited not authoritative if refined: Use hierarchical crossover models for common effects, person-specific contrast uncertainty, season, outdoor pollution and carryover. Prespecify the baseline-exposure interaction and second-season replication test; do not define stable responders from effect direction alone. Report postassignment window opening and exposure changes separately from the primary assignment-based effect, propagate monitor error, and use a narrow outcome hierarchy.; falsification verbatim: With verified particle reduction and adequate precision, failure of the prespecified baseline-exposure relationship to predict second-season blood-pressure contrasts would undermine reproducible personalization. A precise absence of benefit in the prespecified high-exposure group would challenge the vascular hypothesis; a null overall mean alone would not exclude subgroup benefit.; source ids unverified leads: n-of-1-methods-review-2025
wearables-living-review; prospectus precedence: No

E12 — Ventilation, carbon dioxide, and the occupied office

Rank: 181
Cluster: Environmental exposures, air, noise, water, and workplaces
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: CLUSTER_CROSS_FACTOR; allocation or exposure unit: office-period assigned ventilation and gas-filtration conditions; independent sampling unit: office; repeated measurement unit: worker attention task within office-period; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Pollutant mixtures are postassignment variables; no independent causal CO2 effect is isolated; account for occupancy and office-period dependence.
Primary estimand: total ventilation, filtration and prespecified interaction effects under office-period randomization
Candidate primary outcome: blinded standardized attention-task performance
Primary finalized: No
Mathematical model: type: factorial main-effect/interaction planning with actual cross-classified or cluster-period covariance; formulas: E(Y_ict | B_ict, C_ict, X_i) = alpha_t + beta_B B_ict + beta_C C_ict + beta_BC B_ict C_ict + gamma^T X_i + u_i + b_c.
I = mu_11 - mu_10 - mu_01 + mu_00 = beta_BC on the additive scale; main effects are standardized over the other randomized factor.; variables: Y: blinded standardized attention-task performance; i: office; t: worker attention task within office-period; the registered visit/time-window index, not another independent unit; A: office-period assigned ventilation and gas-filtration conditions; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: total ventilation, filtration and prespecified interaction effects under office-period randomization; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; B,C: the two assigned factors named in the per-thesis allocation specification; levels and contrasts must be fixed; mu bc: marginal mean outcome under the specified joint factor levels; I: additive interaction estimand; beta B,beta C,beta BC: factor coefficients; for nonidentity links derive marginal contrasts rather than interpreting coefficients as risk differences; family code: CLUSTER_CROSS_FACTOR
Sample size design: family: factorial main-effect/interaction planning with actual cross-classified or cluster-period covariance; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Pollutant mixtures are postassignment variables; no independent causal CO2 effect is isolated; account for occupancy and office-period dependence.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Factor allocation and coding must match the design; interaction inference is separate from main-effect inference and needs its own precision.
Both factor assignments are at office-period level; office-period effects, sequence, calendar period and carryover are required.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: If assigned ventilation and filtration produce verified environmental contrasts without a sufficiently precise attention benefit, the proposed cognitive advantage is unsupported. Improvement without the intended exposure change would not support the proposed air-exchange mechanism; correlated comfort or noise changes cannot by themselves identify an alternative causal mediator.; preanalysis gate: Pollutant mixtures are postassignment variables; no independent causal CO2 effect is isolated; account for occupancy and office-period dependence.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Pollutant mixtures are postassignment variables; no independent causal CO2 effect is isolated; account for occupancy and office-period dependence.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Pollutant mixtures are postassignment variables; no independent causal CO2 effect is isolated; account for occupancy and office-period dependence.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/180; question: Do enhanced office ventilation and gas-phase filtration improve cognition under ordinary occupancy, and which measured environmental changes accompany their effects?; aims: Characterize pollutant mixtures during ordinary occupancy.
Separate enhanced ventilation and gas-phase filtration in a factorial intervention.
Test whether exposure changes explain independently assessed work-task performance.; design verbatim planning only: Assume 24 offices and approximately 240 workers enter a cluster crossover comparing usual compliant building operation with enhanced ventilation, gas-phase filtration, and both. No condition deliberately exceeds safety standards or changes work schedules.; outcomes verbatim: Primary: blinded standardized attention-task performance. Secondary: headache burden, perceived stuffiness, carbon dioxide, speciated volatile compounds, particles, thermal comfort, and energy consumption.; analysis verbatim inherited not authoritative if refined: Use office-period random effects and randomized assignment for total effects. Treat mediation as exploratory because pollutant mixtures are correlated; adjust workload, occupancy, and outdoor conditions while controlling confirmatory comparisons.; falsification verbatim: If assigned ventilation and filtration produce verified environmental contrasts without a sufficiently precise attention benefit, the proposed cognitive advantage is unsupported. Improvement without the intended exposure change would not support the proposed air-exchange mechanism; correlated comfort or noise changes cannot by themselves identify an alternative causal mediator.; source ids unverified leads: digital-twin-causal-trials-2026
wearables-living-review; prospectus precedence: No

E13 — Quiet workspaces without isolating the worker

Rank: 137
Cluster: Environmental exposures, air, noise, water, and workplaces
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: CLUSTER_CROSS; allocation or exposure unit: office-period acoustic assignment; independent sampling unit: office; repeated measurement unit: worker attention assessment in each acoustic period; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Select one of three period contrasts; hearing-status interaction is separate from main effect; preserve alarm and communication criteria.
Primary estimand: prespecified acoustic intervention versus usual compliant acoustics on attention errors
Candidate primary outcome: standardized sustained-attention error count
Primary finalized: No
Mathematical model: type: paired/crossover contrast precision with period, carryover and within-unit covariance; simulation for actual design; formulas: Y_ipt ~ NegativeBinomial(mu_ipt,phi); log(mu_ipt)=alpha_i+tau A_ip+pi_p+rho A_i,p-1+gamma^T X_i+b_cluster(i,p).
Target the registered within-unit assignment contrast, standardized over period/history; direct and lag effects are distinct parameters.; variables: Y: standardized sustained-attention error count; i: office; t: worker attention assessment in each acoustic period; the registered visit/time-window index, not another independent unit; A: office-period acoustic assignment; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: prespecified acoustic intervention versus usual compliant acoustics on attention errors; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; p: randomized crossover period/session/episode, not the outcome probability; pi p: period effect; include calendar/season/sequence terms supported by the design; rho: prespecified lagged-assignment/carryover parameter; cannot in general identify arbitrary persistent carryover in a short two-period trial; f time(t): prespecified within-period recovery/day trend; g: identity for mean outcomes, log for positive/count means, or logit for binary/proportion outcomes as specified below; mu ipt: mean vigilance lapse/error count; phi: count overdispersion parameter; family code: CLUSTER_CROSS
Sample size design: family: paired/crossover contrast precision with period, carryover and within-unit covariance; simulation for actual design; formula: For a simple two-condition paired mean only: Var(Delta_hat)=sigma_D^2/n_independent; actual cluster/period/lag models require design-specific simulation or c^T(D^T V^(-1)D)^(-1)c.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; sigma D^2: variance of independent-unit paired contrasts including measurement error; n independent: number of independent participants/teams/donors, not repeated samples; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Select one of three period contrasts; hearing-status interaction is separate from main effect; preserve alarm and communication criteria.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Concealed randomized sequence, a scientifically justified recovery interval and no anticipation; period/carryover cannot be repaired solely by a nonsignificant test.
If carryover is plausible, report lag/period and first-period or otherwise valid design sensitivity; do not drop periods based only on a post hoc carryover p-value.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: If subjective quiet improves but attention or speech access worsens, the proposed net functional benefit is rejected.; preanalysis gate: Select one of three period contrasts; hearing-status interaction is separate from main effect; preserve alarm and communication criteria.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Select one of three period contrasts; hearing-status interaction is separate from main effect; preserve alarm and communication criteria.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Compare planned lag/carryover structures and a valid first-period sensitivity where its sample supports inference; assess informative absent periods.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Select one of three period contrasts; hearing-status interaction is separate from main effect; preserve alarm and communication criteria.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/136; question: Which acoustic interventions improve concentrated work while preserving communication, safety, and social connection?; aims: Measure personal acoustic exposure and task demands.
Compare room treatment, optional noise-canceling headphones, and usual acoustics.
Assess persistence and differential effects for workers with hearing or sensory difficulties.; design verbatim planning only: Planning assumption: 18 offices, roughly 216 workers, complete three randomized six-week acoustic periods. Work schedules and lighting remain stable; required alarms and communication standards are independently checked.; outcomes verbatim: Primary: standardized sustained-attention errors. Secondary: work interruption, listening effort, fatigue, perceived inclusion, and objective speech intelligibility.; analysis verbatim inherited not authoritative if refined: Model office clustering, task type, baseline sound sensitivity, and period effects. Prespecify one primary contrast and hearing-status interaction; calibrate noise dosimeters and distinguish acoustic change from self-reported annoyance.; falsification verbatim: If subjective quiet improves but attention or speech access worsens, the proposed net functional benefit is rejected.; source ids unverified leads: wearables-living-review
multisensory-learning-neuroplasticity-2025; prospectus precedence: No

E14 — A healthier commute may trade noise for pollution

Rank: 194
Cluster: Environmental exposures, air, noise, water, and workplaces
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: CROSS; allocation or exposure unit: commuter-journey route recommendation; independent sampling unit: commuter; repeated measurement unit: journey noise/particle measurement; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Weights and units must be fixed before analysis; duration and exertion are separate consequences rather than routine primary adjustment.
Primary estimand: joint-priority versus usual-route assignment effect on the fixed joint exposure burden
Candidate primary outcome: prespecified weighted cumulative journey noise-particle burden
Primary finalized: No
Mathematical model: type: paired/crossover contrast precision with period, carryover and within-unit covariance; simulation for actual design; formulas: g(E[Y_ipt | A_ip,X_i]) = alpha_i + tau A_ip + pi_p + rho A_i,p-1 + gamma^T X_i + f_time(t) + b_cluster(i,p).
Target the registered within-unit assignment contrast, standardized over period/history; direct and lag effects are distinct parameters.; variables: Y: prespecified weighted cumulative journey noise-particle burden; i: commuter; t: journey noise/particle measurement; the registered visit/time-window index, not another independent unit; A: commuter-journey route recommendation; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: joint-priority versus usual-route assignment effect on the fixed joint exposure burden; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; p: randomized crossover period/session/episode, not the outcome probability; pi p: period effect; include calendar/season/sequence terms supported by the design; rho: prespecified lagged-assignment/carryover parameter; cannot in general identify arbitrary persistent carryover in a short two-period trial; f time(t): prespecified within-period recovery/day trend; g: identity for mean outcomes, log for positive/count means, or logit for binary/proportion outcomes as specified below; family code: CROSS
Sample size design: family: paired/crossover contrast precision with period, carryover and within-unit covariance; simulation for actual design; formula: For a simple two-condition paired mean only: Var(Delta_hat)=sigma_D^2/n_independent; actual cluster/period/lag models require design-specific simulation or c^T(D^T V^(-1)D)^(-1)c.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; sigma D^2: variance of independent-unit paired contrasts including measurement error; n independent: number of independent participants/teams/donors, not repeated samples; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Weights and units must be fixed before analysis; duration and exertion are separate consequences rather than routine primary adjustment.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Concealed randomized sequence, a scientifically justified recovery interval and no anticipation; period/carryover cannot be repaired solely by a nonsignificant test.
If carryover is plausible, report lag/period and first-period or otherwise valid design sensitivity; do not drop periods based only on a post hoc carryover p-value.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: If joint optimization only shifts exposure between noise and particles without reducing the prespecified joint cumulative burden, or produces unacceptable increases in travel burden, its practical claim fails. Attenuation after adjustment for route-induced travel time alone does not refute the total effect.; preanalysis gate: Weights and units must be fixed before analysis; duration and exertion are separate consequences rather than routine primary adjustment.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Weights and units must be fixed before analysis; duration and exertion are separate consequences rather than routine primary adjustment.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Compare planned lag/carryover structures and a valid first-period sensitivity where its sample supports inference; assess informative absent periods.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Weights and units must be fixed before analysis; duration and exertion are separate consequences rather than routine primary adjustment.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/193; question: Can route recommendations reduce combined noise and particle exposure without increasing travel burden or unequal access?; aims: Build repeated personal exposure maps using calibrated sensors.
Randomize safe feasible route suggestions within commuters.
Evaluate exposure, functional recovery, and acceptability across mobility and income groups.; design verbatim planning only: Assume 180 adult commuters complete repeated usual, quiet-priority, and joint-exposure-priority journeys over sixteen weeks. Routes are screened for accessibility and safety; no participant is assigned an unusually hazardous route.; outcomes verbatim: Primary: prespecified joint noise-particle exposure burden. Secondary: travel time, exertion, post-commute blood pressure, stress, adherence, and monetary cost.; analysis verbatim inherited not authoritative if refined: Use within-person randomized assignment contrasts with participant and period effects, adjusting for preassignment weather, traffic and planned transport mode. Report each cumulative journey exposure separately alongside the composite, with travel duration and exertion as outcomes. Prespecify composite weights, evaluate their sensitivity, and control secondary testing; any mediator-adjusted analysis is exploratory and distinct from the total route-recommendation effect.; falsification verbatim: If joint optimization only shifts exposure between noise and particles without reducing the prespecified joint cumulative burden, or produces unacceptable increases in travel burden, its practical claim fails. Attenuation after adjustment for route-induced travel time alone does not refute the total effect.; source ids unverified leads: wearables-living-review
wearable-stress-measurement-review-2026; prospectus precedence: No

E15 — Water-filter effectiveness after the installation photograph

Rank: 6
Cluster: Environmental exposures, air, noise, water, and workplaces
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: CLUSTER_CENSOR; allocation or exposure unit: household assigned installation/replacement support; independent sampling unit: household; repeated measurement unit: tap draw and filter-bypass observation; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: censored analytical distribution/change; a treatment effect requires the separately specified randomized assignment and correct sampling frame; specific constraint: Known unsafe supply triggers protection; define assay LOQ and distinguish nondetects, blanks, bypass and missing draws; joint primary rule needed.
Primary estimand: enhanced-support versus standard-protection effect on repeated tap-water lead and bypass frequency
Candidate primary outcome: tap-water lead concentration; bypass indicator is a separately ordered endpoint
Primary finalized: No
Mathematical model: type: censored analytical/audit precision with independent lots/households, supplier/laboratory structure and lower-tail sensitivity; formulas: For latent positive concentration/activity Q with density f_Q, a quantified result contributes f_Q(q); left-censored q<LOQ contributes F_Q(LOQ); an interval result [L,U] contributes F_Q(U)-F_Q(L).
Technical assay failure is a separate missing-result state and cannot be treated as a biological below-LOQ result. Avoid LOQ/2 and arbitrary positive constants unless an explicitly labeled sensitivity model justifies them.
For household-randomized support add arm, baseline/plumbing and household effects to the latent lead model; analyze bypass as a separate registered binary endpoint.; variables: Y: tap-water lead concentration; bypass indicator is a separately ordered endpoint; i: household; t: tap draw and filter-bypass observation; the registered visit/time-window index, not another independent unit; A: household assigned installation/replacement support; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: enhanced-support versus standard-protection effect on repeated tap-water lead and bypass frequency; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; Q,q: latent and measured per-thesis activity/concentration in assay-defined physical units; LOQ: validated, possibly batch-specific lower quantification limit; F Q,f Q: chosen distribution and density for latent quantity; below-limit shape may not be identified; L,U: reported analytical lower/upper bounds; family code: CLUSTER_CENSOR
Sample size design: family: censored analytical/audit precision with independent lots/households, supplier/laboratory structure and lower-tail sensitivity; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Known unsafe supply triggers protection; define assay LOQ and distinguish nondetects, blanks, bypass and missing draws; joint primary rule needed.
Validated assay range, blanks/recovery and category-specific activity units; never pool unlike biological assays as equivalent potency.
Censoring limits are known; unobserved lower-tail shape is assumption dependent; sampling frame/inclusion mechanism governs prevalence inference.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: If improved installation does not reduce verified exposure failures, the proposed implementation mechanism is unsupported.; preanalysis gate: Known unsafe supply triggers protection; define assay LOQ and distinguish nondetects, blanks, bypass and missing draws; joint primary rule needed.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Known unsafe supply triggers protection; define assay LOQ and distinguish nondetects, blanks, bypass and missing draws; joint primary rule needed.
Report nonquantifiable fraction and transparent interval/bounds alongside distribution-model estimates.
Vary lower-tail distributions, blank/recovery error, assay-failure MNAR and supplier/lab clustering.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Known unsafe supply triggers protection; define assay LOQ and distinguish nondetects, blanks, bypass and missing draws; joint primary rule needed.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/5; question: Do verified installation and replacement support improve sustained household lead-exposure control beyond providing certified filters?; aims: Map installation and maintenance failure modes.
Randomize enhanced implementation support on top of standard protection.
Test durability, household cost, and differential benefit in rented housing.; design verbatim planning only: Planning assumption: 240 households already receiving appropriate public-health protection enter a twelve-month implementation trial. Every household receives standard remediation advice and necessary protection; known unsafe supplies trigger immediate action rather than randomized delay.; outcomes verbatim: Primary: repeated independent tap-water lead concentration and filter bypass frequency. Secondary: use consistency, maintenance cost, and clinically obtained blood-lead trajectories where available.; analysis verbatim inherited not authoritative if refined: Stratify by plumbing type and tenancy; use intention-to-treat models with household clustering and laboratory blanks. Separate exposure endpoints from biomarker recovery, which has other sources and different time courses.; falsification verbatim: If improved installation does not reduce verified exposure failures, the proposed implementation mechanism is unsupported.; source ids unverified leads: cdc-lead-clinical-guidance-2025
fda-unapproved-chelation-products; prospectus precedence: No

E16 — Cleaning-product substitution and recurrent workplace symptoms

Rank: 161
Cluster: Environmental exposures, air, noise, water, and workplaces
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: CLUSTER_CROSS; allocation or exposure unit: workplace-period cleaning-product assignment; independent sampling unit: workplace; repeated measurement unit: worker weekly symptom report; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Products may unblind by smell; infection-control equivalence is a separate gate; specify symptom scale before calculating sample size.
Primary estimand: lower-emission versus usual compliant cleaning effect on the selected symptom burden
Candidate primary outcome: weekly headache or respiratory symptom burden, selected before recruitment
Primary finalized: No
Mathematical model: type: paired/crossover contrast precision with period, carryover and within-unit covariance; simulation for actual design; formulas: g(E[Y_ipt | A_ip,X_i]) = alpha_i + tau A_ip + pi_p + rho A_i,p-1 + gamma^T X_i + f_time(t) + b_cluster(i,p).
Target the registered within-unit assignment contrast, standardized over period/history; direct and lag effects are distinct parameters.; variables: Y: weekly headache or respiratory symptom burden, selected before recruitment; i: workplace; t: worker weekly symptom report; the registered visit/time-window index, not another independent unit; A: workplace-period cleaning-product assignment; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: lower-emission versus usual compliant cleaning effect on the selected symptom burden; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; p: randomized crossover period/session/episode, not the outcome probability; pi p: period effect; include calendar/season/sequence terms supported by the design; rho: prespecified lagged-assignment/carryover parameter; cannot in general identify arbitrary persistent carryover in a short two-period trial; f time(t): prespecified within-period recovery/day trend; g: identity for mean outcomes, log for positive/count means, or logit for binary/proportion outcomes as specified below; family code: CLUSTER_CROSS
Sample size design: family: paired/crossover contrast precision with period, carryover and within-unit covariance; simulation for actual design; formula: For a simple two-condition paired mean only: Var(Delta_hat)=sigma_D^2/n_independent; actual cluster/period/lag models require design-specific simulation or c^T(D^T V^(-1)D)^(-1)c.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; sigma D^2: variance of independent-unit paired contrasts including measurement error; n independent: number of independent participants/teams/donors, not repeated samples; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Products may unblind by smell; infection-control equivalence is a separate gate; specify symptom scale before calculating sample size.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Concealed randomized sequence, a scientifically justified recovery interval and no anticipation; period/carryover cannot be repaired solely by a nonsignificant test.
If carryover is plausible, report lag/period and first-period or otherwise valid design sensitivity; do not drop periods based only on a post hoc carryover p-value.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: A precise absence of symptom improvement despite a verified chemical exposure contrast would challenge the clinical benefit hypothesis. Improvement without the intended exposure contrast would not support the proposed chemical mechanism. Report assignment guesses descriptively; benefit confined to correct guessers does not by itself establish an expectancy explanation.; preanalysis gate: Products may unblind by smell; infection-control equivalence is a separate gate; specify symptom scale before calculating sample size.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Products may unblind by smell; infection-control equivalence is a separate gate; specify symptom scale before calculating sample size.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Compare planned lag/carryover structures and a valid first-period sensitivity where its sample supports inference; assess informative absent periods.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Products may unblind by smell; infection-control equivalence is a separate gate; specify symptom scale before calculating sample size.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/160; question: Does verified reduction of cleaning-related volatile compounds reduce recurrent headache and airway symptoms in susceptible workers?; aims: Characterize emissions and cleaning practices without identifying individual workers.
Randomize compliant product substitutions across sites.
Assess whether exposure reductions predict sustained symptom and attendance changes.; design verbatim planning only: Assume 20 workplaces and 240 consenting workers enter a cluster crossover of usual compliant cleaning versus independently evaluated lower-emission products. Infection-control performance and cleaning frequency are maintained, and ventilation is measured.; outcomes verbatim: Primary: weekly headache or respiratory-symptom burden selected before recruitment. Secondary: speciated volatile compounds, rescue-medication use recorded clinically, absenteeism, acceptability, and cleaning efficacy.; analysis verbatim inherited not authoritative if refined: Account for site clustering, season, baseline sensitivity, ventilation, and nonwork exposures. Prespecify symptom hierarchy and negative-control outcomes; sensory unblinding is measured rather than assumed absent.; falsification verbatim: A precise absence of symptom improvement despite a verified chemical exposure contrast would challenge the clinical benefit hypothesis. Improvement without the intended exposure contrast would not support the proposed chemical mechanism. Report assignment guesses descriptively; benefit confined to correct guessers does not by itself establish an expectancy explanation.; source ids unverified leads: nocebo-communication-review-2026
wearables-living-review; prospectus precedence: No

E17 — Wildfire protection as an access problem

Rank: 31
Cluster: Environmental exposures, air, noise, water, and workplaces
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT_LONG; allocation or exposure unit: household randomized to enhanced installation/utility support; independent sampling unit: household; repeated measurement unit: qualifying natural smoke episode and paired monitor interval; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Parallel household assignment with repeated smoke episodes, not crossover; qualifying events do not create independent households; sensor floor and near-zero outdoor denominator need a fixed rule.
Primary estimand: support-offer effect on indoor-to-outdoor particle ratio during prespecified smoke episodes
Candidate primary outcome: log indoor/outdoor particulate ratio with validated positive measurements
Primary finalized: No
Mathematical model: type: contrast precision/power by design-matrix or simulation planning; formulas: E(Y_it | A_i, X_i, Y_i0) = alpha_t + z(A_i)^T theta_t + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Delta = sum_t w_t c^T theta_t; estimate by baseline-adjusted marginal standardization, with unit/cluster-appropriate inference.; variables: Y: log indoor/outdoor particulate ratio with validated positive measurements; i: household; t: qualifying natural smoke episode and paired monitor interval; the registered visit/time-window index, not another independent unit; A: household randomized to enhanced installation/utility support; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: support-offer effect on indoor-to-outdoor particle ratio during prespecified smoke episodes; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; z(A): prespecified indicators for the actual randomized arms; two arms reduce to one indicator; family code: RCT_LONG
Sample size design: family: contrast precision/power by design-matrix or simulation planning; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Parallel household assignment with repeated smoke episodes, not crossover; qualifying events do not create independent households; sensor floor and near-zero outdoor denominator need a fixed rule.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: If assistance raises ownership or satisfaction without improving verified exposure protection, the proposed delivery advantage fails.; preanalysis gate: Parallel household assignment with repeated smoke episodes, not crossover; qualifying events do not create independent households; sensor floor and near-zero outdoor denominator need a fixed rule.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Parallel household assignment with repeated smoke episodes, not crossover; qualifying events do not create independent households; sensor floor and near-zero outdoor denominator need a fixed rule.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Parallel household assignment with repeated smoke episodes, not crossover; qualifying events do not create independent households; sensor floor and near-zero outdoor denominator need a fixed rule.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/30; question: Which delivery model sustains effective clean-air protection among households unable to buy or operate filtration independently?; aims: Map practical barriers before smoke season.
Randomize additional support among households already receiving standard filtration resources.
Measure exposure protection, respiratory burden, and program cost during naturally occurring smoke events.; design verbatim planning only: Planning assumption: 300 lower-income households across two fire-prone regions receive a filter and standard safety information; half receive enhanced installation, maintenance, and utility support. No smoke is induced and emergency protections remain available to all.; outcomes verbatim: Primary: indoor-to-outdoor particulate ratio during prespecified smoke episodes. Secondary: symptoms, device uptime, thermal comfort, displacement, and distribution of benefit.; analysis verbatim inherited not authoritative if refined: Use randomized comparisons with event-level and household clustering; account for outdoor intensity and housing leakage. Calibrate monitors, prespecify event definitions, and report uncertainty if few qualifying episodes occur.; falsification verbatim: If assistance raises ownership or satisfaction without improving verified exposure protection, the proposed delivery advantage fails.; source ids unverified leads: wearables-remote-monitoring-2026
n-of-1-methods-review-2025; prospectus precedence: No

E18 — Safe indoor cooling for vulnerable adults

Rank: 40
Cluster: Environmental exposures, air, noise, water, and workplaces
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT_COUNT; allocation or exposure unit: participant/household assigned additional safe cooling package, assignment level to be fixed; independent sampling unit: household when equipment or occupancy is shared; repeated measurement unit: naturally hot day within summer; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Fix the hot-day definition before randomization; relocation and medically necessary rescue remain strategy consequences; death handling requires decision.
Primary estimand: cooling-package offer effect on heat-related limitation days over two summers
Candidate primary outcome: count of days limiting essential daily activities during the fixed hot-day risk set
Primary finalized: No
Mathematical model: type: overdispersed count contrast planning by simulation, with follow-up/cluster/zero inflation assumptions; formulas: Y_it ~ NegativeBinomial(mu_it,phi); log(mu_it)=alpha_t+z(A_i)^T theta_t+beta Y_i0+gamma^T X_i+b_cluster(i)+u_i.
Delta = standardized E[Y | do(A=1)] - E[Y | do(A=0)] under randomization; report ratio exp(c^T theta) only when the log-mean model supports it.; variables: Y: count of days limiting essential daily activities during the fixed hot-day risk set; i: household when equipment or occupancy is shared; t: naturally hot day within summer; the registered visit/time-window index, not another independent unit; A: participant/household assigned additional safe cooling package, assignment level to be fixed; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: cooling-package offer effect on heat-related limitation days over two summers; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; mu it: conditional mean of the actual per-thesis count; phi: overdispersion parameter, with Var(Y)=mu+mu^2/phi under this parameterization; z(A): prespecified indicators for the actual arms or randomized policy; family code: RCT_COUNT
Sample size design: family: overdispersed count contrast planning by simulation, with follow-up/cluster/zero inflation assumptions; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Fix the hot-day definition before randomization; relocation and medically necessary rescue remain strategy consequences; death handling requires decision.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: If rooms cool but function does not improve, or operating costs displace essential spending, the proposed net benefit is unsupported.; preanalysis gate: Fix the hot-day definition before randomization; relocation and medically necessary rescue remain strategy consequences; death handling requires decision.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Fix the hot-day definition before randomization; relocation and medically necessary rescue remain strategy consequences; death handling requires decision.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Fix the hot-day definition before randomization; relocation and medically necessary rescue remain strategy consequences; death handling requires decision.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/39; question: Can adaptive cooling support improve daily function during ordinary heat events without excessive energy costs or unsafe overcooling?; aims: Characterize indoor exposure and barriers through a summer cohort.
Randomize enhanced cooling support alongside standard heat-health advice.
Estimate functional benefit, operating cost, and equitable reach.; design verbatim planning only: Assume 200 older or chronically ill adults receive a two-summer study with randomized access to an additional safe cooling package, never withholding medically required equipment or emergency assistance. Temperature exposure arises naturally; an independent safety team sets escalation rules.; outcomes verbatim: Primary: days with heat-related limitation in essential daily activities. Secondary: calibrated room temperature, orthostatic symptoms, validated blood pressure, sleep comfort, and utility burden.; analysis verbatim inherited not authoritative if refined: Model outdoor heat, building characteristics, baseline illness, and household occupancy. Use intention-to-treat estimates, prespecified hot-day definitions, and sensitivity analyses for relocation or selective sensor removal.; falsification verbatim: If rooms cool but function does not improve, or operating costs displace essential spending, the proposed net benefit is unsupported.; source ids unverified leads: wearables-remote-monitoring-2026
cuffless-bp-aha-statement-2026; prospectus precedence: No

E19 — Can household microplastic reduction claims survive contamination controls?

Rank: 231
Cluster: Environmental exposures, air, noise, water, and workplaces
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: ENV_CENSOR; allocation or exposure unit: household-period source substitution randomized in crossover sequence, as specified by the source aims; independent sampling unit: household and independent collection batch; repeated measurement unit: polymer-specific sample and split-laboratory assay; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: censored analytical distribution/change; a treatment effect requires the separately specified randomized assignment and correct sampling frame; specific constraint: Source aims specify randomized crossover despite shorter design text; model period/carryover, contamination, recovery and detection limits; external concentration cannot identify absorbed dose or clinical benefit.
Primary estimand: within-household randomized source-substitution effect on external concentration
Candidate primary outcome: blank-corrected polymer-specific particle concentration over the fixed detectable size range
Primary finalized: No
Mathematical model: type: censored analytical/audit precision with independent lots/households, supplier/laboratory structure and lower-tail sensitivity; formulas: For latent positive concentration/activity Q with density f_Q, a quantified result contributes f_Q(q); left-censored q<LOQ contributes F_Q(LOQ); an interval result [L,U] contributes F_Q(U)-F_Q(L).
Technical assay failure is a separate missing-result state and cannot be treated as a biological below-LOQ result. Avoid LOQ/2 and arbitrary positive constants unless an explicitly labeled sensitivity model justifies them.
log(Q_ipt)=alpha_i+tau A_ip+pi_p+rho A_i,p-1+b_collection+epsilon_ipt in the censored likelihood; estimate the registered crossover contrast, with polymer-family multiplicity.; variables: Y: blank-corrected polymer-specific particle concentration over the fixed detectable size range; i: household and independent collection batch; t: polymer-specific sample and split-laboratory assay; the registered visit/time-window index, not another independent unit; A: household-period source substitution randomized in crossover sequence, as specified by the source aims; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: within-household randomized source-substitution effect on external concentration; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; Q,q: latent and measured per-thesis activity/concentration in assay-defined physical units; LOQ: validated, possibly batch-specific lower quantification limit; F Q,f Q: chosen distribution and density for latent quantity; below-limit shape may not be identified; L,U: reported analytical lower/upper bounds; p: randomized household crossover period; pi p: period effect; rho: lagged-assignment/carryover parameter; b collection: independent collection-batch contamination/processing effect; family code: ENV_CENSOR
Sample size design: family: censored analytical/audit precision with independent lots/households, supplier/laboratory structure and lower-tail sensitivity; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Source aims specify randomized crossover despite shorter design text; model period/carryover, contamination, recovery and detection limits; external concentration cannot identify absorbed dose or clinical benefit.
Validated assay range, blanks/recovery and category-specific activity units; never pool unlike biological assays as equivalent potency.
Censoring limits are known; unobserved lower-tail shape is assumption dependent; sampling frame/inclusion mechanism governs prevalence inference.
Source aims specify household randomized crossover; contamination controls, carryover and reference recovery remain required. A log-concentration estimand and mass/count distinction must be fixed; censoring can leave arithmetic-mean change poorly identified.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: If apparent reductions disappear after blanks, size harmonization, or independent laboratory replication, the exposure claim is rejected.; preanalysis gate: Source aims specify randomized crossover despite shorter design text; model period/carryover, contamination, recovery and detection limits; external concentration cannot identify absorbed dose or clinical benefit.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Source aims specify randomized crossover despite shorter design text; model period/carryover, contamination, recovery and detection limits; external concentration cannot identify absorbed dose or clinical benefit.
Report nonquantifiable fraction and transparent interval/bounds alongside distribution-model estimates.
Vary lower-tail distributions, blank/recovery error, assay-failure MNAR and supplier/lab clustering.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Source aims specify randomized crossover despite shorter design text; model period/carryover, contamination, recovery and detection limits; external concentration cannot identify absorbed dose or clinical benefit.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/230; question: Which consumer water and food-contact interventions produce reproducible reductions in externally measured polymer-particle exposure?; aims: Validate sampling blanks, recovery, and polymer identification.
Compare household source substitutions under randomized crossover conditions.
Determine whether reductions reproduce across laboratories and housing contexts.; design verbatim planning only: Planning assumption: 100 households complete repeated sampling of drinking water and selected food-contact pathways before and during safe certified-source substitutions. No novel biological intervention or deliberate exposure is introduced; split samples go to two laboratories.; outcomes verbatim: Primary: blank-corrected polymer-specific particle concentration in a prespecified detectable size range. Secondary: cost, adherence, replacement-product shedding, and interlaboratory agreement.; analysis verbatim inherited not authoritative if refined: Use censored-data models, recovery correction, and blinded laboratory batches. Prespecify polymer families and multiplicity control; report mass and counts separately and avoid equating an external measure with absorbed dose.; falsification verbatim: If apparent reductions disappear after blanks, size harmonization, or independent laboratory replication, the exposure claim is rejected.; source ids unverified leads: detox-diets-critical-review-2014
sweat-metals-systematic-review-2012; prospectus precedence: No

E20 — Exposure dashboards that workers can actually act on

Rank: 77
Cluster: Environmental exposures, air, noise, water, and workplaces
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: CLUSTER; allocation or exposure unit: team randomized monitoring/dashboard/response-pathway offer; independent sampling unit: work team; repeated measurement unit: independent reference exposure samples within worker and shift; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Fix one exposure class/site and comparator; production changes can mediate the offer effect; selective monitor wearing needs independent sampling.
Primary estimand: dashboard-plus-response versus standard compliant monitoring effect on selected exposure class
Candidate primary outcome: independently sampled exposure relative to the relevant occupational standard
Primary finalized: No
Mathematical model: type: contrast precision/power by design-matrix or simulation planning; formulas: E(Y_it | A_i, X_i, Y_i0) = alpha_t + z(A_i)^T theta_t + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Delta = sum_t w_t c^T theta_t; estimate by baseline-adjusted marginal standardization, with unit/cluster-appropriate inference.; variables: Y: independently sampled exposure relative to the relevant occupational standard; i: work team; t: independent reference exposure samples within worker and shift; the registered visit/time-window index, not another independent unit; A: team randomized monitoring/dashboard/response-pathway offer; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: dashboard-plus-response versus standard compliant monitoring effect on selected exposure class; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; z(A): prespecified indicators for the actual randomized arms; two arms reduce to one indicator; family code: CLUSTER
Sample size design: family: contrast precision/power by design-matrix or simulation planning; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Fix one exposure class/site and comparator; production changes can mediate the offer effect; selective monitor wearing needs independent sampling.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Number of randomized clusters, cluster size imbalance and degrees of freedom govern inference; use small-sample correction/randomization inference when appropriate.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: If alerts increase awareness without reducing reference-measured exposure or corrective-action delay, the proposed empowerment mechanism fails.; preanalysis gate: Fix one exposure class/site and comparator; production changes can mediate the offer effect; selective monitor wearing needs independent sampling.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Fix one exposure class/site and comparator; production changes can mediate the offer effect; selective monitor wearing needs independent sampling.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Fix one exposure class/site and comparator; production changes can mediate the offer effect; selective monitor wearing needs independent sampling.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/76; question: Does worker-controlled real-time exposure feedback improve protection beyond standard occupational monitoring and training?; aims: Co-design actionable alerts with workers and occupational hygienists.
Randomize dashboard and response-support strategies above existing protections.
Assess exposure reduction, work disruption, trust, and inequalities in response.; design verbatim planning only: Assume 24 teams in dust- or noise-monitored workplaces enter a cluster trial of standard compliant monitoring, worker dashboards, or dashboards plus a documented management response pathway. Required engineering controls and protective equipment continue in every arm.; outcomes verbatim: Primary: independently sampled exposure relative to relevant occupational standards. Secondary: verified corrective actions, near misses, perceived control, productivity burden, and retaliation concerns.; analysis verbatim inherited not authoritative if refined: Use team-level intention-to-treat contrasts, account for task mix and production volume, and prespecify one exposure class per site. Validate wearable monitors against occupational reference instruments and examine selective wearing.; falsification verbatim: If alerts increase awareness without reducing reference-measured exposure or corrective-action delay, the proposed empowerment mechanism fails.; source ids unverified leads: wearables-living-review
wearables-remote-monitoring-2026; prospectus precedence: No

E21 — A trustworthy digital measure of vasomotor burden

Rank: 143
Cluster: Women's reproductive and sexual health and menopause
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: DIAG; allocation or exposure unit: prospective sensor validation without therapeutic allocation; independent sampling unit: participant; repeated measurement unit: adjudicated hot-flash and nonevent window; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: diagnostic/benchmark agreement or classification validity; reader/interface/navigation effects are causal only in their separately randomized component; specific constraint: Define event matching and nonevent denominators; conduct participant-held-out validation; ordinary sweating is an essential challenge stratum.
Primary estimand: participant-averaged event sensitivity and false-alert burden in home conditions
Candidate primary outcome: adjudicated hot-flash state and sensor alert in prespecified time windows
Primary finalized: No
Mathematical model: type: diagnostic/benchmark sensitivity, false-alert and calibration precision with independent cases/noncases and cluster/reference uncertainty; formulas: Sensitivity = TP/(TP+FN); false-alert burden = FP/(registered nonevent time/opportunities); report participant/family-clustered uncertainty and failed/indeterminate cases.
For probability scores, report calibration and a fixed proper score in participant/source-disjoint evaluation, with uncertain reference labels modeled or bounded.; variables: Y: adjudicated hot-flash state and sensor alert in prespecified time windows; i: participant; t: adjudicated hot-flash and nonevent window; the registered visit/time-window index, not another independent unit; A: prospective sensor validation without therapeutic allocation; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: participant-averaged event sensitivity and false-alert burden in home conditions; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; TP,FN,FP: true-positive, false-negative and false-positive counts relative to the independently defined reference and matching window, not an assumed perfect gold standard; reference: per-thesis independent reference, including adjudication uncertainty; nonevent time/opportunities: fixed at-risk denominator; event specificity and false alerts per hour are distinct targets; family code: DIAG
Sample size design: family: diagnostic/benchmark sensitivity, false-alert and calibration precision with independent cases/noncases and cluster/reference uncertainty; formula: An independent-case rough starting point is n_case ~= z_(1-alpha*/2)^2 Se(1-Se)/h_Se^2; replace with interval-based/clustered simulation for the actual design. Plan noncases/observation time and reference failures separately.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; Se: plausible sensitivity from prior evidence or pilot, not an asserted study result; h Se: desired sensitivity interval half-width; n case: independent positive cases; repeated events within a person require correlation adjustment; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Define event matching and nonevent denominators; conduct participant-held-out validation; ordinary sweating is an essential challenge stratum.
Reference independence, fixed thresholds/window and an explicit denominator; incorporation/verification bias can invalidate apparent accuracy.
Participant/family-level test splits and paired comparisons; report abstention, assay failure and indeterminate truth rather than force every case into a binary label.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: If performance collapses in ordinary heat or misses distressing episodes preferentially, the proposed digital outcome is unsuitable.; preanalysis gate: Define event matching and nonevent denominators; conduct participant-held-out validation; ordinary sweating is an essential challenge stratum.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Define event matching and nonevent denominators; conduct participant-held-out validation; ordinary sweating is an essential challenge stratum.
Vary plausible reference misclassification, verification-selection and event-matching window under a prespecified sensitivity family.
Report subgroup usable-data coverage and calibration; avoid interpreting an imprecise null difference as equivalence.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Define event matching and nonevent denominators; conduct participant-held-out validation; ordinary sweating is an essential challenge stratum.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/142; question: Can unobtrusive sensors distinguish menopausal hot flashes from activity, ambient heat, anxiety, and ordinary sweating?; aims: Create an adjudicated multimodal event dataset.
Validate a locked event model across independent participants and seasons.
Determine which event and distress measures track meaningful treatment response.; design verbatim planning only: Planning assumption: 180 women across the menopausal transition complete supervised validation and six weeks of home monitoring; a subset beginning clinician-selected usual care is followed prospectively. Compare consumer signals with research-grade skin conductance, temperature, and time-stamped reports.; outcomes verbatim: Primary: participant-level sensitivity and false-alert burden for adjudicated events. Secondary: sleep disruption, distress, adherence, and change concordance after routine care.; analysis verbatim inherited not authoritative if refined: Prevent participant leakage between training and testing; model ambient temperature, activity, skin characteristics, and medication use. Prespecify event windows and report calibration as well as discrimination.; falsification verbatim: If performance collapses in ordinary heat or misses distressing episodes preferentially, the proposed digital outcome is unsuitable.; source ids unverified leads: menopause-cognition-sleep-review-2025
wearable-stress-measurement-review-2026
wearables-living-review; prospectus precedence: No

E22 — Is menopausal brain fog changed by treating sleep-related distress?

Rank: 55
Cluster: Women's reproductive and sexual health and menopause
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT; allocation or exposure unit: participant randomized to adapted sleep support or matched comparator; independent sampling unit: participant; repeated measurement unit: alternate-form cognitive testing occasion; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Sleep change, hormone changes and subjective brain fog are not baseline confounders after assignment; blinded objective transfer outcome is primary.
Primary estimand: sleep-intervention assignment effect on the fixed objective attention-memory composite
Candidate primary outcome: externally scored objective attention-memory composite at sixteen weeks
Primary finalized: No
Mathematical model: type: contrast precision/power by design-matrix or simulation planning; formulas: E(Y_it | A_i, X_i, Y_i0) = alpha_t + z(A_i)^T theta_t + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Delta = sum_t w_t c^T theta_t; estimate by baseline-adjusted marginal standardization, with unit/cluster-appropriate inference.; variables: Y: externally scored objective attention-memory composite at sixteen weeks; i: participant; t: alternate-form cognitive testing occasion; the registered visit/time-window index, not another independent unit; A: participant randomized to adapted sleep support or matched comparator; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: sleep-intervention assignment effect on the fixed objective attention-memory composite; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; z(A): prespecified indicators for the actual randomized arms; two arms reduce to one indicator; family code: RCT
Sample size design: family: contrast precision/power by design-matrix or simulation planning; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Sleep change, hormone changes and subjective brain fog are not baseline confounders after assignment; blinded objective transfer outcome is primary.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: If only unblinded symptom ratings improve, the hypothesis of objective cognitive benefit is rejected.; preanalysis gate: Sleep change, hormone changes and subjective brain fog are not baseline confounders after assignment; blinded objective transfer outcome is primary.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Sleep change, hormone changes and subjective brain fog are not baseline confounders after assignment; blinded objective transfer outcome is primary.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Sleep change, hormone changes and subjective brain fog are not baseline confounders after assignment; blinded objective transfer outcome is primary.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/54; question: Does a menopause-adapted behavioral sleep intervention improve objective cognition as well as perceived brain fog?; aims: Map baseline cognitive, mood, and sleep phenotypes.
Randomize menopause-adapted behavioral sleep support against attention-matched health education.
Test whether cognitive changes persist and correspond to independently measured sleep improvement.; design verbatim planning only: Assume 220 perimenopausal women with sleep disturbance and cognitive complaints enter a sixteen-week trial with six-month follow-up. Existing hormone or other clinical treatment remains clinician-directed and is stratified at randomization.; outcomes verbatim: Primary: prespecified objective attention-memory composite. Secondary: brain-fog distress, daily function, insomnia severity, mood, vasomotor burden, and sleep continuity in a research-monitoring subset.; analysis verbatim inherited not authoritative if refined: Use intention-to-treat models, alternate cognitive-test forms, and blinded assessors. Account for baseline treatment and practice effects; treat sleep mediation cautiously and adjust the secondary outcome family.; falsification verbatim: If only unblinded symptom ratings improve, the hypothesis of objective cognitive benefit is rejected.; source ids unverified leads: menopause-cognition-sleep-review-2025
menopause-behavioral-symptoms-review-2026
menopause-hormone-therapy-review-2025; prospectus precedence: No

E23 — Explaining AMH without inventing a fertility score

Rank: 20
Cluster: Women's reproductive and sexual health and menopause
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT; allocation or exposure unit: participant individually randomized to decision aid versus usual counseling; independent sampling unit: participant; repeated measurement unit: comprehension items and follow-up visits; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Pregnancy is descriptive rather than a validation of AMH fertility prediction; aid use and later treatment are postassignment consequences.
Primary estimand: four-week assignment effect on the prespecified 0–100 AMH-understanding score
Candidate primary outcome: AMH comprehension score at four weeks
Primary finalized: No
Mathematical model: type: contrast precision/power by design-matrix or simulation planning; formulas: E(Y_it | A_i, X_i, Y_i0) = alpha_t + z(A_i)^T theta_t + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Delta = sum_t w_t c^T theta_t; estimate by baseline-adjusted marginal standardization, with unit/cluster-appropriate inference.; variables: Y: AMH comprehension score at four weeks; i: participant; t: comprehension items and follow-up visits; the registered visit/time-window index, not another independent unit; A: participant individually randomized to decision aid versus usual counseling; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: four-week assignment effect on the prespecified 0–100 AMH-understanding score; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; z(A): prespecified indicators for the actual randomized arms; two arms reduce to one indicator; family code: RCT
Sample size design: family: contrast precision/power by design-matrix or simulation planning; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Pregnancy is descriptive rather than a validation of AMH fertility prediction; aid use and later treatment are postassignment consequences.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: If knowledge improves but false reassurance, avoidable spending, or delayed referral increase, the proposed decision benefit fails.; preanalysis gate: Pregnancy is descriptive rather than a validation of AMH fertility prediction; aid use and later treatment are postassignment consequences.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Pregnancy is descriptive rather than a validation of AMH fertility prediction; aid use and later treatment are postassignment consequences.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Pregnancy is descriptive rather than a validation of AMH fertility prediction; aid use and later treatment are postassignment consequences.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/19; question: Which risk-communication approach helps people interpret anti-Müllerian hormone results without false reassurance or unnecessary alarm?; aims: Identify misunderstandings in real pretest and post-test consultations.
Co-design explanations distinguishing reserve, response, and reproductive outcomes.
Randomize decision support and assess durable comprehension and subsequent choices.; design verbatim planning only: Planning assumption: 360 adults seeking ovarian-reserve information receive usual counseling or an additional validated decision aid before clinically ordered testing. Follow-up at six months records decisions, care use, and distress; no additional testing is assigned.; outcomes verbatim: Primary: calibrated understanding of what AMH can and cannot predict. Secondary: decisional conflict, out-of-pocket spending, clinically appropriate referral, anxiety, and regret.; analysis verbatim inherited not authoritative if refined: Stratify by infertility history and age; model baseline fertility intentions and numeracy. Prespecify a comprehension scale and decision-quality criteria before unblinding, with sensitivity analyses for missing financial data.; falsification verbatim: If knowledge improves but false reassurance, avoidable spending, or delayed referral increase, the proposed decision benefit fails.; source ids unverified leads: female-fertility-amh-review-2026
belief-effects-health-review-2026; prospectus precedence: Yes; prospectus file: thesis-prospectuses.json; prospectus design and estimand: The analysis unit is the participant. Concealed individual randomization identifies the treatment-policy mean difference in the prespecified 0–100 comprehension score at four weeks, regardless of aid use, result disclosure timing or later care. Usual counseling remains available in both groups. Monitor clinician contamination and document counseling content; substantial contamination would attenuate the offer effect and may motivate a later cluster implementation study.; prospectus analysis plan: Estimate the baseline-adjusted randomized four-week comprehension contrast with clinic and stratification factors, reporting confidence intervals and item-specific misconceptions. Prespecify limited age, language and infertility-history interactions; avoid subgroup fertility prediction. Secondary analyses describe preference congruence, distress and referral timing with multiplicity-aware interpretation. Pregnancy outcomes, if voluntarily reported, are descriptive and cannot establish communication effects on fertility or validate individual AMH forecasts.; precedence note: Expanded prospectus refinements govern this scaffold; verbatim earlier brief material is retained for traceability, not as a contradictory analysis instruction. F15 is explicitly two arms; F30 has no unjustified symptom noninferiority claim.

E24 — Ovarian rejuvenation claims versus cumulative live birth

Rank: 141
Cluster: Women's reproductive and sexual health and menopause
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: OBS_BINARY; allocation or exposure unit: patient independently chooses ovarian PRP or established care; independent sampling unit: baseline-enrolled patient; repeated measurement unit: treatment cycles and pregnancies followed to resolution; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: observational comparison/association; causal interpretation is conditional on explicit exchangeability, positivity, consistency and observation assumptions; specific constraint: All enrolled patients remain in denominator; donor-oocyte birth separate; switching, treatment opportunity and pregnancies extending beyond enrollment window must be tracked.
Primary estimand: standardized eighteen-month cumulative autologous-oocyte live-birth probability difference between baseline strategies, conditional on exchangeability
Candidate primary outcome: at least one autologous-oocyte live birth from attempts initiated within eighteen months
Primary finalized: No
Mathematical model: type: observational overlap-weighted/g-method precision simulation with effective independent units, outcome/event frequency and residual-bias scenarios; formulas: W_i(t) = product_(k<=t) P(A_ik | A_i,<k,X_i0)/P(A_ik | H_ik), optionally multiplied by separately justified observation weights.
m_a = E_X[m(a,X)] for a baseline strategy, or the sequential g-formula for sustained strategies; Delta=m_1-m_0. Fit a weighted, baseline-aligned mean/risk model matching Y rather than assume the treatment coefficient is causal.
m_a = Pr(Y^a=1) by the fixed horizon; retain all baseline-enrolled participants under the chosen switching/competing-event rule.; variables: Y: at least one autologous-oocyte live birth from attempts initiated within eighteen months; i: baseline-enrolled patient; t: treatment cycles and pregnancies followed to resolution; the registered visit/time-window index, not another independent unit; A: patient independently chooses ovarian PRP or established care; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: standardized eighteen-month cumulative autologous-oocyte live-birth probability difference between baseline strategies, conditional on exchangeability; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; H ik: measured predecision history including prior treatment and time-varying covariates; m(a,X): outcome mean/risk model under specified treatment strategy and observed covariates; m a: standardized mean/risk under a, identified only under the listed assumptions; k: ordered treatment decision time; family code: OBS_BINARY
Sample size design: family: observational overlap-weighted/g-method precision simulation with effective independent units, outcome/event frequency and residual-bias scenarios; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: All enrolled patients remain in denominator; donor-oocyte birth separate; switching, treatment opportunity and pregnancies extending beyond enrollment window must be tracked.
Common eligibility/time zero and a defensible active comparator; treatment choice is not randomized.
Conditional exchangeability (including time-varying feedback where relevant), positivity and correct observation/treatment/outcome models are untestable or only partly testable; observed overlap does not prove them.
Do not adjust for posttreatment variables as baseline confounders; define a separate mediation/per-protocol estimand before using g-methods.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: A marker gain without improved live birth or with added harm falsifies the proposed rejuvenation interpretation.; preanalysis gate: All enrolled patients remain in denominator; donor-oocyte birth separate; switching, treatment opportunity and pregnancies extending beyond enrollment window must be tracked.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: All enrolled patients remain in denominator; donor-oocyte birth separate; switching, treatment opportunity and pregnancies extending beyond enrollment window must be tracked.
Overlap/effective-sample-size and weight-tail diagnostics with declared truncation sensitivity; report population changes caused by trimming.
Quantitative unmeasured-confounding and informative-loss sensitivity; negative controls/falsification windows where scientifically defensible.
Compare g-computation/weighting/doubly robust estimates on the same target population and time zero, without interpreting agreement as proof.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
All enrolled patients remain in denominator; donor-oocyte birth separate; switching, treatment opportunity and pregnancies extending beyond enrollment window must be tracked.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/140; question: Do ovarian PRP-associated marker changes predict cumulative live birth after accounting for prognosis and selection into treatment?; aims: Audit endpoints and selective reporting in existing clinical records.
Establish a prospective comparative registry before treatment decisions.
Estimate cumulative live birth, harms, and cost using transparent target-trial assumptions.; design verbatim planning only: Assume 600 patients across six fertility centers are enrolled while considering existing care options; researchers administer no PRP. Compare those independently choosing PRP with clinically comparable patients pursuing established care, followed for eighteen months.; outcomes verbatim: Primary: cumulative autologous-oocyte live birth per enrolled patient from treatment attempts initiated during eighteen months, with all resulting pregnancies followed to resolution. Secondary: donor-oocyte and overall live birth reported separately, pregnancy loss, complications, time to pregnancy, AMH, oocyte yield, and total patient expenditure.; analysis verbatim inherited not authoritative if refined: Align eligibility and time zero, address immortal-time bias, and adjust measured prognosis, prior cycles, age, and access factors. Use sensitivity analyses for unmeasured confounding and missing outcomes; do not claim randomized equivalence. Retain all baseline-enrolled patients in the primary denominator regardless of treatment stopping or switching; distinguish autologous from donor-oocyte outcomes and report the implications of switching and differential treatment opportunity.; falsification verbatim: A marker gain without improved live birth or with added harm falsifies the proposed rejuvenation interpretation.; source ids unverified leads: ovarian-prp-meta-2025
diminished-ovarian-reserve-rct-meta-2025
female-fertility-amh-review-2026; prospectus precedence: No

E25 — Cycle-tracking reliability beyond the textbook cycle

Rank: 131
Cluster: Women's reproductive and sexual health and menopause
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: MULTISTATE_DIAG; allocation or exposure unit: app prediction evaluated against state-appropriate reference; independent sampling unit: participant; repeated measurement unit: cycle, day and state prediction; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: diagnostic/benchmark agreement or classification validity; reader/interface/navigation effects are causal only in their separately randomized component; specific constraint: Reference uncertainty is not a negative event; do not drop irregular cycles or treat withdrawal bleeding as ovulation.
Primary estimand: participant-level phase probability calibration/coverage including no-ovulation, not-applicable and indeterminate states
Candidate primary outcome: reference state drawn from explicit phase/ovulation/applicability categories
Primary finalized: No
Mathematical model: type: diagnostic/benchmark sensitivity, false-alert and calibration precision with independent cases/noncases and cluster/reference uncertainty; formulas: For state k, BS = mean_i sum_k (p_ik - 1{R_i=k})^2; calibrate p_ik against each state including anovulation and not-applicable.
Reference-indeterminate probability is reported and modeled separately; do not relabel an indeterminate reference as a wrong prediction or discard irregular cycles.; variables: Y: reference state drawn from explicit phase/ovulation/applicability categories; i: participant; t: cycle, day and state prediction; the registered visit/time-window index, not another independent unit; A: app prediction evaluated against state-appropriate reference; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: participant-level phase probability calibration/coverage including no-ovulation, not-applicable and indeterminate states; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; TP,FN,FP: true-positive, false-negative and false-positive counts relative to the independently defined reference and matching window, not an assumed perfect gold standard; reference: per-thesis independent reference, including adjudication uncertainty; nonevent time/opportunities: fixed at-risk denominator; event specificity and false alerts per hour are distinct targets; k: state in the preregistered phase/ovulation/applicability state space; p ik: app probability for state k at an eligible prediction time; R i: state-appropriate reference label, with explicit indeterminate handling; BS: multiclass Brier score averaged with participant weights fixed before evaluation; family code: MULTISTATE_DIAG
Sample size design: family: diagnostic/benchmark sensitivity, false-alert and calibration precision with independent cases/noncases and cluster/reference uncertainty; formula: An independent-case rough starting point is n_case ~= z_(1-alpha*/2)^2 Se(1-Se)/h_Se^2; replace with interval-based/clustered simulation for the actual design. Plan noncases/observation time and reference failures separately.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; Se: plausible sensitivity from prior evidence or pilot, not an asserted study result; h Se: desired sensitivity interval half-width; n case: independent positive cases; repeated events within a person require correlation adjustment; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Reference uncertainty is not a negative event; do not drop irregular cycles or treat withdrawal bleeding as ovulation.
Reference independence, fixed thresholds/window and an explicit denominator; incorporation/verification bias can invalidate apparent accuracy.
Participant/family-level test splits and paired comparisons; report abstention, assay failure and indeterminate truth rather than force every case into a binary label.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: If uncertainty estimates are overconfident in irregular cycles or delay clinical help, the improved-tracking claim fails.; preanalysis gate: Reference uncertainty is not a negative event; do not drop irregular cycles or treat withdrawal bleeding as ovulation.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Reference uncertainty is not a negative event; do not drop irregular cycles or treat withdrawal bleeding as ovulation.
Vary plausible reference misclassification, verification-selection and event-matching window under a prespecified sensitivity family.
Report subgroup usable-data coverage and calibration; avoid interpreting an imprecise null difference as equivalence.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Reference uncertainty is not a negative event; do not drop irregular cycles or treat withdrawal bleeding as ovulation.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/130; question: How well do menstrual apps represent uncertainty in irregular cycles, changing contraception status, and the menopausal transition?; aims: Construct a clinically characterized prospective cycle dataset.
Benchmark app predictions against independent biological and bleeding references.
Test whether uncertainty displays improve appropriate symptom reporting and referral decisions.; design verbatim planning only: Planning assumption: 300 adults, intentionally including clinically diagnosed PCOS and perimenopause, contribute six months of app exports and supervised reference assessments in selected cycles. A nested randomized display study compares phase labels with calibrated uncertainty.; outcomes verbatim: Primary: participant-level calibration and coverage of phase predictions against state-appropriate independent references, including explicit no-ovulation, phase-not-applicable, and indeterminate categories. Secondary: bleeding-timing accuracy, missed clinically relevant irregularity, user comprehension, anxiety, and referral appropriateness.; analysis verbatim inherited not authoritative if refined: Separate person-level training and testing; account for hormonal contraception, anovulatory cycles, age, and reference uncertainty. Prespecify subgroups and correct comparisons; do not exclude irregular cycles to improve headline accuracy. Define separate reference targets for bleeding timing, ovulation, and phase; use explicit no-ovulation, not-applicable, or indeterminate states and do not treat withdrawal bleeding as evidence of ovulation.; falsification verbatim: If uncertainty estimates are overconfident in irregular cycles or delay clinical help, the improved-tracking claim fails.; source ids unverified leads: female-menstrual-health-activity-meta-2025
menstrual-energy-intake-meta-2025
n-of-1-methods-review-2025; prospectus precedence: No

E26 — Personalizing low-desire care by distress and context

Rank: 56
Cluster: Women's reproductive and sexual health and menopause
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT; allocation or exposure unit: participant randomized to biopsychosocial assessment/navigation or education; independent sampling unit: participant; repeated measurement unit: sexual-distress assessment visit; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Low desire is not itself a disorder or universal outcome goal; preserve unpartnered participants and baseline phenotype definition.
Primary estimand: assigned care-pathway effect on validated sexual distress at six months
Candidate primary outcome: validated sexual-distress score
Primary finalized: No
Mathematical model: type: contrast precision/power by design-matrix or simulation planning; formulas: E(Y_it | A_i, X_i, Y_i0) = alpha_t + z(A_i)^T theta_t + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Delta = sum_t w_t c^T theta_t; estimate by baseline-adjusted marginal standardization, with unit/cluster-appropriate inference.; variables: Y: validated sexual-distress score; i: participant; t: sexual-distress assessment visit; the registered visit/time-window index, not another independent unit; A: participant randomized to biopsychosocial assessment/navigation or education; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: assigned care-pathway effect on validated sexual distress at six months; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; z(A): prespecified indicators for the actual randomized arms; two arms reduce to one indicator; family code: RCT
Sample size design: family: contrast precision/power by design-matrix or simulation planning; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Low desire is not itself a disorder or universal outcome goal; preserve unpartnered participants and baseline phenotype definition.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: If desire scores rise without reduced distress or personally valued benefit, the proposed care improvement is unsupported.; preanalysis gate: Low desire is not itself a disorder or universal outcome goal; preserve unpartnered participants and baseline phenotype definition.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Low desire is not itself a disorder or universal outcome goal; preserve unpartnered participants and baseline phenotype definition.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Low desire is not itself a disorder or universal outcome goal; preserve unpartnered participants and baseline phenotype definition.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/55; question: Does a biopsychosocial decision pathway improve distressing low sexual desire more than generic sexual-health education?; aims: Identify reproducible clinical and contextual phenotypes.
Co-design a pathway linking priorities to appropriate existing care.
Randomize pathway support against attention-matched education and assess sustained benefit.; design verbatim planning only: Assume 240 women seeking help for distressing low desire enter a pragmatic six-month trial with twelve-month follow-up. Both groups retain clinician-directed treatment access; the intervention improves assessment and navigation rather than assigning hormones or drugs.; outcomes verbatim: Primary: validated sexual-distress change. Secondary: goal attainment, pain, relationship satisfaction where relevant, care uptake, adverse effects, and financial burden.; analysis verbatim inherited not authoritative if refined: Use intention-to-treat models stratified by menopausal status and medication context. Prespecify one phenotype interaction, adjust multiplicity, and handle partner-related missingness without excluding unpartnered participants.; falsification verbatim: If desire scores rise without reduced distress or personally valued benefit, the proposed care improvement is unsupported.; source ids unverified leads: female-hsdd-icsm-2024
belief-effects-health-review-2026; prospectus precedence: No

E27 — Who benefits from pelvic-floor biofeedback?

Rank: 119
Cluster: Women's reproductive and sexual health and menopause
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT_INT; allocation or exposure unit: participant randomized to individualized care with or without appropriate biofeedback; independent sampling unit: participant, with therapist clustering; repeated measurement unit: prespecified patient-priority pain/function assessment visit; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Choose scoring before allocation so outcome selection is not response-driven; phenotype interaction needs within-phenotype randomized overlap.
Primary estimand: biofeedback assignment effect and one baseline muscle-phenotype interaction on the fixed priority outcome hierarchy
Candidate primary outcome: patient-prioritized pain or sexual-function score under a prespecified scoring hierarchy
Primary finalized: No
Mathematical model: type: interaction-specific precision/power with modifier distribution and reliability; simulate the actual interaction; formulas: E(Y_it | A_i, B_i, X_i) = alpha_t + tau_t A_i + eta_t B_i + kappa_t A_i B_i + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Treatment contrast at baseline B=b is tau_t + kappa_t b; the interaction target is kappa_t, not a difference in within-subgroup p-values.; variables: Y: patient-prioritized pain or sexual-function score under a prespecified scoring hierarchy; i: participant, with therapist clustering; t: prespecified patient-priority pain/function assessment visit; the registered visit/time-window index, not another independent unit; A: participant randomized to individualized care with or without appropriate biofeedback; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: biofeedback assignment effect and one baseline muscle-phenotype interaction on the fixed priority outcome hierarchy; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; B: pretreatment clinically assessed pelvic-floor phenotype under a fixed coding hierarchy; centering/coding and any measurement-error model fixed before evaluation; eta: baseline modifier association; kappa: assignment-by-baseline-modifier interaction coefficient on the stated scale; family code: RCT_INT
Sample size design: family: interaction-specific precision/power with modifier distribution and reliability; simulate the actual interaction; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure
modifier prevalence/distribution and variance
modifier reliability and overlap
interaction size or precision target; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Choose scoring before allocation so outcome selection is not response-driven; phenotype interaction needs within-phenotype randomized overlap.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The modifier must be measured before assignment/initiation; interaction is scale dependent and requires support within modifier strata.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: If feedback changes muscle signals without improving symptoms or function, its proposed clinical value is not supported.; preanalysis gate: Choose scoring before allocation so outcome selection is not response-driven; phenotype interaction needs within-phenotype randomized overlap.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Choose scoring before allocation so outcome selection is not response-driven; phenotype interaction needs within-phenotype randomized overlap.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Vary modifier measurement-error assumptions and retain a continuous modifier unless a cutoff was independently fixed.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Choose scoring before allocation so outcome selection is not response-driven; phenotype interaction needs within-phenotype randomized overlap.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/118; question: Does adding instrumented biofeedback to individualized pelvic-floor therapy improve pain and sexual function, and does baseline muscle phenotype matter?; aims: Standardize clinician phenotype assessment and outcome selection.
Randomize adjunctive feedback within phenotype-guided therapy.
Test persistence and whether early skill change predicts later functional improvement.; design verbatim planning only: Planning assumption: 220 women with pelvic-floor-associated sexual symptoms receive therapist-directed individualized care, with or without additional clinically appropriate biofeedback, over sixteen weeks and six-month follow-up. No participant is assigned unsuitable generic strengthening.; outcomes verbatim: Primary: patient-prioritized pain or sexual-function outcome chosen within a prespecified hierarchy. Secondary: distress, urinary symptoms, skill measures, adherence, and treatment burden.; analysis verbatim inherited not authoritative if refined: Stratify by assessed phenotype and menopausal status; estimate intention-to-treat effects with therapist clustering. Prespecify interaction tests, measurement reliability, and multiplicity handling; avoid circularly defining responders by baseline severity.; falsification verbatim: If feedback changes muscle signals without improving symptoms or function, its proposed clinical value is not supported.; source ids unverified leads: female-pelvic-floor-sexual-meta-2024
female-pelvic-floor-postmeno-meta-2025; prospectus precedence: No

E28 — Digital symptom tracking that shortens the route to GSM care

Rank: 57
Cluster: Women's reproductive and sexual health and menopause
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: TIME_CARE; allocation or exposure unit: participant randomized after a separate clinical validation stage; independent sampling unit: participant; repeated measurement unit: care-pathway event and adjudication visit; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Death, competing alternative diagnosis and no plan require explicit event definitions; faster inappropriate treatment cannot count as success.
Primary estimand: difference in cumulative incidence or restricted mean time to an independently appropriate care plan by the fixed horizon
Candidate primary outcome: time from randomization to adjudicated appropriate care plan
Primary finalized: No
Mathematical model: type: randomized competing-risk care-incidence or restricted-mean time precision with adjudication/verification uncertainty; formulas: F_C^a(t)=Pr(T^a<=t,J^a=appropriate care plan); estimate randomized cumulative-incidence difference F_C^1(tau)-F_C^0(tau).
A registered restricted-mean time-without-appropriate-plan contrast is integral_0^tau {[1-F_C^1(t)]-[1-F_C^0(t)]}dt; treat death/other competing events as competing outcomes under the stated definition, not independent censoring.; variables: Y: time from randomization to adjudicated appropriate care plan; i: participant; t: care-pathway event and adjudication visit; the registered visit/time-window index, not another independent unit; A: participant randomized after a separate clinical validation stage; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: difference in cumulative incidence or restricted mean time to an independently appropriate care plan by the fixed horizon; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; T,J: time from randomization and event type, including appropriate plan and competing event; tau: registered care-plan horizon; F C^a: appropriate-plan cumulative incidence under assignment a; does not label faster wrong treatment as success; family code: TIME_CARE
Sample size design: family: randomized competing-risk care-incidence or restricted-mean time precision with adjudication/verification uncertainty; formula: Use design-specific simulation or independent-unit resampling for the stated estimand and target confidence width/power; preserve all allocation, censoring, clustering and measurement-error structure.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Death, competing alternative diagnosis and no plan require explicit event definitions; faster inappropriate treatment cannot count as success.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Blinded independent appropriateness adjudication and common ascertainment; event/care-plan reversal and competing diagnoses require a protocol definition.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: If faster care increases missed alternative diagnoses or unnecessary treatment, the proposed triage benefit fails.; preanalysis gate: Death, competing alternative diagnosis and no plan require explicit event definitions; faster inappropriate treatment cannot count as success.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Death, competing alternative diagnosis and no plan require explicit event definitions; faster inappropriate treatment cannot count as success.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Death, competing alternative diagnosis and no plan require explicit event definitions; faster inappropriate treatment cannot count as success.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/56; question: Can symptom-based digital triage improve recognition of genitourinary syndrome of menopause without misclassifying infection, dermatologic disease, or pelvic pain?; aims: Prospectively map symptom trajectories against clinical diagnoses.
Develop a transparent triage tool with explicit uncertainty.
Randomize supported tracking versus usual appointment preparation.; design verbatim planning only: Assume 320 peri- and postmenopausal patients with vulvovaginal, urinary, or sexual symptoms attend primary-care and gynecology services. A clinical validation cohort precedes a six-month pragmatic randomized trial; adjudicators are blinded to tool assignment.; outcomes verbatim: Primary: time to an independently judged appropriate care plan. Secondary: symptom burden, diagnostic error, unnecessary antibiotics or purchases, sexual distress, and health-service cost.; analysis verbatim inherited not authoritative if refined: Use intention-to-treat survival analysis, prespecified diagnostic categories, and site stratification. Adjust the diagnostic accuracy family for multiplicity and assess verification bias when follow-up examination is incomplete.; falsification verbatim: If faster care increases missed alternative diagnoses or unnecessary treatment, the proposed triage benefit fails.; source ids unverified leads: female-gsm-icsm-2026
female-gsm-guideline-2025; prospectus precedence: No

E29 — Long-term outcomes after marketed vaginal rejuvenation

Rank: 198
Cluster: Women's reproductive and sexual health and menopause
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: OBS_BINARY; allocation or exposure unit: patient independently chooses marketed procedure or conservative care; independent sampling unit: patient; repeated measurement unit: confidential follow-up and adjudicated complication; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: observational comparison/association; causal interpretation is conditional on explicit exchangeability, positivity, consistency and observation assumptions; specific constraint: Selection of clinic, procedure and follow-up can confound apparent satisfaction; classify durability and independently obtain outcomes.
Primary estimand: standardized risk of persistent pain/impairment/complication across observed care strategies, with causal interpretation conditional on exchangeability
Candidate primary outcome: prespecified persistent clinically adjudicated complication indicator
Primary finalized: No
Mathematical model: type: observational overlap-weighted/g-method precision simulation with effective independent units, outcome/event frequency and residual-bias scenarios; formulas: W_i(t) = product_(k<=t) P(A_ik | A_i,<k,X_i0)/P(A_ik | H_ik), optionally multiplied by separately justified observation weights.
m_a = E_X[m(a,X)] for a baseline strategy, or the sequential g-formula for sustained strategies; Delta=m_1-m_0. Fit a weighted, baseline-aligned mean/risk model matching Y rather than assume the treatment coefficient is causal.
m_a = Pr(Y^a=1) by the fixed horizon; retain all baseline-enrolled participants under the chosen switching/competing-event rule.; variables: Y: prespecified persistent clinically adjudicated complication indicator; i: patient; t: confidential follow-up and adjudicated complication; the registered visit/time-window index, not another independent unit; A: patient independently chooses marketed procedure or conservative care; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: standardized risk of persistent pain/impairment/complication across observed care strategies, with causal interpretation conditional on exchangeability; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; H ik: measured predecision history including prior treatment and time-varying covariates; m(a,X): outcome mean/risk model under specified treatment strategy and observed covariates; m a: standardized mean/risk under a, identified only under the listed assumptions; k: ordered treatment decision time; family code: OBS_BINARY
Sample size design: family: observational overlap-weighted/g-method precision simulation with effective independent units, outcome/event frequency and residual-bias scenarios; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Selection of clinic, procedure and follow-up can confound apparent satisfaction; classify durability and independently obtain outcomes.
Common eligibility/time zero and a defensible active comparator; treatment choice is not randomized.
Conditional exchangeability (including time-varying feedback where relevant), positivity and correct observation/treatment/outcome models are untestable or only partly testable; observed overlap does not prove them.
Do not adjust for posttreatment variables as baseline confounders; define a separate mediation/per-protocol estimand before using g-methods.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: If sustained improvement is robust, complications rare, and results remain favorable after independent follow-up, the hypothesized satisfaction inflation is unsupported.; preanalysis gate: Selection of clinic, procedure and follow-up can confound apparent satisfaction; classify durability and independently obtain outcomes.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Selection of clinic, procedure and follow-up can confound apparent satisfaction; classify durability and independently obtain outcomes.
Overlap/effective-sample-size and weight-tail diagnostics with declared truncation sensitivity; report population changes caused by trimming.
Quantitative unmeasured-confounding and informative-loss sensitivity; negative controls/falsification windows where scientifically defensible.
Compare g-computation/weighting/doubly robust estimates on the same target population and time zero, without interpreting agreement as proof.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Selection of clinic, procedure and follow-up can confound apparent satisfaction; classify durability and independently obtain outcomes.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/197; question: What are the durable patient-reported outcomes, complications, and financial consequences after elective energy-based vaginal procedures?; aims: Audit consent claims and routine outcome collection.
Recruit an independent prospective cohort before independently chosen procedures or conservative care.
Compare durable function, harms, costs, and expectations over two years.; design verbatim planning only: Planning assumption: 400 patients considering existing marketed care are recruited through multiple independent channels; researchers provide no procedure and no treatment recommendation. Follow-up includes confidential reports and clinically adjudicated complications, with a conservative-care comparison cohort.; outcomes verbatim: Primary: persistent pain, functional impairment, or other clinically adjudicated complications. Secondary: symptom relief, satisfaction, retreatment, regret, sexual distress, and out-of-pocket cost.; analysis verbatim inherited not authoritative if refined: Align baseline and follow-up windows; adjust measured indication severity, expectations, finances, and prior treatment. Use inverse-probability follow-up weighting and sensitivity analyses for unmeasured confounding and selective clinic participation.; falsification verbatim: If sustained improvement is robust, complications rare, and results remain favorable after independent follow-up, the hypothesized satisfaction inflation is unsupported.; source ids unverified leads: fda-vaginal-energy-device-warning-2018
female-gsm-guideline-2025
female-gsm-icsm-2026; prospectus precedence: No

E30 — Whose menopause treatment evidence transports to ordinary care?

Rank: 97
Cluster: Women's reproductive and sexual health and menopause
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: OBS_LONG; allocation or exposure unit: clinician-selected menopause care strategy with treatment changes; independent sampling unit: patient within data system; repeated measurement unit: longitudinal symptom/function and treatment history; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: observational comparison/association; causal interpretation is conditional on explicit exchangeability, positivity, consistency and observation assumptions; specific constraint: Define target population and active comparator; treatment-confounder feedback and positivity failure prevent simple baseline-only causal adjustment.
Primary estimand: overlap-population and transported mean function change under defined care strategies, conditional on sequential exchangeability
Candidate primary outcome: validated symptom-related daily-function score at eighteen months
Primary finalized: No
Mathematical model: type: observational overlap-weighted/g-method precision simulation with effective independent units, outcome/event frequency and residual-bias scenarios; formulas: W_i(t) = product_(k<=t) P(A_ik | A_i,<k,X_i0)/P(A_ik | H_ik), optionally multiplied by separately justified observation weights.
m_a = E_X[m(a,X)] for a baseline strategy, or the sequential g-formula for sustained strategies; Delta=m_1-m_0. Fit a weighted, baseline-aligned mean/risk model matching Y rather than assume the treatment coefficient is causal.; variables: Y: validated symptom-related daily-function score at eighteen months; i: patient within data system; t: longitudinal symptom/function and treatment history; the registered visit/time-window index, not another independent unit; A: clinician-selected menopause care strategy with treatment changes; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: overlap-population and transported mean function change under defined care strategies, conditional on sequential exchangeability; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; H ik: measured predecision history including prior treatment and time-varying covariates; m(a,X): outcome mean/risk model under specified treatment strategy and observed covariates; m a: standardized mean/risk under a, identified only under the listed assumptions; k: ordered treatment decision time; family code: OBS_LONG
Sample size design: family: observational overlap-weighted/g-method precision simulation with effective independent units, outcome/event frequency and residual-bias scenarios; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Define target population and active comparator; treatment-confounder feedback and positivity failure prevent simple baseline-only causal adjustment.
Common eligibility/time zero and a defensible active comparator; treatment choice is not randomized.
Conditional exchangeability (including time-varying feedback where relevant), positivity and correct observation/treatment/outcome models are untestable or only partly testable; observed overlap does not prove them.
Do not adjust for posttreatment variables as baseline confounders; define a separate mediation/per-protocol estimand before using g-methods.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: If estimated differences disappear after indication alignment or cannot transport to underrepresented groups, broad personalization claims fail.; preanalysis gate: Define target population and active comparator; treatment-confounder feedback and positivity failure prevent simple baseline-only causal adjustment.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Define target population and active comparator; treatment-confounder feedback and positivity failure prevent simple baseline-only causal adjustment.
Overlap/effective-sample-size and weight-tail diagnostics with declared truncation sensitivity; report population changes caused by trimming.
Quantitative unmeasured-confounding and informative-loss sensitivity; negative controls/falsification windows where scientifically defensible.
Compare g-computation/weighting/doubly robust estimates on the same target population and time zero, without interpreting agreement as proof.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Define target population and active comparator; treatment-confounder feedback and positivity failure prevent simple baseline-only causal adjustment.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/96; question: How well do symptom and safety estimates for clinician-selected menopause care generalize across socioeconomic position, comorbidity, and treatment access?; aims: Map representation and missing outcomes in routine-care datasets.
Emulate a sequence of clinically relevant new-user treatment comparisons.
Validate patient-reported symptom and functional outcomes in an inclusive prospective subcohort.; design verbatim planning only: Assume linkage of an established regional health-record cohort with approximately 600 newly recruited symptomatic patients followed for eighteen months. Compare clinician-selected care strategies only where treatment indications and adequate overlap permit credible analysis.; outcomes verbatim: Primary: validated symptom-related daily-function change. Secondary: vasomotor burden, treatment discontinuation, costs, and clinically documented adverse outcomes; rare safety events require larger external linkage.; analysis verbatim inherited not authoritative if refined: Specify target trials, active comparators, time zero, and treatment-confounder feedback. Use propensity weighting and transportability analyses, report positivity failures, and label residual-confounding sensitivity rather than asserting randomized certainty.; falsification verbatim: If estimated differences disappear after indication alignment or cannot transport to underrepresented groups, broad personalization claims fail.; source ids unverified leads: menopause-hormone-therapy-review-2025
menopause-cognition-sleep-review-2025
digital-twin-causal-trials-2026; prospectus precedence: No

E31 — When a home LED mask improves the photograph

Rank: 186
Cluster: Skin, hair, vision, hearing, and sensory function
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT; allocation or exposure unit: participant randomized home light device versus credible comparator; independent sampling unit: participant; repeated measurement unit: three-dimensional imaging visit; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Choose metric and timing; photography standardization, pigment-specific measurement error and independently verified device output precede efficacy analysis.
Primary estimand: device-assignment effect on independently scored three-dimensional surface change and its prespecified persistence contrast
Candidate primary outcome: blinded three-dimensional skin-surface metric
Primary finalized: No
Mathematical model: type: contrast precision/power by design-matrix or simulation planning; formulas: E(Y_it | A_i, X_i, Y_i0) = alpha_t + z(A_i)^T theta_t + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Delta = sum_t w_t c^T theta_t; estimate by baseline-adjusted marginal standardization, with unit/cluster-appropriate inference.; variables: Y: blinded three-dimensional skin-surface metric; i: participant; t: three-dimensional imaging visit; the registered visit/time-window index, not another independent unit; A: participant randomized home light device versus credible comparator; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: device-assignment effect on independently scored three-dimensional surface change and its prespecified persistence contrast; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; z(A): prespecified indicators for the actual randomized arms; two arms reduce to one indicator; family code: RCT
Sample size design: family: contrast precision/power by design-matrix or simulation planning; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Choose metric and timing; photography standardization, pigment-specific measurement error and independently verified device output precede efficacy analysis.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: If improvement appears only in consumer photographs or disappears after hydration and imaging standardization, the structural cosmetic claim fails.; preanalysis gate: Choose metric and timing; photography standardization, pigment-specific measurement error and independently verified device output precede efficacy analysis.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Choose metric and timing; photography standardization, pigment-specific measurement error and independently verified device output precede efficacy analysis.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Choose metric and timing; photography standardization, pigment-specific measurement error and independently verified device output precede efficacy analysis.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/185; question: Do home light-device effects on skin appearance persist under standardized imaging, diverse pigmentation, and independent device verification?; aims: Independently characterize device output, heat, fit, and sham credibility.
Run a randomized sham-controlled clinical evaluation.
Test durability and agreement between instrumental and patient-prioritized outcomes.; design verbatim planning only: Planning assumption: 200 adults with photoaging concerns enter a dermatologist-supervised sixteen-week trial with six-month follow-up. Randomization is stratified by measured pigmentation; usual photoprotection continues and no aggressive home procedures are added.; outcomes verbatim: Primary: blinded three-dimensional skin-surface change. Secondary: patient-valued appearance, irritation, pigment change, adherence, and durability after device cessation.; analysis verbatim inherited not authoritative if refined: Use standardized capture conditions, assessor masking, batch-balanced imaging, and prespecified facial regions. Model baseline pigmentation, age, sun exposure, and multiplicity; quantify whether image algorithms have unequal error.; falsification verbatim: If improvement appears only in consumer photographs or disappears after hydration and imaging standardization, the structural cosmetic claim fails.; source ids unverified leads: skin-led-rct-2025
skin-photoaging-systematic-review-2025; prospectus precedence: No

E32 — UV feedback that works across skin tones and budgets

Rank: 58
Cluster: Skin, hair, vision, hearing, and sensory function
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT; allocation or exposure unit: participant randomized to personalized UV feedback versus generic messages; independent sampling unit: participant; repeated measurement unit: dosimeter day and protection record; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Dosimetry alone is not skin exposure under clothing; algorithm validity and unlogged protection require sensitivity checks.
Primary estimand: feedback-offer effect on independently assessed unprotected UV exposure over the fixed horizon
Candidate primary outcome: reference UV dose combined with prespecified protection record algorithm
Primary finalized: No
Mathematical model: type: contrast precision/power by design-matrix or simulation planning; formulas: E(Y_it | A_i, X_i, Y_i0) = alpha_t + z(A_i)^T theta_t + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Delta = sum_t w_t c^T theta_t; estimate by baseline-adjusted marginal standardization, with unit/cluster-appropriate inference.; variables: Y: reference UV dose combined with prespecified protection record algorithm; i: participant; t: dosimeter day and protection record; the registered visit/time-window index, not another independent unit; A: participant randomized to personalized UV feedback versus generic messages; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: feedback-offer effect on independently assessed unprotected UV exposure over the fixed horizon; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; z(A): prespecified indicators for the actual randomized arms; two arms reduce to one indicator; family code: RCT
Sample size design: family: contrast precision/power by design-matrix or simulation planning; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Dosimetry alone is not skin exposure under clothing; algorithm validity and unlogged protection require sensitivity checks.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: If alerts lower outdoor activity without reducing unprotected exposure or worsen disparities, the proposed practical benefit fails.; preanalysis gate: Dosimetry alone is not skin exposure under clothing; algorithm validity and unlogged protection require sensitivity checks.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Dosimetry alone is not skin exposure under clothing; algorithm validity and unlogged protection require sensitivity checks.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Dosimetry alone is not skin exposure under clothing; algorithm validity and unlogged protection require sensitivity checks.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/57; question: Can personalized ultraviolet feedback improve everyday photoprotection without discouraging beneficial outdoor activity?; aims: Validate wearable exposure estimates across clothing and skin characteristics.
Co-design culturally appropriate protection messages.
Randomize feedback strategies and test sustained behavior and burden.; design verbatim planning only: Assume 300 adults, including outdoor workers and diverse pigmentation, receive a summer-long trial of standard advice, UV alerts, or alerts plus practical protection support. All groups retain usual sun-safety resources; no deliberate UV exposure occurs.; outcomes verbatim: Primary: independently assessed unprotected UV exposure, combining dosimetry with validated clothing and protection records. Secondary: sunburn, pigmentation changes, outdoor activity, anxiety, cost, and adherence.; analysis verbatim inherited not authoritative if refined: Use intention-to-treat analysis, calibrate sensor placement, and model occupation, weather, baseline habits, and missing wear. Prespecify one primary contrast and pigmentation interaction; avoid interpreting short-term exposure reduction as proven cancer prevention.; falsification verbatim: If alerts lower outdoor activity without reducing unprotected exposure or worsen disparities, the proposed practical benefit fails.; source ids unverified leads: skin-photoaging-systematic-review-2025
wearables-living-review; prospectus precedence: No

E33 — Does simplifying a skincare stack restore barrier function?

Rank: 79
Cluster: Skin, hair, vision, hearing, and sensory function
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT; allocation or exposure unit: participant randomized to structured simplification or comparator care; independent sampling unit: participant; repeated measurement unit: irritation-related quality-of-life assessment; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Product count and transepidermal water loss are secondary/process measures; rescue treatment remains in the strategy estimand.
Primary estimand: simplification-strategy effect on irritation-related quality of life
Candidate primary outcome: validated irritation-related quality-of-life score
Primary finalized: No
Mathematical model: type: contrast precision/power by design-matrix or simulation planning; formulas: E(Y_it | A_i, X_i, Y_i0) = alpha_t + z(A_i)^T theta_t + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Delta = sum_t w_t c^T theta_t; estimate by baseline-adjusted marginal standardization, with unit/cluster-appropriate inference.; variables: Y: validated irritation-related quality-of-life score; i: participant; t: irritation-related quality-of-life assessment; the registered visit/time-window index, not another independent unit; A: participant randomized to structured simplification or comparator care; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: simplification-strategy effect on irritation-related quality of life; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; z(A): prespecified indicators for the actual randomized arms; two arms reduce to one indicator; family code: RCT
Sample size design: family: contrast precision/power by design-matrix or simulation planning; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Product count and transepidermal water loss are secondary/process measures; rescue treatment remains in the strategy estimand.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: If fewer products reduce spending but leave symptoms and barrier function unchanged, the hypothesized dermatologic benefit is unsupported.; preanalysis gate: Product count and transepidermal water loss are secondary/process measures; rescue treatment remains in the strategy estimand.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Product count and transepidermal water loss are secondary/process measures; rescue treatment remains in the strategy estimand.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Product count and transepidermal water loss are secondary/process measures; rescue treatment remains in the strategy estimand.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/78; question: Among people with persistent irritation and complex cosmetic routines, does structured simplification improve skin function and well-being?; aims: Characterize ingredient overlap, usage patterns, and baseline barrier function.
Randomize routine-simplification support against general skin-health education.
Assess durable symptom relief, spending, and gradual reintroduction decisions.; design verbatim planning only: Planning assumption: 240 adults with mild chronic irritation receive twelve weeks of support and six-month follow-up. Both groups continue clinically necessary treatment; dermatologists screen for eczema, infection, and other conditions requiring specific care.; outcomes verbatim: Primary: validated irritation-related quality-of-life change. Secondary: transepidermal water loss, redness assessed across pigmentation, product count, patient spending, and recurrence.; analysis verbatim inherited not authoritative if refined: Use intention-to-treat models with assessor masking for instrumental outcomes. Account for season, baseline disease, concurrent prescriptions, and sun exposure; prespecify irritation as primary and correct mechanistic comparisons.; falsification verbatim: If fewer products reduce spending but leave symptoms and barrier function unchanged, the hypothesized dermatologic benefit is unsupported.; source ids unverified leads: skin-photoaging-systematic-review-2025
tretinoin-photoaging-meta-2025; prospectus precedence: No

E34 — The hair-growth photograph as a measurement instrument

Rank: 174
Cluster: Skin, hair, vision, hearing, and sensory function
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: FALSEPOS; allocation or exposure unit: image acquisition condition randomized within participant/session; independent sampling unit: participant and independent reference image set; repeated measurement unit: image/rater classification; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: diagnostic/benchmark agreement or classification validity; reader/interface/navigation effects are causal only in their separately randomized component; specific constraint: Separate lighting/styling effects from demographic associations; reference repeatability determines what counts as no change.
Primary estimand: false-positive improvement probability relative to unchanged reference phototrichogram under acquisition changes
Candidate primary outcome: indicator of apparent improvement when reference change is absent under the fixed tolerance
Primary finalized: No
Mathematical model: type: diagnostic/benchmark sensitivity, false-alert and calibration precision with independent cases/noncases and cluster/reference uncertainty; formulas: Sensitivity = TP/(TP+FN); false-alert burden = FP/(registered nonevent time/opportunities); report participant/family-clustered uncertainty and failed/indeterminate cases.
For probability scores, report calibration and a fixed proper score in participant/source-disjoint evaluation, with uncertain reference labels modeled or bounded.
FPR_k = Pr(apparent improvement=1 | reference change within fixed no-change tolerance, acquisition condition k); compare conditions under their randomization.; variables: Y: indicator of apparent improvement when reference change is absent under the fixed tolerance; i: participant and independent reference image set; t: image/rater classification; the registered visit/time-window index, not another independent unit; A: image acquisition condition randomized within participant/session; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: false-positive improvement probability relative to unchanged reference phototrichogram under acquisition changes; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; TP,FN,FP: true-positive, false-negative and false-positive counts relative to the independently defined reference and matching window, not an assumed perfect gold standard; reference: per-thesis independent reference, including adjudication uncertainty; nonevent time/opportunities: fixed at-risk denominator; event specificity and false alerts per hour are distinct targets; family code: FALSEPOS
Sample size design: family: diagnostic/benchmark sensitivity, false-alert and calibration precision with independent cases/noncases and cluster/reference uncertainty; formula: An independent-case rough starting point is n_case ~= z_(1-alpha*/2)^2 Se(1-Se)/h_Se^2; replace with interval-based/clustered simulation for the actual design. Plan noncases/observation time and reference failures separately.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; Se: plausible sensitivity from prior evidence or pilot, not an asserted study result; h Se: desired sensitivity interval half-width; n case: independent positive cases; repeated events within a person require correlation adjustment; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Separate lighting/styling effects from demographic associations; reference repeatability determines what counts as no change.
Reference independence, fixed thresholds/window and an explicit denominator; incorporation/verification bias can invalidate apparent accuracy.
Participant/family-level test splits and paired comparisons; report abstention, assay failure and indeterminate truth rather than force every case into a binary label.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: If standardization only improves average accuracy while retaining large subgroup errors, the proposed broadly usable instrument fails.; preanalysis gate: Separate lighting/styling effects from demographic associations; reference repeatability determines what counts as no change.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Separate lighting/styling effects from demographic associations; reference repeatability determines what counts as no change.
Vary plausible reference misclassification, verification-selection and event-matching window under a prespecified sensitivity family.
Report subgroup usable-data coverage and calibration; avoid interpreting an imprecise null difference as equivalence.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Separate lighting/styling effects from demographic associations; reference repeatability determines what counts as no change.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/173; question: How much apparent hair improvement can be generated by ordinary changes in lighting, styling, camera geometry, and scalp pigmentation?; aims: Quantify image artifacts under controlled capture variations.
Develop a practical standardized photography and quality-control workflow.
Validate change detection against independent scalp-density measurements.; design verbatim planning only: Assume 200 adults with varied hair characteristics complete repeated imaging while hair status is stable; 120 patients receiving existing dermatologist-directed care are then followed for twelve months. No treatment is assigned or altered.; outcomes verbatim: Primary: false-positive improvement classification relative to reference phototrichograms. Secondary: change agreement, interrater reliability, subgroup error, usability, and minimum detectable density change.; analysis verbatim inherited not authoritative if refined: Split participants between development and validation; model repeated regions, reference uncertainty, and baseline density. Prespecify hair-characteristic subgroups and error thresholds, with blinded images and multiplicity control.; falsification verbatim: If standardization only improves average accuracy while retaining large subgroup errors, the proposed broadly usable instrument fails.; source ids unverified leads: hair-minoxidil-5ari-nma-2025
hair-prp-vs-minoxidil-meta-2026
hair-lllt-minoxidil-meta-2025; prospectus precedence: No

E35 — Hair-loss treatment support beyond daily reminders

Rank: 138
Cluster: Skin, hair, vision, hearing, and sensory function
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT; allocation or exposure unit: participant randomized to expectation/adverse-effect support versus usual support; independent sampling unit: participant; repeated measurement unit: goal and burden assessment visit; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Goal and burden rule must be registered before treatment; persistence alone is not success; prescribing remains independent.
Primary estimand: support-offer effect on prespecified goal attainment with acceptable burden
Candidate primary outcome: patient-defined goal attainment under a fixed acceptable-burden rule
Primary finalized: No
Mathematical model: type: contrast precision/power by design-matrix or simulation planning; formulas: E(Y_it | A_i, X_i, Y_i0) = alpha_t + z(A_i)^T theta_t + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Delta = sum_t w_t c^T theta_t; estimate by baseline-adjusted marginal standardization, with unit/cluster-appropriate inference.; variables: Y: patient-defined goal attainment under a fixed acceptable-burden rule; i: participant; t: goal and burden assessment visit; the registered visit/time-window index, not another independent unit; A: participant randomized to expectation/adverse-effect support versus usual support; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: support-offer effect on prespecified goal attainment with acceptable burden; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; z(A): prespecified indicators for the actual randomized arms; two arms reduce to one indicator; family code: RCT
Sample size design: family: contrast precision/power by design-matrix or simulation planning; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Goal and burden rule must be registered before treatment; persistence alone is not success; prescribing remains independent.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: If tailored support increases persistence without improving patient-defined goals or increases adverse-effect burden relative to the prespecified comparator, the proposed support benefit fails; no advantage over reminders would reject its added value.; preanalysis gate: Goal and burden rule must be registered before treatment; persistence alone is not success; prescribing remains independent.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Goal and burden rule must be registered before treatment; persistence alone is not success; prescribing remains independent.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Goal and burden rule must be registered before treatment; persistence alone is not success; prescribing remains independent.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/137; question: Does adherence support that addresses expectations and adverse-effect concerns improve outcomes from established alopecia care?; aims: Identify modifiable reasons for early discontinuation.
Randomize tailored support, generic reminders, and usual follow-up.
Link adherence trajectories to independently measured hair and well-being outcomes.; design verbatim planning only: Planning assumption: 270 adults with dermatologist-confirmed androgenetic alopecia starting clinically indicated established care are followed for eighteen months. Clinicians select and adjust treatment independently; study support provides education and navigation without dosing recommendations.; outcomes verbatim: Primary: patient-defined goal attainment with acceptable treatment burden. Secondary: standardized hair density, persistence, adverse effects, spending, distress, and appropriate clinician-contact rates.; analysis verbatim inherited not authoritative if refined: Use intention-to-treat comparisons with a prespecified primary support contrast. Stratify by treatment class and baseline severity, use independent adherence measures where feasible, and distinguish medically appropriate discontinuation from nonadherence.; falsification verbatim: If tailored support increases persistence without improving patient-defined goals or increases adverse-effect burden relative to the prespecified comparator, the proposed support benefit fails; no advantage over reminders would reject its added value.; source ids unverified leads: hair-minoxidil-5ari-nma-2025
hair-lllt-minoxidil-meta-2025
belief-effects-health-review-2026; prospectus precedence: No

E36 — When normal postpartum hair recovery looks like a supplement effect

Rank: 228
Cluster: Skin, hair, vision, hearing, and sensory function
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: OBS_SURV; allocation or exposure unit: supplement use and start time observed during postpartum recovery; independent sampling unit: participant; repeated measurement unit: reference density assessment, yielding interval-censored recovery time; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: observational comparison/association; causal interpretation is conditional on explicit exchangeability, positivity, consistency and observation assumptions; specific constraint: Natural recovery and initiation timing create confounding and immortal-time risks; do not claim supplement efficacy from before-after recovery.
Primary estimand: supplement-associated time to sustained reference-density recovery conditional on baseline postpartum timing and prognosis
Candidate primary outcome: time to sustained density recovery, observed between scheduled assessments
Primary finalized: No
Mathematical model: type: interval-censored observational time-to-recovery precision/power with baseline timing, confounding, censoring and competing-event sensitivity; formulas: W_i(t) = product_(k<=t) P(A_ik | A_i,<k,X_i0)/P(A_ik | H_ik), optionally multiplied by separately justified observation weights.
m_a = E_X[m(a,X)] for a baseline strategy, or the sequential g-formula for sustained strategies; Delta=m_1-m_0. Fit a weighted, baseline-aligned mean/risk model matching Y rather than assume the treatment coefficient is causal.
For interval-censored recovery T_i in (L_i,R_i], likelihood contribution is S(L_i | history)-S(R_i | history); right-censored contribution is S(L_i). Treatment initiation may be time-varying and must not create immortal time.; variables: Y: time to sustained density recovery, observed between scheduled assessments; i: participant; t: reference density assessment, yielding interval-censored recovery time; the registered visit/time-window index, not another independent unit; A: supplement use and start time observed during postpartum recovery; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: supplement-associated time to sustained reference-density recovery conditional on baseline postpartum timing and prognosis; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; H ik: measured predecision history including prior treatment and time-varying covariates; m(a,X): outcome mean/risk model under specified treatment strategy and observed covariates; m a: standardized mean/risk under a, identified only under the listed assumptions; k: ordered treatment decision time; L i,R i: last not-recovered and first sustained-recovery assessments; S(t | history): conditional probability recovery has not occurred by t; define postpartum baseline/time origin and competing events; family code: OBS_SURV
Sample size design: family: interval-censored observational time-to-recovery precision/power with baseline timing, confounding, censoring and competing-event sensitivity; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Natural recovery and initiation timing create confounding and immortal-time risks; do not claim supplement efficacy from before-after recovery.
Common eligibility/time zero and a defensible active comparator; treatment choice is not randomized.
Conditional exchangeability (including time-varying feedback where relevant), positivity and correct observation/treatment/outcome models are untestable or only partly testable; observed overlap does not prove them.
Do not adjust for posttreatment variables as baseline confounders; define a separate mediation/per-protocol estimand before using g-methods.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: If observed supplement associations remain robust across stage alignment and independent density outcomes, the natural-recovery explanation is weakened.; preanalysis gate: Natural recovery and initiation timing create confounding and immortal-time risks; do not claim supplement efficacy from before-after recovery.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Natural recovery and initiation timing create confounding and immortal-time risks; do not claim supplement efficacy from before-after recovery.
Overlap/effective-sample-size and weight-tail diagnostics with declared truncation sensitivity; report population changes caused by trimming.
Quantitative unmeasured-confounding and informative-loss sensitivity; negative controls/falsification windows where scientifically defensible.
Compare g-computation/weighting/doubly robust estimates on the same target population and time zero, without interpreting agreement as proof.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Natural recovery and initiation timing create confounding and immortal-time risks; do not claim supplement efficacy from before-after recovery.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/227; question: How much of perceived postpartum hair-supplement benefit is explained by natural recovery, baseline timing, and selective reporting?; aims: Establish prospective natural-history trajectories after childbirth.
Record independently chosen supplement initiation and expectations before follow-up.
Compare stage-aligned recovery, cost, and distress without assigning products.; design verbatim planning only: Assume 300 postpartum participants recruited through maternity services contribute standardized scalp images and symptom reports for eighteen months. Dermatologists assess shedding and alternative diagnoses; clinically indicated testing and treatment remain independent.; outcomes verbatim: Primary: time to sustained reference-measured density recovery. Secondary: perceived shedding, distress, supplement expenditure, adverse effects, and referral for persistent or atypical loss.; analysis verbatim inherited not authoritative if refined: Align time since childbirth and onset of shedding; adjust breastfeeding, illness, nutritional diagnoses, thyroid disease, and baseline severity. Use new-user comparisons with explicit residual-confounding sensitivity and account for selective photo submission.; falsification verbatim: If observed supplement associations remain robust across stage alignment and independent density outcomes, the natural-recovery explanation is weakened.; source ids unverified leads: hair-minoxidil-5ari-nma-2025
nih-ods-b6-2023
belief-effects-health-review-2026; prospectus precedence: No

E37 — Digital eyestrain: blinking support, workstation change, or both?

Rank: 70
Cluster: Skin, hair, vision, hearing, and sensory function
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: FACTOR; allocation or exposure unit: participant randomized to blinking support and workstation change factorially; independent sampling unit: participant; repeated measurement unit: ocular-discomfort assessment; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Interactions need a scale and dedicated precision planning; neither component is isolated by comparing significance across arms.
Primary estimand: factorial blinking/workstation main effects and their interaction on discomfort
Candidate primary outcome: validated ocular-discomfort burden
Primary finalized: No
Mathematical model: type: factorial main-effect/interaction planning with actual cross-classified or cluster-period covariance; formulas: E(Y_ict | B_ict, C_ict, X_i) = alpha_t + beta_B B_ict + beta_C C_ict + beta_BC B_ict C_ict + gamma^T X_i + u_i + b_c.
I = mu_11 - mu_10 - mu_01 + mu_00 = beta_BC on the additive scale; main effects are standardized over the other randomized factor.; variables: Y: validated ocular-discomfort burden; i: participant; t: ocular-discomfort assessment; the registered visit/time-window index, not another independent unit; A: participant randomized to blinking support and workstation change factorially; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: factorial blinking/workstation main effects and their interaction on discomfort; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; B,C: the two assigned factors named in the per-thesis allocation specification; levels and contrasts must be fixed; mu bc: marginal mean outcome under the specified joint factor levels; I: additive interaction estimand; beta B,beta C,beta BC: factor coefficients; for nonidentity links derive marginal contrasts rather than interpreting coefficients as risk differences; family code: FACTOR
Sample size design: family: factorial main-effect/interaction planning with actual cross-classified or cluster-period covariance; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Interactions need a scale and dedicated precision planning; neither component is isolated by comparing significance across arms.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Factor allocation and coding must match the design; interaction inference is separate from main-effect inference and needs its own precision.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: If symptoms improve only with unblinded expectations and objective or functional measures worsen, the proposed mechanism and practical benefit fail.; preanalysis gate: Interactions need a scale and dedicated precision planning; neither component is isolated by comparing significance across arms.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Interactions need a scale and dedicated precision planning; neither component is isolated by comparing significance across arms.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Interactions need a scale and dedicated precision planning; neither component is isolated by comparing significance across arms.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/69; question: Which component of common screen-use advice improves ocular symptoms and work function when tested independently?; aims: Phenotype baseline ocular and task-related contributors.
Separate blinking support and workstation adjustment in a factorial trial.
Test whether benefits persist after prompts are withdrawn.; design verbatim planning only: Planning assumption: 240 symptomatic screen workers undergo optometric assessment, then receive twelve weeks of blinking support, workstation adjustment, both, or attention-matched general education. Existing clinical treatment continues; lighting schedules and circadian timing are not manipulated.; outcomes verbatim: Primary: validated ocular-discomfort burden. Secondary: clinician-assessed tear-film stability, blink behavior, reading performance, neck discomfort, task interruption, and symptom recurrence.; analysis verbatim inherited not authoritative if refined: Use factorial intention-to-treat estimates with prespecified interaction testing and blinded ocular assessment. Adjust baseline dry-eye status, corrected vision, contact-lens use, and workload; model objective sensor error and control secondary comparisons.; falsification verbatim: If symptoms improve only with unblinded expectations and objective or functional measures worsen, the proposed mechanism and practical benefit fail.; source ids unverified leads: multisensory-learning-neuroplasticity-2025
wearables-living-review; prospectus precedence: No

E38 — Personal-audio exposure feedback that reflects the ear

Rank: 120
Cluster: Skin, hair, vision, hearing, and sensory function
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT; allocation or exposure unit: participant randomized to calibrated dose alerts or generic listening messages; independent sampling unit: participant; repeated measurement unit: listening-day reference acoustic dose; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Define dose unit and capped/reference range; missing playback streams do not equal silence; symptoms are separate secondary endpoints.
Primary estimand: alert-offer effect on cumulative reference acoustic dose within the validated method range
Candidate primary outcome: independently estimated cumulative acoustic dose
Primary finalized: No
Mathematical model: type: contrast precision/power by design-matrix or simulation planning; formulas: E(Y_it | A_i, X_i, Y_i0) = alpha_t + z(A_i)^T theta_t + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Delta = sum_t w_t c^T theta_t; estimate by baseline-adjusted marginal standardization, with unit/cluster-appropriate inference.; variables: Y: independently estimated cumulative acoustic dose; i: participant; t: listening-day reference acoustic dose; the registered visit/time-window index, not another independent unit; A: participant randomized to calibrated dose alerts or generic listening messages; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: alert-offer effect on cumulative reference acoustic dose within the validated method range; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; z(A): prespecified indicators for the actual randomized arms; two arms reduce to one indicator; family code: RCT
Sample size design: family: contrast precision/power by design-matrix or simulation planning; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Define dose unit and capped/reference range; missing playback streams do not equal silence; symptoms are separate secondary endpoints.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: If dashboard dose falls without corresponding reference-measured reduction, the proposed protective feedback effect is falsified.; preanalysis gate: Define dose unit and capped/reference range; missing playback streams do not equal silence; symptoms are separate secondary endpoints.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Define dose unit and capped/reference range; missing playback streams do not equal silence; symptoms are separate secondary endpoints.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Define dose unit and capped/reference range; missing playback streams do not equal silence; symptoms are separate secondary endpoints.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/119; question: Do calibrated listening-dose alerts reduce actual acoustic exposure more than generic safe-listening messages?; aims: Quantify error across device and headphone combinations using acoustic reference methods.
Validate feasible calibration and uncertainty flags.
Randomize calibrated feedback versus generic messaging during ordinary listening.; design verbatim planning only: Assume 240 frequent adult headphone users undergo baseline audiology and device calibration before a six-month randomized field study. Listening is self-selected; researchers never assign loud exposure or ask participants to increase volume.; outcomes verbatim: Primary: independently estimated cumulative acoustic dose within the reference method range. Secondary: alert adherence, listening satisfaction, tinnitus symptoms, speech-in-noise performance, and differential device access.; analysis verbatim inherited not authoritative if refined: Use intention-to-treat analysis, propagate calibration error, and model background noise and selective headphone switching. Prespecify exposure as primary; audiometric changes are exploratory because duration and sample size cannot establish hearing-loss prevention.; falsification verbatim: If dashboard dose falls without corresponding reference-measured reduction, the proposed protective feedback effect is falsified.; source ids unverified leads: wearables-living-review
multisensory-learning-neuroplasticity-2025; prospectus precedence: No

E39 — Sensory substitution that transfers beyond the training room

Rank: 96
Cluster: Skin, hair, vision, hearing, and sensory function
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT; allocation or exposure unit: participant randomized to context-rich training versus equal device drills; independent sampling unit: participant; repeated measurement unit: unfamiliar mobility route/task; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Repeated routes are not independent participants; screen route difficulty and keep test routes outside training.
Primary estimand: training assignment effect on safe independent completion of unfamiliar mobility tasks
Candidate primary outcome: safe unfamiliar-task completion under fixed independence and near-miss rules
Primary finalized: No
Mathematical model: type: contrast precision/power by design-matrix or simulation planning; formulas: E(Y_it | A_i, X_i, Y_i0) = alpha_t + z(A_i)^T theta_t + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Delta = sum_t w_t c^T theta_t; estimate by baseline-adjusted marginal standardization, with unit/cluster-appropriate inference.; variables: Y: safe unfamiliar-task completion under fixed independence and near-miss rules; i: participant; t: unfamiliar mobility route/task; the registered visit/time-window index, not another independent unit; A: participant randomized to context-rich training versus equal device drills; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: training assignment effect on safe independent completion of unfamiliar mobility tasks; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; z(A): prespecified indicators for the actual randomized arms; two arms reduce to one indicator; family code: RCT
Sample size design: family: contrast precision/power by design-matrix or simulation planning; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Repeated routes are not independent participants; screen route difficulty and keep test routes outside training.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: If participants improve practiced tasks but gain no unfamiliar-route safety or daily participation, the proposed transferable augmentation fails.; preanalysis gate: Repeated routes are not independent participants; screen route difficulty and keep test routes outside training.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Repeated routes are not independent participants; screen route difficulty and keep test routes outside training.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Repeated routes are not independent participants; screen route difficulty and keep test routes outside training.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/95; question: Can context-rich training with an affordable tactile or auditory navigation aid improve unfamiliar-route mobility more than equally intensive device drills in people with visual impairment?; aims: Co-design meaningful mobility goals and accessible device interfaces.
Randomize context-rich training versus equally intensive device drills.
Assess unfamiliar-route transfer, daily participation, and sustained voluntary use.; design verbatim planning only: Planning assumption: 120 adults with visual impairment receive a noninvasive aid alongside their usual mobility supports, with twelve weeks of supervised training and six-month follow-up. Independent mobility specialists evaluate unfamiliar controlled and community routes.; outcomes verbatim: Primary: safe completion of unfamiliar mobility tasks. Secondary: collisions or near misses, cognitive workload, confidence, community participation, hearing interference, and abandonment.; analysis verbatim inherited not authoritative if refined: Use blinded outcome adjudication where feasible, stratify residual vision and hearing, and model trainer clustering. Prespecify the transfer endpoint; report task learning separately and examine missing follow-up from device abandonment.; falsification verbatim: If participants improve practiced tasks but gain no unfamiliar-route safety or daily participation, the proposed transferable augmentation fails.; source ids unverified leads: multisensory-learning-neuroplasticity-2025
ibci-outcome-measures-2025; prospectus precedence: No

E40 — Tinnitus feedback: useful coping signal or symptom amplifier?

Rank: 71
Cluster: Skin, hair, vision, hearing, and sensory function
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT; allocation or exposure unit: participant randomized tinnitus-tracking frequency or comparator format; independent sampling unit: participant; repeated measurement unit: functional-impairment visit and symptom entries; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Frequent symptom reports may alter salience and missingness; equal independent scheduled outcomes are required across schedules.
Primary estimand: continuous-tracking offer effect on tinnitus-related functional impairment
Candidate primary outcome: validated tinnitus-related functional-impairment score
Primary finalized: No
Mathematical model: type: contrast precision/power by design-matrix or simulation planning; formulas: E(Y_it | A_i, X_i, Y_i0) = alpha_t + z(A_i)^T theta_t + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Delta = sum_t w_t c^T theta_t; estimate by baseline-adjusted marginal standardization, with unit/cluster-appropriate inference.; variables: Y: validated tinnitus-related functional-impairment score; i: participant; t: functional-impairment visit and symptom entries; the registered visit/time-window index, not another independent unit; A: participant randomized tinnitus-tracking frequency or comparator format; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: continuous-tracking offer effect on tinnitus-related functional impairment; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; z(A): prespecified indicators for the actual randomized arms; two arms reduce to one indicator; family code: RCT
Sample size design: family: contrast precision/power by design-matrix or simulation planning; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Frequent symptom reports may alter salience and missingness; equal independent scheduled outcomes are required across schedules.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: If lower-burden tracking only reduces recorded symptoms without improving independent function or distress, the proposed coping advantage fails.; preanalysis gate: Frequent symptom reports may alter salience and missingness; equal independent scheduled outcomes are required across schedules.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Frequent symptom reports may alter salience and missingness; equal independent scheduled outcomes are required across schedules.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Frequent symptom reports may alter salience and missingness; equal independent scheduled outcomes are required across schedules.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/70; question: Does continuous tinnitus tracking improve coping, or can frequent symptom attention increase distress without improving hearing-related function?; aims: Map relationships between logging intensity, salience, and distress.
Randomize tracking frequency and goal-oriented feedback above standard education.
Assess durable functioning after logging is withdrawn.; design verbatim planning only: Assume 240 adults with persistent clinically assessed tinnitus receive twelve weeks of frequent symptom logging, low-burden goal-oriented tracking, or minimal tracking, followed by six months of observation. All groups retain usual audiologic care and evidence-based support access.; outcomes verbatim: Primary: validated tinnitus-related functional impairment. Secondary: distress, symptom salience, hearing-related participation, sleep disruption, app burden, and care use.; analysis verbatim inherited not authoritative if refined: Use intention-to-treat comparisons, baseline hearing and anxiety stratification, and prespecified primary contrasts. Separate sampling reactivity from missingness using brief independent assessments; control the secondary outcome family.; falsification verbatim: If lower-burden tracking only reduces recorded symptoms without improving independent function or distress, the proposed coping advantage fails.; source ids unverified leads: belief-effects-health-review-2026
nocebo-communication-review-2026
wearable-stress-measurement-review-2026; prospectus precedence: No

F01 — Teaching endpoint transportability in biohacking claims

Rank: 134
Cluster: Community evidence, incentives, ethics and policy
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT; allocation or exposure unit: participant randomized scaffold/source-links/general-literacy text; independent sampling unit: participant with claim/topic effects; repeated measurement unit: claim response and follow-up topic; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Held-out topics test transfer; source links alone do not validate the claim summaries used in the experiment.
Primary estimand: scaffold versus length-matched literacy assignment effect on unsupported-extrapolation detection score
Candidate primary outcome: blinded unsupported-extrapolation detection score
Primary finalized: No
Mathematical model: type: contrast precision/power by design-matrix or simulation planning; formulas: E(Y_it | A_i, X_i, Y_i0) = alpha_t + z(A_i)^T theta_t + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Delta = sum_t w_t c^T theta_t; estimate by baseline-adjusted marginal standardization, with unit/cluster-appropriate inference.; variables: Y: blinded unsupported-extrapolation detection score; i: participant with claim/topic effects; t: claim response and follow-up topic; the registered visit/time-window index, not another independent unit; A: participant randomized scaffold/source-links/general-literacy text; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: scaffold versus length-matched literacy assignment effect on unsupported-extrapolation detection score; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; z(A): prespecified indicators for the actual randomized arms; two arms reduce to one indicator; family code: RCT
Sample size design: family: contrast precision/power by design-matrix or simulation planning; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Held-out topics test transfer; source links alone do not validate the claim summaries used in the experiment.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Reject the mechanism if source links perform equally well or improvement disappears on unfamiliar topics.; preanalysis gate: Held-out topics test transfer; source links alone do not validate the claim summaries used in the experiment.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Held-out topics test transfer; source links alone do not validate the claim summaries used in the experiment.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Held-out topics test transfer; source links alone do not validate the claim summaries used in the experiment.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/133; question: Does teaching population and endpoint matching improve durable interpretation of enhancement claims?; aims: Construct an adjudicated claim set with explicit population and endpoint boundaries.
Randomize competing explanations and measure immediate inference accuracy.
Test retention and transfer to unfamiliar intervention categories.; design verbatim planning only: Recruit approximately 600 adult volunteers for a preregistered online experiment comparing a structured scaffold, source links, and length-matched general literacy text. Use invented claim presentations grounded in inherited source summaries; assign different topics at follow-up. Recruitment numbers are planning assumptions.; outcomes verbatim: Primary: blinded scoring of unsupported extrapolation detection; secondary: confidence calibration and information-seeking choices.; analysis verbatim inherited not authoritative if refined: Stratify randomization by baseline literacy; use intention-to-treat models, attrition weighting, and a prespecified secondary-outcome hierarchy.; falsification verbatim: Reject the mechanism if source links perform equally well or improvement disappears on unfamiliar topics.; source ids unverified leads: suvorexant-csf-biomarkers-rct-2023
mc-detection-systematic-review-2025
belief-effects-health-review-2026; prospectus precedence: No

F02 — The missing denominator in community success stories

Rank: 202
Cluster: Community evidence, incentives, ethics and policy
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: SELECT; allocation or exposure unit: participant randomized reporting invitation; intervention choice remains self-selected; independent sampling unit: participant; repeated measurement unit: confidential outcome and consented shareable summary; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: reporting selection and invitation effects; self-selected behavior efficacy remains unidentified; specific constraint: Invitation randomization identifies an invitation effect, not intervention efficacy or the outcomes of nonresponders without assumptions.
Primary estimand: difference in improvement prevalence/mean between shareable reports and confidential cohort reports within intervention category
Candidate primary outcome: prespecified reported improvement measure
Primary finalized: No
Mathematical model: type: selection-difference and bounded-missingness precision with intervention-category stratification; formulas: Selection difference = E[Y | shareable report available]-E[Y | confidential cohort frame], within fixed intervention category.
For bounded Y in [L,U], unknown missing mass m gives mean bounds (1-m)mu_observed+m L through (1-m)mu_observed+m U; randomized invitation effects are estimated separately by assignment.; variables: Y: prespecified reported improvement measure; i: participant; t: confidential outcome and consented shareable summary; the registered visit/time-window index, not another independent unit; A: participant randomized reporting invitation; intervention choice remains self-selected; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: difference in improvement prevalence/mean between shareable reports and confidential cohort reports within intervention category; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; shareable report available: consent-compatible reporting/selection indicator, not random treatment uptake; L,U: validated lower/upper outcome bounds; must be set from the instrument, not invented; m: unobserved proportion in the defined cohort frame; mu observed: observed outcome mean under the stated reporting process; family code: SELECT
Sample size design: family: selection-difference and bounded-missingness precision with intervention-category stratification; formula: Use design-specific simulation or independent-unit resampling for the stated estimand and target confidence width/power; preserve all allocation, censoring, clustering and measurement-error structure.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Invitation randomization identifies an invitation effect, not intervention efficacy or the outcomes of nonresponders without assumptions.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Confidential responses may also be selected; weighting identifies cohort means only under explicit response assumptions.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: The hypothesis fails if shareability is unrelated to improvement and prompts do not change reporting balance.; preanalysis gate: Invitation randomization identifies an invitation effect, not intervention efficacy or the outcomes of nonresponders without assumptions.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Invitation randomization identifies an invitation effect, not intervention efficacy or the outcomes of nonresponders without assumptions.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Invitation randomization identifies an invitation effect, not intervention efficacy or the outcomes of nonresponders without assumptions.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/201; question: How much does selective reporting distort perceived benefit in voluntary biohacking cohorts?; aims: Enroll participants before an independently chosen behavior change begins.
Compare confidential outcomes with participants' voluntary public-sharing decisions.
Randomize reporting prompts to test whether neutral check-ins reduce selection.; design verbatim planning only: Follow a planning cohort of 400 adults undertaking low-risk, self-selected sleep or activity changes for six months. Researchers neither prescribe interventions nor require public posts. Randomize neutral reporting invitations versus usual follow-up; compare all confidential responses with consented shareable summaries.; outcomes verbatim: Primary: selection difference in reported improvement; secondary: discontinuation, adverse experiences, and denominator comprehension.; analysis verbatim inherited not authoritative if refined: Model missingness jointly with outcomes; use inverse-probability weights and sensitivity bounds for unobserved dissatisfied participants. Separate intervention categories.; falsification verbatim: The hypothesis fails if shareability is unrelated to improvement and prompts do not change reporting balance.; source ids unverified leads: rwd-digital-platform-cohort-2026
n-of-1-methods-review-2025
digital-n-of-1-physiology-2026; prospectus precedence: No

F03 — Commercial disclosure versus evidential disclosure

Rank: 135
Cluster: Community evidence, incentives, ethics and policy
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: ITEM_FACTOR; allocation or exposure unit: participant and claim-card financial/methodological disclosure assignment; independent sampling unit: participant and independently sampled claim; repeated measurement unit: repeated fictional claim-card judgment; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Evidence strength must vary independently of disclosure; correct crossed person/claim inference and avoid treating repeated cards as people.
Primary estimand: financial- and methodological-disclosure effects and their interaction on strong-versus-weak evidence discrimination
Candidate primary outcome: prespecified within-person evidence-discrimination score
Primary finalized: No
Mathematical model: type: factorial main-effect/interaction planning with actual cross-classified or cluster-period covariance; formulas: E(Y_ict | B_ict, C_ict, X_i) = alpha_t + beta_B B_ict + beta_C C_ict + beta_BC B_ict C_ict + gamma^T X_i + u_i + b_c. Include an independently sampled claim-item effect as well as the participant effect.
I = mu_11 - mu_10 - mu_01 + mu_00 = beta_BC on the additive scale; main effects are standardized over the other randomized factor.; variables: Y: prespecified within-person evidence-discrimination score; i: participant and independently sampled claim; t: repeated fictional claim-card judgment; the registered visit/time-window index, not another independent unit; A: participant and claim-card financial/methodological disclosure assignment; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: financial- and methodological-disclosure effects and their interaction on strong-versus-weak evidence discrimination; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; B,C: the two assigned factors named in the per-thesis allocation specification; levels and contrasts must be fixed; mu bc: marginal mean outcome under the specified joint factor levels; I: additive interaction estimand; beta B,beta C,beta BC: factor coefficients; for nonidentity links derive marginal contrasts rather than interpreting coefficients as risk differences; family code: ITEM_FACTOR
Sample size design: family: factorial main-effect/interaction planning with actual cross-classified or cluster-period covariance; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Evidence strength must vary independently of disclosure; correct crossed person/claim inference and avoid treating repeated cards as people.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Factor allocation and coding must match the design; interaction inference is separate from main-effect inference and needs its own precision.
Financial disclosure and methodological disclosure are distinct randomized factors; strong/weak underlying evidence defines the discrimination contrast.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Reject synergy if combined disclosure adds no discrimination or merely lowers trust equally across evidence levels.; preanalysis gate: Evidence strength must vary independently of disclosure; correct crossed person/claim inference and avoid treating repeated cards as people.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Evidence strength must vary independently of disclosure; correct crossed person/claim inference and avoid treating repeated cards as people.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Evidence strength must vary independently of disclosure; correct crossed person/claim inference and avoid treating repeated cards as people.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/134; question: Which disclosure helps readers evaluate commercial health claims without dismissing valid evidence?; aims: Develop matched claim cards crossing sponsorship and study quality.
Estimate separate and interactive disclosure effects on judgments.
Test delayed discrimination between strong and weak evidence.; design verbatim planning only: Use a preregistered factorial experiment with approximately 800 adults and repeated, randomized fictional claim cards. Compare financial disclosure, methodological disclosure, both, and neither across substantively identical claims. A separate 200-person cohort assesses replication; counts are planning assumptions.; outcomes verbatim: Primary: discrimination between stronger and weaker evidence; secondary: calibrated confidence, recall, and willingness to consult clinicians.; analysis verbatim inherited not authoritative if refined: Fit participant- and claim-level mixed models; adjust for financial literacy and topic familiarity, with multiplicity control across prespecified interactions.; falsification verbatim: Reject synergy if combined disclosure adds no discrimination or merely lowers trust equally across evidence levels.; source ids unverified leads: nicotine-cognition-review-2024
belief-effects-health-review-2026
fda-compounding-questions-2026; prospectus precedence: No

F04 — Recommendation objectives and risky health-information exposure

Rank: 208
Cluster: Community evidence, incentives, ethics and policy
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT_PROP; allocation or exposure unit: participant randomized recommendation-ranking policy; independent sampling unit: participant with content/source clustering; repeated measurement unit: session and selected claim; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Selection is downstream of policy; an unsupported-selection proportion requires a denominator and zero-selection handling; report amount selected too.
Primary estimand: engagement versus evidence-diverse ranking effect on the probability a selected claim is unsupported
Candidate primary outcome: unsupported selections divided by all selections under a fixed empty-session rule
Primary finalized: No
Mathematical model: type: proportion/binomial design with opportunity denominators, overdispersion and independent-person precision; formulas: K_it | n_it, p_it ~ Binomial(n_it,p_it) with beta-binomial or clustered dependence if needed; logit(p_it)=alpha_t+z(A_i)^T theta_t+gamma^T X_i+u_i.
Report the chosen participant-average or opportunity-average probability difference; define n_it=0 before analysis.; variables: Y: unsupported selections divided by all selections under a fixed empty-session rule; i: participant with content/source clustering; t: session and selected claim; the registered visit/time-window index, not another independent unit; A: participant randomized recommendation-ranking policy; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: engagement versus evidence-diverse ranking effect on the probability a selected claim is unsupported; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; K it: number of primary successes/events within the registered opportunity set; n it: number of eligible opportunities, defined independently of observed success; p it: success/event probability under that explicitly defined opportunity set; z(A): prespecified indicators for the actual arms or randomized policy; family code: RCT_PROP
Sample size design: family: proportion/binomial design with opportunity denominators, overdispersion and independent-person precision; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Selection is downstream of policy; an unsupported-selection proportion requires a denominator and zero-selection handling; report amount selected too.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
If the intervention changes opportunity counts, participant-average and pooled-opportunity targets differ; choose and report the intended target.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: The incentive hypothesis weakens if ordering changes engagement but not unsupported-claim selection or consensus judgments.; preanalysis gate: Selection is downstream of policy; an unsupported-selection proportion requires a denominator and zero-selection handling; report amount selected too.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Selection is downstream of policy; an unsupported-selection proportion requires a denominator and zero-selection handling; report amount selected too.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Selection is downstream of policy; an unsupported-selection proportion requires a denominator and zero-selection handling; report amount selected too.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/207; question: Do engagement-oriented recommendations increase unsupported escalation narratives compared with diversity-oriented recommendations?; aims: Create a consented, expert-rated library of nonoperational health-information summaries.
Randomize ranking objectives in a research feed.
Estimate whether social-proof exposure explains subsequent claim selection.; design verbatim planning only: Recruit a planning sample of 500 adults for four weekly sessions in an isolated research platform. Randomly assign engagement-based, chronological, or evidence-diverse ordering using the same content pool. Measure simulated reading and sharing choices without directing participants to products or external posts.; outcomes verbatim: Primary: proportion of unsupported claims selected; secondary: risk comprehension, reading diversity, and perceived consensus.; analysis verbatim inherited not authoritative if refined: Use intention-to-treat multilevel models, logged exposure, and preregistered mediation sensitivity analyses; correct secondary comparisons across ranking arms.; falsification verbatim: The incentive hypothesis weakens if ordering changes engagement but not unsupported-claim selection or consensus judgments.; source ids unverified leads: belief-effects-health-review-2026
ai-biotech-governance-2026
ai-biology-national-academies-2025; prospectus precedence: No

F05 — Registration friction and the credibility of personal experiments

Rank: 102
Cluster: Community evidence, incentives, ethics and policy
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT_BINARY; allocation or exposure unit: participant randomized registration format; independent sampling unit: participant experiment with reviewer clustering; repeated measurement unit: registration/reviewer and repeat experiment; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Missing registration is a failure category, not a deleted observation; detailed format is a corrected secondary contrast.
Primary estimand: concise-registration versus free-diary effect on estimand recoverability
Candidate primary outcome: indicator that blinded reviewers recover a prespecified estimand under the fixed rubric
Primary finalized: No
Mathematical model: type: binary risk-difference precision/power, simulated with actual cluster/trajectory structure; formulas: logit Pr(Y_i=1 | A_i,X_i) = alpha + z(A_i)^T theta + gamma^T X_i + b_cluster(i).
RD = E_X[p_hat(1,X) - p_hat(0,X)]; report absolute risk difference as well as any odds ratio.; variables: Y: indicator that blinded reviewers recover a prespecified estimand under the fixed rubric; i: participant experiment with reviewer clustering; t: registration/reviewer and repeat experiment; the registered visit/time-window index, not another independent unit; A: participant randomized registration format; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: concise-registration versus free-diary effect on estimand recoverability; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; logit(p): log[p/(1-p)]; p hat(a,X): fitted outcome probability under arm a standardized to the prespecified population; RD: marginal risk difference; z(A): prespecified indicators for the actual arms or randomized policy; family code: RCT_BINARY
Sample size design: family: binary risk-difference precision/power, simulated with actual cluster/trajectory structure; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Missing registration is a failure category, not a deleted observation; detailed format is a corrected secondary contrast.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Reject superiority if concise forms do not improve interpretation or if gains disappear after accounting for noncompletion.; preanalysis gate: Missing registration is a failure category, not a deleted observation; detailed format is a corrected secondary contrast.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Missing registration is a failure category, not a deleted observation; detailed format is a corrected secondary contrast.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Missing registration is a failure category, not a deleted observation; detailed format is a corrected secondary contrast.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/101; question: Can lightweight registration improve interpretability of benign personal experiments without excluding less technical participants?; aims: Identify common ambiguities in simulated personal-experiment plans.
Randomize registration interfaces during supervised benign research tasks.
Assess whether independent reviewers recover intended comparisons and outcomes.; design verbatim planning only: Enroll approximately 240 adults in researcher-supervised comparisons of alternative reminder schedules for a nonmedical learning task. Randomize a concise registration, detailed registration, or free-form diary. Independently blinded reviewers evaluate plans before accessing outcomes; repeat the task with a new comparison.; outcomes verbatim: Primary: recoverability of the prespecified estimand; secondary: completion, outcome switching, and reviewer agreement.; analysis verbatim inherited not authoritative if refined: Prespecify literacy and disability interactions; use intention-to-treat models and report missing registrations as outcomes rather than deleting them.; falsification verbatim: Reject superiority if concise forms do not improve interpretation or if gains disappear after accounting for noncompletion.; source ids unverified leads: n-of-1-methods-review-2025
digital-n-of-1-physiology-2026
belief-effects-health-review-2026; prospectus precedence: No

F06 — Regulatory vocabulary and mistaken product equivalence

Rank: 103
Cluster: Community evidence, incentives, ethics and policy
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT_PROP; allocation or exposure unit: participant randomized explanation format; independent sampling unit: participant and independently sampled scenario; repeated measurement unit: category-assignment item at baseline/follow-up; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Reference categories and reading-time match precede unblinding; category knowledge does not prove a product safe or effective.
Primary estimand: category-matrix versus prose effect on correct regulatory-category assignment
Candidate primary outcome: proportion of scenario categories correctly assigned
Primary finalized: No
Mathematical model: type: proportion/binomial design with opportunity denominators, overdispersion and independent-person precision; formulas: K_it | n_it, p_it ~ Binomial(n_it,p_it) with beta-binomial or clustered dependence if needed; logit(p_it)=alpha_t+z(A_i)^T theta_t+gamma^T X_i+u_i.
Report the chosen participant-average or opportunity-average probability difference; define n_it=0 before analysis.; variables: Y: proportion of scenario categories correctly assigned; i: participant and independently sampled scenario; t: category-assignment item at baseline/follow-up; the registered visit/time-window index, not another independent unit; A: participant randomized explanation format; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: category-matrix versus prose effect on correct regulatory-category assignment; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; K it: number of primary successes/events within the registered opportunity set; n it: number of eligible opportunities, defined independently of observed success; p it: success/event probability under that explicitly defined opportunity set; z(A): prespecified indicators for the actual arms or randomized policy; family code: RCT_PROP
Sample size design: family: proportion/binomial design with opportunity denominators, overdispersion and independent-person precision; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Reference categories and reading-time match precede unblinding; category knowledge does not prove a product safe or effective.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
If the intervention changes opportunity counts, participant-average and pooled-opportunity targets differ; choose and report the intended target.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: The proposed mechanism fails if the matrix improves terminology recall but leaves equivalence judgments unchanged.; preanalysis gate: Reference categories and reading-time match precede unblinding; category knowledge does not prove a product safe or effective.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Reference categories and reading-time match precede unblinding; category knowledge does not prove a product safe or effective.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Reference categories and reading-time match precede unblinding; category knowledge does not prove a product safe or effective.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/102; question: Which explanation best prevents readers from equating compounded, approved, off-label, and research-use products?; aims: Map regulatory confusions through consented interviews.
Randomize educational formats using hypothetical product descriptions.
Test whether learning transfers across drug and peptide examples.; design verbatim planning only: Interview 40 adults, then recruit a planning sample of 600 for a three-arm trial of a category matrix, standard prose, or brief warning. Use fictional products, matched reading time, and blinded adjudication of scenario responses at baseline and six weeks.; outcomes verbatim: Primary: correct category assignment; secondary: recognition of evidence limits, clinician-question quality, and confidence calibration.; analysis verbatim inherited not authoritative if refined: Adjust for baseline literacy and prior compounding experience; prespecify one primary contrast and control false discovery across transfer outcomes.; falsification verbatim: The proposed mechanism fails if the matrix improves terminology recall but leaves equivalence judgments unchanged.; source ids unverified leads: fda-compounding-questions-2026
fda-off-label-use-2026
fda-unapproved-drugs-2026; prospectus precedence: No

F07 — Proportionate oversight for community biology spaces

Rank: 220
Cluster: Community evidence, incentives, ethics and policy
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: REVIEW_CROSS; allocation or exposure unit: reviewer-scenario review condition randomized in counterbalanced order; independent sampling unit: reviewer and independently sampled scenario; partner space for shadowing; repeated measurement unit: reviewer-panel comparison; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: diagnostic/benchmark agreement or classification validity; reader/interface/navigation effects are causal only in their separately randomized component; specific constraint: Panel disagreement is uncertainty in the criterion; shadowed administrative reviews estimate transportability descriptively and are not randomized biological oversight outcomes.
Primary estimand: tiered-framework versus usual-review effect on panel agreement in the scenario frame
Candidate primary outcome: agreement with independent multidisciplinary panel for the specified scenario decisions
Primary finalized: No
Mathematical model: type: diagnostic/benchmark sensitivity, false-alert and calibration precision with independent cases/noncases and cluster/reference uncertainty; formulas: logit Pr(agreement_ij=1)=alpha+tau framework_ij+u_reviewer,i+v_scenario,j+pi_period; quantify independent-panel disagreement separately.
Randomization-based scenario-review contrast is distinct from nonrandom shadow-site transport findings.; variables: Y: agreement with independent multidisciplinary panel for the specified scenario decisions; i: reviewer and independently sampled scenario; partner space for shadowing; t: reviewer-panel comparison; the registered visit/time-window index, not another independent unit; A: reviewer-scenario review condition randomized in counterbalanced order; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: tiered-framework versus usual-review effect on panel agreement in the scenario frame; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; TP,FN,FP: true-positive, false-negative and false-positive counts relative to the independently defined reference and matching window, not an assumed perfect gold standard; reference: per-thesis independent reference, including adjudication uncertainty; nonevent time/opportunities: fixed at-risk denominator; event specificity and false alerts per hour are distinct targets; agreement ij: reviewer decision matches the prespecified panel criterion for scenario j; framework ij: randomized framework assignment within reviewer/scenario schedule; u reviewer,v scenario: crossed reviewer and scenario effects; pi period: review order/period effect; family code: REVIEW_CROSS
Sample size design: family: diagnostic/benchmark sensitivity, false-alert and calibration precision with independent cases/noncases and cluster/reference uncertainty; formula: An independent-case rough starting point is n_case ~= z_(1-alpha*/2)^2 Se(1-Se)/h_Se^2; replace with interval-based/clustered simulation for the actual design. Plan noncases/observation time and reference failures separately.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; Se: plausible sensitivity from prior evidence or pilot, not an asserted study result; h Se: desired sensitivity interval half-width; n case: independent positive cases; repeated events within a person require correlation adjustment; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Panel disagreement is uncertainty in the criterion; shadowed administrative reviews estimate transportability descriptively and are not randomized biological oversight outcomes.
Reference independence, fixed thresholds/window and an explicit denominator; incorporation/verification bias can invalidate apparent accuracy.
Participant/family-level test splits and paired comparisons; report abstention, assay failure and indeterminate truth rather than force every case into a binary label.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Reject the framework if agreement does not improve or smaller sites incur disproportionate burden without better detection.; preanalysis gate: Panel disagreement is uncertainty in the criterion; shadowed administrative reviews estimate transportability descriptively and are not randomized biological oversight outcomes.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Panel disagreement is uncertainty in the criterion; shadowed administrative reviews estimate transportability descriptively and are not randomized biological oversight outcomes.
Vary plausible reference misclassification, verification-selection and event-matching window under a prespecified sensitivity family.
Report subgroup usable-data coverage and calibration; avoid interpreting an imprecise null difference as equivalence.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Panel disagreement is uncertainty in the criterion; shadowed administrative reviews estimate transportability descriptively and are not randomized biological oversight outcomes.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/219; question: Can a tiered governance framework improve oversight quality without disproportionately excluding small community laboratories?; aims: Translate inherited governance principles into auditable decision criteria.
Compare review formats on sanitized laboratory-governance scenarios.
Evaluate feasibility and burden with community and institutional reviewers.; design verbatim planning only: Construct 60 nontechnical scenarios spanning consent, waste accountability, data access, and incident reporting. Approximately 120 reviewers from community laboratories and universities complete randomized crossover reviews. A second phase shadows voluntary administrative reviews at eight partner spaces without observing or directing biological procedures.; outcomes verbatim: Primary: agreement with an independent multidisciplinary panel; secondary: missed escalation, review time, and smaller-site burden.; analysis verbatim inherited not authoritative if refined: Model reviewer and scenario effects, counterbalance sequence, and prespecify subgroup analyses; quantify panel disagreement instead of treating consensus as objective truth.; falsification verbatim: Reject the framework if agreement does not improve or smaller sites incur disproportionate burden without better detection.; source ids unverified leads: synthetic-cell-governance-2026
ai-biotech-governance-2026
ai-biology-national-academies-2025; prospectus precedence: No

F08 — Dynamic consent and the representativeness of health-data donation

Rank: 59
Cluster: Community evidence, incentives, ethics and policy
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT; allocation or exposure unit: participant randomized granular revisable or conventional study consent; independent sampling unit: participant; repeated measurement unit: comprehension assessment and permission revision; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Withdrawal changes the consented training population; selection weighting cannot identify unavailable data under arbitrary MNAR; model fairness is secondary.
Primary estimand: consent-interface assignment effect on informed-choice comprehension
Candidate primary outcome: prespecified informed-consent comprehension score
Primary finalized: No
Mathematical model: type: contrast precision/power by design-matrix or simulation planning; formulas: E(Y_it | A_i, X_i, Y_i0) = alpha_t + z(A_i)^T theta_t + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Delta = sum_t w_t c^T theta_t; estimate by baseline-adjusted marginal standardization, with unit/cluster-appropriate inference.; variables: Y: prespecified informed-consent comprehension score; i: participant; t: comprehension assessment and permission revision; the registered visit/time-window index, not another independent unit; A: participant randomized granular revisable or conventional study consent; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: consent-interface assignment effect on informed-choice comprehension; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; z(A): prespecified indicators for the actual randomized arms; two arms reduce to one indicator; family code: RCT
Sample size design: family: contrast precision/power by design-matrix or simulation planning; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Withdrawal changes the consented training population; selection weighting cannot identify unavailable data under arbitrary MNAR; model fairness is secondary.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Reject the proposed benefit if comprehension and trust remain unchanged or subgroup prediction deteriorates despite weighting.; preanalysis gate: Withdrawal changes the consented training population; selection weighting cannot identify unavailable data under arbitrary MNAR; model fairness is secondary.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Withdrawal changes the consented training population; selection weighting cannot identify unavailable data under arbitrary MNAR; model fairness is secondary.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Withdrawal changes the consented training population; selection weighting cannot identify unavailable data under arbitrary MNAR; model fairness is secondary.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/58; question: Does granular ongoing consent improve data-donation trust while changing who remains represented?; aims: Co-design understandable consent choices with underrepresented participants.
Randomize consent interfaces for a low-risk longitudinal dataset.
Measure effects on retention, missingness, and model transportability.; design verbatim planning only: Recruit approximately 500 adults across socioeconomic strata for one year of optional activity and symptom surveys. Compare granular revisable permissions with conventional study-specific consent; both allow withdrawal. Train identical benign activity-prediction models on resulting consented datasets and test them in a separately recruited 150-person cohort.; outcomes verbatim: Primary: informed-choice comprehension; secondary: retention, withdrawal patterns, subgroup calibration, and trust.; analysis verbatim inherited not authoritative if refined: Use randomized interface effects for consent outcomes; model selection with weighting and sensitivity bounds. Prespecify fairness metrics and multiplicity adjustments.; falsification verbatim: Reject the proposed benefit if comprehension and trust remain unchanged or subgroup prediction deteriorates despite weighting.; source ids unverified leads: rwd-digital-platform-cohort-2026
digital-phenotyping-relapse-review-2026
ai-biology-national-academies-2025; prospectus precedence: No

F09 — Risk communication for people excluded from optimization culture

Rank: 80
Cluster: Community evidence, incentives, ethics and policy
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT; allocation or exposure unit: participant randomized resource-aware or standard accurate explanation; independent sampling unit: participant and independently sampled allocation scenario; repeated measurement unit: hypothetical allocation and diary occasion; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Scenario allocations are hypothetical; income/disability modification must be prespecified and is not a claim about actual health benefit.
Primary estimand: resource-aware explanation effect on evidence-consistent allocation
Candidate primary outcome: allocation score based on a prespecified scenario-specific evidence rubric
Primary finalized: No
Mathematical model: type: contrast precision/power by design-matrix or simulation planning; formulas: E(Y_it | A_i, X_i, Y_i0) = alpha_t + z(A_i)^T theta_t + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Delta = sum_t w_t c^T theta_t; estimate by baseline-adjusted marginal standardization, with unit/cluster-appropriate inference.; variables: Y: allocation score based on a prespecified scenario-specific evidence rubric; i: participant and independently sampled allocation scenario; t: hypothetical allocation and diary occasion; the registered visit/time-window index, not another independent unit; A: participant randomized resource-aware or standard accurate explanation; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: resource-aware explanation effect on evidence-consistent allocation; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; z(A): prespecified indicators for the actual randomized arms; two arms reduce to one indicator; family code: RCT
Sample size design: family: contrast precision/power by design-matrix or simulation planning; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Scenario allocations are hypothetical; income/disability modification must be prespecified and is not a claim about actual health benefit.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: The hypothesis fails if spending intentions change solely through fear or if agency declines materially.; preanalysis gate: Scenario allocations are hypothetical; income/disability modification must be prespecified and is not a claim about actual health benefit.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Scenario allocations are hypothetical; income/disability modification must be prespecified and is not a claim about actual health benefit.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Scenario allocations are hypothetical; income/disability modification must be prespecified and is not a claim about actual health benefit.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/79; question: Do resource-aware health explanations improve decisions more than optimization-focused explanations among financially constrained adults?; aims: Identify constraints through interviews spanning income and disability.
Develop matched explanations of testing and supplement claims.
Randomize formats and examine both comprehension and budget allocation.; design verbatim planning only: Interview 45 adults, then recruit a planning sample of 700 stratified by income for an online randomized study. Compare explanations including opportunity costs and accessible alternatives with equally accurate standard explanations. Use hypothetical spending allocations and a six-week decision diary; no products are offered.; outcomes verbatim: Primary: allocation consistent with scenario-specific evidence; secondary: trust, agency, delayed recall, and anticipated financial burden.; analysis verbatim inherited not authoritative if refined: Prespecify income and disability interactions, use partial pooling, and correct secondary comparisons; treat diary nonresponse with weighted sensitivity analyses.; falsification verbatim: The hypothesis fails if spending intentions change solely through fear or if agency declines materially.; source ids unverified leads: endocrine-vitamin-d-guideline-2024
mc-ed-primary-care-guidance-2025
belief-effects-health-review-2026; prospectus precedence: No

F10 — When retrieval makes a health chatbot confidently wrong

Rank: 203
Cluster: Community evidence, incentives, ethics and policy
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: BENCH_BINARY; allocation or exposure unit: fixed model configuration applied to the same held-out prompt set; separate reader randomization; independent sampling unit: source family/prompt family; participant for reader experiment; repeated measurement unit: system answer and adjudicator rating; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: diagnostic/benchmark agreement or classification validity; reader/interface/navigation effects are causal only in their separately randomized component; specific constraint: System benchmark and reader effects are distinct estimands; freeze model versions, retrieval corpus, answer generation and source-entailment rubric.
Primary estimand: paired unsupported-recommendation rate difference between boundary-checking and retrieval-only systems
Candidate primary outcome: adjudicated unsupported-recommendation indicator per prompt
Primary finalized: No
Mathematical model: type: diagnostic/benchmark sensitivity, false-alert and calibration precision with independent cases/noncases and cluster/reference uncertainty; formulas: For paired prompt i, D_i=U_i,new-U_i,base with U=1 for unsupported recommendation; Delta=mean_i D_i with source-family-clustered uncertainty. Reader comprehension is estimated separately by reader randomization.; variables: Y: adjudicated unsupported-recommendation indicator per prompt; i: source family/prompt family; participant for reader experiment; t: system answer and adjudicator rating; the registered visit/time-window index, not another independent unit; A: fixed model configuration applied to the same held-out prompt set; separate reader randomization; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: paired unsupported-recommendation rate difference between boundary-checking and retrieval-only systems; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; TP,FN,FP: true-positive, false-negative and false-positive counts relative to the independently defined reference and matching window, not an assumed perfect gold standard; reference: per-thesis independent reference, including adjudication uncertainty; nonevent time/opportunities: fixed at-risk denominator; event specificity and false alerts per hour are distinct targets; U i,new,U i,base: blinded adjudicated unsupported-recommendation indicators for the same held-out prompt under the two locked systems; D i: paired unsupported-rate difference; family code: BENCH_BINARY
Sample size design: family: diagnostic/benchmark sensitivity, false-alert and calibration precision with independent cases/noncases and cluster/reference uncertainty; formula: An independent-case rough starting point is n_case ~= z_(1-alpha*/2)^2 Se(1-Se)/h_Se^2; replace with interval-based/clustered simulation for the actual design. Plan noncases/observation time and reference failures separately.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; Se: plausible sensitivity from prior evidence or pilot, not an asserted study result; h Se: desired sensitivity interval half-width; n case: independent positive cases; repeated events within a person require correlation adjustment; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: System benchmark and reader effects are distinct estimands; freeze model versions, retrieval corpus, answer generation and source-entailment rubric.
Reference independence, fixed thresholds/window and an explicit denominator; incorporation/verification bias can invalidate apparent accuracy.
Participant/family-level test splits and paired comparisons; report abstention, assay failure and indeterminate truth rather than force every case into a binary label.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Reject added value if gains vanish on held-out sources or fewer recommendations merely reflect indiscriminate refusal.; preanalysis gate: System benchmark and reader effects are distinct estimands; freeze model versions, retrieval corpus, answer generation and source-entailment rubric.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: System benchmark and reader effects are distinct estimands; freeze model versions, retrieval corpus, answer generation and source-entailment rubric.
Vary plausible reference misclassification, verification-selection and event-matching window under a prespecified sensitivity family.
Report subgroup usable-data coverage and calibration; avoid interpreting an imprecise null difference as equivalence.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
System benchmark and reader effects are distinct estimands; freeze model versions, retrieval corpus, answer generation and source-entailment rubric.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/202; question: Does explicit source-boundary checking reduce unsupported enhancement recommendations beyond citation retrieval alone?; aims: Create a clinician-adjudicated benchmark of claim-source mismatches.
Compare fixed chatbot configurations under blinded evaluation.
Test whether readers recognize residual errors after using each system.; design verbatim planning only: Build approximately 300 nonoperational prompts from inherited sources and 100 held-out prompts from later independently acquired evidence. Compare retrieval-only, boundary-checking, and abstention-calibrated configurations at fixed model versions. Randomize 240 adult readers to answer sets, measuring comprehension rather than treatment uptake.; outcomes verbatim: Primary: unsupported recommendation rate; secondary: citation entailment, appropriate abstention, and reader error detection.; analysis verbatim inherited not authoritative if refined: Cluster uncertainty by source family; preregister scoring rubrics, adjudicator agreement, and multiplicity across system contrasts. Separate benchmark performance from reader effects.; falsification verbatim: Reject added value if gains vanish on held-out sources or fewer recommendations merely reflect indiscriminate refusal.; source ids unverified leads: ai-drug-discovery-practical-guidance-2026
fda-fdalabel-drug-labeling-2026
fda-off-label-use-2026; prospectus precedence: No

F11 — Does an immune-age panel improve prediction of vaccine response and infection risk?

Rank: 210
Cluster: Immune aging, infection and postviral recovery
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: PRED; allocation or exposure unit: routine clinician-directed vaccination, no immune intervention assigned; independent sampling unit: participant with site-held-out validation; repeated measurement unit: baseline features and scheduled immune assay; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: prediction, calibration or model transportability; prediction gain is not clinical efficacy, mechanistic necessity or causal treatment selection; specific constraint: Baseline antigen-specific immunity and vaccine history are essential comparators; predicting an assay response does not establish infection protection.
Primary estimand: external calibration/loss gain of locked immune-age panel beyond a complete routine predictor model
Candidate primary outcome: prespecified antigen-specific immune-response measure at a fixed postvaccination time
Primary finalized: No
Mathematical model: type: external/nested predictive-validation precision with target prevalence/variance, independent-unit holdout coverage and model complexity; formulas: Gain = mean_test[L(Y,f_base(X)) - L(Y,f_new(X))], with both predictions frozen before test outcomes; bootstrap independent held-out units.
For a binary target assess E[Y | p_hat=p] versus p and calibration intercept/slope logit Pr(Y=1)=a_cal+b_cal logit(p_hat); for a continuous target use Y=a_cal+b_cal y_hat+error plus scale-appropriate calibration/loss.; variables: Y: prespecified antigen-specific immune-response measure at a fixed postvaccination time; i: participant with site-held-out validation; t: baseline features and scheduled immune assay; the registered visit/time-window index, not another independent unit; A: routine clinician-directed vaccination, no immune intervention assigned; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: external calibration/loss gain of locked immune-age panel beyond a complete routine predictor model; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; L: one proper or scientifically justified primary loss/scoring rule fixed for the actual outcome type; positive Gain favors new model; f base,f new: locked baseline/new prediction functions using only information available at the registered prediction time; test: untouched person/family/lab/site/time set as specified per thesis; p hat,y hat: frozen probability or continuous prediction; a cal,b cal: calibration intercept and slope, estimated on evaluation data for assessment; using them to update the model requires another evaluation set; family code: PRED
Sample size design: family: external/nested predictive-validation precision with target prevalence/variance, independent-unit holdout coverage and model complexity; formula: Simulate or bootstrap the independent-unit test loss/calibration/decision criterion; plan cases/events, noncases and independent sites/families/programs separately. No universal events-per-variable or participant N is asserted.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Baseline antigen-specific immunity and vaccine history are essential comparators; predicting an assay response does not establish infection protection.
Train/tune/test separation at the actual biological/participant/family/time level; no future, postoutcome or target-derived predictors.
Validation support must cover intended deployment; calibration and missing/failing measurements are reported with discrimination/loss.
A validation subset selected by culture success, assay viability, consent or data completeness is a different target unless the full strategy is explicitly evaluated.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: The panel fails its primary prediction claim if calibration or discrimination for the prespecified immune response does not improve meaningfully over the full routine-predictor benchmark in the held-out site. Useful immune-response prediction without precise infection validation remains a narrower result than predicting protection.; preanalysis gate: Baseline antigen-specific immunity and vaccine history are essential comparators; predicting an assay response does not establish infection protection.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Baseline antigen-specific immunity and vaccine history are essential comparators; predicting an assay response does not establish infection protection.
Repeat loss/calibration with independent-cluster resampling, prespecified subgroup/transport shifts and plausible missingness.
Compare fixed simple baselines, calibration and uncertainty coverage; report abstention/failure/turnaround rather than deleting hard cases.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Frozen prediction time/horizon, proper loss, baseline model, feature availability, independent split/holdout and target deployment population.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Baseline antigen-specific immunity and vaccine history are essential comparators; predicting an assay response does not establish infection protection.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/209; question: Do immune-age signatures predict vaccine response and subsequent infection better than ordinary clinical variables?; aims: Benchmark candidate signatures against routine clinical predictors.
Validate response and waning predictions in a held-out clinic.
Test associations with adjudicated infections and meaningful functional disruption.; design verbatim planning only: Prospectively enroll a planning cohort of 600 adults aged 60 or older receiving routine clinician-directed vaccination at three sites. Collect baseline and scheduled follow-up blood samples over 18 months; compare locked biomarker models with age, frailty, comorbidity, and vaccination-history models.; outcomes verbatim: Primary: external calibration for prespecified immune response; secondary: documented infections, symptom days, and lost usual activity.; analysis verbatim inherited not authoritative if refined: Use nested development and site-held-out validation. Benchmark the locked panel against routine predictors including baseline antigen-specific immunity, prior infection, vaccine product and vaccination history, age, frailty, comorbidity and immunosuppressants. Prespecify the immune-response target and control biomarker-family multiplicity. Analyze documented infections separately with uncertainty and ascertainment limitations; immune-response prediction does not establish protection prediction or clinical utility.; falsification verbatim: The panel fails its primary prediction claim if calibration or discrimination for the prespecified immune response does not improve meaningfully over the full routine-predictor benchmark in the held-out site. Useful immune-response prediction without precise infection validation remains a narrower result than predicting protection.; source ids unverified leads: immunosenescence-vaccine-review-2025
vaccine-efficacy-aging-review-2025
serum-proteome-aging-cohort-2025; prospectus precedence: No

F12 — Sleep regularity as a modifiable vaccine-response determinant

Rank: 148
Cluster: Immune aging, infection and postviral recovery
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT; allocation or exposure unit: candidate participant allocation to schedule support versus education; random sequence must be specified; independent sampling unit: participant; repeated measurement unit: immune assay and sleep-monitoring occasion; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Inherited intention-to-treat analysis implies an assignment design but randomization details need confirmation; vaccine timing/product remain clinician-directed; sleep mediation is secondary.
Primary estimand: sleep-support offer effect on the prespecified antigen-specific response
Candidate primary outcome: fixed postvaccination antigen-specific immune-response measure
Primary finalized: No
Mathematical model: type: contrast precision/power by design-matrix or simulation planning; formulas: E(Y_it | A_i, X_i, Y_i0) = alpha_t + z(A_i)^T theta_t + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Delta = sum_t w_t c^T theta_t; estimate by baseline-adjusted marginal standardization, with unit/cluster-appropriate inference.; variables: Y: fixed postvaccination antigen-specific immune-response measure; i: participant; t: immune assay and sleep-monitoring occasion; the registered visit/time-window index, not another independent unit; A: candidate participant allocation to schedule support versus education; random sequence must be specified; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: sleep-support offer effect on the prespecified antigen-specific response; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; z(A): prespecified indicators for the actual randomized arms; two arms reduce to one indicator; family code: RCT
Sample size design: family: contrast precision/power by design-matrix or simulation planning; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Inherited intention-to-treat analysis implies an assignment design but randomization details need confirmation; vaccine timing/product remain clinician-directed; sleep mediation is secondary.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Reject the causal hypothesis if sleep regularity improves but immune response remains unchanged within clinically meaningful uncertainty bounds.; preanalysis gate: Inherited intention-to-treat analysis implies an assignment design but randomization details need confirmation; vaccine timing/product remain clinician-directed; sleep mediation is secondary.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Inherited intention-to-treat analysis implies an assignment design but randomization details need confirmation; vaccine timing/product remain clinician-directed; sleep mediation is secondary.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Inherited intention-to-treat analysis implies an assignment design but randomization details need confirmation; vaccine timing/product remain clinician-directed; sleep mediation is secondary.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/147; question: Does improving sleep regularity before routine vaccination improve immune response in older adults?; aims: Establish reliable sleep-regularity measurement in older participants.
Randomize behavioral support before routine vaccination.
Test whether immune-response changes persist beyond the initial follow-up.; design verbatim planning only: Recruit approximately 180 adults aged 60–80 from vaccination clinics for an investigator-supervised behavioral trial. Compare personalized schedule support with general health education; vaccination type and timing remain routine clinical decisions. Measure actigraphy, diaries, immune responses, and three-month persistence. Numbers are planning assumptions.; outcomes verbatim: Primary: prespecified antigen-specific response; secondary: sleep regularity, daytime function, adherence, and adverse experiences.; analysis verbatim inherited not authoritative if refined: Use intention-to-treat analysis, stratify by vaccine type, adjust baseline immunity, and prespecify mediation assumptions and multiplicity across immune assays.; falsification verbatim: Reject the causal hypothesis if sleep regularity improves but immune response remains unchanged within clinically meaningful uncertainty bounds.; source ids unverified leads: immunosenescence-vaccine-review-2025
vaccine-efficacy-aging-review-2025
exercise-sleep-network-meta-2026; prospectus precedence: No

F13 — Food-access support, micronutrient adequacy, and vaccine response

Rank: 63
Cluster: Immune aging, infection and postviral recovery
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT_INT; allocation or exposure unit: candidate participant allocation to grocery/dietitian support versus existing resources; independent sampling unit: participant with clinic clustering; repeated measurement unit: immune assay and food-access assessment; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Confirm concealed randomization before causal assignment analysis; corrected deficiency and food uptake are postassignment pathways, not baseline adjustments.
Primary estimand: food-access offer effect and one baseline-inadequacy interaction on immune response
Candidate primary outcome: prespecified postvaccination immune-response measure
Primary finalized: No
Mathematical model: type: interaction-specific precision/power with modifier distribution and reliability; simulate the actual interaction; formulas: E(Y_it | A_i, B_i, X_i) = alpha_t + tau_t A_i + eta_t B_i + kappa_t A_i B_i + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Treatment contrast at baseline B=b is tau_t + kappa_t b; the interaction target is kappa_t, not a difference in within-subgroup p-values.; variables: Y: prespecified postvaccination immune-response measure; i: participant with clinic clustering; t: immune assay and food-access assessment; the registered visit/time-window index, not another independent unit; A: candidate participant allocation to grocery/dietitian support versus existing resources; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: food-access offer effect and one baseline-inadequacy interaction on immune response; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; B: baseline nutritional-inadequacy measure/category, fixed independently of immune-response outcomes; centering/coding and any measurement-error model fixed before evaluation; eta: baseline modifier association; kappa: assignment-by-baseline-modifier interaction coefficient on the stated scale; family code: RCT_INT
Sample size design: family: interaction-specific precision/power with modifier distribution and reliability; simulate the actual interaction; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure
modifier prevalence/distribution and variance
modifier reliability and overlap
interaction size or precision target; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Confirm concealed randomization before causal assignment analysis; corrected deficiency and food uptake are postassignment pathways, not baseline adjustments.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The modifier must be measured before assignment/initiation; interaction is scale dependent and requires support within modifier strata.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: The hypothesis fails if access and nutritional adequacy improve without any corresponding response difference or baseline-status interaction.; preanalysis gate: Confirm concealed randomization before causal assignment analysis; corrected deficiency and food uptake are postassignment pathways, not baseline adjustments.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Confirm concealed randomization before causal assignment analysis; corrected deficiency and food uptake are postassignment pathways, not baseline adjustments.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Vary modifier measurement-error assumptions and retain a continuous modifier unless a cutoff was independently fixed.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Confirm concealed randomization before causal assignment analysis; corrected deficiency and food uptake are postassignment pathways, not baseline adjustments.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/62; question: Can improving food access enhance vaccine response among nutritionally vulnerable older adults?; aims: Characterize food insecurity and clinically assessed nutritional status.
Randomize a food-access support package versus usual resources.
Test nutritional adequacy as a mediator and baseline effect modifier.; design verbatim planning only: Enroll a planning sample of 240 adults aged over 60 through community clinics before routine vaccination. Offer either grocery-access assistance and dietitian navigation or existing community resources; continue all usual clinical care. Follow dietary adequacy, routine deficiency markers, vaccine response, and daily function for six months.; outcomes verbatim: Primary: prespecified immune-response measure; secondary: food security, corrected inadequacy, function, and participant burden.; analysis verbatim inherited not authoritative if refined: Prespecify baseline inadequacy interaction, adjust vaccine history and site, and use intention-to-treat mediation sensitivity analyses. Control multiple nutrient comparisons.; falsification verbatim: The hypothesis fails if access and nutritional adequacy improve without any corresponding response difference or baseline-status interaction.; source ids unverified leads: endocrine-vitamin-d-guideline-2024
nih-ods-b12-2025
immunosenescence-vaccine-review-2025; prospectus precedence: No

F14 — NAD target engagement versus recovery in long COVID

Rank: 99
Cluster: Immune aging, infection and postviral recovery
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: MIXED_EVIDENCE; allocation or exposure unit: original trial randomization plus separate observational usual-care cohort; independent sampling unit: participant within original trial/cohort, never pooled as exchangeable arms; repeated measurement unit: recovery and biomarker visits; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Trial randomization cannot make biomarker change randomized or identify a cohort treatment effect; selective follow-up requires sensitivity bounds.
Primary estimand: randomized functional-recovery assignment contrast in the trial; separate biomarker-recovery association in the cohort
Candidate primary outcome: patient-reported functional-recovery score at the fixed six-month time
Primary finalized: No
Mathematical model: type: contrast precision/power by design-matrix or simulation planning; formulas: E(Y_it | A_i, X_i, Y_i0) = alpha_t + z(A_i)^T theta_t + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Delta = sum_t w_t c^T theta_t; estimate by baseline-adjusted marginal standardization, with unit/cluster-appropriate inference.; variables: Y: patient-reported functional-recovery score at the fixed six-month time; i: participant within original trial/cohort, never pooled as exchangeable arms; t: recovery and biomarker visits; the registered visit/time-window index, not another independent unit; A: original trial randomization plus separate observational usual-care cohort; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: randomized functional-recovery assignment contrast in the trial; separate biomarker-recovery association in the cohort; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; z(A): prespecified indicators for the actual randomized arms; two arms reduce to one indicator; family code: MIXED_EVIDENCE
Sample size design: family: contrast precision/power by design-matrix or simulation planning; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Trial randomization cannot make biomarker change randomized or identify a cohort treatment effect; selective follow-up requires sensitivity bounds.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The formula applies to the original randomized comparison only; observational validation/cohort associations receive separate labels and cannot be pooled as randomized arms.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Reject the proposed disconnect if robust biomarker-mediated functional benefit persists across plausible missingness analyses and independent validation.; preanalysis gate: Trial randomization cannot make biomarker change randomized or identify a cohort treatment effect; selective follow-up requires sensitivity bounds.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Trial randomization cannot make biomarker change randomized or identify a cohort treatment effect; selective follow-up requires sensitivity bounds.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Trial randomization cannot make biomarker change randomized or identify a cohort treatment effect; selective follow-up requires sensitivity bounds.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/98; question: Do NAD-related biomarker changes track meaningful recovery independently of selective follow-up in long COVID?; aims: Reconstruct prespecified outcomes and attrition in accessible trial data.
Estimate treatment effects under explicit missingness assumptions.
Prospectively validate biomarker-recovery associations in an independent clinical cohort.; design verbatim planning only: Seek individual data from the inherited randomized trial and enroll approximately 150 long-COVID clinic patients for observational follow-up during usual care. Compare randomized groups in the trial and biomarker trajectories in the cohort; do not initiate supplementation. Use repeated cognition, fatigue, function, and blood measures over six months.; outcomes verbatim: Primary: patient-reported functional recovery; secondary: fatigue, validated cognition, biomarker change, and dropout patterns.; analysis verbatim inherited not authoritative if refined: Use joint longitudinal-dropout models, pattern-mixture sensitivity bounds, and prespecified causal mediation assumptions; adjust illness duration, sleep, depression, and concurrent care.; falsification verbatim: Reject the proposed disconnect if robust biomarker-mediated functional benefit persists across plausible missingness analyses and independent validation.; source ids unverified leads: nr-long-covid-rct-2025
wearables-post-acute-2026
aging-clocks-critical-review-2026; prospectus precedence: No

F15 — Do recovery dashboards help or overrule lived symptoms?

Rank: 14
Cluster: Immune aging, infection and postviral recovery
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT; allocation or exposure unit: participant randomized uncertainty-aware dashboard versus symptom diary (two-arm expanded prospectus); independent sampling unit: participant; repeated measurement unit: scheduled participation and worsening assessments; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Expanded two-arm prospectus takes precedence over the earlier three-arm brief; both receive clinical pacing support; include use discontinuation/additional care and report worsening separately.
Primary estimand: 24-week uncertainty-dashboard versus diary assignment effect on participation
Candidate primary outcome: functional-participation score at twenty-four weeks
Primary finalized: No
Mathematical model: type: contrast precision/power by design-matrix or simulation planning; formulas: E(Y_it | A_i, X_i, Y_i0) = alpha_t + z(A_i)^T theta_t + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Delta = sum_t w_t c^T theta_t; estimate by baseline-adjusted marginal standardization, with unit/cluster-appropriate inference.; variables: Y: functional-participation score at twenty-four weeks; i: participant; t: scheduled participation and worsening assessments; the registered visit/time-window index, not another independent unit; A: participant randomized uncertainty-aware dashboard versus symptom diary (two-arm expanded prospectus); code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: 24-week uncertainty-dashboard versus diary assignment effect on participation; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; z(A): prespecified indicators for the actual randomized arms; two arms reduce to one indicator; family code: RCT
Sample size design: family: contrast precision/power by design-matrix or simulation planning; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Expanded two-arm prospectus takes precedence over the earlier three-arm brief; both receive clinical pacing support; include use discontinuation/additional care and report worsening separately.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Reject benefit if comprehension improves but function does not, or if any gain accompanies more symptom worsening.; preanalysis gate: Expanded two-arm prospectus takes precedence over the earlier three-arm brief; both receive clinical pacing support; include use discontinuation/additional care and report worsening separately.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Expanded two-arm prospectus takes precedence over the earlier three-arm brief; both receive clinical pacing support; include use discontinuation/additional care and report worsening separately.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Expanded two-arm prospectus takes precedence over the earlier three-arm brief; both receive clinical pacing support; include use discontinuation/additional care and report worsening separately.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/13; question: Does adding uncertainty-aware wearable feedback to clinical pacing support improve postviral function?; aims: Validate within-person concordance of wearable signals and symptoms.
Co-design an uncertainty-aware dashboard with affected patients.
Randomize dashboard formats during clinician-supported recovery care.; design verbatim planning only: Follow approximately 180 adults with persistent postviral symptoms for baseline calibration, then randomize uncertainty-aware feedback, conventional score display, or symptom diaries. All receive usual clinical pacing support; no exertion targets are assigned by the algorithm. Follow participants for six months using accessible remote assessments.; outcomes verbatim: Primary: functional participation; secondary: post-exertional worsening, anxiety, autonomy, healthcare use, and alert burden.; analysis verbatim inherited not authoritative if refined: Use intention-to-treat repeated-measures models, stratify baseline severity and device type, and treat worsening as a prespecified safety endpoint. Analyze missingness by symptom severity.; falsification verbatim: Reject benefit if comprehension improves but function does not, or if any gain accompanies more symptom worsening.; source ids unverified leads: wearables-post-acute-2026
wearable-hrv-inflammation-2026
wearables-remote-monitoring-2026; prospectus precedence: Yes; prospectus file: thesis-prospectuses.json; prospectus design and estimand: The primary estimand is the treatment-policy mean difference in 24-week participation under assignment to dashboard versus diary, regardless of dashboard use, additional care or device discontinuation. The analysis unit is the participant. Randomization is stratified by baseline severity and clinic. Both groups retain symptom-led clinical support; neither receives algorithmic exertion targets or automated clearance to exercise.; prospectus analysis plan: Use baseline-adjusted regression for the 24-week primary contrast with clinic and stratification covariates, supported by repeated-measures estimates. Report confidence intervals and absolute symptom-worsening differences alongside benefit. Predefine one primary efficacy contrast; severity interactions are exploratory. Relate pre-randomization wearable discordance to later outcomes descriptively. A safety committee reviews worsening and burden using prespecified pause criteria; the study is not powered to exclude rare harm.; precedence note: Expanded prospectus refinements govern this scaffold; verbatim earlier brief material is retained for traceability, not as a contradictory analysis instruction. F15 is explicitly two arms; F30 has no unjustified symptom noninferiority claim.

F16 — Separating autonomic variation from changes in circulating inflammatory markers

Rank: 179
Cluster: Immune aging, infection and postviral recovery
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: WITHIN_PRED; allocation or exposure unit: natural postviral recovery with no assigned exposure; independent sampling unit: participant and independent validation site; repeated measurement unit: repeated HRV and circulating-marker window; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: association or prospective prediction; temporal order and model adjustment alone do not identify a mechanism or treatment effect; specific constraint: Within-person and between-person associations must be separated; circulating-marker prediction is not a measure of global immune recovery.
Primary estimand: held-out incremental prediction of the fixed circulating-marker panel from HRV beyond sleep/breathing/orthostasis
Candidate primary outcome: prespecified marker-specific concentration or change
Primary finalized: No
Mathematical model: type: independent-person/site prospective prediction-gain and calibration precision, with within-person dependence; formulas: Y_it = alpha + u_i + beta_W (B_it-Bbar_i) + beta_B Bbar_i + gamma^T X_it + f_time(t) + b_study + epsilon_it; u_i is a random intercept with a declared conditional mean/covariance model, not unrestricted fixed person effects.
B_observed,it = B_true,it + e_it; estimate measurement variance from replicates/validation or report sensitivity rather than assume it known.
Primary prediction target: Gain=mean_test[L(Y,f_base(X))-L(Y,f_new(X,B_history))], with locked models and proper outcome-specific loss; report calibration as well as gain.
For new held-out persons/sites, predict using the training-estimated population random-effect distribution or only genuinely available preprediction calibration; never fit a person intercept using that test person's future outcome.; variables: Y: prespecified marker-specific concentration or change; i: participant and independent validation site; t: repeated HRV and circulating-marker window; the registered visit/time-window index, not another independent unit; A: natural postviral recovery with no assigned exposure; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: held-out incremental prediction of the fixed circulating-marker panel from HRV beyond sleep/breathing/orthostasis; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; B it: prespecified wearable HRV history available before each circulating-marker target; Bbar i: person/donor mean over a prespecified observation schedule; beta W,beta B: within-unit and between-unit association coefficients, not causal effects; X it: explicitly time-ordered baseline/premeasurement covariates; posttreatment adjustment changes interpretation; f time(t): prespecified sampling/recovery/calendar trajectory; b study: study/site/batch effect where relevant; B true,e it: latent predictor and measurement error, requiring identification constraints; Gain: held-out improvement over the complete prespecified routine/sleep/autonomic or clinical baseline; L: registered proper/validated outcome-specific loss; f base,f new: models frozen in development, with all feature selection internal to training; B history: predictor history available before the registered prediction time; family code: WITHIN_PRED
Sample size design: family: independent-person/site prospective prediction-gain and calibration precision, with within-person dependence; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Within-person and between-person associations must be separated; circulating-marker prediction is not a measure of global immune recovery.
Associations need a clear sampling/time frame; causal interpretation is not supplied by this model.
Measurement error, reverse causation, time-varying confounding and selected observation may remain despite adjustment.
All feature selection and time normalization occur inside training splits; later outcome windows are untouched and only information available at prediction is used.
For the secondary within/between association model, specify the conditional distribution of random intercept u_i; fixed person effects cannot separately estimate beta_B for time-invariant person means.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Reproducibly useful incremental prediction of the prespecified circulating markers across sites would challenge the hypothesized limited monitoring value. Poor prediction would limit HRV as a monitor of those markers, without demonstrating absence of inflammation or excluding unmeasured inflammatory processes.; preanalysis gate: Within-person and between-person associations must be separated; circulating-marker prediction is not a measure of global immune recovery.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Within-person and between-person associations must be separated; circulating-marker prediction is not a measure of global immune recovery.
Vary lag order/window and measurement-error reliability within a prespecified family; include negative-control windows when meaningful.
Separate within- and between-unit effects, missingness patterns and source-study replication; do not use cell/assay counts as independent donor counts.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Within-person and between-person associations must be separated; circulating-marker prediction is not a measure of global immune recovery.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/178; question: Do wearable HRV trajectories improve prediction of selected circulating inflammatory markers beyond sleep, breathing and orthostatic measures during postviral recovery?; aims: Quantify repeatability across standardized and free-living HRV measurements.
Partition associations with inflammatory and autonomic measures.
Validate a restrained interpretation model in an independent clinic.; design verbatim planning only: Recruit approximately 220 adults with persistent postviral symptoms and 100 recovered comparators. Collect repeated wearable data, supervised nonprovocative autonomic assessments, symptoms, sleep diaries, and routine inflammatory markers for nine months. Compare within-person trajectories and test a locked model at a second site.; outcomes verbatim: Primary: incremental inflammatory-marker prediction beyond sleep and autonomic variables; secondary: symptom concordance and subgroup calibration.; analysis verbatim inherited not authoritative if refined: Use multilevel within-between decomposition, adjust medications, infection recurrence and activity, and control assay-family multiplicity. Prespecify the circulating-marker panel and account for repeatability and measurement error when comparing predictor contributions. Interpret findings as marker-specific associations; neither marker prediction nor its absence establishes global inflammatory recovery, immune state or biological causation.; falsification verbatim: Reproducibly useful incremental prediction of the prespecified circulating markers across sites would challenge the hypothesized limited monitoring value. Poor prediction would limit HRV as a monitor of those markers, without demonstrating absence of inflammation or excluding unmeasured inflammatory processes.; source ids unverified leads: wearable-hrv-inflammation-2026
wearables-post-acute-2026
slow-breathing-hrv-review-2025; prospectus precedence: No

F17 — Testing behavior as a hidden driver of immune-health claims

Rank: 125
Cluster: Immune aging, infection and postviral recovery
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: ASCERTAIN; allocation or exposure unit: common offered assessment procedure; testing uptake and lifestyle observed; independent sampling unit: participant; repeated measurement unit: symptomatic episode captured by study and/or routine record; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: difference in capture of symptomatic episodes; no identification of unrecognized/asymptomatic infection incidence; specific constraint: Neither record system captures all infection; scheduled common access only partly standardizes ascertainment; asymptomatic total incidence is outside target.
Primary estimand: difference in adjudicated symptomatic-episode capture under study procedure versus routine records
Candidate primary outcome: paired episode ascertainment/capture indicator within the fixed season
Primary finalized: No
Mathematical model: type: paired episode-capture discrepancy precision with participant clustering and missing-episode bounds; formulas: D_i=N_i,study-N_i,routine over the common season; estimate mean D and paired episode capture probabilities standardized to the prespecified population.
Observed capture probability depends on symptom reporting/testing uptake; use inverse-observation weighting only with supported observation probabilities and bound missed episodes.; variables: Y: paired episode ascertainment/capture indicator within the fixed season; i: participant; t: symptomatic episode captured by study and/or routine record; the registered visit/time-window index, not another independent unit; A: common offered assessment procedure; testing uptake and lifestyle observed; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: difference in adjudicated symptomatic-episode capture under study procedure versus routine records; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; N i,study,N i,routine: matched adjudicated symptomatic-episode counts under study ascertainment and routine records; D i: paired capture discrepancy, not total infection incidence; family code: ASCERTAIN
Sample size design: family: paired episode-capture discrepancy precision with participant clustering and missing-episode bounds; formula: Use design-specific simulation or independent-unit resampling for the stated estimand and target confidence width/power; preserve all allocation, censoring, clustering and measurement-error structure.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Neither record system captures all infection; scheduled common access only partly standardizes ascertainment; asymptomatic total incidence is outside target.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Neither source is perfect truth; event matching and common symptom criteria fixed; common offered access does not ensure equal uptake.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Similar associations under the common symptom-assessment procedure, routine records and symptom-burden outcomes would weaken ascertainment bias as the main explanation, subject to residual differences in symptom recognition, reporting and testing uptake.; preanalysis gate: Neither record system captures all infection; scheduled common access only partly standardizes ascertainment; asymptomatic total incidence is outside target.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Neither record system captures all infection; scheduled common access only partly standardizes ascertainment; asymptomatic total incidence is outside target.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Neither record system captures all infection; scheduled common access only partly standardizes ascertainment; asymptomatic total incidence is outside target.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/124; question: How much do testing behavior and healthcare access change observed associations between optimization habits and detected symptomatic infections?; aims: Measure baseline testing propensity and healthcare access.
Compare routine-recorded infection with standardized symptom-triggered assessment.
Estimate how ascertainment correction changes lifestyle-infection associations.; design verbatim planning only: Enroll a planning cohort of 700 adults spanning tracking intensity and income for one respiratory-infection season. Use the same scheduled symptom check-ins and prespecified symptom-triggered assessment access for all, while recording completion and uptake. Distinguish study-supplied assessments from concurrent routine clinical-record episodes. No infection exposure is assigned and no pathogen methods are studied.; outcomes verbatim: Primary: discrepancy between prospectively adjudicated symptomatic episodes under the study procedure and episodes captured in routine clinical records. Secondary: symptom days, functional disruption, check-in and testing uptake, and healthcare access. Total infection incidence, including unrecognized asymptomatic infections, is outside the primary estimand.; analysis verbatim inherited not authoritative if refined: Use inverse-observation weighting, negative-control healthcare contacts and quantitative bias analysis; adjust occupation, household exposure, vaccination, comorbidity and prior infection. Model differential symptom reporting and testing uptake, with sensitivity bounds for missed episodes. Treat common access as partial ascertainment standardization and avoid identifying all residual lifestyle associations as immune effects.; falsification verbatim: Similar associations under the common symptom-assessment procedure, routine records and symptom-burden outcomes would weaken ascertainment bias as the main explanation, subject to residual differences in symptom recognition, reporting and testing uptake.; source ids unverified leads: rwd-digital-platform-cohort-2026
immunosenescence-vaccine-review-2025
wearables-post-acute-2026; prospectus precedence: No

F18 — Periodontal care as a test of oral-systemic immune causality

Rank: 162
Cluster: Immune aging, infection and postviral recovery
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT; allocation or exposure unit: candidate participant allocation to facilitated early dental appointment versus standard scheduling; independent sampling unit: participant; repeated measurement unit: inflammatory and vaccine assay visit; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Confirm random allocation; urgent care preserved; receipt-based IV requires exclusion/monotonicity not supplied by randomization alone.
Primary estimand: facilitated-scheduling offer effect on the fixed inflammatory composite
Candidate primary outcome: prespecified inflammatory-composite score
Primary finalized: No
Mathematical model: type: contrast precision/power by design-matrix or simulation planning; formulas: E(Y_it | A_i, X_i, Y_i0) = alpha_t + z(A_i)^T theta_t + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Delta = sum_t w_t c^T theta_t; estimate by baseline-adjusted marginal standardization, with unit/cluster-appropriate inference.; variables: Y: prespecified inflammatory-composite score; i: participant; t: inflammatory and vaccine assay visit; the registered visit/time-window index, not another independent unit; A: candidate participant allocation to facilitated early dental appointment versus standard scheduling; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: facilitated-scheduling offer effect on the fixed inflammatory composite; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; z(A): prespecified indicators for the actual randomized arms; two arms reduce to one indicator; family code: RCT
Sample size design: family: contrast precision/power by design-matrix or simulation planning; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Confirm random allocation; urgent care preserved; receipt-based IV requires exclusion/monotonicity not supplied by randomization alone.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Reject the systemic mechanism if substantial periodontal improvement occurs without inflammatory change; vaccine effects remain separately falsifiable.; preanalysis gate: Confirm random allocation; urgent care preserved; receipt-based IV requires exclusion/monotonicity not supplied by randomization alone.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Confirm random allocation; urgent care preserved; receipt-based IV requires exclusion/monotonicity not supplied by randomization alone.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Confirm random allocation; urgent care preserved; receipt-based IV requires exclusion/monotonicity not supplied by randomization alone.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/161; question: Does expedited periodontal care alter inflammatory burden and routine vaccine response in older adults?; aims: Characterize oral disease and systemic inflammatory profiles.
Randomize expedited access to indicated periodontal care.
Test whether oral improvement precedes immune and functional changes.; design verbatim planning only: Recruit approximately 180 adults aged over 60 with nonurgent periodontal treatment needs. Compare facilitated early appointments with standard scheduling, retaining urgent-care access for everyone. Coordinate sampling around routine clinician-directed vaccination and follow participants for six months; clinical treatment follows existing dental standards.; outcomes verbatim: Primary: prespecified inflammatory composite; secondary: periodontal status, antigen-specific response, oral function, and quality of life.; analysis verbatim inherited not authoritative if refined: Use intention-to-treat estimates and instrumental-variable sensitivity analyses for treatment receipt; adjust smoking, diabetes, and baseline immunity, with hierarchical testing across endpoints.; falsification verbatim: Reject the systemic mechanism if substantial periodontal improvement occurs without inflammatory change; vaccine effects remain separately falsifiable.; source ids unverified leads: oral-microbiome-systemic-review-2026
oral-gut-axis-review-2026
immunosenescence-vaccine-review-2025; prospectus precedence: No

F19 — Immune safety during clinician-prescribed off-label mTOR inhibition

Rank: 199
Cluster: Immune aging, infection and postviral recovery
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: OBS_RATE; allocation or exposure unit: clinician-selected new off-label mTOR initiation versus eligible noninitiation; independent sampling unit: participant; repeated measurement unit: recurrent clinically significant infection and exposure interval; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: observational comparison/association; causal interpretation is conditional on explicit exchangeability, positivity, consistency and observation assumptions; specific constraint: Time zero, switching, indication and health behavior matter; death terminates infection opportunity and is not ordinary independent dropout.
Primary estimand: standardized clinically significant infection rate or cumulative count under specified observed strategies, conditional on exchangeability
Candidate primary outcome: adjudicated clinically significant infection event process
Primary finalized: No
Mathematical model: type: observational overlap-weighted/g-method precision simulation with effective independent units, outcome/event frequency and residual-bias scenarios; formulas: W_i(t) = product_(k<=t) P(A_ik | A_i,<k,X_i0)/P(A_ik | H_ik), optionally multiplied by separately justified observation weights.
m_a = E_X[m(a,X)] for a baseline strategy, or the sequential g-formula for sustained strategies; Delta=m_1-m_0. Fit a weighted, baseline-aligned mean/risk model matching Y rather than assume the treatment coefficient is causal.
Model recurrent N_i(t) jointly with terminal death or report mean cumulative infection count M_a(t)=E[N^a(t)] with deaths retained under an explicit terminal-event strategy; incidence density alone may change with survival.; variables: Y: adjudicated clinically significant infection event process; i: participant; t: recurrent clinically significant infection and exposure interval; the registered visit/time-window index, not another independent unit; A: clinician-selected new off-label mTOR initiation versus eligible noninitiation; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: standardized clinically significant infection rate or cumulative count under specified observed strategies, conditional on exchangeability; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; H ik: measured predecision history including prior treatment and time-varying covariates; m(a,X): outcome mean/risk model under specified treatment strategy and observed covariates; m a: standardized mean/risk under a, identified only under the listed assumptions; k: ordered treatment decision time; N i(t): cumulative adjudicated clinically significant infections by t; M a(t): strategy-specific cumulative infection count under the chosen death strategy; family code: OBS_RATE
Sample size design: family: observational overlap-weighted/g-method precision simulation with effective independent units, outcome/event frequency and residual-bias scenarios; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure
recurrent-event intensity and overdispersion
death/competing event rate
observation opportunity and exposure history; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Time zero, switching, indication and health behavior matter; death terminates infection opportunity and is not ordinary independent dropout.
Common eligibility/time zero and a defensible active comparator; treatment choice is not randomized.
Conditional exchangeability (including time-varying feedback where relevant), positivity and correct observation/treatment/outcome models are untestable or only partly testable; observed overlap does not prove them.
Do not adjust for posttreatment variables as baseline confounders; define a separate mediation/per-protocol estimand before using g-methods.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: The selection hypothesis weakens if consistent infection benefit remains after robust adjustment and independent-site replication.; preanalysis gate: Time zero, switching, indication and health behavior matter; death terminates infection opportunity and is not ordinary independent dropout.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Time zero, switching, indication and health behavior matter; death terminates infection opportunity and is not ordinary independent dropout.
Overlap/effective-sample-size and weight-tail diagnostics with declared truncation sensitivity; report population changes caused by trimming.
Quantitative unmeasured-confounding and informative-loss sensitivity; negative controls/falsification windows where scientifically defensible.
Compare g-computation/weighting/doubly robust estimates on the same target population and time zero, without interpreting agreement as proof.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Time zero, switching, indication and health behavior matter; death terminates infection opportunity and is not ordinary independent dropout.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/198; question: Are immune-health claims about off-label mTOR inhibition consistent with prospectively observed infection and discontinuation outcomes?; aims: Establish a multiclinic registry before treatment decisions.
Emulate a new-user trial against comparable patients deferring treatment.
Compare laboratory changes with adjudicated infection and discontinuation outcomes.; design verbatim planning only: Recruit a planning cohort of 350 adults evaluated at clinics where off-label prescribing already occurs, independent of study participation. Follow initiators and clinically comparable noninitiators for 18 months, recording indication, cointerventions, routine laboratory monitoring, infections, and function. Researchers neither recommend nor supply medication.; outcomes verbatim: Primary: clinically significant infection rate; secondary: treatment discontinuation, adverse events, function, and exploratory immune biomarkers.; analysis verbatim inherited not authoritative if refined: Specify time zero, eligibility, and exposure strategies prospectively; use overlap weighting, time-varying treatment models, and sensitivity analyses for unmeasured health behavior.; falsification verbatim: The selection hypothesis weakens if consistent infection benefit remains after robust adjustment and independent-site replication.; source ids unverified leads: pearl-rapamycin-healthspan-rct-2025
immunosenescence-microbiota-mtor-review-2026
fda-off-label-use-2026; prospectus precedence: No

F20 — Postviral microbiome signatures: consequence, predictor, or mediator?

Rank: 224
Cluster: Immune aging, infection and postviral recovery
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: WITHIN_PRED; allocation or exposure unit: natural microbiome/diet/care trajectories during postviral recovery; independent sampling unit: participant with clinic-held-out validation; repeated measurement unit: stool/metabolite and later functional visit; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: association or prospective prediction; temporal order and model adjustment alone do not identify a mechanism or treatment effect; specific constraint: Prediction is distinct from a longitudinal causal microbiome effect; illness affects diet, sampling and antibiotics, creating time-varying feedback.
Primary estimand: incremental prospective prediction of later functional recovery using only earlier microbial features
Candidate primary outcome: later prespecified functional-recovery score
Primary finalized: No
Mathematical model: type: independent-person/site prospective prediction-gain and calibration precision, with within-person dependence; formulas: Y_it = alpha + u_i + beta_W (B_it-Bbar_i) + beta_B Bbar_i + gamma^T X_it + f_time(t) + b_study + epsilon_it; u_i is a random intercept with a declared conditional mean/covariance model, not unrestricted fixed person effects.
B_observed,it = B_true,it + e_it; estimate measurement variance from replicates/validation or report sensitivity rather than assume it known.
Primary prediction target: Gain=mean_test[L(Y,f_base(X))-L(Y,f_new(X,B_history))], with locked models and proper outcome-specific loss; report calibration as well as gain.
For new held-out persons/sites, predict using the training-estimated population random-effect distribution or only genuinely available preprediction calibration; never fit a person intercept using that test person's future outcome.; variables: Y: later prespecified functional-recovery score; i: participant with clinic-held-out validation; t: stool/metabolite and later functional visit; the registered visit/time-window index, not another independent unit; A: natural microbiome/diet/care trajectories during postviral recovery; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: incremental prospective prediction of later functional recovery using only earlier microbial features; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; B it: prespecified compositional microbial functional features measured before the later recovery target; Bbar i: person/donor mean over a prespecified observation schedule; beta W,beta B: within-unit and between-unit association coefficients, not causal effects; X it: explicitly time-ordered baseline/premeasurement covariates; posttreatment adjustment changes interpretation; f time(t): prespecified sampling/recovery/calendar trajectory; b study: study/site/batch effect where relevant; B true,e it: latent predictor and measurement error, requiring identification constraints; Gain: held-out improvement over the complete prespecified routine/sleep/autonomic or clinical baseline; L: registered proper/validated outcome-specific loss; f base,f new: models frozen in development, with all feature selection internal to training; B history: predictor history available before the registered prediction time; family code: WITHIN_PRED
Sample size design: family: independent-person/site prospective prediction-gain and calibration precision, with within-person dependence; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Prediction is distinct from a longitudinal causal microbiome effect; illness affects diet, sampling and antibiotics, creating time-varying feedback.
Associations need a clear sampling/time frame; causal interpretation is not supplied by this model.
Measurement error, reverse causation, time-varying confounding and selected observation may remain despite adjustment.
All feature selection and time normalization occur inside training splits; later outcome windows are untouched and only information available at prediction is used.
For the secondary within/between association model, specify the conditional distribution of random intercept u_i; fixed person effects cannot separately estimate beta_B for time-invariant person means.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Reject the residual-signature hypothesis if associations vanish under dietary adjustment or fail at the held-out clinic.; preanalysis gate: Prediction is distinct from a longitudinal causal microbiome effect; illness affects diet, sampling and antibiotics, creating time-varying feedback.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Prediction is distinct from a longitudinal causal microbiome effect; illness affects diet, sampling and antibiotics, creating time-varying feedback.
Vary lag order/window and measurement-error reliability within a prespecified family; include negative-control windows when meaningful.
Separate within- and between-unit effects, missingness patterns and source-study replication; do not use cell/assay counts as independent donor counts.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Prediction is distinct from a longitudinal causal microbiome effect; illness affects diet, sampling and antibiotics, creating time-varying feedback.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/223; question: Do microbiome changes precede postviral recovery after accounting for diet, antibiotics, and illness-driven behavior?; aims: Measure microbiome, dietary, medication, and symptom trajectories together.
Test lagged prediction against reverse-direction models.
Validate the strongest restricted signature in an independent cohort.; design verbatim planning only: Enroll approximately 250 adults early in persistent postviral illness and 100 recovered comparators from two clinics. Obtain repeated noninvasive samples, diet records, medication histories, and functional assessments over one year. Compare within-person changes and hold out one clinic for locked validation; no microbial products are administered.; outcomes verbatim: Primary: incremental prediction of later functional recovery; secondary: metabolite concordance, symptom burden, and reverse temporal associations.; analysis verbatim inherited not authoritative if refined: Use prespecified compositional methods, marginal structural models for time-varying confounding, and false-discovery control. Report residual confounding and avoid equating prediction with mediation.; falsification verbatim: Reject the residual-signature hypothesis if associations vanish under dietary adjustment or fail at the held-out clinic.; source ids unverified leads: gut-brain-dysbiosis-review-2026
immunosenescence-microbiota-mtor-review-2026
nr-long-covid-rct-2025; prospectus precedence: No

F21 — Therapeutic stimulant benefit beyond perceived productivity

Rank: 190
Cluster: Clinical repurposing and neuropsychiatric claims
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: OBS_LONG_INT; allocation or exposure unit: clinician-selected new ADHD stimulant, alternative strategy or deferred medicine; independent sampling unit: participant; repeated measurement unit: sleep/function/treatment assessment; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: observational strategy association; a causal interpretation additionally requires sequential exchangeability, positivity and valid treatment/observation models; specific constraint: Postinitiation sleep/autonomic change can mediate or confound later treatment; a baseline-sleep interaction differs from mediation.
Primary estimand: overlap-population functional-effect contrast and baseline-sleep modification under defined strategies, conditional on exchangeability
Candidate primary outcome: validated daily functional-impairment score at nine months
Primary finalized: No
Mathematical model: type: interaction-specific precision/power with modifier distribution and reliability; simulate the actual interaction; formulas: E(Y_it | A_i, B_i, X_i) = alpha_t + tau_t A_i + eta_t B_i + kappa_t A_i B_i + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Treatment contrast at baseline B=b is tau_t + kappa_t b; the interaction target is kappa_t, not a difference in within-subgroup p-values.
W_i(t) = product_(k<=t) P(A_ik | A_i,<k, X_i0) / P(A_ik | H_ik), multiplied by observation/censoring weights if their assumptions are defensible.
For baseline overlap target define h(X)=[sum_a 1/e_a(X)]^(-1), e_a(X)=Pr(A=a|X); baseline arm weight is h(X)/e_A(X). For binary treatment the target tilt is h(X)=e(X)[1-e(X)] and arm weights remain h(X)/e_A(X). Standardize effects/moderation to this h-tilted population; sequential weights, if justified, are additional and do not themselves define overlap.; variables: Y: validated daily functional-impairment score at nine months; i: participant; t: sleep/function/treatment assessment; the registered visit/time-window index, not another independent unit; A: clinician-selected new ADHD stimulant, alternative strategy or deferred medicine; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: overlap-population functional-effect contrast and baseline-sleep modification under defined strategies, conditional on exchangeability; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; B: baseline sleep measure fixed before independently selected ADHD care starts; centering/coding and any measurement-error model fixed before evaluation; eta: baseline modifier association; kappa: assignment-by-baseline-modifier interaction coefficient on the stated scale; H ik: observed predecision treatment/covariate history at k, excluding future measurements; e a(X): baseline propensity for the actual two/multiple treatment strategies; h(X): prespecified common-support target density tilt; normalization and multiarm comparisons fixed before analysis; family code: OBS_LONG_INT
Sample size design: family: interaction-specific precision/power with modifier distribution and reliability; simulate the actual interaction; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure
modifier prevalence/distribution and variance
modifier reliability and overlap
interaction size or precision target; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Postinitiation sleep/autonomic change can mediate or confound later treatment; a baseline-sleep interaction differs from mediation.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The modifier must be measured before assignment/initiation; interaction is scale dependent and requires support within modifier strata.
Apply the observational longitudinal weighting/g-formula conditions; the displayed regression alone does not eliminate treatment-selection or treatment-confounder feedback bias.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Reject the proposed mismatch if productivity closely tracks independent function and sleep does not modify benefit.; preanalysis gate: Postinitiation sleep/autonomic change can mediate or confound later treatment; a baseline-sleep interaction differs from mediation.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Postinitiation sleep/autonomic change can mediate or confound later treatment; a baseline-sleep interaction differs from mediation.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Vary modifier measurement-error assumptions and retain a continuous modifier unless a cutoff was independently fixed.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Fix overlap target and baseline treatment coding, then decide whether/how sequential treatment weights preserve that target.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Postinitiation sleep/autonomic change can mediate or confound later treatment; a baseline-sleep interaction differs from mediation.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/189; question: How do sleep and autonomic changes modify functional benefit after clinician-initiated ADHD stimulant treatment?; aims: Characterize sleep and functional impairment before treatment decisions.
Estimate longitudinal benefit and adverse-outcome differences between clinical treatment strategies.
Validate whether perceived productivity tracks functional change.; design verbatim planning only: Prospectively follow approximately 300 newly assessed adults with ADHD whose clinicians independently choose stimulant treatment, another indicated strategy, or deferred medication. Collect baseline and nine-month validated symptom, sleep, blood-pressure, and functional assessments. Emulate a new-user trial with comparable eligibility and time zero; numbers are planning assumptions.; outcomes verbatim: Primary: daily functional impairment; secondary: ADHD symptoms, sleep, cardiovascular tolerability, and productivity-rating calibration.; analysis verbatim inherited not authoritative if refined: Adjust severity, comorbidity, occupation, and treatment preference using overlap weighting; model switching and informative dropout. Prespecify the sleep interaction and outcome hierarchy.; falsification verbatim: Reject the proposed mismatch if productivity closely tracks independent function and sleep does not modify benefit.; source ids unverified leads: neurofeedback-adhd-meta-2025
fda-off-label-use-2026
digital-phenotyping-relapse-review-2026; prospectus precedence: No

F22 — Nicotine cognition during cessation: withdrawal relief or durable gain?

Rank: 144
Cluster: Clinical repurposing and neuropsychiatric claims
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: OBS_LONG; allocation or exposure unit: routine nicotine-replacement or non-nicotine cessation care; independent sampling unit: participant enrolled before care pathway; repeated measurement unit: fixed six-month attention test plus earlier visits; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: observational comparison/association; causal interpretation is conditional on explicit exchangeability, positivity, consistency and observation assumptions; specific constraint: Do not restrict primary analysis to abstainers; abstinence and withdrawal are time-varying selected consequences; no never-user nicotine-exposure inference.
Primary estimand: full-cohort six-month attention change and treatment-pathway association including relapse
Candidate primary outcome: sustained-attention score at the fixed six-month assessment
Primary finalized: No
Mathematical model: type: observational overlap-weighted/g-method precision simulation with effective independent units, outcome/event frequency and residual-bias scenarios; formulas: W_i(t) = product_(k<=t) P(A_ik | A_i,<k,X_i0)/P(A_ik | H_ik), optionally multiplied by separately justified observation weights.
m_a = E_X[m(a,X)] for a baseline strategy, or the sequential g-formula for sustained strategies; Delta=m_1-m_0. Fit a weighted, baseline-aligned mean/risk model matching Y rather than assume the treatment coefficient is causal.; variables: Y: sustained-attention score at the fixed six-month assessment; i: participant enrolled before care pathway; t: fixed six-month attention test plus earlier visits; the registered visit/time-window index, not another independent unit; A: routine nicotine-replacement or non-nicotine cessation care; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: full-cohort six-month attention change and treatment-pathway association including relapse; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; H ik: measured predecision history including prior treatment and time-varying covariates; m(a,X): outcome mean/risk model under specified treatment strategy and observed covariates; m a: standardized mean/risk under a, identified only under the listed assumptions; k: ordered treatment decision time; family code: OBS_LONG
Sample size design: family: observational overlap-weighted/g-method precision simulation with effective independent units, outcome/event frequency and residual-bias scenarios; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Do not restrict primary analysis to abstainers; abstinence and withdrawal are time-varying selected consequences; no never-user nicotine-exposure inference.
Common eligibility/time zero and a defensible active comparator; treatment choice is not randomized.
Conditional exchangeability (including time-varying feedback where relevant), positivity and correct observation/treatment/outcome models are untestable or only partly testable; observed overlap does not prove them.
Do not adjust for posttreatment variables as baseline confounders; define a separate mediation/per-protocol estimand before using g-methods.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Durable attention gains weakly related to withdrawal under prespecified longitudinal and selection sensitivity analyses would weaken withdrawal relief as the dominant explanation. Gains observed only among successful abstainers would not by themselves distinguish durable nicotine benefit from cessation-related or selection effects.; preanalysis gate: Do not restrict primary analysis to abstainers; abstinence and withdrawal are time-varying selected consequences; no never-user nicotine-exposure inference.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Do not restrict primary analysis to abstainers; abstinence and withdrawal are time-varying selected consequences; no never-user nicotine-exposure inference.
Overlap/effective-sample-size and weight-tail diagnostics with declared truncation sensitivity; report population changes caused by trimming.
Quantitative unmeasured-confounding and informative-loss sensitivity; negative controls/falsification windows where scientifically defensible.
Compare g-computation/weighting/doubly robust estimates on the same target population and time zero, without interpreting agreement as proof.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Do not restrict primary analysis to abstainers; abstinence and withdrawal are time-varying selected consequences; no never-user nicotine-exposure inference.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/143; question: Does perceived cognitive enhancement during nicotine treatment reflect withdrawal relief rather than durable improvement?; aims: Measure cognition, withdrawal, and sleep before a cessation attempt.
Follow trajectories during clinician-selected evidence-based cessation care.
Compare early perceived benefits with cognition after sustained abstinence.; design verbatim planning only: Enroll approximately 250 nicotine-dependent adults already entering cessation services. Observe routine clinician-selected nicotine-replacement and non-nicotine treatment pathways without assigning products. Assess cognition at baseline, early cessation and a fixed six-month visit for everyone, including participants who relapse. Compare within-person trajectories and clinically comparable treatment groups; never-users are not exposed.; outcomes verbatim: Primary: Sustained-attention change from baseline to the fixed six-month assessment in the full enrolled cohort. Secondary: Attention trajectories among biochemically confirmed sustained abstainers, withdrawal, perceived focus, sleep, dependence and cessation success.; analysis verbatim inherited not authoritative if refined: Use longitudinal mediation with explicit assumptions and time-varying smoking exposure. Weight for baseline treatment selection and informative assessment loss; retain relapse as an observed time-varying event. Label stable-abstainer analyses as selected subgroup descriptions unless a defensible selection estimand and sensitivity analysis are specified. Prespecify cognitive-domain multiplicity.; falsification verbatim: Durable attention gains weakly related to withdrawal under prespecified longitudinal and selection sensitivity analyses would weaken withdrawal relief as the dominant explanation. Gains observed only among successful abstainers would not by themselves distinguish durable nicotine benefit from cessation-related or selection effects.; source ids unverified leads: nicotine-patch-cognition-meta-2021
nicotine-cognition-review-2024
fda-smoking-cessation-medicines-2025
fda-nicotine-addiction-2025; prospectus precedence: No

F23 — Functional unblinding in supervised psychedelic trials

Rank: 200
Cluster: Clinical repurposing and neuropsychiatric claims
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: UNBLIND; allocation or exposure unit: original legally authorized trial randomization; independent sampling unit: participant within independent trial and indication; repeated measurement unit: symptom visit, expectancy and allocation-guess report; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized total-effect estimation plus assumption-indexed partial-identification/sensitivity analysis; pharmacology and expectancy are not automatically separable; specific constraint: Allocation guess is postassignment; regression adjustment for guesses does not isolate pharmacology or expectancy; do not pool unrelated indications as one endpoint.
Primary estimand: randomized indication-specific symptom effect and its identified/sensitivity interval under explicit unblinding assumptions
Candidate primary outcome: indication-specific symptom score at the registered trial endpoint
Primary finalized: No
Mathematical model: type: original trial total-effect precision plus sensitivity/partial-identification interval width across unblinding assumptions; formulas: Within each compatible indication/trial estimate Delta_total=E[Y|randomized active]-E[Y|randomized comparator], with baseline adjustment and original allocation.
For an explicitly postulated additive differential reporting/measurement-bias parameter b in [b_L,b_U], the sensitivity contrast lies in [Delta_total-b_U,Delta_total-b_L], with sampling uncertainty also propagated. This does not by itself identify a pharmacological direct effect.
Postrandomization treatment guesses can be affected by symptom response and sensations; no ordinary adjustment for guesses or expectancy establishes causal separation.; variables: Y: indication-specific symptom score at the registered trial endpoint; i: participant within independent trial and indication; t: symptom visit, expectancy and allocation-guess report; the registered visit/time-window index, not another independent unit; A: original legally authorized trial randomization; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: randomized indication-specific symptom effect and its identified/sensitivity interval under explicit unblinding assumptions; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; Delta total: original randomized total effect on the fixed indication-specific symptom scale; b: explicitly assumed differential reporting/measurement contribution on the same outcome scale, not estimated as a known drug mechanism; b L,b U: externally justified sensitivity bounds; values and interpretation require registration; family code: UNBLIND
Sample size design: family: original trial total-effect precision plus sensitivity/partial-identification interval width across unblinding assumptions; formula: Use design-specific simulation or independent-unit resampling for the stated estimand and target confidence width/power; preserve all allocation, censoring, clustering and measurement-error structure.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Allocation guess is postassignment; regression adjustment for guesses does not isolate pharmacology or expectancy; do not pool unrelated indications as one endpoint.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Outcome/indication compatibility, baseline expectancy timing and a causal diagram are required; posttreatment unblinding measures can be both mediators and consequences of response.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Reject strong expectancy sensitivity if clinically meaningful benefit remains stable across plausible bounds and independent assessments.; preanalysis gate: Allocation guess is postassignment; regression adjustment for guesses does not isolate pharmacology or expectancy; do not pool unrelated indications as one endpoint.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Allocation guess is postassignment; regression adjustment for guesses does not isolate pharmacology or expectancy; do not pool unrelated indications as one endpoint.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Allocation guess is postassignment; regression adjustment for guesses does not isolate pharmacology or expectancy; do not pool unrelated indications as one endpoint.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/199; question: How much do treatment guesses and expectancy alter estimated psychiatric benefit in supervised psychedelic research?; aims: Harmonize expectancy and allocation-guess measures in accessible supervised trials.
Compare outcome patterns across subjective and independently assessed endpoints.
Develop and prospectively validate an unblinding sensitivity framework.; design verbatim planning only: Seek individual data from several completed, legally authorized psychiatric trials and prospectively embed assessment in an existing supervised trial, targeting approximately 150 participants if access permits. Preserve original randomization, psychological support, and clinical protocols; compare randomized arms without changing treatment delivery.; outcomes verbatim: Primary: robustness of indication-specific symptom effects to unblinding assumptions; secondary: functioning, adverse events, and assessor agreement.; analysis verbatim inherited not authoritative if refined: Use causal diagrams, baseline expectancy stratification, partial-identification bounds, and outcome-family multiplicity control. Avoid interpreting post-randomization adjustment as definitive causal separation.; falsification verbatim: Reject strong expectancy sensitivity if clinically meaningful benefit remains stable across plausible bounds and independent assessments.; source ids unverified leads: psychedelic-living-review-2025
belief-effects-health-review-2026
nocebo-communication-review-2026; prospectus precedence: No

F24 — Do microdosing claims transfer beyond practiced cognitive tasks?

Rank: 232
Cluster: Clinical repurposing and neuropsychiatric claims
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: META_RCT; allocation or exposure unit: original randomized controlled microdosing trial exposure, if lawful data access obtained; independent sampling unit: independent trial and participant; repeated measurement unit: unpracticed cognitive-task item and follow-up; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Additional supervised trial is conditional on access; task transfer and ordinary daily function are distinct; small-study/publication sensitivity needed.
Primary estimand: randomized effect on fixed unpracticed cognitive-control composite, synthesized only across compatible trials
Candidate primary outcome: externally scaled unpracticed cognitive-control composite
Primary finalized: No
Mathematical model: type: individual-trial and compatible-trial synthesis precision; independent trial count and selection sensitivity; formulas: E(Y_it | A_i, X_i, Y_i0) = alpha_t + z(A_i)^T theta_t + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Delta = sum_t w_t c^T theta_t; estimate by baseline-adjusted marginal standardization, with unit/cluster-appropriate inference.
Delta_hat_s | Delta_s ~ Normal(Delta_s, SE_s^2); Delta_s ~ Normal(mu_Delta, tau_trial^2), only for outcome-compatible trials.; variables: Y: externally scaled unpracticed cognitive-control composite; i: independent trial and participant; t: unpracticed cognitive-task item and follow-up; the registered visit/time-window index, not another independent unit; A: original randomized controlled microdosing trial exposure, if lawful data access obtained; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: randomized effect on fixed unpracticed cognitive-control composite, synthesized only across compatible trials; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; z(A): prespecified indicators for the actual randomized arms; two arms reduce to one indicator; s: independent trial; SE s: sampling standard error of trial-specific randomized contrast; tau trial^2: between-trial effect variance; mu Delta: mean compatible-trial assignment effect; family code: META_RCT
Sample size design: family: individual-trial and compatible-trial synthesis precision; independent trial count and selection sensitivity; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Additional supervised trial is conditional on access; task transfer and ordinary daily function are distinct; small-study/publication sensitivity needed.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Reject the hypothesis if reproducible improvements appear on held-out tasks and daily function without cognitive-control costs.; preanalysis gate: Additional supervised trial is conditional on access; task transfer and ordinary daily function are distinct; small-study/publication sensitivity needed.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Additional supervised trial is conditional on access; task transfer and ordinary daily function are distinct; small-study/publication sensitivity needed.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Additional supervised trial is conditional on access; task transfer and ordinary daily function are distinct; small-study/publication sensitivity needed.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/231; question: Do cognitive changes reported in controlled microdosing research generalize to unfamiliar tasks and daily functioning?; aims: Map controlled-study task families and reporting completeness.
Reanalyze accessible data using practice-aware hierarchical models.
Embed transfer assessments in an independently authorized supervised study.; design verbatim planning only: Conduct a registered evidence synthesis and individual-data analysis, then partner with an existing legal, ethics-approved supervised trial for approximately 100 participants if feasible. Compare its randomized exposure groups using additional nonoperational cognitive assessments and follow-up; investigators do not recruit or direct informal self-dosing.; outcomes verbatim: Primary: unpracticed cognitive-control composite; secondary: near transfer, daily-function ratings, expectancy, and adverse experiences.; analysis verbatim inherited not authoritative if refined: Model task-level clustering and repeated practice, use robust small-study sensitivity analyses, and correct task-family multiplicity. Separate confirmatory transfer outcomes from exploratory creativity measures.; falsification verbatim: Reject the hypothesis if reproducible improvements appear on held-out tasks and daily function without cognitive-control costs.; source ids unverified leads: psychedelic-microdosing-cognition-meta-2026
psychedelic-living-review-2025
belief-effects-health-review-2026; prospectus precedence: No

F25 — Esketamine response, dissociation, and durable daily function

Rank: 126
Cluster: Clinical repurposing and neuropsychiatric claims
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: PRED; allocation or exposure unit: clinician-selected esketamine or specialist comparator care; independent sampling unit: participant with clinic-held-out validation; repeated measurement unit: early-response landmark and later function visit; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: prediction, calibration or model transportability; prediction gain is not clinical efficacy, mechanistic necessity or causal treatment selection; specific constraint: Fix a landmark, include only predictors available then and account for early dropout; dissociation is not randomized; comparative treatment effects remain observational.
Primary estimand: calibration and incremental prediction of six-month functional recovery from frozen early-response variables
Candidate primary outcome: prespecified six-month functional-recovery status or score
Primary finalized: No
Mathematical model: type: external/nested predictive-validation precision with target prevalence/variance, independent-unit holdout coverage and model complexity; formulas: Gain = mean_test[L(Y,f_base(X)) - L(Y,f_new(X))], with both predictions frozen before test outcomes; bootstrap independent held-out units.
For a binary target assess E[Y | p_hat=p] versus p and calibration intercept/slope logit Pr(Y=1)=a_cal+b_cal logit(p_hat); for a continuous target use Y=a_cal+b_cal y_hat+error plus scale-appropriate calibration/loss.; variables: Y: prespecified six-month functional-recovery status or score; i: participant with clinic-held-out validation; t: early-response landmark and later function visit; the registered visit/time-window index, not another independent unit; A: clinician-selected esketamine or specialist comparator care; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: calibration and incremental prediction of six-month functional recovery from frozen early-response variables; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; L: one proper or scientifically justified primary loss/scoring rule fixed for the actual outcome type; positive Gain favors new model; f base,f new: locked baseline/new prediction functions using only information available at the registered prediction time; test: untouched person/family/lab/site/time set as specified per thesis; p hat,y hat: frozen probability or continuous prediction; a cal,b cal: calibration intercept and slope, estimated on evaluation data for assessment; using them to update the model requires another evaluation set; family code: PRED
Sample size design: family: external/nested predictive-validation precision with target prevalence/variance, independent-unit holdout coverage and model complexity; formula: Simulate or bootstrap the independent-unit test loss/calibration/decision criterion; plan cases/events, noncases and independent sites/families/programs separately. No universal events-per-variable or participant N is asserted.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Fix a landmark, include only predictors available then and account for early dropout; dissociation is not randomized; comparative treatment effects remain observational.
Train/tune/test separation at the actual biological/participant/family/time level; no future, postoutcome or target-derived predictors.
Validation support must cover intended deployment; calibration and missing/failing measurements are reported with discrimination/loss.
A validation subset selected by culture success, assay viability, consent or data completeness is a different target unless the full strategy is explicitly evaluated.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Reject the proposed prognostic limit if early symptom and dissociation measures accurately calibrate durable function across sites.; preanalysis gate: Fix a landmark, include only predictors available then and account for early dropout; dissociation is not randomized; comparative treatment effects remain observational.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Fix a landmark, include only predictors available then and account for early dropout; dissociation is not randomized; comparative treatment effects remain observational.
Repeat loss/calibration with independent-cluster resampling, prespecified subgroup/transport shifts and plausible missingness.
Compare fixed simple baselines, calibration and uncertainty coverage; report abstention/failure/turnaround rather than deleting hard cases.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Frozen prediction time/horizon, proper loss, baseline model, feature availability, independent split/holdout and target deployment population.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Fix a landmark, include only predictors available then and account for early dropout; dissociation is not randomized; comparative treatment effects remain observational.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/125; question: Does early symptom improvement during supervised esketamine care predict sustained functional recovery independently of dissociation?; aims: Measure acute effects and baseline expectations prospectively.
Track symptom, functional, and treatment-continuation trajectories.
Validate a parsimonious early prediction model across clinics.; design verbatim planning only: Follow approximately 280 adults receiving clinically indicated esketamine within licensed monitoring pathways at three clinics for one year. Include a clinically comparable group receiving another specialist treatment for resistant depression. Researchers observe existing care and compare locked early-response models in a held-out clinic.; outcomes verbatim: Primary: six-month functional recovery; secondary: symptom remission, discontinuation, cognition, dissociation, and serious adverse events.; analysis verbatim inherited not authoritative if refined: Use treatment-selection weighting, joint dropout models, and site-held-out validation; adjust concomitant antidepressants and psychosocial care. Prespecify predictor selection and outcome multiplicity.; falsification verbatim: Reject the proposed prognostic limit if early symptom and dissociation measures accurately calibrate durable function across sites.; source ids unverified leads: ketamine-esketamine-ipd-review-2025
fda-spravato-label-2024
digital-phenotyping-relapse-review-2026; prospectus precedence: No

F26 — Metformin, B12 status, and apparent cognitive aging

Rank: 145
Cluster: Clinical repurposing and neuropsychiatric claims
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: OBS_LONG; allocation or exposure unit: metformin new initiation for separately specified indications versus matched noninitiation; independent sampling unit: participant; repeated measurement unit: B12/function/medication visit; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: observational comparison/association; causal interpretation is conditional on explicit exchangeability, positivity, consistency and observation assumptions; specific constraint: Do not pool indicated and off-label initiators without target-population justification; B12 treatment is time-varying usual care, not withheld or randomized.
Primary estimand: indication-specific standardized fatigue-related function contrast under initiation strategies, conditional on exchangeability
Candidate primary outcome: fatigue-related functional score over eighteen months
Primary finalized: No
Mathematical model: type: observational overlap-weighted/g-method precision simulation with effective independent units, outcome/event frequency and residual-bias scenarios; formulas: W_i(t) = product_(k<=t) P(A_ik | A_i,<k,X_i0)/P(A_ik | H_ik), optionally multiplied by separately justified observation weights.
m_a = E_X[m(a,X)] for a baseline strategy, or the sequential g-formula for sustained strategies; Delta=m_1-m_0. Fit a weighted, baseline-aligned mean/risk model matching Y rather than assume the treatment coefficient is causal.; variables: Y: fatigue-related functional score over eighteen months; i: participant; t: B12/function/medication visit; the registered visit/time-window index, not another independent unit; A: metformin new initiation for separately specified indications versus matched noninitiation; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: indication-specific standardized fatigue-related function contrast under initiation strategies, conditional on exchangeability; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; H ik: measured predecision history including prior treatment and time-varying covariates; m(a,X): outcome mean/risk model under specified treatment strategy and observed covariates; m a: standardized mean/risk under a, identified only under the listed assumptions; k: ordered treatment decision time; family code: OBS_LONG
Sample size design: family: observational overlap-weighted/g-method precision simulation with effective independent units, outcome/event frequency and residual-bias scenarios; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Do not pool indicated and off-label initiators without target-population justification; B12 treatment is time-varying usual care, not withheld or randomized.
Common eligibility/time zero and a defensible active comparator; treatment choice is not randomized.
Conditional exchangeability (including time-varying feedback where relevant), positivity and correct observation/treatment/outcome models are untestable or only partly testable; observed overlap does not prove them.
Do not adjust for posttreatment variables as baseline confounders; define a separate mediation/per-protocol estimand before using g-methods.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Reject the deficiency pathway if B12 trajectories do not predict symptoms or correction is unrelated to later improvement under sensitivity analysis.; preanalysis gate: Do not pool indicated and off-label initiators without target-population justification; B12 treatment is time-varying usual care, not withheld or randomized.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Do not pool indicated and off-label initiators without target-population justification; B12 treatment is time-varying usual care, not withheld or randomized.
Overlap/effective-sample-size and weight-tail diagnostics with declared truncation sensitivity; report population changes caused by trimming.
Quantitative unmeasured-confounding and informative-loss sensitivity; negative controls/falsification windows where scientifically defensible.
Compare g-computation/weighting/doubly robust estimates on the same target population and time zero, without interpreting agreement as proof.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Do not pool indicated and off-label initiators without target-population justification; B12 treatment is time-varying usual care, not withheld or randomized.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/144; question: Does B12 depletion explain part of cognitive or fatigue change attributed to off-label metformin use?; aims: Establish baseline B12, cognition, fatigue, and indication.
Compare trajectories in new users and matched nonusers.
Test whether clinically managed deficiency correction alters the observed association.; design verbatim planning only: Enroll approximately 350 adults beginning metformin for established or independently chosen off-label indications, plus 200 matched noninitiators. Observe routine care for 18 months with baseline and repeated functional, cognitive, and laboratory assessments. Record clinician-directed deficiency treatment without assigning medication or withholding supplementation.; outcomes verbatim: Primary: fatigue-related functional change; secondary: validated cognition, B12 status, metabolic markers, and treatment discontinuation.; analysis verbatim inherited not authoritative if refined: Use indication-stratified target-trial emulation, time-varying confounding adjustment, and sensitivity analyses for diet, renal function, and health behavior; prespecify mediation and outcome hierarchies.; falsification verbatim: Reject the deficiency pathway if B12 trajectories do not predict symptoms or correction is unrelated to later improvement under sensitivity analysis.; source ids unverified leads: nih-ods-b12-2025
metformin-antiaging-uncertainty-2025
fda-off-label-use-2026; prospectus precedence: No

F27 — Next-day function after off-label sedative prescribing

Rank: 112
Cluster: Clinical repurposing and neuropsychiatric claims
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: OBS_LONG_INT; allocation or exposure unit: clinician-selected new trazodone versus CBT-I initiation; independent sampling unit: participant; repeated measurement unit: daytime-function and insomnia visit; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: observational strategy association; a causal interpretation additionally requires sequential exchangeability, positivity and valid treatment/observation models; specific constraint: Access, mood and preferences may confound comparator choice; switching and falls cannot be dismissed as missingness; subjective sleep is a secondary outcome.
Primary estimand: overlap-population daytime-impairment contrast and baseline-sleepiness interaction under defined strategies
Candidate primary outcome: validated daytime functional-impairment score at six months
Primary finalized: No
Mathematical model: type: interaction-specific precision/power with modifier distribution and reliability; simulate the actual interaction; formulas: E(Y_it | A_i, B_i, X_i) = alpha_t + tau_t A_i + eta_t B_i + kappa_t A_i B_i + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Treatment contrast at baseline B=b is tau_t + kappa_t b; the interaction target is kappa_t, not a difference in within-subgroup p-values.
W_i(t) = product_(k<=t) P(A_ik | A_i,<k, X_i0) / P(A_ik | H_ik), multiplied by observation/censoring weights if their assumptions are defensible.
For baseline overlap target define h(X)=[sum_a 1/e_a(X)]^(-1), e_a(X)=Pr(A=a|X); baseline arm weight is h(X)/e_A(X). For binary treatment the target tilt is h(X)=e(X)[1-e(X)] and arm weights remain h(X)/e_A(X). Standardize effects/moderation to this h-tilted population; sequential weights, if justified, are additional and do not themselves define overlap.; variables: Y: validated daytime functional-impairment score at six months; i: participant; t: daytime-function and insomnia visit; the registered visit/time-window index, not another independent unit; A: clinician-selected new trazodone versus CBT-I initiation; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: overlap-population daytime-impairment contrast and baseline-sleepiness interaction under defined strategies; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; B: baseline sleepiness score before trazodone/CBT-I initiation; centering/coding and any measurement-error model fixed before evaluation; eta: baseline modifier association; kappa: assignment-by-baseline-modifier interaction coefficient on the stated scale; H ik: observed predecision treatment/covariate history at k, excluding future measurements; e a(X): baseline propensity for the actual two/multiple treatment strategies; h(X): prespecified common-support target density tilt; normalization and multiarm comparisons fixed before analysis; family code: OBS_LONG_INT
Sample size design: family: interaction-specific precision/power with modifier distribution and reliability; simulate the actual interaction; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure
modifier prevalence/distribution and variance
modifier reliability and overlap
interaction size or precision target; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Access, mood and preferences may confound comparator choice; switching and falls cannot be dismissed as missingness; subjective sleep is a secondary outcome.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The modifier must be measured before assignment/initiation; interaction is scale dependent and requires support within modifier strata.
Apply the observational longitudinal weighting/g-formula conditions; the displayed regression alone does not eliminate treatment-selection or treatment-confounder feedback bias.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Reject discordance if sleep improvements consistently predict daytime function equally across clinical strategies.; preanalysis gate: Access, mood and preferences may confound comparator choice; switching and falls cannot be dismissed as missingness; subjective sleep is a secondary outcome.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Access, mood and preferences may confound comparator choice; switching and falls cannot be dismissed as missingness; subjective sleep is a secondary outcome.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Vary modifier measurement-error assumptions and retain a continuous modifier unless a cutoff was independently fixed.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Fix overlap target and baseline treatment coding, then decide whether/how sequential treatment weights preserve that target.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Access, mood and preferences may confound comparator choice; switching and falls cannot be dismissed as missingness; subjective sleep is a secondary outcome.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/111; question: Do improvements in perceived sleep translate into next-day function after off-label trazodone prescribing?; aims: Characterize reasons for treatment choice and baseline sleepiness.
Compare trajectories after clinician-selected trazodone or CBT-I initiation.
Identify whether perceived sleep change predicts independent daytime performance.; design verbatim planning only: Prospectively enroll approximately 300 adults with chronic insomnia starting clinically chosen care at sleep and primary-care clinics. Compare new off-label trazodone users with CBT-I initiators over six months, preserving usual care and medication decisions. Obtain diaries, validated daytime-function measures, brief attention tasks, and adverse-event assessments.; outcomes verbatim: Primary: daytime functional impairment; secondary: insomnia severity, attention, falls or near-falls, discontinuation, and subjective-objective discordance.; analysis verbatim inherited not authoritative if refined: Use overlap weighting for severity, mood, access, and preferences; model switching and informative dropout. Prespecify baseline-sleepiness interaction and secondary-outcome control.; falsification verbatim: Reject discordance if sleep improvements consistently predict daytime function equally across clinical strategies.; source ids unverified leads: va-dod-insomnia-guideline-2025
belief-effects-health-review-2026
fda-off-label-use-2026; prospectus precedence: No

F28 — Orexin antagonists: separating sleep treatment from prevention claims

Rank: 233
Cluster: Clinical repurposing and neuropsychiatric claims
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: ASSOC_EIV; allocation or exposure unit: clinician-selected orexin-antagonist or CBT-I care; independent sampling unit: participant; repeated measurement unit: paired biomarker/function change and technical assay replicate; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: association or prospective prediction; temporal order and model adjustment alone do not identify a mechanism or treatment effect; specific constraint: Association cannot establish dementia prevention, causal mediation or biomarker surrogacy; optional CSF sampling creates selection.
Primary estimand: error-adjusted association between prespecified biomarker change and functional change
Candidate primary outcome: functional-change score paired with biomarker-change concentration
Primary finalized: No
Mathematical model: type: longitudinal association/reliability or held-out prediction precision simulation, not causal-efficacy power; formulas: Y_it = alpha + u_i + beta_W (B_it-Bbar_i) + beta_B Bbar_i + gamma^T X_it + f_time(t) + b_study + epsilon_it; u_i is a random intercept with a declared conditional mean/covariance model, not unrestricted fixed person effects.
B_observed,it = B_true,it + e_it; estimate measurement variance from replicates/validation or report sensitivity rather than assume it known.; variables: Y: functional-change score paired with biomarker-change concentration; i: participant; t: paired biomarker/function change and technical assay replicate; the registered visit/time-window index, not another independent unit; A: clinician-selected orexin-antagonist or CBT-I care; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: error-adjusted association between prespecified biomarker change and functional change; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; B it: change in the one prespecified exploratory biomarker concentration, in its assay unit; Bbar i: person/donor mean over a prespecified observation schedule; beta W,beta B: within-unit and between-unit association coefficients, not causal effects; X it: explicitly time-ordered baseline/premeasurement covariates; posttreatment adjustment changes interpretation; f time(t): prespecified sampling/recovery/calendar trajectory; b study: study/site/batch effect where relevant; B true,e it: latent predictor and measurement error, requiring identification constraints; family code: ASSOC_EIV
Sample size design: family: longitudinal association/reliability or held-out prediction precision simulation, not causal-efficacy power; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Association cannot establish dementia prevention, causal mediation or biomarker surrogacy; optional CSF sampling creates selection.
Associations need a clear sampling/time frame; causal interpretation is not supplied by this model.
Measurement error, reverse causation, time-varying confounding and selected observation may remain despite adjustment.
Specify E[u_i | Bbar_i,X] and its covariance or a correlated-random-effects formulation; an unrestricted fixed person intercept would absorb the between-person term.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Reject the proposed dissociation if functional and biomarker changes show reproducible, precisely estimated co-variation across prespecified measurements. Concordance alone would not establish later prognosis, central nervous system target engagement from blood alone, or dementia prevention.; preanalysis gate: Association cannot establish dementia prevention, causal mediation or biomarker surrogacy; optional CSF sampling creates selection.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Association cannot establish dementia prevention, causal mediation or biomarker surrogacy; optional CSF sampling creates selection.
Vary lag order/window and measurement-error reliability within a prespecified family; include negative-control windows when meaningful.
Separate within- and between-unit effects, missingness patterns and source-study replication; do not use cell/assay counts as independent donor counts.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Association cannot establish dementia prevention, causal mediation or biomarker surrogacy; optional CSF sampling creates selection.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/232; question: Do clinical sleep benefits and exploratory neurodegeneration biomarkers change together during indicated orexin-antagonist treatment?; aims: Establish repeated functional and biomarker measurement reliability.
Compare clinically chosen orexin-antagonist and behavioral treatment trajectories.
Evaluate whether functional and biomarker changes co-vary after accounting for measurement reliability and baseline clinical differences.; design verbatim planning only: Follow approximately 180 adults with clinically diagnosed insomnia starting an orexin antagonist or CBT-I through usual care. Collect sleep, daytime function, and research blood biomarkers for six months, with an optional nested cerebrospinal-fluid substudy only where independently approved. Compare baseline-adjusted trajectories; no prevention prescribing is initiated.; outcomes verbatim: Primary: association between functional change and prespecified biomarker change; secondary: insomnia response, attention, tolerability, and assay repeatability.; analysis verbatim inherited not authoritative if refined: Adjust baseline cognition, age, sleep apnea, and treatment preference; use errors-in-variables models and biomarker-family multiplicity control. Explicitly exclude dementia-incidence inference.; falsification verbatim: Reject the proposed dissociation if functional and biomarker changes show reproducible, precisely estimated co-variation across prespecified measurements. Concordance alone would not establish later prognosis, central nervous system target engagement from blood alone, or dementia prevention.; source ids unverified leads: suvorexant-csf-biomarkers-rct-2023
daridorexant-phase3-2022
daridorexant-dailymed-label-2026; prospectus precedence: No

F29 — CBD anxiety claims under real-world medication interactions

Rank: 211
Cluster: Clinical repurposing and neuropsychiatric claims
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: OBS_LONG_INT; allocation or exposure unit: clinician-selected new CBD or usual anxiety-care initiation; independent sampling unit: participant; repeated measurement unit: function, co-medication and product-exposure visit; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: observational strategy association; a causal interpretation additionally requires sequential exchangeability, positivity and valid treatment/observation models; specific constraint: Prevalent users are descriptive only; uncertain products and changing co-medications require exposure-error and sequential-confounding sensitivity.
Primary estimand: new-user standardized functional contrast and one baseline interaction index under comparable strategies
Candidate primary outcome: validated daily-function score at six months
Primary finalized: No
Mathematical model: type: interaction-specific precision/power with modifier distribution and reliability; simulate the actual interaction; formulas: E(Y_it | A_i, B_i, X_i) = alpha_t + tau_t A_i + eta_t B_i + kappa_t A_i B_i + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Treatment contrast at baseline B=b is tau_t + kappa_t b; the interaction target is kappa_t, not a difference in within-subgroup p-values.
W_i(t) = product_(k<=t) P(A_ik | A_i,<k, X_i0) / P(A_ik | H_ik), multiplied by observation/censoring weights if their assumptions are defensible.; variables: Y: validated daily-function score at six months; i: participant; t: function, co-medication and product-exposure visit; the registered visit/time-window index, not another independent unit; A: clinician-selected new CBD or usual anxiety-care initiation; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: new-user standardized functional contrast and one baseline interaction index under comparable strategies; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; B: one baseline medication-interaction index with definition fixed before analysis; centering/coding and any measurement-error model fixed before evaluation; eta: baseline modifier association; kappa: assignment-by-baseline-modifier interaction coefficient on the stated scale; H ik: observed predecision treatment/covariate history at k, excluding future measurements; family code: OBS_LONG_INT
Sample size design: family: interaction-specific precision/power with modifier distribution and reliability; simulate the actual interaction; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure
modifier prevalence/distribution and variance
modifier reliability and overlap
interaction size or precision target; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Prevalent users are descriptive only; uncertain products and changing co-medications require exposure-error and sequential-confounding sensitivity.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The modifier must be measured before assignment/initiation; interaction is scale dependent and requires support within modifier strata.
Apply the observational longitudinal weighting/g-formula conditions; the displayed regression alone does not eliminate treatment-selection or treatment-confounder feedback bias.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Reject the proposed discordance if anxiety improvement predicts functional benefit similarly regardless of sedation and interaction burden.; preanalysis gate: Prevalent users are descriptive only; uncertain products and changing co-medications require exposure-error and sequential-confounding sensitivity.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Prevalent users are descriptive only; uncertain products and changing co-medications require exposure-error and sequential-confounding sensitivity.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Vary modifier measurement-error assumptions and retain a continuous modifier unless a cutoff was independently fixed.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Prevalent users are descriptive only; uncertain products and changing co-medications require exposure-error and sequential-confounding sensitivity.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/210; question: Does reported anxiety relief from clinically supervised CBD use coincide with better function or with sedation and interaction burden?; aims: Document clinician-verified products and concurrent medication exposure.
Track anxiety, sedation, cognition, and daily functioning prospectively.
Validate whether interaction burden predicts discordant subjective and functional outcomes.; design verbatim planning only: Recruit approximately 240 adults before independently chosen initiation of clinically supervised CBD or another usual anxiety-care strategy, using common eligibility criteria and a baseline assessment before either strategy starts. Observe care for six months; collect verified product records where available, medication reconciliation, validated anxiety scales, remote attention tasks and functional assessments. Analyze any separately recruited prevalent CBD users descriptively rather than include them in the new-user comparison.; outcomes verbatim: Primary: daily function; secondary: anxiety, sedation, cognition, adverse events, and clinically indicated liver-test abnormalities.; analysis verbatim inherited not authoritative if refined: Align eligibility, baseline and time zero for both new-user groups. Use indication and treatment-preference weighting, time-varying co-medication models and missingness sensitivity analyses; assess comparator overlap and uncertain product exposure. Prespecify one interaction index and control secondary testing.; falsification verbatim: Reject the proposed discordance if anxiety improvement predicts functional benefit similarly regardless of sedation and interaction burden.; source ids unverified leads: cbd-anxiety-meta-2024
fda-cbd-consumer-update-2022
cannabis-cognition-methods-review-2025; prospectus precedence: No

F30 — Medication reconciliation for enhancement-oriented polypharmacy

Rank: 12
Cluster: Clinical repurposing and neuropsychiatric claims
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT_COUNT; allocation or exposure unit: participant individually randomized pharmacist-clinician reconciliation versus usual review; independent sampling unit: participant with clinic effects; repeated measurement unit: blinded problem adjudication at baseline and six months; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Do not equate theoretical alerts with clinical harm; recommendations may be refused; no symptom noninferiority claim without a separately justified design.
Primary estimand: six-month assignment effect on mean unresolved actionable medication-related problems
Candidate primary outcome: count of unresolved clinically actionable problems under a fixed blinded rubric
Primary finalized: No
Mathematical model: type: overdispersed count contrast planning by simulation, with follow-up/cluster/zero inflation assumptions; formulas: Y_it ~ NegativeBinomial(mu_it,phi); log(mu_it)=alpha_t+z(A_i)^T theta_t+beta Y_i0+gamma^T X_i+b_cluster(i)+u_i.
Delta = standardized E[Y | do(A=1)] - E[Y | do(A=0)] under randomization; report ratio exp(c^T theta) only when the log-mean model supports it.; variables: Y: count of unresolved clinically actionable problems under a fixed blinded rubric; i: participant with clinic effects; t: blinded problem adjudication at baseline and six months; the registered visit/time-window index, not another independent unit; A: participant individually randomized pharmacist-clinician reconciliation versus usual review; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: six-month assignment effect on mean unresolved actionable medication-related problems; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; mu it: conditional mean of the actual per-thesis count; phi: overdispersion parameter, with Var(Y)=mu+mu^2/phi under this parameterization; z(A): prespecified indicators for the actual arms or randomized policy; family code: RCT_COUNT
Sample size design: family: overdispersed count contrast planning by simulation, with follow-up/cluster/zero inflation assumptions; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Do not equate theoretical alerts with clinical harm; recommendations may be refused; no symptom noninferiority claim without a separately justified design.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: No reduction in unresolved actionable problems, poor adjudication reliability, or a clinically concerning symptom/function deterioration would undermine the offered service. Symptom deterioration is a safety boundary; this sample is not assumed to establish formal symptom noninferiority.; preanalysis gate: Do not equate theoretical alerts with clinical harm; recommendations may be refused; no symptom noninferiority claim without a separately justified design.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Do not equate theoretical alerts with clinical harm; recommendations may be refused; no symptom noninferiority claim without a separately justified design.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Do not equate theoretical alerts with clinical harm; recommendations may be refused; no symptom noninferiority claim without a separately justified design.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/11; question: Can pharmacist-led evidence reconciliation improve safety and function among adults using multiple enhancement-oriented medicines?; aims: Map treatment goals, evidence gaps, and interaction burden.
Randomize enhanced reconciliation versus usual medication review.
Test whether goal-aligned medication changes mediate functional improvement.; design verbatim planning only: Enroll approximately 220 adults already using at least three health-optimization products, including a prescription medicine, through primary-care clinics. Compare pharmacist-clinician reconciliation with usual review for six months. All prescribing and deprescribing remain shared clinical decisions; researchers do not initiate drugs or require discontinuation.; outcomes verbatim: Primary: six-month count of unresolved clinically actionable medication-related problems under blinded standardized adjudication. Secondary: patient-prioritized symptoms, general function, agency, costs, health-care use, recommendation acceptance and adverse events.; analysis verbatim inherited not authoritative if refined: Use intention-to-treat models with clinic effects, blinded adjudication, and prespecified symptom noninferiority margins justified before recruitment. Control secondary outcomes and examine baseline burden.; falsification verbatim: No reduction in unresolved actionable problems, poor adjudication reliability, or a clinically concerning symptom/function deterioration would undermine the offered service. Symptom deterioration is a safety boundary; this sample is not assumed to establish formal symptom noninferiority.; source ids unverified leads: fda-fdalabel-drug-labeling-2026
fda-compounding-questions-2026
fda-off-label-use-2026; prospectus precedence: Yes; prospectus file: thesis-prospectuses.json; prospectus design and estimand: Use a parallel individual randomized trial stratified by clinic and baseline problem burden. A dedicated intervention pharmacist conducts structured goal and evidence reconciliation, communicates with the treating clinician, and follows agreed actions. The comparator is usual review documented prospectively; equal research inventory visits do not duplicate the intervention discussion. The estimand is the six-month mean difference in unresolved actionable problems under assignment, including refusal of recommendations and clinician decisions to continue treatment. The trial estimates the service package, including its extra expertise and contact.; prospectus analysis plan: Model unresolved problem counts using an appropriate overdispersed count model, baseline burden and clinic effects, and report adjusted absolute differences as well as ratios. Avoid equating theoretical interaction alerts with clinical harm. Analyze function and symptoms in a registered secondary hierarchy and report deterioration proportions with intervals. Changes in product use are mediators, not baseline confounders. Any mediation analysis explicitly addresses post-randomization confounding and remains hypothesis-generating. Do not claim symptom noninferiority unless a separate justified design and sample size support it.; precedence note: Expanded prospectus refinements govern this scaffold; verbatim earlier brief material is retained for traceability, not as a contradictory analysis instruction. F15 is explicitly two arms; F30 has no unjustified symptom noninferiority claim.

F31 — Do biology foundation models survive historical and family holdouts?

Rank: 175
Cluster: AI biology, RNA and synthetic-biology validation
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: PRED; allocation or exposure unit: locked model applied to temporal/family/laboratory holdout records; independent sampling unit: protein family and independent laboratory/reference sample; repeated measurement unit: protein prediction and assay replicate; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: prediction, calibration or model transportability; prediction gain is not clinical efficacy, mechanistic necessity or causal treatment selection; specific constraint: Training overlap is uncertain; performance decline does not prove leakage; whole families/laboratories must stay together in resampling.
Primary estimand: external functional-prediction loss and calibration gain of foundation model versus fixed baseline
Candidate primary outcome: independently measured benign protein function
Primary finalized: No
Mathematical model: type: external/nested predictive-validation precision with target prevalence/variance, independent-unit holdout coverage and model complexity; formulas: Gain = mean_test[L(Y,f_base(X)) - L(Y,f_new(X))], with both predictions frozen before test outcomes; bootstrap independent held-out units.
For a binary target assess E[Y | p_hat=p] versus p and calibration intercept/slope logit Pr(Y=1)=a_cal+b_cal logit(p_hat); for a continuous target use Y=a_cal+b_cal y_hat+error plus scale-appropriate calibration/loss.; variables: Y: independently measured benign protein function; i: protein family and independent laboratory/reference sample; t: protein prediction and assay replicate; the registered visit/time-window index, not another independent unit; A: locked model applied to temporal/family/laboratory holdout records; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: external functional-prediction loss and calibration gain of foundation model versus fixed baseline; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; L: one proper or scientifically justified primary loss/scoring rule fixed for the actual outcome type; positive Gain favors new model; f base,f new: locked baseline/new prediction functions using only information available at the registered prediction time; test: untouched person/family/lab/site/time set as specified per thesis; p hat,y hat: frozen probability or continuous prediction; a cal,b cal: calibration intercept and slope, estimated on evaluation data for assessment; using them to update the model requires another evaluation set; family code: PRED
Sample size design: family: external/nested predictive-validation precision with target prevalence/variance, independent-unit holdout coverage and model complexity; formula: Simulate or bootstrap the independent-unit test loss/calibration/decision criterion; plan cases/events, noncases and independent sites/families/programs separately. No universal events-per-variable or participant N is asserted.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Training overlap is uncertain; performance decline does not prove leakage; whole families/laboratories must stay together in resampling.
Train/tune/test separation at the actual biological/participant/family/time level; no future, postoutcome or target-derived predictors.
Validation support must cover intended deployment; calibration and missing/failing measurements are reported with discrimination/loss.
A validation subset selected by culture success, assay viability, consent or data completeness is a different target unless the full strategy is explicitly evaluated.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Reject the predicted collapse of model advantage if foundation models retain meaningful calibrated gains across strict holdouts and external assays. Neither retained performance nor a decline alone proves the presence or absence of training leakage.; preanalysis gate: Training overlap is uncertain; performance decline does not prove leakage; whole families/laboratories must stay together in resampling.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Training overlap is uncertain; performance decline does not prove leakage; whole families/laboratories must stay together in resampling.
Repeat loss/calibration with independent-cluster resampling, prespecified subgroup/transport shifts and plausible missingness.
Compare fixed simple baselines, calibration and uncertainty coverage; report abstention/failure/turnaround rather than deleting hard cases.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Frozen prediction time/horizon, proper loss, baseline model, feature availability, independent split/holdout and target deployment population.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Training overlap is uncertain; performance decline does not prove leakage; whole families/laboratories must stay together in resampling.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/174; question: How much reported molecular-prediction performance survives joint temporal, biological-family, and laboratory holdouts?; aims: Audit dataset provenance and potential pretraining overlap.
Benchmark locked models under increasingly strict holdouts.
Validate uncertainty and abstention on independently generated benign reference data.; design verbatim planning only: Assemble approximately 10,000 licensed or openly available nonhazardous protein-function records, then obtain a planning set of 120 independently measured benign reference samples through partner laboratories. Compare selected foundation models with homology, structure-only, and regularized statistical baselines. Freeze evaluation plans before accessing external outcomes.; outcomes verbatim: Primary: externally calibrated functional prediction; secondary: ranking utility, uncertainty coverage, subgroup performance, and computational cost.; analysis verbatim inherited not authoritative if refined: Bootstrap by protein family and laboratory, document uncertain training overlap and correct comparisons across model families. Report missing-data and assay-quality sensitivity analyses. Interpret holdout performance as transportability evidence; attribute changes specifically to leakage only where the provenance audit independently supports overlap.; falsification verbatim: Reject the predicted collapse of model advantage if foundation models retain meaningful calibrated gains across strict holdouts and external assays. Neither retained performance nor a decline alone proves the presence or absence of training leakage.; source ids unverified leads: ai-drug-discovery-foundation-models-2025
ai-drug-discovery-practical-guidance-2026
ai-protein-design-validation-review-2026; prospectus precedence: No

F32 — Protein-design confidence versus experimentally measured specificity

Rank: 221
Cluster: AI biology, RNA and synthetic-biology validation
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: BUDGET; allocation or exposure unit: fixed ranking policy applied to the same independently evaluated candidate panel; independent sampling unit: protein family and independent candidate preparation/laboratory; repeated measurement unit: candidate assay, replicate and selected budget set; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: budget-matched ranking performance on the approved reference panel, not therapeutic efficacy or causal biological design superiority; specific constraint: Budget and confirmation rule fixed before outcomes; candidate overlap creates paired comparisons; no therapeutic protein-design efficacy claim.
Primary estimand: difference in reproducibly specific candidates found at an identical evaluation budget
Candidate primary outcome: number of independently confirmed specific candidates selected by each ranking
Primary finalized: No
Mathematical model: type: paired budget-yield precision simulation with independent candidate families and laboratory confirmation; formulas: K_p(B)=sum_j 1{j in frozen selection set S_p(B)} R_j; Delta_B=K_uncertainty(B)-K_confidence(B).
R_j=1 only if independent functional/specificity confirmation and prespecified cross-laboratory reproducibility criteria pass; compare paired equal-budget sets with family/laboratory-aware resampling.; variables: Y: number of independently confirmed specific candidates selected by each ranking; i: protein family and independent candidate preparation/laboratory; t: candidate assay, replicate and selected budget set; the registered visit/time-window index, not another independent unit; A: fixed ranking policy applied to the same independently evaluated candidate panel; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: difference in reproducibly specific candidates found at an identical evaluation budget; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; B: fixed evaluation budget, including confirmation costs under an identical accounting rule; p: ranking policy; j: independent candidate preparation, with biological-family structure; S p(B): selection set determined without seeing confirmatory outcomes; R j: reproducibly specific candidate indicator; K p(B): confirmed-candidate yield under policy p at budget B; family code: BUDGET
Sample size design: family: paired budget-yield precision simulation with independent candidate families and laboratory confirmation; formula: Use design-specific simulation or independent-unit resampling for the stated estimand and target confidence width/power; preserve all allocation, censoring, clustering and measurement-error structure.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Budget and confirmation rule fixed before outcomes; candidate overlap creates paired comparisons; no therapeutic protein-design efficacy claim.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
All policies evaluated on a common eligible panel; shared candidates and assay resources create paired dependence.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Reject superiority if uncertainty-aware selection finds no more reproducible candidates or fails on an unseen reference family.; preanalysis gate: Budget and confirmation rule fixed before outcomes; candidate overlap creates paired comparisons; no therapeutic protein-design efficacy claim.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Budget and confirmation rule fixed before outcomes; candidate overlap creates paired comparisons; no therapeutic protein-design efficacy claim.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Budget and confirmation rule fixed before outcomes; candidate overlap creates paired comparisons; no therapeutic protein-design efficacy claim.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/220; question: Can calibrated uncertainty predict which AI-ranked benign protein candidates retain specificity outside the training assay?; aims: Define a blinded nontherapeutic specificity benchmark with partner laboratories.
Compare candidate-selection strategies under equal evaluation budgets.
Test whether uncertainty predicts cross-assay and cross-laboratory failures.; design verbatim planning only: Use a planning set of 150 pre-existing, institutionally approved benign reference proteins or archived candidates, excluding pathogens, toxins, and therapeutic optimization. Randomize coded samples across three laboratories. Compare locked confidence-only, uncertainty-aware, and simple baseline rankings against independent functional and specificity measurements; no operational design methods are disseminated.; outcomes verbatim: Primary: reproducibly specific candidates identified per evaluation budget; secondary: calibration, stability, reproducibility, and failure-mode classification.; analysis verbatim inherited not authoritative if refined: Use paired budget-matched comparisons, laboratory random effects, and family-clustered uncertainty. Prespecify success criteria and multiplicity across assay families.; falsification verbatim: Reject superiority if uncertainty-aware selection finds no more reproducible candidates or fails on an unseen reference family.; source ids unverified leads: ai-protein-design-validation-review-2026
ai-biologic-discovery-review-2026
ai-drug-discovery-practical-guidance-2026; prospectus precedence: No

F33 — RNA delivery success across human tissue-model diversity

Rank: 215
Cluster: AI biology, RNA and synthetic-biology validation
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: LAB_RANK; allocation or exposure unit: coded reference delivery system applied to approved donor tissue models; independent sampling unit: donor with laboratory-held-out validation; repeated measurement unit: tissue-system culture and assay replicate; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: prediction, calibration or model transportability; prediction gain is not clinical efficacy, mechanistic necessity or causal treatment selection; specific constraint: Uptake is not expression; viability/inflammation are separate gates; specify randomized culture allocation before calling contrasts causal in vitro.
Primary estimand: cross-context agreement and uncertainty of functional-expression rankings across tissue/donor/laboratory
Candidate primary outcome: independent functional-expression readout for each reference system
Primary finalized: No
Mathematical model: type: external/nested predictive-validation precision with target prevalence/variance, independent-unit holdout coverage and model complexity; formulas: Gain = mean_test[L(Y,f_base(X)) - L(Y,f_new(X))], with both predictions frozen before test outcomes; bootstrap independent held-out units.
For a binary target assess E[Y | p_hat=p] versus p and calibration intercept/slope logit Pr(Y=1)=a_cal+b_cal logit(p_hat); for a continuous target use Y=a_cal+b_cal y_hat+error plus scale-appropriate calibration/loss.
Within context k, compute ranking R_k of the fixed condition-response vector; report rank concordance across held-out contexts with uncertainty propagated from independent donors/materials, not just point-rank correlations.; variables: Y: independent functional-expression readout for each reference system; i: donor with laboratory-held-out validation; t: tissue-system culture and assay replicate; the registered visit/time-window index, not another independent unit; A: coded reference delivery system applied to approved donor tissue models; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: cross-context agreement and uncertainty of functional-expression rankings across tissue/donor/laboratory; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; L: one proper or scientifically justified primary loss/scoring rule fixed for the actual outcome type; positive Gain favors new model; f base,f new: locked baseline/new prediction functions using only information available at the registered prediction time; test: untouched person/family/lab/site/time set as specified per thesis; p hat,y hat: frozen probability or continuous prediction; a cal,b cal: calibration intercept and slope, estimated on evaluation data for assessment; using them to update the model requires another evaluation set; k: donor/tissue/laboratory/model-route context; R k: ranking of fixed reference systems/preparations under one independently specified functional readout; family code: LAB_RANK
Sample size design: family: external/nested predictive-validation precision with target prevalence/variance, independent-unit holdout coverage and model complexity; formula: Simulate or bootstrap the independent-unit test loss/calibration/decision criterion; plan cases/events, noncases and independent sites/families/programs separately. No universal events-per-variable or participant N is asserted.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Uptake is not expression; viability/inflammation are separate gates; specify randomized culture allocation before calling contrasts causal in vitro.
Train/tune/test separation at the actual biological/participant/family/time level; no future, postoutcome or target-derived predictors.
Validation support must cover intended deployment; calibration and missing/failing measurements are reported with discrimination/loss.
A validation subset selected by culture success, assay viability, consent or data completeness is a different target unless the full strategy is explicitly evaluated.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Reject context dependence if rankings and calibrated uptake-expression relationships remain stable across tissues, donors, and laboratories.; preanalysis gate: Uptake is not expression; viability/inflammation are separate gates; specify randomized culture allocation before calling contrasts causal in vitro.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Uptake is not expression; viability/inflammation are separate gates; specify randomized culture allocation before calling contrasts causal in vitro.
Repeat loss/calibration with independent-cluster resampling, prespecified subgroup/transport shifts and plausible missingness.
Compare fixed simple baselines, calibration and uncertainty coverage; report abstention/failure/turnaround rather than deleting hard cases.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Frozen prediction time/horizon, proper loss, baseline model, feature availability, independent split/holdout and target deployment population.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Uptake is not expression; viability/inflammation are separate gates; specify randomized culture allocation before calling contrasts causal in vitro.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/214; question: Does apparent RNA uptake predict functional expression and tolerability across donors and tissue models?; aims: Harmonize noninfectious reference-model measurement across laboratories.
Separate uptake, expression, and inflammatory readouts across donors.
Validate whether donor-aware models improve external ranking and uncertainty.; design verbatim planning only: Use archived, institutionally approved human tissue models from approximately 30 consenting donors across two relevant tissue classes and a small panel of pre-existing reference delivery systems. Compare coded systems with established laboratory controls under independent institutional protocols. Hold out one laboratory and donor subgroup; no human administration or formulation instructions are involved.; outcomes verbatim: Primary: cross-context agreement of functional-expression rankings; secondary: uptake-expression discordance, inflammatory response, viability, and donor heterogeneity.; analysis verbatim inherited not authoritative if refined: Use hierarchical donor-by-system models, batch adjustment, and prespecified tissue contrasts; control readout-family multiplicity and validate uncertainty on held-out donors.; falsification verbatim: Reject context dependence if rankings and calibrated uptake-expression relationships remain stable across tissues, donors, and laboratories.; source ids unverified leads: rna-delivery-review-2025
mrna-pharmacology-framework-2025
mrna-clinical-development-review-2025; prospectus precedence: No

F34 — Can mRNA pharmacology models predict durable clinical target engagement?

Rank: 230
Cluster: AI biology, RNA and synthetic-biology validation
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: PRED_CURVE; allocation or exposure unit: locked pharmacology model evaluated in existing therapeutic mRNA programs; independent sampling unit: participant and independent program; repeated measurement unit: sparse sampled target-engagement time point; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: prediction, calibration or model transportability; prediction gain is not clinical efficacy, mechanistic necessity or causal treatment selection; specific constraint: Aggregate reconstruction cannot recover participant-level heterogeneity; informative sampling and dropout matter; target engagement is not clinical benefit.
Primary estimand: late-timepoint predictive calibration/loss of mechanistic versus simpler pharmacology models
Candidate primary outcome: target-engagement concentration/effect at a prespecified late horizon
Primary finalized: No
Mathematical model: type: external/nested predictive-validation precision with target prevalence/variance, independent-unit holdout coverage and model complexity; formulas: Gain = mean_test[L(Y,f_base(X)) - L(Y,f_new(X))], with both predictions frozen before test outcomes; bootstrap independent held-out units.
For a binary target assess E[Y | p_hat=p] versus p and calibration intercept/slope logit Pr(Y=1)=a_cal+b_cal logit(p_hat); for a continuous target use Y=a_cal+b_cal y_hat+error plus scale-appropriate calibration/loss.
Y_ist = f_model(t;theta_is)+epsilon_ist; compare predictions at locked late times, holding entire programs out where transport is claimed.; variables: Y: target-engagement concentration/effect at a prespecified late horizon; i: participant and independent program; t: sparse sampled target-engagement time point; the registered visit/time-window index, not another independent unit; A: locked pharmacology model evaluated in existing therapeutic mRNA programs; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: late-timepoint predictive calibration/loss of mechanistic versus simpler pharmacology models; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; L: one proper or scientifically justified primary loss/scoring rule fixed for the actual outcome type; positive Gain favors new model; f base,f new: locked baseline/new prediction functions using only information available at the registered prediction time; test: untouched person/family/lab/site/time set as specified per thesis; p hat,y hat: frozen probability or continuous prediction; a cal,b cal: calibration intercept and slope, estimated on evaluation data for assessment; using them to update the model requires another evaluation set; s: independent therapeutic program; theta is: participant/program pharmacology parameters, with hierarchical assumptions; f model: candidate pharmacology function fixed before late-time evaluation; family code: PRED_CURVE
Sample size design: family: external/nested predictive-validation precision with target prevalence/variance, independent-unit holdout coverage and model complexity; formula: Simulate or bootstrap the independent-unit test loss/calibration/decision criterion; plan cases/events, noncases and independent sites/families/programs separately. No universal events-per-variable or participant N is asserted.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Aggregate reconstruction cannot recover participant-level heterogeneity; informative sampling and dropout matter; target engagement is not clinical benefit.
Train/tune/test separation at the actual biological/participant/family/time level; no future, postoutcome or target-derived predictors.
Validation support must cover intended deployment; calibration and missing/failing measurements are reported with discrimination/loss.
A validation subset selected by culture success, assay viability, consent or data completeness is a different target unless the full strategy is explicitly evaluated.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Reject the mechanistic advantage if simpler models predict late outcomes equally well and retain calibration across programs.; preanalysis gate: Aggregate reconstruction cannot recover participant-level heterogeneity; informative sampling and dropout matter; target engagement is not clinical benefit.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Aggregate reconstruction cannot recover participant-level heterogeneity; informative sampling and dropout matter; target engagement is not clinical benefit.
Repeat loss/calibration with independent-cluster resampling, prespecified subgroup/transport shifts and plausible missingness.
Compare fixed simple baselines, calibration and uncertainty coverage; report abstention/failure/turnaround rather than deleting hard cases.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Frozen prediction time/horizon, proper loss, baseline model, feature availability, independent split/holdout and target deployment population.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Aggregate reconstruction cannot recover participant-level heterogeneity; informative sampling and dropout matter; target engagement is not clinical benefit.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/229; question: Which pharmacology assumptions best predict the duration and variability of target engagement in therapeutic mRNA trials?; aims: Reconstruct model assumptions and sampling schedules from accessible studies.
Compare mechanistic and empirical models on withheld follow-up.
Test transportability across indications without pooling incompatible endpoints.; design verbatim planning only: Seek deidentified data from two or three existing noninfectious therapeutic mRNA programs, targeting roughly 150 participants where available. Compare locked model families using original randomized comparators when present and later-timepoint prediction otherwise. Aggregate-data reconstruction provides a narrower fallback, explicitly distinguished from participant-level inference.; outcomes verbatim: Primary: late-timepoint predictive calibration; secondary: uncertainty coverage, subgroup errors, target-engagement duration, and sensitivity to sparse sampling.; analysis verbatim inherited not authoritative if refined: Fit hierarchical indication-specific models, address informative sampling and dropout, and prespecify model-comparison penalties. Do not infer clinical benefit from target engagement alone.; falsification verbatim: Reject the mechanistic advantage if simpler models predict late outcomes equally well and retain calibration across programs.; source ids unverified leads: mrna-pharmacology-framework-2025
mrna-clinical-development-review-2025
rna-delivery-review-2025; prospectus precedence: No

F35 — Independent detection limits in genome-editing safety assessment

Rank: 176
Cluster: AI biology, RNA and synthetic-biology validation
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: BENCH_DIAG; allocation or exposure unit: locked safety pipeline applied to coded existing or digital reference dataset; independent sampling unit: reference sample with laboratory/batch structure; repeated measurement unit: variant call by pipeline and adjudicator; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: diagnostic/benchmark agreement or classification validity; reader/interface/navigation effects are causal only in their separately randomized component; specific constraint: Ground truth is itself uncertain; digital variants may not reproduce biological artifacts; tune and test on disjoint samples.
Primary estimand: detection probability and false-discovery rate near fixed reporting limits
Candidate primary outcome: reference-adjudicated rare-variant call state and pipeline detection
Primary finalized: No
Mathematical model: type: diagnostic/benchmark sensitivity, false-alert and calibration precision with independent cases/noncases and cluster/reference uncertainty; formulas: Sensitivity = TP/(TP+FN); false-alert burden = FP/(registered nonevent time/opportunities); report participant/family-clustered uncertainty and failed/indeterminate cases.
For probability scores, report calibration and a fixed proper score in participant/source-disjoint evaluation, with uncertain reference labels modeled or bounded.
FDP=FP/(TP+FP) when discoveries>0, with a prespecified zero-discovery rule; FDR=E[FDP] under the declared repeated benchmark/sample process. Report empirical FDP/precision and its uncertainty without confusing FP per nonevent with FDR.; variables: Y: reference-adjudicated rare-variant call state and pipeline detection; i: reference sample with laboratory/batch structure; t: variant call by pipeline and adjudicator; the registered visit/time-window index, not another independent unit; A: locked safety pipeline applied to coded existing or digital reference dataset; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: detection probability and false-discovery rate near fixed reporting limits; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; TP,FN,FP: true-positive, false-negative and false-positive counts relative to the independently defined reference and matching window, not an assumed perfect gold standard; reference: per-thesis independent reference, including adjudication uncertainty; nonevent time/opportunities: fixed at-risk denominator; event specificity and false alerts per hour are distinct targets; FDP: false-discovery proportion among reported positive calls; FDR: expected FDP under the specified benchmark/randomness process; family code: BENCH_DIAG
Sample size design: family: diagnostic/benchmark sensitivity, false-alert and calibration precision with independent cases/noncases and cluster/reference uncertainty; formula: An independent-case rough starting point is n_case ~= z_(1-alpha*/2)^2 Se(1-Se)/h_Se^2; replace with interval-based/clustered simulation for the actual design. Plan noncases/observation time and reference failures separately.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; Se: plausible sensitivity from prior evidence or pilot, not an asserted study result; h Se: desired sensitivity interval half-width; n case: independent positive cases; repeated events within a person require correlation adjustment; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Ground truth is itself uncertain; digital variants may not reproduce biological artifacts; tune and test on disjoint samples.
Reference independence, fixed thresholds/window and an explicit denominator; incorporation/verification bias can invalidate apparent accuracy.
Participant/family-level test splits and paired comparisons; report abstention, assay failure and indeterminate truth rather than force every case into a binary label.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Reject the overconfidence hypothesis if standard pipelines retain nominal coverage and error control under all held-out shifts.; preanalysis gate: Ground truth is itself uncertain; digital variants may not reproduce biological artifacts; tune and test on disjoint samples.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Ground truth is itself uncertain; digital variants may not reproduce biological artifacts; tune and test on disjoint samples.
Vary plausible reference misclassification, verification-selection and event-matching window under a prespecified sensitivity family.
Report subgroup usable-data coverage and calibration; avoid interpreting an imprecise null difference as equivalence.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Ground truth is itself uncertain; digital variants may not reproduce biological artifacts; tune and test on disjoint samples.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/175; question: Do genome-editing safety-analysis pipelines remain calibrated when rare signals, batch artifacts, and incomplete references coexist?; aims: Assemble licensed reference datasets and nonoperational digital stress tests.
Compare locked pipelines under blinded laboratory and frequency shifts.
Validate whether uncertainty-aware review reduces missed and spurious safety findings.; design verbatim planning only: Analyze existing deidentified sequencing panels and digitally simulated variants without editing or generating biological constructs. A planning benchmark includes 200 coded datasets from three laboratories. Compare standard automated calls, uncertainty-calibrated calls, and blinded expert review against independently adjudicated reference truth.; outcomes verbatim: Primary: calibrated detection and false-discovery rates near prespecified reporting limits; secondary: review burden and reproducibility.; analysis verbatim inherited not authoritative if refined: Stratify by variant class and batch quality, bootstrap by reference sample, and separate tuning from locked evaluation. Control pipeline-family multiplicity and quantify uncertain ground truth.; falsification verbatim: Reject the overconfidence hypothesis if standard pipelines retain nominal coverage and error control under all held-out shifts.; source ids unverified leads: fda-gene-editing-guidance
fda-genome-editing-safety-2026
ai-drug-discovery-practical-guidance-2026; prospectus precedence: No

F36 — Ancestry representation in genome-editing safety evidence

Rank: 206
Cluster: AI biology, RNA and synthetic-biology validation
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: COVERAGE; allocation or exposure unit: pooled versus donor-aware evaluation of consent-compatible reference genomes; independent sampling unit: donor, not individual variants; repeated measurement unit: variant/class uncertainty interval and unresolved call; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: diagnostic/benchmark agreement or classification validity; reader/interface/navigation effects are causal only in their separately randomized component; specific constraint: Technical quality and platform may explain differences; social race is not interchangeable with genetic ancestry; sparse support prevents universal parity claims.
Primary estimand: ancestry-continuous/subgroup uncertainty coverage and unresolved-call disparity under fixed quality adjustment
Candidate primary outcome: indicator of truth within reported interval plus unresolved-call burden
Primary finalized: No
Mathematical model: type: diagnostic/benchmark sensitivity, false-alert and calibration precision with independent cases/noncases and cluster/reference uncertainty; formulas: Coverage_g = mean_(i in g) 1{truth_i in interval_i}; Unresolved_g=mean_(i in g) 1{call_i unresolved}; compare standardized estimates across quality/ancestry support.
Report donor-clustered uncertainty and ancestry-continuous trends; coverage only among resolved calls is a conditional, potentially selected target.; variables: Y: indicator of truth within reported interval plus unresolved-call burden; i: donor, not individual variants; t: variant/class uncertainty interval and unresolved call; the registered visit/time-window index, not another independent unit; A: pooled versus donor-aware evaluation of consent-compatible reference genomes; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: ancestry-continuous/subgroup uncertainty coverage and unresolved-call disparity under fixed quality adjustment; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; TP,FN,FP: true-positive, false-negative and false-positive counts relative to the independently defined reference and matching window, not an assumed perfect gold standard; reference: per-thesis independent reference, including adjudication uncertainty; nonevent time/opportunities: fixed at-risk denominator; event specificity and false alerts per hour are distinct targets; g: prespecified supported ancestry/technical stratum, not an interchangeable social-race label; truth i: independent reference truth with quantified uncertainty; interval i: pipeline uncertainty interval/set at its registered nominal level; family code: COVERAGE
Sample size design: family: diagnostic/benchmark sensitivity, false-alert and calibration precision with independent cases/noncases and cluster/reference uncertainty; formula: An independent-case rough starting point is n_case ~= z_(1-alpha*/2)^2 Se(1-Se)/h_Se^2; replace with interval-based/clustered simulation for the actual design. Plan noncases/observation time and reference failures separately.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; Se: plausible sensitivity from prior evidence or pilot, not an asserted study result; h Se: desired sensitivity interval half-width; n case: independent positive cases; repeated events within a person require correlation adjustment; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Technical quality and platform may explain differences; social race is not interchangeable with genetic ancestry; sparse support prevents universal parity claims.
Reference independence, fixed thresholds/window and an explicit denominator; incorporation/verification bias can invalidate apparent accuracy.
Participant/family-level test splits and paired comparisons; report abstention, assay failure and indeterminate truth rather than force every case into a binary label.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Reject the coverage hypothesis if subgroup differences disappear after technical-quality adjustment and donor-aware evaluation adds no reliability.; preanalysis gate: Technical quality and platform may explain differences; social race is not interchangeable with genetic ancestry; sparse support prevents universal parity claims.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Technical quality and platform may explain differences; social race is not interchangeable with genetic ancestry; sparse support prevents universal parity claims.
Vary plausible reference misclassification, verification-selection and event-matching window under a prespecified sensitivity family.
Report subgroup usable-data coverage and calibration; avoid interpreting an imprecise null difference as equivalence.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Technical quality and platform may explain differences; social race is not interchangeable with genetic ancestry; sparse support prevents universal parity claims.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/205; question: Do safety-assessment reference resources support equally reliable inference across genetic ancestry and structural-variation backgrounds?; aims: Audit ancestry and variation coverage in accessible reference datasets.
Quantify subgroup calibration and unresolved-call burden.
Test alternative reference-allocation policies under fixed validation resources.; design verbatim planning only: Use licensed, consent-compatible genomic reference datasets representing a planning target of 1,000 donors, with no new genome editing. Compare conventional pooled benchmarking with leave-group-out and donor-aware evaluation. Simulate resource allocation for additional validation using benign digital reference problems; confirm key findings in an independent dataset.; outcomes verbatim: Primary: subgroup uncertainty coverage and unresolved-call burden; secondary: error disparity, resource efficiency, and reference completeness.; analysis verbatim inherited not authoritative if refined: Model ancestry continuously where appropriate, distinguish ancestry from social race, and adjust sequencing quality and platform. Use hierarchical estimates and prespecified disparity measures.; falsification verbatim: Reject the coverage hypothesis if subgroup differences disappear after technical-quality adjustment and donor-aware evaluation adds no reliability.; source ids unverified leads: fda-gene-editing-guidance
fda-genome-editing-safety-2026
polygenic-cardiovascular-consensus-2025; prospectus precedence: No

F37 — Younger molecular clocks versus preserved cellular identity

Rank: 239
Cluster: AI biology, RNA and synthetic-biology validation
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: ASSOC_EIV; allocation or exposure unit: published/archived nonclinical reprogramming states observed; independent sampling unit: independent donor line and study; repeated measurement unit: single cell, time point and clock/identity assay; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: association or prospective prediction; temporal order and model adjustment alone do not identify a mechanism or treatment effect; specific constraint: Cell count is not donor replication; cell composition/stress and study selection can explain clock change; correlated clocks are not independent confirmations.
Primary estimand: association of molecular-age change with preserved identity and independently measured function on held-out donors/studies
Candidate primary outcome: independent functional readout alongside age and identity change
Primary finalized: No
Mathematical model: type: longitudinal association/reliability or held-out prediction precision simulation, not causal-efficacy power; formulas: Y_it = alpha + u_i + beta_W (B_it-Bbar_i) + beta_B Bbar_i + gamma^T X_it + f_time(t) + b_study + epsilon_it; u_i is a random intercept with a declared conditional mean/covariance model, not unrestricted fixed person effects.
B_observed,it = B_true,it + e_it; estimate measurement variance from replicates/validation or report sensitivity rather than assume it known.; variables: Y: independent functional readout alongside age and identity change; i: independent donor line and study; t: single cell, time point and clock/identity assay; the registered visit/time-window index, not another independent unit; A: published/archived nonclinical reprogramming states observed; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: association of molecular-age change with preserved identity and independently measured function on held-out donors/studies; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; B it: fixed molecular-age change with independently scored identity and function; donor/study-held-out analysis; Bbar i: person/donor mean over a prespecified observation schedule; beta W,beta B: within-unit and between-unit association coefficients, not causal effects; X it: explicitly time-ordered baseline/premeasurement covariates; posttreatment adjustment changes interpretation; f time(t): prespecified sampling/recovery/calendar trajectory; b study: study/site/batch effect where relevant; B true,e it: latent predictor and measurement error, requiring identification constraints; family code: ASSOC_EIV
Sample size design: family: longitudinal association/reliability or held-out prediction precision simulation, not causal-efficacy power; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Cell count is not donor replication; cell composition/stress and study selection can explain clock change; correlated clocks are not independent confirmations.
Associations need a clear sampling/time frame; causal interpretation is not supplied by this model.
Measurement error, reverse causation, time-varying confounding and selected observation may remain despite adjustment.
Specify E[u_i | Bbar_i,X] and its covariance or a correlated-random-effects formulation; an unrestricted fixed person intercept would absorb the between-person term.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Reject identity confounding if clock reductions consistently predict independent functional improvement after lineage adjustment across held-out studies.; preanalysis gate: Cell count is not donor replication; cell composition/stress and study selection can explain clock change; correlated clocks are not independent confirmations.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Cell count is not donor replication; cell composition/stress and study selection can explain clock change; correlated clocks are not independent confirmations.
Vary lag order/window and measurement-error reliability within a prespecified family; include negative-control windows when meaningful.
Separate within- and between-unit effects, missingness patterns and source-study replication; do not use cell/assay counts as independent donor counts.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Cell count is not donor replication; cell composition/stress and study selection can explain clock change; correlated clocks are not independent confirmations.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/238; question: Do molecular age reductions in reprogramming datasets coincide with preserved identity and improved cellular function?; aims: Audit clock behavior across archived reprogramming datasets.
Separate age-associated expression changes from lineage and stress signals.
Validate a locked identity-preserving score against archived functional measurements.; design verbatim planning only: Integrate consent-compatible single-cell and epigenetic datasets from published nonclinical studies, targeting 20 independent donor lines where accessible. Compare clock-only, lineage-aware, and stress-adjusted models using donor- and study-held-out validation. Partner laboratories may provide archived samples or existing functional records; no reprogramming procedures are performed.; outcomes verbatim: Primary: concordance of age change with preserved identity and independent function; secondary: heterogeneity, stress signatures, and cross-study calibration.; analysis verbatim inherited not authoritative if refined: Control cell composition, donor age, batch, and measurement timing; use multilevel models and pathway-family multiplicity control. Avoid treating correlated clocks as independent confirmations.; falsification verbatim: Reject identity confounding if clock reductions consistently predict independent functional improvement after lineage adjustment across held-out studies.; source ids unverified leads: reprogramming-genomic-tech-2026
aging-clocks-critical-review-2026
epigenetic-interventions-review-2026; prospectus precedence: No

F38 — When organoid drug sensitivity fails to predict patient response

Rank: 229
Cluster: AI biology, RNA and synthetic-biology validation
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: PRED_FALLBACK; allocation or exposure unit: patient starting independently selected standard oncology treatment; blinded organoid score; independent sampling unit: all enrolled patients; repeated measurement unit: culture attempt, timed prediction and response adjudication; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: prediction, calibration or model transportability; prediction gain is not clinical efficacy, mechanistic necessity or causal treatment selection; specific constraint: Use routine-model fallback whenever organoid result is absent/late; no random imputation of nonexistent scores; external validation requires another independent cohort.
Primary estimand: internally validated improvement of the all-patient prediction strategy over routine prediction by the fixed response time
Candidate primary outcome: adjudicated response by fixed assessment time with prespecified early progression/death rule
Primary finalized: No
Mathematical model: type: external/nested predictive-validation precision with target prevalence/variance, independent-unit holdout coverage and model complexity; formulas: Gain = mean_test[L(Y,f_base(X)) - L(Y,f_new(X))], with both predictions frozen before test outcomes; bootstrap independent held-out units.
For a binary target assess E[Y | p_hat=p] versus p and calibration intercept/slope logit Pr(Y=1)=a_cal+b_cal logit(p_hat); for a continuous target use Y=a_cal+b_cal y_hat+error plus scale-appropriate calibration/loss.
f_strategy(X,O,M)=f_combined(X,O) if a valid organoid O is timely (M=1), otherwise f_routine(X); evaluate every enrolled patient.
Early progression/death must enter the registered response definition or an explicit competing/ordinal outcome; missing assessments among living patients are separate.; variables: Y: adjudicated response by fixed assessment time with prespecified early progression/death rule; i: all enrolled patients; t: culture attempt, timed prediction and response adjudication; the registered visit/time-window index, not another independent unit; A: patient starting independently selected standard oncology treatment; blinded organoid score; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: internally validated improvement of the all-patient prediction strategy over routine prediction by the fixed response time; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; L: one proper or scientifically justified primary loss/scoring rule fixed for the actual outcome type; positive Gain favors new model; f base,f new: locked baseline/new prediction functions using only information available at the registered prediction time; test: untouched person/family/lab/site/time set as specified per thesis; p hat,y hat: frozen probability or continuous prediction; a cal,b cal: calibration intercept and slope, estimated on evaluation data for assessment; using them to update the model requires another evaluation set; O: organoid score only when actually produced and available by the prediction deadline; M: valid timely-culture indicator; f strategy: all-patient operational prediction strategy, including routine-model fallback; family code: PRED_FALLBACK
Sample size design: family: external/nested predictive-validation precision with target prevalence/variance, independent-unit holdout coverage and model complexity; formula: Simulate or bootstrap the independent-unit test loss/calibration/decision criterion; plan cases/events, noncases and independent sites/families/programs separately. No universal events-per-variable or participant N is asserted.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Use routine-model fallback whenever organoid result is absent/late; no random imputation of nonexistent scores; external validation requires another independent cohort.
Train/tune/test separation at the actual biological/participant/family/time level; no future, postoutcome or target-derived predictors.
Validation support must cover intended deployment; calibration and missing/failing measurements are reported with discrimination/loss.
A validation subset selected by culture success, assay viability, consent or data completeness is a different target unless the full strategy is explicitly evaluated.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Reject added value if organoid scores do not improve calibration beyond routine predictors or gains vanish after accounting for culture failures.; preanalysis gate: Use routine-model fallback whenever organoid result is absent/late; no random imputation of nonexistent scores; external validation requires another independent cohort.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Use routine-model fallback whenever organoid result is absent/late; no random imputation of nonexistent scores; external validation requires another independent cohort.
Repeat loss/calibration with independent-cluster resampling, prespecified subgroup/transport shifts and plausible missingness.
Compare fixed simple baselines, calibration and uncertainty coverage; report abstention/failure/turnaround rather than deleting hard cases.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Frozen prediction time/horizon, proper loss, baseline model, feature availability, independent split/holdout and target deployment population.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Use routine-model fallback whenever organoid result is absent/late; no random imputation of nonexistent scores; external validation requires another independent cohort.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/228; question: Does donor-specific organoid testing add prospective response information beyond routine molecular and clinical predictors?; aims: Quantify culture success and molecular fidelity in consecutively enrolled patients.
Compare locked response predictions with standard clinical models.
Identify transportability failures related to heterogeneity and missing cultures.; design verbatim planning only: Partner with an oncology biobank to enroll approximately 120 patients with metastatic colorectal cancer starting independently selected standard treatment. Generate predictions through existing approved workflows, keeping scores blinded to clinicians. Compare predictions with adjudicated response and patient-reported function; include all enrolled patients in feasibility analyses.; outcomes verbatim: Primary: Internally validated incremental prediction of response by a prespecified clinical assessment time, using the routine clinical model whenever no timely organoid result is available. Secondary: Culture coverage, calibration, turnaround, progression, function and conditional prediction performance among patients with timely successful cultures.; analysis verbatim inherited not authoritative if refined: Use nested validation for the combined all-patient prediction strategy, with adjustment for treatment and baseline disease burden. Prespecify response ascertainment, handling of progression or death before assessment and sensitivity analyses for missing assessments. Report culture success and turnaround for every enrolled patient; evaluate the routine-model fallback rather than impute organoid scores as if culture failure were random. Analyze successful-culture performance conditionally, with selection sensitivity analyses and restricted heterogeneity interactions. Claim external validation only after a genuinely independent cohort is evaluated; distinguish prediction from causal treatment selection.; falsification verbatim: Reject added value if organoid scores do not improve calibration beyond routine predictors or gains vanish after accounting for culture failures.; source ids unverified leads: patient-derived-organoids-colorectal-2026
ai-drug-discovery-practical-guidance-2026
clinicaltrials-registry; prospectus precedence: No

F39 — Autonomous laboratories under drift and proxy failure

Rank: 226
Cluster: AI biology, RNA and synthetic-biology validation
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: TRAJECTORY; allocation or exposure unit: independently initialized task trajectory randomized within task block to selection policy; independent sampling unit: independent trajectory; laboratory is an additional transport unit; repeated measurement unit: adaptive measurements within each isolated trajectory; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Fix binary trajectory success before registration; fractions may be secondary with declared denominators; adaptive measurements are not independent policy replications.
Primary estimand: policy-assignment effect on independent reproducibility of conclusions at an equal measurement budget
Candidate primary outcome: trajectory-level binary reproducibility success under a prespecified blinded reserved-reference criterion
Primary finalized: No
Mathematical model: type: binary risk-difference precision/power, simulated with actual cluster/trajectory structure; formulas: logit Pr(Y_i=1 | A_i,X_i) = alpha + z(A_i)^T theta + gamma^T X_i + b_cluster(i).
RD = E_X[p_hat(1,X) - p_hat(0,X)]; report absolute risk difference as well as any odds ratio.; variables: Y: trajectory-level binary reproducibility success under a prespecified blinded reserved-reference criterion; i: independent trajectory; laboratory is an additional transport unit; t: adaptive measurements within each isolated trajectory; the registered visit/time-window index, not another independent unit; A: independently initialized task trajectory randomized within task block to selection policy; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: policy-assignment effect on independent reproducibility of conclusions at an equal measurement budget; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; logit(p): log[p/(1-p)]; p hat(a,X): fitted outcome probability under arm a standardized to the prespecified population; RD: marginal risk difference; z(A): prespecified indicators for the actual arms or randomized policy; family code: TRAJECTORY
Sample size design: family: binary risk-difference precision/power, simulated with actual cluster/trajectory structure; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Fix binary trajectory success before registration; fractions may be secondary with declared denominators; adaptive measurements are not independent policy replications.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Trajectory is the randomized experimental unit; equal-budget confirmatory reproducibility is computed once per trajectory or analyzed with trajectory-level dependence.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Reject the reliability tradeoff if score-maximizing selection matches reproducibility and calibration across all transfer conditions.; preanalysis gate: Fix binary trajectory success before registration; fractions may be secondary with declared denominators; adaptive measurements are not independent policy replications.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Fix binary trajectory success before registration; fractions may be secondary with declared denominators; adaptive measurements are not independent policy replications.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Fix binary trajectory success before registration; fractions may be secondary with declared denominators; adaptive measurements are not independent policy replications.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/225; question: Do autonomous experiment-selection systems retain reliable conclusions when assay drift separates proxy success from true measurement quality?; aims: Define benign reference tasks with separate proxy and reference readouts.
Compare locked autonomous and conventional experiment-selection policies.
Test reproducibility after instrument and laboratory transfer.; design verbatim planning only: Use pre-existing nonliving commercial enzyme-reference assays at two automation laboratories, with a planning total of 300 coded measurements. Allocate this budget across multiple independently initialized task trajectories randomized in blocks to score-maximizing, uncertainty-aware or conventional balanced selection. Give each trajectory an equal measurement budget and keep its history separate. Introduce only approved measurement-calibration changes; no biological engineering, live organisms or hazardous optimization tasks are included.; outcomes verbatim: Primary: independent reproducibility of selected conclusions; secondary: calibration, false discoveries, resource use, and human-review burden.; analysis verbatim inherited not authoritative if refined: Analyze independent trajectory-level outcomes, with task-block randomization and instrument/day effects; cluster uncertainty within each adaptive trajectory. Reserve blinded reference measurements for each policy and transfer condition. Prespecify drift scenarios, the number of independent trajectories and corrected policy comparisons; do not count within-trajectory measurements as independent policy replications.; falsification verbatim: Reject the reliability tradeoff if score-maximizing selection matches reproducibility and calibration across all transfer conditions.; source ids unverified leads: self-driving-labs-review-2025
ai-drug-discovery-practical-guidance-2026
ai-biology-national-academies-2025; prospectus precedence: No

F40 — Synthetic-cell claims and interlaboratory model validity

Rank: 235
Cluster: AI biology, RNA and synthetic-biology validation
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: VAR_AGREE; allocation or exposure unit: coded nonliving reference material assigned across existing laboratory procedures; independent sampling unit: independent material/reference sample and laboratory; repeated measurement unit: instrument/analyst/property replicate; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: measurement agreement/reliability; condition contrasts are causal only where the condition itself was randomized; specific constraint: Four laboratories give limited lab-population inference; censored/missing physical readouts need explicit models; no claim about living cells or organisms.
Primary estimand: cross-laboratory agreement and variance partition for prespecified physical properties
Candidate primary outcome: permeability or stability property in its physical unit, selected before confirmation
Primary finalized: No
Mathematical model: type: paired hierarchical agreement/bias/limits precision with independent participants/donors/labs and replicate allocation; formulas: D_ij = S_ij - R_ij = beta_0 + beta_context^T C_ij + u_i + b_lab + epsilon_ij.
Context-specific agreement limits use beta_0 + beta_context^T C +/- z_(1-alpha/2) sqrt(Var(D_new | target context)); distinguish a new person from a repeat reading.
If reference error is material, S_ij = T_ij + b_S + e_S,ij and R_ij = T_ij + b_R + e_R,ij; identifiability needs reference validation/replicates or stated constraints.
Partition material, instrument, analyst and laboratory covariance; with few laboratories distinguish a finite-laboratory contrast from a population variance claim.; variables: Y: permeability or stability property in its physical unit, selected before confirmation; i: independent material/reference sample and laboratory; t: instrument/analyst/property replicate; the registered visit/time-window index, not another independent unit; A: coded nonliving reference material assigned across existing laboratory procedures; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: cross-laboratory agreement and variance partition for prespecified physical properties; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; S ij: sensor/test result in the stated endpoint unit; R ij: paired validated reference result in the same unit; D ij: paired test-minus-reference error, matching the per-thesis outcome; C ij: prespecified context, firmware/device and subgroup features; T ij: latent underlying quantity; not automatically identifiable; b S,b R: test and reference biases; e S,e R: test and reference measurement errors; b lab: laboratory effect when multiple laboratories actually sampled; D new: prediction error for the prespecified future measurement context; family code: VAR_AGREE
Sample size design: family: paired hierarchical agreement/bias/limits precision with independent participants/donors/labs and replicate allocation; formula: Plan confidence width for bias and agreement limits by hierarchical simulation; n readings is not n independent people. Acceptance limits in physical units must be independently justified.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Four laboratories give limited lab-population inference; censored/missing physical readouts need explicit models; no claim about living cells or organisms.
Reference calibration, matched time/unit and appropriately modeled reference error; correlation is not agreement.
Future-context prediction assumptions must match held-out persons/labs/conditions; missing/failing readings are outcomes of the measurement system.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Reject the measurement-variation hypothesis if performance rankings remain stable and agreement is high under local methods.; preanalysis gate: Four laboratories give limited lab-population inference; censored/missing physical readouts need explicit models; no claim about living cells or organisms.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Four laboratories give limited lab-population inference; censored/missing physical readouts need explicit models; no claim about living cells or organisms.
Assess proportional/heteroscedastic error, reference uncertainty and subgroup-specific failure, not just a global correlation.
Repeat estimates with and without questionable quality strata under predeclared rules; report excluded/failing measurements explicitly.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Four laboratories give limited lab-population inference; censored/missing physical readouts need explicit models; no claim about living cells or organisms.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/234; question: Do nonliving synthetic-cell reference models support reproducible permeability and stability claims across laboratories?; aims: Develop a shared nonoperational reference-reporting ontology.
Conduct a blinded interlaboratory comparison of existing nonliving models.
Test which reported properties predict reproducibility across contexts.; design verbatim planning only: Coordinate four laboratories using a planning panel of 60 commercially available or archived genome-free membrane-model reference samples. Compare local reporting with a harmonized analysis of coded measurements under existing institutional procedures. Use independent reference materials and hold out one laboratory for validation; no self-replication, biological release, or new organism construction is studied.; outcomes verbatim: Primary: cross-laboratory agreement for prespecified physical properties; secondary: uncertainty coverage, reporting completeness, and transfer prediction.; analysis verbatim inherited not authoritative if refined: Partition material, instrument, analyst, and laboratory variance; prespecify acceptable agreement and control multiple property comparisons. Model missing or censored measurements explicitly.; falsification verbatim: Reject the measurement-variation hypothesis if performance rankings remain stable and agreement is high under local methods.; source ids unverified leads: synthetic-cell-governance-2026
self-driving-labs-review-2025
ai-biology-national-academies-2025; prospectus precedence: No

L01 — Validated ex-vivo muscle autophagic flux and restored physical reserve

Rank: 45
Cluster: Longevity/geroscience
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT_MECHANISM; allocation or exposure unit: participant exercise assignment; separately randomized donor-culture allocation; independent sampling unit: participant; donor for ex-vivo validation; repeated measurement unit: challenge, biopsy/culture and assay replicate; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Biopsy consent and viable-assay selection limit mechanism inference; flux is a postassignment mediator, not a randomized explanatory variable.
Primary estimand: exercise assignment effect on walking reserve after the standardized challenge
Candidate primary outcome: walking-reserve endpoint after a fixed safe challenge
Primary finalized: No
Mathematical model: type: contrast precision/power by design-matrix or simulation planning; formulas: E(Y_it | A_i, X_i, Y_i0) = alpha_t + z(A_i)^T theta_t + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Delta = sum_t w_t c^T theta_t; estimate by baseline-adjusted marginal standardization, with unit/cluster-appropriate inference.; variables: Y: walking-reserve endpoint after a fixed safe challenge; i: participant; donor for ex-vivo validation; t: challenge, biopsy/culture and assay replicate; the registered visit/time-window index, not another independent unit; A: participant exercise assignment; separately randomized donor-culture allocation; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: exercise assignment effect on walking reserve after the standardized challenge; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; z(A): prespecified indicators for the actual randomized arms; two arms reduce to one indicator; family code: RCT_MECHANISM
Sample size design: family: contrast precision/power by design-matrix or simulation planning; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Biopsy consent and viable-assay selection limit mechanism inference; flux is a postassignment mediator, not a randomized explanatory variable.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Failed viability or dynamic-assay validation precludes flux inference; functional gains without validated flux change reject the proposed mediation.; preanalysis gate: Biopsy consent and viable-assay selection limit mechanism inference; flux is a postassignment mediator, not a randomized explanatory variable.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Biopsy consent and viable-assay selection limit mechanism inference; flux is a postassignment mediator, not a randomized explanatory variable.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Biopsy consent and viable-assay selection limit mechanism inference; flux is a postassignment mediator, not a randomized explanatory variable.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/44; question: When exercise improves older adults’ physical reserve, does validated ex-vivo muscle autophagic flux help explain improvement beyond static markers?; aims: Validate dynamic turnover in fresh viable muscle; restrict banked frozen samples to static-marker measurements.
Measure repeated ex-vivo flux and reserve changes during supervised randomized exercise versus attention control.
Test mediation with participant-derived myotubes and independent cohort replication.; design verbatim planning only: Population model: Adults aged 65–80, fresh viable muscle biopsies with demonstrated assay viability, and derived muscle cultures; separate biopsy consent. Comparator: Progressive supervised exercise versus attention control. Causal replication: Randomization, validated dynamic turnover assays, blinded readouts, culture perturbation, and external cohort replication; ex-vivo findings require in-vivo interpretation caution. Sample planning: Plan approximately 100 participants, 40 paired biopsies; pilot variance determines final recruitment.; outcomes verbatim: Primary: walking reserve after a standardized challenge. Secondary: muscle flux, strength recovery, falls, and training tolerance.; analysis verbatim inherited not authoritative if refined: Prespecify mediation assumptions and adjust baseline activity, nutrition, and medications; control endpoint multiplicity and randomize paired specimens across assay batches.; falsification verbatim: Failed viability or dynamic-assay validation precludes flux inference; functional gains without validated flux change reject the proposed mediation.; source ids unverified leads: autophagy-exercise-human-meta-2023
resistance-sarcopenia-meta-2026; prospectus precedence: No

L02 — Tissue NAD target engagement and the limits of blood response

Rank: 69
Cluster: Longevity/geroscience
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: MIXED_EVIDENCE; allocation or exposure unit: original regulated precursor-trial assignment; validation cohort not necessarily randomized; independent sampling unit: participant within original trial or separate validation cohort; repeated measurement unit: exertion/recovery assay and visit; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: A blood NAD increase cannot identify an individual muscle benefit; mediation and validation require distinct estimands and consent-compatible specimen access.
Primary estimand: original-trial assignment effect on exertional fatigability, plus separately labeled biomarker/energetics association
Candidate primary outcome: standardized exertional-fatigability score
Primary finalized: No
Mathematical model: type: contrast precision/power by design-matrix or simulation planning; formulas: E(Y_it | A_i, X_i, Y_i0) = alpha_t + z(A_i)^T theta_t + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Delta = sum_t w_t c^T theta_t; estimate by baseline-adjusted marginal standardization, with unit/cluster-appropriate inference.; variables: Y: standardized exertional-fatigability score; i: participant within original trial or separate validation cohort; t: exertion/recovery assay and visit; the registered visit/time-window index, not another independent unit; A: original regulated precursor-trial assignment; validation cohort not necessarily randomized; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: original-trial assignment effect on exertional fatigability, plus separately labeled biomarker/energetics association; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; z(A): prespecified indicators for the actual randomized arms; two arms reduce to one indicator; family code: MIXED_EVIDENCE
Sample size design: family: contrast precision/power by design-matrix or simulation planning; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: A blood NAD increase cannot identify an individual muscle benefit; mediation and validation require distinct estimands and consent-compatible specimen access.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The formula applies to the original randomized comparison only; observational validation/cohort associations receive separate labels and cannot be pooled as randomized arms.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Uniform blood–muscle coupling or absent treatment-by-impairment interaction would undermine the proposed explanation for heterogeneous functional response.; preanalysis gate: A blood NAD increase cannot identify an individual muscle benefit; mediation and validation require distinct estimands and consent-compatible specimen access.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: A blood NAD increase cannot identify an individual muscle benefit; mediation and validation require distinct estimands and consent-compatible specimen access.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
A blood NAD increase cannot identify an individual muscle benefit; mediation and validation require distinct estimands and consent-compatible specimen access.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/68; question: Does a blood NAD increase identify older adults whose skeletal muscle energetics and exertional fatigue actually improve?; aims: Quantify paired blood and muscle NAD-pathway variability using existing trial samples.
Relate randomized precursor-associated tissue changes to energetics and fatigue trajectories.
Test baseline impairment as a moderator in an independently recruited validation cohort.; design verbatim planning only: Population model: Older adults with measurable fatigability, using an existing regulated randomized precursor study and validation cohort. Comparator: Placebo and baseline-matched unimpaired participants. Causal replication: Treatment assignment anchors mediation; blinded tissue assays and external moderator validation. Sample planning: Assume 120 archived participants and 50 validation volunteers; availability and pilot precision govern scope.; outcomes verbatim: Primary: exertional fatigability. Secondary: muscle energetic recovery, tissue metabolites, walking endurance, and persistent adverse events.; analysis verbatim inherited not authoritative if refined: Address renal function, diet, activity, and baseline metabolic status; limit moderator tests, correct secondary multiplicity, and balance treatment groups across analytical batches.; falsification verbatim: Uniform blood–muscle coupling or absent treatment-by-impairment interaction would undermine the proposed explanation for heterogeneous functional response.; source ids unverified leads: nad-wellness-review-2026
nad-boosters-comparison-rct-2026
nmn-safety-meta-2026; prospectus precedence: No

L03 — Multimorbidity endpoints when death and ascertainment alter apparent geroscience benefit

Rank: 75
Cluster: Longevity/geroscience
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: ENDPOINT_COMPARE; allocation or exposure unit: original trial allocation and separately simulated known-truth processes; independent sampling unit: independent trial/cohort and simulation replicate; repeated measurement unit: diagnosis, disability and terminal event; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: endpoint operating-characteristic comparison under explicit simulated truth and observed archived trials; specific constraint: Death changes disease opportunity and diagnostic intensity changes recorded disease; patient priorities define the reference, not a universal disease-count truth.
Primary estimand: frequency of contradictory treatment conclusions across fixed endpoint definitions under known scenarios and archived trial contrasts
Candidate primary outcome: indicator that an endpoint conclusion conflicts with the prespecified joint survival/function criterion
Primary finalized: No
Mathematical model: type: terminal/recurrent-event simulation and independent-trial paired endpoint-comparison precision; formulas: C_hat=(1/B) sum_b 1{conclusion(method,b) conflicts with registered joint survival/function criterion in scenario b}.
For archived trial s, compare treatment estimates under each endpoint using shared participant bootstrap/covariance; separate conflicting point-estimate signs from conflicting clinically meaningful decisions.; variables: Y: indicator that an endpoint conclusion conflicts with the prespecified joint survival/function criterion; i: independent trial/cohort and simulation replicate; t: diagnosis, disability and terminal event; the registered visit/time-window index, not another independent unit; A: original trial allocation and separately simulated known-truth processes; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: frequency of contradictory treatment conclusions across fixed endpoint definitions under known scenarios and archived trial contrasts; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; B: independent simulated scenarios/replicates with specified truth and priority rule; s: independent archived trial; C hat: scenario-specific contradiction frequency; criterion: prespecified patient-informed survival/function ordering, not an assumed universal disease-count ground truth; family code: ENDPOINT_COMPARE
Sample size design: family: terminal/recurrent-event simulation and independent-trial paired endpoint-comparison precision; formula: Use design-specific simulation or independent-unit resampling for the stated estimand and target confidence width/power; preserve all allocation, censoring, clustering and measurement-error structure.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Death changes disease opportunity and diagnostic intensity changes recorded disease; patient priorities define the reference, not a universal disease-count truth.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Known truth exists only in simulations; trial archives show agreement/conflict, not an unquestionable truth about which endpoint is correct.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Stable first-event conclusions across mortality and surveillance scenarios, with equivalent functional alignment, would refute the proposed endpoint advantage.; preanalysis gate: Death changes disease opportunity and diagnostic intensity changes recorded disease; patient priorities define the reference, not a universal disease-count truth.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Death changes disease opportunity and diagnostic intensity changes recorded disease; patient priorities define the reference, not a universal disease-count truth.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Death changes disease opportunity and diagnostic intensity changes recorded disease; patient priorities define the reference, not a universal disease-count truth.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/74; question: Can common multimorbidity endpoints imply geroscience benefit when differential mortality or diagnostic intensity actually explains fewer recorded diseases?; aims: Define outcome priorities with diverse older adults and map candidate multimorbidity estimands.
Quantify endpoint reversals under known mortality, recurrence, and ascertainment mechanisms using calibrated simulations.
Validate selected estimands against functional trajectories and randomized prevention-trial contrasts in independent archives.; design verbatim planning only: Population model: Existing longitudinal aging cohorts and randomized prevention-trial archives with diagnoses, death, function, and healthcare-contact histories. Comparator: First-event composites, recurrent-event summaries, and terminal-event-aware disability-free survival. Causal replication: Known-truth simulations, randomized treatment contrasts, blinded adjudication, and held-out-cohort validation. Sample planning: Seek 5,000–10,000 archived participants and 40–60 priority-elicitation volunteers; event counts and access determine precision.; outcomes verbatim: Primary: frequency of clinically contradictory treatment conclusions. Secondary: functional alignment, ascertainment sensitivity, subgroup calibration, and patient-priority agreement.; analysis verbatim inherited not authoritative if refined: Address baseline multimorbidity, diagnostic intensity, informative dropout, and competing death; prespecify estimands, correct endpoint multiplicity, and harmonize coding and laboratory batch changes.; falsification verbatim: Stable first-event conclusions across mortality and surveillance scenarios, with equivalent functional alignment, would refute the proposed endpoint advantage.; source ids unverified leads: select-semaglutide-cardiovascular-2023
pearl-rapamycin-healthspan-rct-2025
metformin-antiaging-uncertainty-2025; prospectus precedence: No

L04 — mTOR modulation and recovery from anabolic stress in aging muscle

Rank: 107
Cluster: Longevity/geroscience
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT_MECHANISM; allocation or exposure unit: original rapamycin/placebo trial assignment; separate older-donor culture allocation; independent sampling unit: participant; independent donor for cultures; repeated measurement unit: strength-recovery time point and specimen assay; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Physiology add-on consent may select participants; resting healthspan and culture signaling cannot substitute for human functional recovery.
Primary estimand: original-trial assignment effect on the fixed strength-recovery contrast after standardized loading
Candidate primary outcome: strength recovery at the registered postchallenge time or prespecified recovery-curve summary
Primary finalized: No
Mathematical model: type: contrast precision/power by design-matrix or simulation planning; formulas: E(Y_it | A_i, X_i, Y_i0) = alpha_t + z(A_i)^T theta_t + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Delta = sum_t w_t c^T theta_t; estimate by baseline-adjusted marginal standardization, with unit/cluster-appropriate inference.; variables: Y: strength recovery at the registered postchallenge time or prespecified recovery-curve summary; i: participant; independent donor for cultures; t: strength-recovery time point and specimen assay; the registered visit/time-window index, not another independent unit; A: original rapamycin/placebo trial assignment; separate older-donor culture allocation; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: original-trial assignment effect on the fixed strength-recovery contrast after standardized loading; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; z(A): prespecified indicators for the actual randomized arms; two arms reduce to one indicator; family code: RCT_MECHANISM
Sample size design: family: contrast precision/power by design-matrix or simulation planning; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Physiology add-on consent may select participants; resting healthspan and culture signaling cannot substitute for human functional recovery.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Equivalent recovery despite verified pathway engagement, or no relation between recovery and function, would reject the proposed hidden vulnerability.; preanalysis gate: Physiology add-on consent may select participants; resting healthspan and culture signaling cannot substitute for human functional recovery.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Physiology add-on consent may select participants; resting healthspan and culture signaling cannot substitute for human functional recovery.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Physiology add-on consent may select participants; resting healthspan and culture signaling cannot substitute for human functional recovery.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/106; question: Does experimental mTOR modulation alter muscle recovery capacity even when resting healthspan measures show little change?; aims: Develop reproducible recovery endpoints from supervised resistance challenges in older adults.
Embed validated recovery endpoints in an investigator-led experimental rapamycin trial add-on.
Use donor-derived muscle systems to separate pathway engagement from contractile repair capacity.; design verbatim planning only: Population model: Investigator-led rapamycin trial participants consenting to a physiology add-on, plus older-donor muscle cultures. Comparator: Trial placebo and matched untreated cultures. Causal replication: Existing randomized assignment, repeated challenges, blinded readouts, and independent donor replication. Sample planning: Plan 60–100 trial participants and 20 donors; add-on access and pilot recovery variance determine precision.; outcomes verbatim: Primary: recovery of strength after standardized loading. Secondary: anabolic signaling, lean tissue trajectory, fatigability, and clinical adverse events.; analysis verbatim inherited not authoritative if refined: Adjust baseline training, energy intake, sex, and metabolic disease; preregister the recovery contrast, control secondary multiplicity, and balance specimens across batches.; falsification verbatim: Equivalent recovery despite verified pathway engagement, or no relation between recovery and function, would reject the proposed hidden vulnerability.; source ids unverified leads: pearl-rapamycin-healthspan-rct-2025
exercise-mimetics-review-2026
resistance-sarcopenia-meta-2026; prospectus precedence: No

L05 — Exercise-specific cross-species transportability of functional recovery

Rank: 159
Cluster: Longevity/geroscience
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: SPECIES_TRANSPORT; allocation or exposure unit: exercise assignment randomized separately within mouse and human study; independent sampling unit: mouse (or cage if allocation/exposure shared); human participant; independent study/lab for transport; repeated measurement unit: repeated challenge within organism; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: cross-species transport for the tested exercise contrast only; one contrast cannot identify a general intervention-ranking map; specific constraint: Human-mouse scale linking and transport assumptions are unverified; one exercise contrast cannot validate cross-intervention rankings.
Primary estimand: error of transported human functional treatment-effect predictions for the tested exercise contrast
Candidate primary outcome: species-specific functional effect mapped to a prespecified common scale
Primary finalized: No
Mathematical model: type: two-species randomized-contrast and transport-link precision with independent-laboratory validation; formulas: Within each species s, Delta_s=E[Y_s|exercise]-E[Y_s|control] under that species randomization and prespecified scale.
Prediction error for the tested contrast = L(Delta_human, g(Delta_mouse,R_mouse;phi)); fit/lock the linking function using independent data and propagate uncertainty in both species contrasts.; variables: Y: species-specific functional effect mapped to a prespecified common scale; i: mouse (or cage if allocation/exposure shared); human participant; independent study/lab for transport; t: repeated challenge within organism; the registered visit/time-window index, not another independent unit; A: exercise assignment randomized separately within mouse and human study; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: error of transported human functional treatment-effect predictions for the tested exercise contrast; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; s: mouse or human study population; Delta s: species-specific randomized functional treatment effect; R mouse: prespecified mouse recovery dimensions; g,phi: externally identified/locked cross-species scale/transport map and its parameters; L: registered prediction loss on the human functional scale; family code: SPECIES_TRANSPORT
Sample size design: family: two-species randomized-contrast and transport-link precision with independent-laboratory validation; formula: Use design-specific simulation or independent-unit resampling for the stated estimand and target confidence width/power; preserve all allocation, censoring, clustering and measurement-error structure.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Human-mouse scale linking and transport assumptions are unverified; one exercise contrast cannot validate cross-intervention rankings.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Scale/measurement invariance and transport map are substantive assumptions; cage exposure may define mouse experimental units.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Absent measurement invariance, discordant intervention effects, or no advantage over resting benchmarks would reject the proposed translational recovery framework.; preanalysis gate: Human-mouse scale linking and transport assumptions are unverified; one exercise contrast cannot validate cross-intervention rankings.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Human-mouse scale linking and transport assumptions are unverified; one exercise contrast cannot validate cross-intervention rankings.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Human-mouse scale linking and transport assumptions are unverified; one exercise contrast cannot validate cross-intervention rankings.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/158; question: Within supervised exercise studies, do recovery dimensions in aged mice predict human functional effects more accurately than resting frailty or molecular aging measures?; aims: Establish comparable recovery dimensions using gait, activity, and energetic measurements in aged mice and older adults.
Test randomized supervised-exercise effects within each species and evaluate prespecified cross-species transport predictions.
Replicate conserved dimensions in an independent animal laboratory and held-out human cohort.; design verbatim planning only: Population model: Aged mice and screened adults aged 65–85, with repeated noninjurious exertional challenges. Comparator: Exercise versus usual-activity controls within each species; resting frailty and molecular-score benchmarks. Causal replication: Within-species randomization, blinded outcomes, explicit transport assumptions, and independent laboratory/cohort replication. Sample planning: Plan 60–80 mice and approximately 100 adults; pilot reliability and species-interaction uncertainty determine final allocation.; outcomes verbatim: Primary: accuracy of transported functional-effect predictions for the supervised-exercise contrast. Secondary: recovery repeatability, mobility change, energetic recovery, measurement invariance, and adverse events.; analysis verbatim inherited not authoritative if refined: Model sex, baseline reserve, disease burden, and species-specific scaling; prespecify recovery dimensions, control interaction multiplicity, and calibrate device, specimen, and laboratory batches. Restrict transport conclusions to the tested exercise contrast; cross-intervention ranking requires additional independent intervention mechanisms.; falsification verbatim: Absent measurement invariance, discordant intervention effects, or no advantage over resting benchmarks would reject the proposed translational recovery framework.; source ids unverified leads: exercise-mimetics-review-2026
resistance-sarcopenia-meta-2026
autophagy-exercise-human-meta-2023; prospectus precedence: No

L06 — Selective senescence removal versus repair competence in aged connective tissue

Rank: 39
Cluster: Longevity/geroscience
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: LAB; allocation or exposure unit: within-donor culture/explant allocated to state-enriched or control conditions; independent sampling unit: independent tissue donor, with independently allocated explant/culture units; repeated measurement unit: mechanical and repair assay/time replicate; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: controlled in-vitro/ex-vivo allocation contrast when culture allocation is randomized; no direct human therapeutic-efficacy claim; specific constraint: Multiple explants do not increase donor replication; depletion can affect non-senescent cells; mechanism requires orthogonal specificity controls.
Primary estimand: within-donor condition effect on mechanical recovery, with repair preserved as a separate gate
Candidate primary outcome: prespecified tissue mechanical-recovery endpoint
Primary finalized: No
Mathematical model: type: multilevel donor-by-preparation/condition experimental contrast precision; independent-donor and independent-batch allocation simulation; formulas: Y_djbrt = alpha + z(A_djb)^T theta + gamma^T X_d + u_d + v_b + q_dj + epsilon_djbrt.
Delta_model = c^T theta standardized to sampled donors/conditions; r is a technical repeat, not another donor or manufacturing batch.; variables: Y: prespecified tissue mechanical-recovery endpoint; i: independent tissue donor, with independently allocated explant/culture units; t: mechanical and repair assay/time replicate; the registered visit/time-window index, not another independent unit; A: within-donor culture/explant allocated to state-enriched or control conditions; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: within-donor condition effect on mechanical recovery, with repair preserved as a separate gate; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; d: independent tissue/cell donor; j: independently allocated culture/explant/construct unit within donor; b: independent preparation/manufacturing/differentiation batch as appropriate; r: technical assay repeat or cell observation within allocated culture; z(A): indicators for the named controlled preparation/state/reference conditions; u d,v b,q dj: donor, batch and allocated-culture effects; crossed donor/preparation structure must follow actual allocation; X d: pretreatment donor/model attributes and baseline function only; family code: LAB
Sample size design: family: multilevel donor-by-preparation/condition experimental contrast precision; independent-donor and independent-batch allocation simulation; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Multiple explants do not increase donor replication; depletion can affect non-senescent cells; mechanism requires orthogonal specificity controls.
Randomize independently allocated cultures/explants within blocking factors; donor and preparation/batch are the biological replication units for transport.
Technical replicates/cells/fields reduce measurement error but do not increase independent donor count.
Culture viability, identity loss, assay failure and lineage selection are outcomes/selection processes; do not silently condition treatment comparisons on successful posttreatment models.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Equal repair loss after selective and broad depletion, or absent mechanics improvement, would reject the proposed therapeutic separation.; preanalysis gate: Multiple explants do not increase donor replication; depletion can affect non-senescent cells; mechanism requires orthogonal specificity controls.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Multiple explants do not increase donor replication; depletion can affect non-senescent cells; mechanism requires orthogonal specificity controls.
Leave donors and manufacturing batches out; compare crossed/nested covariance structures supported by the design.
Report viable/failed model coverage, censored assays, donor-context heterogeneity and robustness to contamination/handling controls.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Confirmed donor/preparation access and allocated culture design; validated functional assay, dynamic range, technical error, detection limits and identity/viability gates.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Multiple explants do not increase donor replication; depletion can affect non-senescent cells; mechanism requires orthogonal specificity controls.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/38; question: Does reducing senescent fibroblast burden improve connective-tissue mechanics without compromising repair after a standardized injury?; aims: Define fibroblast states associated with stiff aged human connective tissue.
Compare selective and broad senescence-targeting strategies in matched tissue explants.
Test whether state-specific depletion predicts mechanical recovery across independent donors.; design verbatim planning only: Population model: Discarded surgical tendon and dermal tissue from older adults, with matched organotypic cultures. Comparator: Vehicle, nonselective depletion, and state-enriched experimental conditions. Causal replication: Within-donor randomized allocation, blinded mechanics, orthogonal senescence assays, and independent donor replication. Sample planning: Plan 35–50 donors across two tissues; pilot variance guides allocation.; outcomes verbatim: Primary: tissue mechanical recovery. Secondary: wound closure, matrix architecture, viable cell identity, and persistent senescence signatures.; analysis verbatim inherited not authoritative if refined: Control donor diabetes, medication exposure, tissue site, and baseline damage; prespecify state contrasts, adjust molecular multiplicity, and block processing by donor and batch.; falsification verbatim: Equal repair loss after selective and broad depletion, or absent mechanics improvement, would reject the proposed therapeutic separation.; source ids unverified leads: senolytics-clinical-translation-2025
collagen-umbrella-review-2026
collagen-tendon-rct-2025; prospectus precedence: No

L07 — Separating vascular adaptation from expectancy in hyperbaric aging research

Rank: 182
Cluster: Longevity/geroscience
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT_MECHANISM; allocation or exposure unit: participant randomized supervised hyperbaric versus specialist sham; independent sampling unit: participant; repeated measurement unit: exercise/perfusion assessment visit; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Sham may be physiologically active; postassignment expectancy or perfusion adjustment does not isolate a pure vascular or pressure mechanism.
Primary estimand: assignment effect on objectively measured exercise capacity at the durable follow-up
Candidate primary outcome: prespecified objective exercise-capacity endpoint
Primary finalized: No
Mathematical model: type: contrast precision/power by design-matrix or simulation planning; formulas: E(Y_it | A_i, X_i, Y_i0) = alpha_t + z(A_i)^T theta_t + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Delta = sum_t w_t c^T theta_t; estimate by baseline-adjusted marginal standardization, with unit/cluster-appropriate inference.; variables: Y: prespecified objective exercise-capacity endpoint; i: participant; t: exercise/perfusion assessment visit; the registered visit/time-window index, not another independent unit; A: participant randomized supervised hyperbaric versus specialist sham; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: assignment effect on objectively measured exercise capacity at the durable follow-up; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; z(A): prespecified indicators for the actual randomized arms; two arms reduce to one indicator; family code: RCT_MECHANISM
Sample size design: family: contrast precision/power by design-matrix or simulation planning; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Sham may be physiologically active; postassignment expectancy or perfusion adjustment does not isolate a pure vascular or pressure mechanism.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Capacity changes without perfusion changes, or similar improvement under sham, would reject the vascular mediation hypothesis.; preanalysis gate: Sham may be physiologically active; postassignment expectancy or perfusion adjustment does not isolate a pure vascular or pressure mechanism.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Sham may be physiologically active; postassignment expectancy or perfusion adjustment does not isolate a pure vascular or pressure mechanism.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Sham may be physiologically active; postassignment expectancy or perfusion adjustment does not isolate a pure vascular or pressure mechanism.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/181; question: Which objective vascular changes explain reported improvements in older-adult exercise capacity after supervised hyperbaric exposure?; aims: Reanalyze available trial cardiopulmonary and perfusion data with blinded endpoints.
Validate perfusion–capacity coupling in a sham-controlled mechanistic substudy.
Test persistence and baseline vascular impairment as prespecified effect modifiers.; design verbatim planning only: Population model: Older adults eligible for investigator-led supervised hyperbaric research, excluding contraindications. Comparator: Specialist-designed sham with measured expectancy and physiological activity. Causal replication: Random assignment, blinded physiological assessment, expectancy measurement, and independent-site endpoint replication. Sample planning: Plan 70–90 participants; sham feasibility and pilot perfusion variance determine final sample.; outcomes verbatim: Primary: objectively measured exercise capacity. Secondary: peripheral perfusion, exertional symptoms, durability, adverse events, and treatment burden.; analysis verbatim inherited not authoritative if refined: Prespecify vascular mediation; address baseline fitness, anemia, medication use, and expectancy, control multiple perfusion endpoints, and calibrate instruments across sites and batches.; falsification verbatim: Capacity changes without perfusion changes, or similar improvement under sham, would reject the vascular mediation hypothesis.; source ids unverified leads: hbot-physical-performance-rct-2024
exercise-mimetics-review-2026; prospectus precedence: No

L08 — Proteostatic recovery kinetics as an early marker of functional aging

Rank: 29
Cluster: Longevity/geroscience
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: PRED_SURV; allocation or exposure unit: natural recovery physiology measured before later mobility decline; independent sampling unit: participant in development or independent validation cohort; repeated measurement unit: challenge protein time point and outcome follow-up; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: prediction, calibration or model transportability; prediction gain is not clinical efficacy, mechanistic necessity or causal treatment selection; specific constraint: Death competes with decline and illness affects sampling; recovery half-times need reliable sampling and do not by themselves identify a causal pathway.
Primary estimand: incremental calibrated prediction of incident mobility decline beyond resting protein benchmarks
Candidate primary outcome: incident mobility decline by a fixed horizon or time to decline
Primary finalized: No
Mathematical model: type: landmark/competing-risk prediction precision with event-free risk set, event counts, censoring and external calibration; formulas: At fixed landmark ell, use only H_i(ell) among alive event-free participants: F_1(t | H(ell))=Pr(T<=t,J=event | T>ell,H(ell)).
Evaluate calibrated cause-specific cumulative-incidence predictions and proper censoring-aware Brier/loss over (ell,tau]; do not treat death as independent censoring for the event risk.; variables: Y: incident mobility decline by a fixed horizon or time to decline; i: participant in development or independent validation cohort; t: challenge protein time point and outcome follow-up; the registered visit/time-window index, not another independent unit; A: natural recovery physiology measured before later mobility decline; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: incremental calibrated prediction of incident mobility decline beyond resting protein benchmarks; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; L: one proper or scientifically justified primary loss/scoring rule fixed for the actual outcome type; positive Gain favors new model; f base,f new: locked baseline/new prediction functions using only information available at the registered prediction time; test: untouched person/family/lab/site/time set as specified per thesis; p hat,y hat: frozen probability or continuous prediction; a cal,b cal: calibration intercept and slope, estimated on evaluation data for assessment; using them to update the model requires another evaluation set; ell: predeclared prediction landmark; H i(ell): history available no later than ell; T,J: first relevant event time and event type, including competing death; tau: fixed prediction horizon; F 1: conditional cumulative incidence for the primary event; early/prelandmark events reported separately; family code: PRED_SURV
Sample size design: family: landmark/competing-risk prediction precision with event-free risk set, event counts, censoring and external calibration; formula: Simulate or bootstrap the independent-unit test loss/calibration/decision criterion; plan cases/events, noncases and independent sites/families/programs separately. No universal events-per-variable or participant N is asserted.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Death competes with decline and illness affects sampling; recovery half-times need reliable sampling and do not by themselves identify a causal pathway.
Train/tune/test separation at the actual biological/participant/family/time level; no future, postoutcome or target-derived predictors.
Validation support must cover intended deployment; calibration and missing/failing measurements are reported with discrimination/loss.
A validation subset selected by culture success, assay viability, consent or data completeness is a different target unless the full strategy is explicitly evaluated.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: No improvement over resting proteins in held-out participants, or poor challenge repeatability, would falsify the resilience-marker claim.; preanalysis gate: Death competes with decline and illness affects sampling; recovery half-times need reliable sampling and do not by themselves identify a causal pathway.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Death competes with decline and illness affects sampling; recovery half-times need reliable sampling and do not by themselves identify a causal pathway.
Repeat loss/calibration with independent-cluster resampling, prespecified subgroup/transport shifts and plausible missingness.
Compare fixed simple baselines, calibration and uncertainty coverage; report abstention/failure/turnaround rather than deleting hard cases.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Frozen prediction time/horizon, proper loss, baseline model, feature availability, independent split/holdout and target deployment population.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Death competes with decline and illness affects sampling; recovery half-times need reliable sampling and do not by themselves identify a causal pathway.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/28; question: Do recovery kinetics after a safe physiological challenge predict functional decline better than resting aging-associated protein concentrations?; aims: Identify reproducible protein recovery modules after standardized supervised exercise.
Test whether recovery modules predict subsequent mobility decline beyond baseline protein levels.
Replicate prediction in an independent aging cohort with functional follow-up.; design verbatim planning only: Population model: Community-dwelling adults aged 60–85 spanning mobility levels; no experimental medicines. Comparator: Age-matched high-reserve participants and each participant’s resting trajectory. Causal replication: Repeated within-person challenges, temporal ordering, negative-control modules, and external validation; prediction is distinguished from causation. Sample planning: Plan 180 participants with 12–24-month follow-up and 60 independent validation participants.; outcomes verbatim: Primary: incident mobility decline. Secondary: recovery half-times, fatigue, strength, and incremental predictive calibration.; analysis verbatim inherited not authoritative if refined: Adjust disease burden, activity, kidney function, and recent illness; prespecify modules, control protein-level discovery multiplicity, and randomize longitudinal samples across batches.; falsification verbatim: No improvement over resting proteins in held-out participants, or poor challenge repeatability, would falsify the resilience-marker claim.; source ids unverified leads: serum-proteome-aging-cohort-2025
autophagy-exercise-human-meta-2023
aging-clocks-critical-review-2026; prospectus precedence: No

L09 — Endocrine transition, muscle quality, and the interpretation of normal aging

Rank: 113
Cluster: Longevity/geroscience
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: OBS_TRAJECTORY; allocation or exposure unit: naturally occurring endocrine transition, no hormone assignment; independent sampling unit: participant within independent cohort; repeated measurement unit: repeat hormones, muscle image and activity epoch; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: association or prospective prediction; temporal order and model adjustment alone do not identify a mechanism or treatment effect; specific constraint: Chronological age, transition timing and activity may be collinear; time-varying adjustment can remove pathways or induce bias; negative-control windows support falsification only.
Primary estimand: within-person transition-associated change in muscle quality after prespecified baseline/time adjustment
Candidate primary outcome: imaging-derived muscle-quality measure
Primary finalized: No
Mathematical model: type: longitudinal association/reliability or held-out prediction precision simulation, not causal-efficacy power; formulas: Y_it = alpha + u_i + beta_W (B_it-Bbar_i) + beta_B Bbar_i + gamma^T X_it + f_time(t) + b_study + epsilon_it; u_i is a random intercept with a declared conditional mean/covariance model, not unrestricted fixed person effects.
B_observed,it = B_true,it + e_it; estimate measurement variance from replicates/validation or report sensitivity rather than assume it known.; variables: Y: imaging-derived muscle-quality measure; i: participant within independent cohort; t: repeat hormones, muscle image and activity epoch; the registered visit/time-window index, not another independent unit; A: naturally occurring endocrine transition, no hormone assignment; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: within-person transition-associated change in muscle quality after prespecified baseline/time adjustment; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; B it: natural endocrine-transition indicator/continuous stage within a fixed window, with age/calendar adjustment and uncertain stage timing; Bbar i: person/donor mean over a prespecified observation schedule; beta W,beta B: within-unit and between-unit association coefficients, not causal effects; X it: explicitly time-ordered baseline/premeasurement covariates; posttreatment adjustment changes interpretation; f time(t): prespecified sampling/recovery/calendar trajectory; b study: study/site/batch effect where relevant; B true,e it: latent predictor and measurement error, requiring identification constraints; family code: OBS_TRAJECTORY
Sample size design: family: longitudinal association/reliability or held-out prediction precision simulation, not causal-efficacy power; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Chronological age, transition timing and activity may be collinear; time-varying adjustment can remove pathways or induce bias; negative-control windows support falsification only.
Associations need a clear sampling/time frame; causal interpretation is not supplied by this model.
Measurement error, reverse causation, time-varying confounding and selected observation may remain despite adjustment.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: No transition-linked departure from age trends, or disappearance under within-person controls, would refute a distinct transition effect.; preanalysis gate: Chronological age, transition timing and activity may be collinear; time-varying adjustment can remove pathways or induce bias; negative-control windows support falsification only.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Chronological age, transition timing and activity may be collinear; time-varying adjustment can remove pathways or induce bias; negative-control windows support falsification only.
Vary lag order/window and measurement-error reliability within a prespecified family; include negative-control windows when meaningful.
Separate within- and between-unit effects, missingness patterns and source-study replication; do not use cell/assay counts as independent donor counts.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Chronological age, transition timing and activity may be collinear; time-varying adjustment can remove pathways or induce bias; negative-control windows support falsification only.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/112; question: How much of midlife muscle-quality decline reflects endocrine transition rather than chronological aging or reduced physical activity?; aims: Characterize hormone, muscle-quality, and activity trajectories across midlife.
Test mediation by body composition and objectively measured activity.
Validate transition-specific risk estimates across sexes and an independent clinical cohort.; design verbatim planning only: Population model: Existing midlife longitudinal cohorts with repeat hormones and muscle imaging, augmented by wearable measurements. Comparator: Age-matched participants outside transition windows and clinically confirmed deficiency strata. Causal replication: Within-person change models, negative-control time windows, and external replication; no hormone assignment. Sample planning: Seek approximately 500 existing trajectories and 100 deeply phenotyped validation participants.; outcomes verbatim: Primary: imaging-derived muscle quality. Secondary: strength, mobility, activity, body composition, and clinically assessed endocrine symptoms.; analysis verbatim inherited not authoritative if refined: Address adiposity, reproductive stage, illness, medication, and socioeconomic confounding; prespecify sex-specific hypotheses, correct interaction multiplicity, and harmonize hormone assays across batches.; falsification verbatim: No transition-linked departure from age trends, or disappearance under within-person controls, would refute a distinct transition effect.; source ids unverified leads: testosterone-endocrine-statement-2026
testosterone-hypogonadism-review-2026
resistance-sarcopenia-meta-2026; prospectus precedence: No

L10 — Competing-risk phenotypes for healthy-aging intervention selection

Rank: 19
Cluster: Longevity/geroscience
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: CATE_TRIAL; allocation or exposure unit: original allocation within independent diet/exercise/geroscience trials; independent sampling unit: independent trial and participant; repeated measurement unit: functional benefit/harm assessment; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: No observed individual benefit-harm label; trial count and phenotype overlap govern transport; baseline phenotypes only, no posttreatment stratification.
Primary estimand: held-out calibration of average randomized benefit and harm effects within frozen baseline reserve phenotypes
Candidate primary outcome: separately defined functional-benefit and harm outcomes
Primary finalized: No
Mathematical model: type: independent-trial/phenotype coverage and held-out average-effect calibration precision; participant totals alone are inadequate; formulas: Within trial s and frozen baseline phenotype g, estimate Delta_sg=E[Y | A=1,g,s]-E[Y | A=0,g,s] under randomized allocation; validate predictions of Delta_sg in held-out trials.
Model Delta_hat_sg with its covariance matrix, shrink interactions and assess benefit and harm outcomes separately; do not label individuals as observed beneficiaries.; variables: Y: separately defined functional-benefit and harm outcomes; i: independent trial and participant; t: functional benefit/harm assessment; the registered visit/time-window index, not another independent unit; A: original allocation within independent diet/exercise/geroscience trials; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: held-out calibration of average randomized benefit and harm effects within frozen baseline reserve phenotypes; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; P i: frozen pre-evaluation personal/phenotype effect prediction (or pretrial acute-response contrast for P19), with scale fixed before evaluation; eta: prognostic main effect of prediction; kappa: treatment-by-frozen-prediction coefficient, distinguishing predictive modification from prognosis; X it: registered predecision or baseline covariates; no future measurements; f time(t): predeclared practice/calendar trend; s: independent trial; g: baseline phenotype fixed before outcomes; Delta sg: average randomized effect in trial/phenotype, with each endpoint separately defined; Delta hat sg: noisy trial/phenotype effect estimate with sampling covariance; family code: CATE_TRIAL
Sample size design: family: independent-trial/phenotype coverage and held-out average-effect calibration precision; participant totals alone are inadequate; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure
distribution/reliability of frozen predictions
true effect heterogeneity scenarios
within-person randomization schedule
independent validation sample/cluster coverage; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: No observed individual benefit-harm label; trial count and phenotype overlap govern transport; baseline phenotypes only, no posttreatment stratification.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Predictions are frozen before evaluation; noisy person-specific estimates are not observed counterfactual truth; training/evaluation dependence must be propagated.
A nonzero interaction does not by itself establish useful calibration or superiority over a population-only policy.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: No held-out improvement over simple clinical variables, or unstable phenotype assignments, would reject the proposed selection framework.; preanalysis gate: No observed individual benefit-harm label; trial count and phenotype overlap govern transport; baseline phenotypes only, no posttreatment stratification.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: No observed individual benefit-harm label; trial count and phenotype overlap govern transport; baseline phenotypes only, no posttreatment stratification.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
No observed individual benefit-harm label; trial count and phenotype overlap govern transport; baseline phenotypes only, no posttreatment stratification.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/18; question: Can a small set of functional reserve phenotypes identify when apparently beneficial geroscience interventions create countervailing risks?; aims: Construct reserve phenotypes from harmonized randomized lifestyle and geroscience datasets.
Estimate phenotype-specific average randomized effects on prespecified functional benefit and harm outcomes.
Externally test calibration of phenotype-specific average effects and decision utility, with uncertainty for each intervention contrast.; design verbatim planning only: Population model: Individual participant data from at least three older-adult intervention cohorts spanning diet, exercise, and investigational geroscience. Comparator: Single-score prediction and unstratified treatment-effect models. Causal replication: Trial-specific randomized effects, held-out-study validation, and negative-control outcome checks. Sample planning: Plan 600–1,000 existing participants; minimum cohort coverage and endpoint overlap determine feasibility.; outcomes verbatim: Primary: held-out calibration of phenotype-specific average randomized effects on prespecified functional benefit and harm outcomes. Secondary: treatment burden, lean mass, mobility, and subgroup uncertainty.; analysis verbatim inherited not authoritative if refined: Avoid post-treatment stratification; address missingness, trial selection, and baseline medication confounding, penalize interactions, control endpoint multiplicity, and harmonize measurement batches. Evaluate average treatment effects within prespecified baseline phenotypes rather than assigning observed individual benefit–harm labels; report the assumptions needed for any individual counterfactual predictions.; falsification verbatim: No held-out improvement over simple clinical variables, or unstable phenotype assignments, would reject the proposed selection framework.; source ids unverified leads: calerie-organ-aging-rct-2026
pearl-rapamycin-healthspan-rct-2025
metformin-antiaging-uncertainty-2025; prospectus precedence: No

L11 — Minimum detectable biological change in commercial aging-clock assays

Rank: 16
Cluster: Epigenomics/multiomics/aging measurement
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RELIABILITY; allocation or exposure unit: specimen handling condition randomized; donor collection repeated; independent sampling unit: donor with independent laboratory/batch bridge; repeated measurement unit: day, split specimen and clock-assay replicate; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: variance/reliability and response-reproducibility estimation; noisy change labels do not establish stable individual causal effects; specific constraint: Reliable change measures assay/day noise, not true rejuvenation; biological fluctuation and laboratory drift must be separately estimated.
Primary estimand: clock-specific reliable-change threshold and false-rejuvenation probability under stable conditions
Candidate primary outcome: clock score in its stated age-score units
Primary finalized: No
Mathematical model: type: variance-component/reliable-change interval precision by hierarchical simulation; independent donors/persons and replicate allocation; formulas: Y_ibr = mu + u_i + beta_b + v_ib + e_ibr; estimate donor/person, day/block, person-by-block and technical residual components using the actual crossed/nested replication.
For two repeated measurements, reliable-change threshold = z_(1-alpha*/2) sqrt[Var(Y_i2-Y_i1 | no persistent change)]; include covariance instead of assuming independent errors.; variables: Y: clock score in its stated age-score units; i: donor with independent laboratory/batch bridge; t: day, split specimen and clock-assay replicate; the registered visit/time-window index, not another independent unit; A: specimen handling condition randomized; donor collection repeated; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: clock-specific reliable-change threshold and false-rejuvenation probability under stable conditions; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; b: collection day or repeated training block; r: technical/retest replicate; u i: stable unit-level component; beta b: fixed day/block/period shift; v ib: unit-by-day/block deviation; e ibr: technical/retest error; alpha*: clock-family/reliability error rate after registered multiplicity rule; family code: RELIABILITY
Sample size design: family: variance-component/reliable-change interval precision by hierarchical simulation; independent donors/persons and replicate allocation; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Reliable change measures assay/day noise, not true rejuvenation; biological fluctuation and laboratory drift must be separately estimated.
Enough independent units and repeated occasions to separate variance components; fit/identifiability cannot be assumed from a planning sample count.
Reliable change excludes modeled short-term noise, not all biological confounding or regression to the mean.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Near-zero handling effects and stable small-change classification across laboratories would reject substantial measurement-induced false change.; preanalysis gate: Reliable change measures assay/day noise, not true rejuvenation; biological fluctuation and laboratory drift must be separately estimated.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Reliable change measures assay/day noise, not true rejuvenation; biological fluctuation and laboratory drift must be separately estimated.
Profile/bootstrap variance-component intervals, boundary estimates and covariance structures; report false-positive classifications under negative controls.
Vary missingness/retest reliability and avoid declaring responders from effect direction alone.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Reliable change measures assay/day noise, not true rejuvenation; biological fluctuation and laboratory drift must be separately estimated.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/15; question: How much apparent individual rejuvenation is explainable by specimen handling, assay variation, and ordinary short-term within-person fluctuation?; aims: Partition collection, storage, platform, and within-person variance using replicated specimens.
Derive prespecified reliable-change intervals for common clock families.
Validate intervals using blinded repeat samples from an independent laboratory.; design verbatim planning only: Population model: Adults across age, ancestry, and sex strata with repeated blood collection. Comparator: Split specimens, replicate assays, and independently processed reference specimens. Causal replication: Randomized handling conditions and blinded cross-laboratory replication. Sample planning: Plan 90 donors sampled three times; pilot variance sets replicate allocation.; outcomes verbatim: Primary: clock-specific reliable-change thresholds. Secondary: false rejuvenation calls, interlaboratory agreement, cost, and subgroup measurement error.; analysis verbatim inherited not authoritative if refined: Model nested donor, day, and batch effects; prespecify clock families, correct multiple reliability comparisons, and assess medication and recent-illness confounding.; falsification verbatim: Near-zero handling effects and stable small-change classification across laboratories would reject substantial measurement-induced false change.; source ids unverified leads: aging-clocks-critical-review-2026
epigenetic-interventions-review-2026
clinical-aging-clock-2025; prospectus precedence: No

L12 — Separating cell-composition shifts from intrinsic epigenetic aging

Rank: 65
Cluster: Epigenomics/multiomics/aging measurement
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: MIXTURE; allocation or exposure unit: randomized artificial mixtures; original trial assignment for longitudinal archive; independent sampling unit: independent donor and trial participant; repeated measurement unit: cell-type molecular profile, mixture and technical replicate; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: deconvolution error and within-cell validation; DNA-only bulk data do not identify cell-resolved aging; specific constraint: For nonlinear clocks, reconstruct methylation mixtures before applying the clock; DNA-only archives cannot identify within-cell change without validated cell-resolved material.
Primary estimand: error in inferred within-cell methylation-clock change relative to directly measured cell-resolved change
Candidate primary outcome: clock change from fixed cell-type molecular profiles
Primary finalized: No
Mathematical model: type: donor/cell-type mixture-validation precision with randomized mixtures and independent longitudinal reference specimens; formulas: At methylation locus l, B_itl=sum_k p_itk C_itkl+e_itl, where p_itk>=0 and sum_k p_itk=1.
Apply the fixed clock function h to molecular profiles/constructed mixtures: compare inferred within-cell change Delta h(C_ik) with directly measured cell-specific change. In general h(sum p C) != sum p h(C).; variables: Y: clock change from fixed cell-type molecular profiles; i: independent donor and trial participant; t: cell-type molecular profile, mixture and technical replicate; the registered visit/time-window index, not another independent unit; A: randomized artificial mixtures; original trial assignment for longitudinal archive; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: error in inferred within-cell methylation-clock change relative to directly measured cell-resolved change; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; l: molecular locus; k: cell type; B itl: measured bulk methylation at locus l; p itk: cell proportions with reference uncertainty; C itkl: cell-type-specific methylation; e itl: measurement/model error; h: fixed possibly nonlinear clock function; family code: MIXTURE
Sample size design: family: donor/cell-type mixture-validation precision with randomized mixtures and independent longitudinal reference specimens; formula: Use design-specific simulation or independent-unit resampling for the stated estimand and target confidence width/power; preserve all allocation, censoring, clustering and measurement-error structure.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: For nonlinear clocks, reconstruct methylation mixtures before applying the clock; DNA-only archives cannot identify within-cell change without validated cell-resolved material.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Reference cell profiles/material must be available and preserved; mixture nonidentification and nonlinear clock mapping cannot be fixed with sample count alone.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Accurate unadjusted bulk interpretation across deliberate mixtures, or no functional distinction between components, would weaken the decomposition premise.; preanalysis gate: For nonlinear clocks, reconstruct methylation mixtures before applying the clock; DNA-only archives cannot identify within-cell change without validated cell-resolved material.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: For nonlinear clocks, reconstruct methylation mixtures before applying the clock; DNA-only archives cannot identify within-cell change without validated cell-resolved material.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
For nonlinear clocks, reconstruct methylation mixtures before applying the clock; DNA-only archives cannot identify within-cell change without validated cell-resolved material.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/64; question: When blood methylation age changes, how much reflects altered cell proportions versus within-cell epigenetic remodeling?; aims: Construct matched bulk, sorted-cell, and single-cell methylation reference panels.
Decompose longitudinal clock change in existing intervention samples.
Validate inferred components against independently measured function and cell composition.; design verbatim planning only: Population model: Banked longitudinal trial blood with preserved cell fractions or a validated cell- or nucleus-resolved methylation workflow, plus fresh adult reference donors; DNA-only archives support bulk measurements only. Comparator: Bulk scores versus directly measured cell-specific methylation and reconstructed mixtures. Causal replication: Randomized mixture experiments, longitudinal trial assignment, and independent cohort validation. Sample planning: Plan 60 reference donors and 120 archived paired specimens; matched cell-resolved material and pilot preservation quality gate longitudinal within-cell validation.; outcomes verbatim: Primary: error in inferred within-cell clock change. Secondary: functional associations and portability across collection conditions.; analysis verbatim inherited not authoritative if refined: Account for smoking, illness, and medication confounding; lock deconvolution models before validation, correct cell-type multiplicity, and balance timepoints across batches.; falsification verbatim: Accurate unadjusted bulk interpretation across deliberate mixtures, or no functional distinction between components, would weaken the decomposition premise.; source ids unverified leads: epigenetic-interventions-review-2026
multiomics-healthy-prevention-2025
aging-clocks-critical-review-2026; prospectus precedence: No

L13 — Are aging clocks valid treatment-effect surrogates across intervention classes?

Rank: 33
Cluster: Epigenomics/multiomics/aging measurement
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: SURROGATE; allocation or exposure unit: randomized treatment contrast within each independent trial; independent sampling unit: independent trial/intervention class; participants estimate within-trial effects; repeated measurement unit: paired clock and function measurements; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: trial-level treatment-effect prediction/surrogate validation candidate, not causal mediation or individual aging reversal; specific constraint: Trial-level surrogacy is distinct from individual association; preserve covariance of estimated effects; trial count/classes govern information and surrogate paradox remains possible.
Primary estimand: leave-intervention-class-out absolute prediction error for twelve-month functional treatment effects from earlier clock effects
Candidate primary outcome: standardized twelve-month randomized physical-function effect in the fixed compatible scale family
Primary finalized: No
Mathematical model: type: independent-trial and intervention-class surrogacy prediction precision; bivariate errors-in-variables simulation; formulas: For independent trial s, (dS_hat_s,dY_hat_s)^T ~ Normal[(dS_s,dY_s)^T, Sigma_s]; retain within-trial covariance and measurement error.
dY_s=alpha+v_class(s)+beta dS_s+u_s with hierarchical zero-mean v_class and training-estimated variance; for a held-out class integrate over its new-class effect distribution without fitting it to held-out function outcomes. Evaluate mean absolute prediction error abs(dY_hat_s-dY_pred,-class(s)) with uncertainty in both effects and predictions.
dS is an earlier fixed clock treatment effect (preferably six months); dY is the compatible standardized twelve-month randomized function effect. Leave the entire intervention class out of model fitting.; variables: Y: standardized twelve-month randomized physical-function effect in the fixed compatible scale family; i: independent trial/intervention class; participants estimate within-trial effects; t: paired clock and function measurements; the registered visit/time-window index, not another independent unit; A: randomized treatment contrast within each independent trial; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: leave-intervention-class-out absolute prediction error for twelve-month functional treatment effects from earlier clock effects; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; s: independent randomized trial, not a participant; dS s,dY s: true trial-level randomized clock and function effects on fixed compatible scales; dS hat s,dY hat s: estimated trial contrasts; Sigma s: their within-trial sampling covariance matrix; alpha,v class,beta,u s: training-population intercept, hierarchical class effect, cross-trial association and residual trial effect; new-class predictive uncertainty must be propagated; dY pred,-class(s): function-effect prediction trained without any trial from the held-out intervention class; family code: SURROGATE
Sample size design: family: independent-trial and intervention-class surrogacy prediction precision; bivariate errors-in-variables simulation; formula: Use design-specific simulation or independent-unit resampling for the stated estimand and target confidence width/power; preserve all allocation, censoring, clustering and measurement-error structure.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Trial-level surrogacy is distinct from individual association; preserve covariance of estimated effects; trial count/classes govern information and surrogate paradox remains possible.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Compatible function family and timing, reliable trial contrasts and representative independent intervention classes; selective trial/endpoint access can invalidate transport.
No participant-level clock/function correlation can substitute for cross-trial validation; surrogate paradox and class dependence remain possible.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Consistent accurate held-out treatment-effect prediction would reject the hypothesis of poor cross-class surrogacy.; preanalysis gate: Trial-level surrogacy is distinct from individual association; preserve covariance of estimated effects; trial count/classes govern information and surrogate paradox remains possible.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Trial-level surrogacy is distinct from individual association; preserve covariance of estimated effects; trial count/classes govern information and surrogate paradox remains possible.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Enough independently accessible randomized trials/classes with paired compatible earlier clock and later function outcomes; trial-level covariance or recoverable IPD.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Trial-level surrogacy is distinct from individual association; preserve covariance of estimated effects; trial count/classes govern information and surrogate paradox remains possible.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/32; question: Do randomized treatment effects on aging clocks predict treatment effects on meaningful function across different intervention mechanisms?; aims: Harmonize participant-level clocks and functional endpoints from randomized trials.
Estimate individual-level and trial-level surrogacy separately.
Test transport to held-out intervention classes with explicit uncertainty bounds.; design verbatim planning only: Population model: At least eight independent randomized trials with paired aging measures and physical outcomes. Comparator: Chronological age, routine clinical scores, and no-surrogate models. Causal replication: Randomized treatment contrasts, leave-class-out validation, and selective-reporting sensitivity analyses. Sample planning: Seek 1,000–2,000 archived participants; trial count, not individuals alone, governs feasibility.; outcomes verbatim: Primary: out-of-class treatment-effect prediction error. Secondary: surrogate paradox frequency, calibration, and functional outcome coverage.; analysis verbatim inherited not authoritative if refined: Address trial selection, mediator–outcome confounding, missing endpoints, and intervention adherence; use hierarchical multiplicity control and harmonize assay batches without leaking treatment labels.; falsification verbatim: Consistent accurate held-out treatment-effect prediction would reject the hypothesis of poor cross-class surrogacy.; source ids unverified leads: epigenetic-interventions-review-2026
calerie-organ-aging-rct-2026
clinical-aging-clock-2025; prospectus precedence: Yes; prospectus file: thesis-prospectuses.json; prospectus design and estimand: The primary endpoint is leave-class-out absolute prediction error for the standardized randomized treatment effect on physical function at 12 months, using a prespecified allowable assessment window. The predictor is clock treatment effect at an earlier harmonized assessment, preferably six months. Fix one compatible function family and one or two clock versions before outcome analysis; do not combine unrelated scales without validated linking. Contemporaneous-only datasets receive a separately labeled weaker analysis.; prospectus analysis plan: Use intention-to-treat contrasts within trials, then hierarchical bivariate models that propagate uncertainty in both treatment effects and their covariance. Report predictive intervals, calibration, absolute prediction error and trial-level explained variation with small-trial-count uncertainty. Compare held-out predictions with intercept-only and available clinical-score benchmarks. Keep individual association analyses separate. Model class heterogeneity cautiously; no causal mediation claim follows without additional untestable assumptions.; precedence note: Expanded prospectus refinements govern this scaffold; verbatim earlier brief material is retained for traceability, not as a contradictory analysis instruction. F15 is explicitly two arms; F30 has no unjustified symptom noninferiority claim.

L14 — Transportable biological-age estimation across ancestry and material deprivation

Rank: 30
Cluster: Epigenomics/multiomics/aging measurement
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: PRED; allocation or exposure unit: fixed score evaluated in independent cohort/system/ancestry-deprivation strata; independent sampling unit: participant and external cohort; bridge donor for assay alignment; repeated measurement unit: functional outcome and repeat score specimen; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: prediction, calibration or model transportability; prediction gain is not clinical efficacy, mechanistic necessity or causal treatment selection; specific constraint: Calibration refers to observed function risk, not an unobservable true biological age; care-access selection and missing tests can prevent transport.
Primary estimand: subgroup calibration of fixed-horizon functional-decline risk predicted from age scores
Candidate primary outcome: fixed-horizon functional-decline indicator
Primary finalized: No
Mathematical model: type: external/nested predictive-validation precision with target prevalence/variance, independent-unit holdout coverage and model complexity; formulas: Gain = mean_test[L(Y,f_base(X)) - L(Y,f_new(X))], with both predictions frozen before test outcomes; bootstrap independent held-out units.
For a binary target assess E[Y | p_hat=p] versus p and calibration intercept/slope logit Pr(Y=1)=a_cal+b_cal logit(p_hat); for a continuous target use Y=a_cal+b_cal y_hat+error plus scale-appropriate calibration/loss.; variables: Y: fixed-horizon functional-decline indicator; i: participant and external cohort; bridge donor for assay alignment; t: functional outcome and repeat score specimen; the registered visit/time-window index, not another independent unit; A: fixed score evaluated in independent cohort/system/ancestry-deprivation strata; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: subgroup calibration of fixed-horizon functional-decline risk predicted from age scores; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; L: one proper or scientifically justified primary loss/scoring rule fixed for the actual outcome type; positive Gain favors new model; f base,f new: locked baseline/new prediction functions using only information available at the registered prediction time; test: untouched person/family/lab/site/time set as specified per thesis; p hat,y hat: frozen probability or continuous prediction; a cal,b cal: calibration intercept and slope, estimated on evaluation data for assessment; using them to update the model requires another evaluation set; family code: PRED
Sample size design: family: external/nested predictive-validation precision with target prevalence/variance, independent-unit holdout coverage and model complexity; formula: Simulate or bootstrap the independent-unit test loss/calibration/decision criterion; plan cases/events, noncases and independent sites/families/programs separately. No universal events-per-variable or participant N is asserted.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Calibration refers to observed function risk, not an unobservable true biological age; care-access selection and missing tests can prevent transport.
Train/tune/test separation at the actual biological/participant/family/time level; no future, postoutcome or target-derived predictors.
Validation support must cover intended deployment; calibration and missing/failing measurements are reported with discrimination/loss.
A validation subset selected by culture success, assay viability, consent or data completeness is a different target unless the full strategy is explicitly evaluated.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Equivalent original calibration across contexts or no external improvement would reject the need for the proposed adjustment.; preanalysis gate: Calibration refers to observed function risk, not an unobservable true biological age; care-access selection and missing tests can prevent transport.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Calibration refers to observed function risk, not an unobservable true biological age; care-access selection and missing tests can prevent transport.
Repeat loss/calibration with independent-cluster resampling, prespecified subgroup/transport shifts and plausible missingness.
Compare fixed simple baselines, calibration and uncertainty coverage; report abstention/failure/turnaround rather than deleting hard cases.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Frozen prediction time/horizon, proper loss, baseline model, feature availability, independent split/holdout and target deployment population.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Calibration refers to observed function risk, not an unobservable true biological age; care-access selection and missing tests can prevent transport.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/29; question: Do biological-age scores retain calibration across ancestry, deprivation, and health-system contexts without encoding unequal access to care?; aims: Audit age-score calibration in diverse cohorts with linked functional outcomes.
Separate assay, healthcare-access, and ancestry-associated contributions to error.
Externally validate context-aware recalibration and subgroup uncertainty intervals.; design verbatim planning only: Population model: Three cohorts spanning ancestry, socioeconomic conditions, and laboratory systems. Comparator: Original scores, pooled recalibration, and context-aware models. Causal replication: External-site holdouts, healthcare-use negative controls, and matched assay bridge samples; causal claims remain limited. Sample planning: Seek 5,000 existing records and 150 bridge specimens, contingent on subgroup coverage.; outcomes verbatim: Primary: subgroup calibration of fixed-horizon functional-decline risk predicted from the age scores. Secondary: uncertainty coverage for repeat assay scores, incremental functional prediction, referral disparities, and transportability under missing tests.; analysis verbatim inherited not authoritative if refined: Model medication, comorbidity, and testing-selection confounding; preregister intersectional contrasts, control subgroup multiplicity, and use bridge samples to estimate batch effects. Fix the functional-decline definition and prediction horizon before validation; distinguish outcome-risk calibration from agreement with chronological age or an unobservable true biological age.; falsification verbatim: Equivalent original calibration across contexts or no external improvement would reject the need for the proposed adjustment.; source ids unverified leads: aging-clocks-critical-review-2026
clinical-aging-clock-2025
multiomics-healthy-prevention-2025; prospectus precedence: No

L15 — Reversible physiological states that masquerade as accelerated epigenetic aging

Rank: 146
Cluster: Epigenomics/multiomics/aging measurement
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: OBS_TRAJECTORY; allocation or exposure unit: natural schedule/physiological transition and recovered period; independent sampling unit: participant and independent archived trajectory; repeated measurement unit: clock specimen over transition/recovery; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: association or prospective prediction; temporal order and model adjustment alone do not identify a mechanism or treatment effect; specific constraint: No induced illness or sleep deprivation; persistence window and latent decomposition require identification constraints; short-term state association is not irreversible aging.
Primary estimand: persistent component of clock-score change after a fixed recovery window
Candidate primary outcome: fixed clock score over natural-state and recovery times
Primary finalized: No
Mathematical model: type: longitudinal association/reliability or held-out prediction precision simulation, not causal-efficacy power; formulas: Y_it=mu_i+P_it+T_it+e_it, separating persistent state P, transient state T and assay error e under a fully specified state-space/recovery model.
Primary candidate Delta_P=E[P_i,recovered-P_i,baseline]; an identifiable weaker alternative is the prespecified recovered-window score mean minus baseline-window mean, explicitly labeled persistence-by-window rather than a uniquely identified latent biological state.; variables: Y: fixed clock score over natural-state and recovery times; i: participant and independent archived trajectory; t: clock specimen over transition/recovery; the registered visit/time-window index, not another independent unit; A: natural schedule/physiological transition and recovered period; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: persistent component of clock-score change after a fixed recovery window; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; B it: prespecified molecular/physiological/state predictor paired with the per-thesis endpoint; exact definition/timing fixed before analysis; Bbar i: person/donor mean over a prespecified observation schedule; beta W,beta B: within-unit and between-unit association coefficients, not causal effects; X it: explicitly time-ordered baseline/premeasurement covariates; posttreatment adjustment changes interpretation; f time(t): prespecified sampling/recovery/calendar trajectory; b study: study/site/batch effect where relevant; B true,e it: latent predictor and measurement error, requiring identification constraints; P it: persistent score component under preregistered dynamics/constraints; T it: transient component constrained by a validated recovery window/decay model; e it: assay error estimated from reference replicates; mu i: unit baseline level under identified centering constraints; Delta P: persistent postrecovery-minus-baseline component contrast, provisional until dynamics identified; family code: OBS_TRAJECTORY
Sample size design: family: longitudinal association/reliability or held-out prediction precision simulation, not causal-efficacy power; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: No induced illness or sleep deprivation; persistence window and latent decomposition require identification constraints; short-term state association is not irreversible aging.
Associations need a clear sampling/time frame; causal interpretation is not supplied by this model.
Measurement error, reverse causation, time-varying confounding and selected observation may remain despite adjustment.
Without constraints and sufficient recovered sampling, persistent/transient decomposition is nonidentified; choose a registered recovered-window contrast if latent components cannot be distinguished.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: No reversible excursions beyond technical error, or identical prediction by transient and persistent components, would undermine the model.; preanalysis gate: No induced illness or sleep deprivation; persistence window and latent decomposition require identification constraints; short-term state association is not irreversible aging.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: No induced illness or sleep deprivation; persistence window and latent decomposition require identification constraints; short-term state association is not irreversible aging.
Vary lag order/window and measurement-error reliability within a prespecified family; include negative-control windows when meaningful.
Separate within- and between-unit effects, missingness patterns and source-study replication; do not use cell/assay counts as independent donor counts.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Fix state-space dynamics, centering, recovery-window/decay assumptions and observation frequency, or select the weaker recovered-window target.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
No induced illness or sleep deprivation; persistence window and latent decomposition require identification constraints; short-term state association is not irreversible aging.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/145; question: Which short-term physiological states shift aging clocks without indicating persistent deterioration or subsequent loss of function?; aims: Measure dense within-person trajectories around naturally occurring sleep disruption and recovery.
Separate transient and persistent methylomic–metabolomic components using prespecified state-space models.
Validate functional prediction in an independent longitudinal aging cohort.; design verbatim planning only: Population model: Adults with variable work schedules plus established aging-cohort specimens; no induced illness or sleep deprivation. Comparator: Stable-schedule participants and within-person recovered periods. Causal replication: Repeated natural transitions, negative-control intervals, and external validation; residual confounding is acknowledged. Sample planning: Plan 120 intensively sampled participants and approximately 400 archived trajectories.; outcomes verbatim: Primary: persistence-adjusted clock change. Secondary: false acceleration calls, recovery time, functional prediction, and participant distress.; analysis verbatim inherited not authoritative if refined: Adjust smoking, recent illness, menstrual stage, medications, and sampling time; prespecify latent dimensions, control clock multiplicity, and interleave longitudinal batches.; falsification verbatim: No reversible excursions beyond technical error, or identical prediction by transient and persistent components, would undermine the model.; source ids unverified leads: epigenetic-interventions-review-2026
aging-clocks-critical-review-2026
multiomics-healthy-prevention-2025; prospectus precedence: No

L16 — Organ-age discordance as a mechanistic phenotype rather than a single age

Rank: 147
Cluster: Epigenomics/multiomics/aging measurement
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: PRED; allocation or exposure unit: organ-score discordance observed before functional outcomes; independent sampling unit: participant with independent cohort validation; repeated measurement unit: organ-specific assay and functional follow-up; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: prediction, calibration or model transportability; prediction gain is not clinical efficacy, mechanistic necessity or causal treatment selection; specific constraint: Lock organ pairs before validation; shuffled organ assignments are a negative-control benchmark, not proof of mechanism; score covariance matters.
Primary estimand: incremental organ-relevant functional prediction using repeatability-corrected discordance
Candidate primary outcome: prespecified organ-relevant function outcome at a fixed horizon
Primary finalized: No
Mathematical model: type: external/nested predictive-validation precision with target prevalence/variance, independent-unit holdout coverage and model complexity; formulas: Gain = mean_test[L(Y,f_base(X)) - L(Y,f_new(X))], with both predictions frozen before test outcomes; bootstrap independent held-out units.
For a binary target assess E[Y | p_hat=p] versus p and calibration intercept/slope logit Pr(Y=1)=a_cal+b_cal logit(p_hat); for a continuous target use Y=a_cal+b_cal y_hat+error plus scale-appropriate calibration/loss.; variables: Y: prespecified organ-relevant function outcome at a fixed horizon; i: participant with independent cohort validation; t: organ-specific assay and functional follow-up; the registered visit/time-window index, not another independent unit; A: organ-score discordance observed before functional outcomes; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: incremental organ-relevant functional prediction using repeatability-corrected discordance; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; L: one proper or scientifically justified primary loss/scoring rule fixed for the actual outcome type; positive Gain favors new model; f base,f new: locked baseline/new prediction functions using only information available at the registered prediction time; test: untouched person/family/lab/site/time set as specified per thesis; p hat,y hat: frozen probability or continuous prediction; a cal,b cal: calibration intercept and slope, estimated on evaluation data for assessment; using them to update the model requires another evaluation set; family code: PRED
Sample size design: family: external/nested predictive-validation precision with target prevalence/variance, independent-unit holdout coverage and model complexity; formula: Simulate or bootstrap the independent-unit test loss/calibration/decision criterion; plan cases/events, noncases and independent sites/families/programs separately. No universal events-per-variable or participant N is asserted.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Lock organ pairs before validation; shuffled organ assignments are a negative-control benchmark, not proof of mechanism; score covariance matters.
Train/tune/test separation at the actual biological/participant/family/time level; no future, postoutcome or target-derived predictors.
Validation support must cover intended deployment; calibration and missing/failing measurements are reported with discrimination/loss.
A validation subset selected by culture success, assay viability, consent or data completeness is a different target unless the full strategy is explicitly evaluated.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Poor repeatability or no prediction beyond mean age would falsify discordance as a useful biological phenotype.; preanalysis gate: Lock organ pairs before validation; shuffled organ assignments are a negative-control benchmark, not proof of mechanism; score covariance matters.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Lock organ pairs before validation; shuffled organ assignments are a negative-control benchmark, not proof of mechanism; score covariance matters.
Repeat loss/calibration with independent-cluster resampling, prespecified subgroup/transport shifts and plausible missingness.
Compare fixed simple baselines, calibration and uncertainty coverage; report abstention/failure/turnaround rather than deleting hard cases.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Frozen prediction time/horizon, proper loss, baseline model, feature availability, independent split/holdout and target deployment population.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Lock organ pairs before validation; shuffled organ assignments are a negative-control benchmark, not proof of mechanism; score covariance matters.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/146; question: Do stable discrepancies between organ-specific aging scores identify distinct biological constraints and future functional vulnerabilities?; aims: Estimate reliability and covariance of organ-specific scores across repeat samples.
Link persistent discordance patterns to organ-relevant physiological measurements.
Validate prospective prediction and intervention responsiveness in independent data.; design verbatim planning only: Population model: Existing longitudinal aging cohorts and archived caloric-restriction trial specimens. Comparator: Mean biological age, conventional risk scores, and shuffled organ assignments. Causal replication: Repeat measurements distinguish noise; external validation and randomized trial contrasts test responsiveness. Sample planning: Seek 1,000 longitudinal participants and approximately 180 trial participants; precision depends on event coverage.; outcomes verbatim: Primary: organ-relevant functional prediction. Secondary: discordance repeatability, treatment-effect heterogeneity, and incremental calibration.; analysis verbatim inherited not authoritative if refined: Adjust comorbidity, medications, socioeconomic status, and body composition; separate discovery from validation, correct organ-pair multiplicity, and model assay batch covariance.; falsification verbatim: Poor repeatability or no prediction beyond mean age would falsify discordance as a useful biological phenotype.; source ids unverified leads: calerie-organ-aging-rct-2026
serum-proteome-aging-cohort-2025
aging-clocks-critical-review-2026; prospectus precedence: No

L17 — Cell-free DNA aging signals versus transient tissue injury

Rank: 155
Cluster: Epigenomics/multiomics/aging measurement
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: ASSOC_EIV; allocation or exposure unit: elective procedure and injury severity observed, not randomized; independent sampling unit: participant and independent control participant; repeated measurement unit: preprocedure, injury and recovered cfDNA specimen; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: association or prospective prediction; temporal order and model adjustment alone do not identify a mechanism or treatment effect; specific constraint: Perioperative cointerventions and renal clearance confound injury associations; no identified causal attributable fraction or occult-disease diagnostic claim.
Primary estimand: within-person cfDNA age-signal association with independently measured injury/recovery
Candidate primary outcome: prespecified tissue-associated cfDNA age signal
Primary finalized: No
Mathematical model: type: longitudinal association/reliability or held-out prediction precision simulation, not causal-efficacy power; formulas: Y_it = alpha + u_i + beta_W (B_it-Bbar_i) + beta_B Bbar_i + gamma^T X_it + f_time(t) + b_study + epsilon_it; u_i is a random intercept with a declared conditional mean/covariance model, not unrestricted fixed person effects.
B_observed,it = B_true,it + e_it; estimate measurement variance from replicates/validation or report sensitivity rather than assume it known.; variables: Y: prespecified tissue-associated cfDNA age signal; i: participant and independent control participant; t: preprocedure, injury and recovered cfDNA specimen; the registered visit/time-window index, not another independent unit; A: elective procedure and injury severity observed, not randomized; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: within-person cfDNA age-signal association with independently measured injury/recovery; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; B it: independently measured procedural-injury severity/trajectory before or paired with each cfDNA sample; Bbar i: person/donor mean over a prespecified observation schedule; beta W,beta B: within-unit and between-unit association coefficients, not causal effects; X it: explicitly time-ordered baseline/premeasurement covariates; posttreatment adjustment changes interpretation; f time(t): prespecified sampling/recovery/calendar trajectory; b study: study/site/batch effect where relevant; B true,e it: latent predictor and measurement error, requiring identification constraints; family code: ASSOC_EIV
Sample size design: family: longitudinal association/reliability or held-out prediction precision simulation, not causal-efficacy power; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Perioperative cointerventions and renal clearance confound injury associations; no identified causal attributable fraction or occult-disease diagnostic claim.
Associations need a clear sampling/time frame; causal interpretation is not supplied by this model.
Measurement error, reverse causation, time-varying confounding and selected observation may remain despite adjustment.
Specify E[u_i | Bbar_i,X] and its covariance or a correlated-random-effects formulation; an unrestricted fixed person intercept would absorb the between-person term.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: No age-signal association with independently verified injury or persistent specificity after recovery would weaken the injury explanation.; preanalysis gate: Perioperative cointerventions and renal clearance confound injury associations; no identified causal attributable fraction or occult-disease diagnostic claim.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Perioperative cointerventions and renal clearance confound injury associations; no identified causal attributable fraction or occult-disease diagnostic claim.
Vary lag order/window and measurement-error reliability within a prespecified family; include negative-control windows when meaningful.
Separate within- and between-unit effects, missingness patterns and source-study replication; do not use cell/assay counts as independent donor counts.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Perioperative cointerventions and renal clearance confound injury associations; no identified causal attributable fraction or occult-disease diagnostic claim.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/154; question: How strongly do age-associated cell-free-DNA methylation signals change with independently measured transient procedural injury and subsequent recovery?; aims: Benchmark tissue-origin deconvolution against reference methylomes and clinical injury markers.
Track age-associated signals before and after naturally occurring clinical recovery.
Validate a joint injury–aging model in an independent biobank.; design verbatim planning only: Population model: Consented clinical biobank samples around elective procedures and stable community controls. Comparator: Preprocedure baseline, recovered samples, and injury-marker-negative controls. Causal replication: Within-person trajectories, negative-control tissues, blinded deconvolution, and external replication; procedure effects are not randomized. Sample planning: Plan 150 clinical trajectories and 150 controls, subject to paired-sample availability.; outcomes verbatim: Primary: within-person age-signal change associated with independently measured procedural injury and recovery. Secondary: tissue specificity, persistence, false-positive age-change classification, and recovery concordance.; analysis verbatim inherited not authoritative if refined: Adjust cancer history, renal clearance, procedure type, and medications; lock tissue panels, control tissue-level multiplicity, and randomize samples across sequencing batches. Treat the injury contribution as an association with sensitivity bounds for perioperative confounding; do not claim an identified causal attributable fraction or validated discrimination of occult disease.; falsification verbatim: No age-signal association with independently verified injury or persistent specificity after recovery would weaken the injury explanation.; source ids unverified leads: mc-ed-ai-liquid-biopsy-review-2026
epigenetic-interventions-review-2026
aging-clocks-critical-review-2026; prospectus precedence: No

L18 — Which omics layer adds durable information about functional aging?

Rank: 98
Cluster: Epigenomics/multiomics/aging measurement
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: PRED; allocation or exposure unit: locked clinical/single-/multiomic model evaluated prospectively; independent sampling unit: participant with independent cohort testing; repeated measurement unit: omics layer, specimen and physical-function follow-up; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: prediction, calibration or model transportability; prediction gain is not clinical efficacy, mechanistic necessity or causal treatment selection; specific constraint: All feature selection and imputation must be inside development folds; bridge specimens estimate batch effects without test-outcome leakage.
Primary estimand: external functional-decline prediction gain of multiomics versus fixed clinical/single-layer benchmarks
Candidate primary outcome: prespecified later functional-decline outcome
Primary finalized: No
Mathematical model: type: external/nested predictive-validation precision with target prevalence/variance, independent-unit holdout coverage and model complexity; formulas: Gain = mean_test[L(Y,f_base(X)) - L(Y,f_new(X))], with both predictions frozen before test outcomes; bootstrap independent held-out units.
For a binary target assess E[Y | p_hat=p] versus p and calibration intercept/slope logit Pr(Y=1)=a_cal+b_cal logit(p_hat); for a continuous target use Y=a_cal+b_cal y_hat+error plus scale-appropriate calibration/loss.; variables: Y: prespecified later functional-decline outcome; i: participant with independent cohort testing; t: omics layer, specimen and physical-function follow-up; the registered visit/time-window index, not another independent unit; A: locked clinical/single-/multiomic model evaluated prospectively; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: external functional-decline prediction gain of multiomics versus fixed clinical/single-layer benchmarks; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; L: one proper or scientifically justified primary loss/scoring rule fixed for the actual outcome type; positive Gain favors new model; f base,f new: locked baseline/new prediction functions using only information available at the registered prediction time; test: untouched person/family/lab/site/time set as specified per thesis; p hat,y hat: frozen probability or continuous prediction; a cal,b cal: calibration intercept and slope, estimated on evaluation data for assessment; using them to update the model requires another evaluation set; family code: PRED
Sample size design: family: external/nested predictive-validation precision with target prevalence/variance, independent-unit holdout coverage and model complexity; formula: Simulate or bootstrap the independent-unit test loss/calibration/decision criterion; plan cases/events, noncases and independent sites/families/programs separately. No universal events-per-variable or participant N is asserted.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: All feature selection and imputation must be inside development folds; bridge specimens estimate batch effects without test-outcome leakage.
Train/tune/test separation at the actual biological/participant/family/time level; no future, postoutcome or target-derived predictors.
Validation support must cover intended deployment; calibration and missing/failing measurements are reported with discrimination/loss.
A validation subset selected by culture success, assay viability, consent or data completeness is a different target unless the full strategy is explicitly evaluated.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Consistent large external gains from full-stack models would reject the sparse-equivalence hypothesis; absent gains over routine measures would challenge multiomic utility.; preanalysis gate: All feature selection and imputation must be inside development folds; bridge specimens estimate batch effects without test-outcome leakage.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: All feature selection and imputation must be inside development folds; bridge specimens estimate batch effects without test-outcome leakage.
Repeat loss/calibration with independent-cluster resampling, prespecified subgroup/transport shifts and plausible missingness.
Compare fixed simple baselines, calibration and uncertainty coverage; report abstention/failure/turnaround rather than deleting hard cases.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Frozen prediction time/horizon, proper loss, baseline model, feature availability, independent split/holdout and target deployment population.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
All feature selection and imputation must be inside development folds; bridge specimens estimate batch effects without test-outcome leakage.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/97; question: Does combining proteomics, metabolomics, and methylation improve prospective functional prediction after accounting for repeatability, cost, and missingness?; aims: Quantify redundant and complementary signals across matched omics layers.
Develop sparse reliability-weighted models with prespecified clinical benchmarks.
Validate function prediction, missing-layer robustness, and cost-adjusted utility externally.; design verbatim planning only: Population model: Two longitudinal adult cohorts with matched omics and repeated physical-function assessment. Comparator: Routine clinical variables, single-omic models, and unrestricted multiomic models. Causal replication: Nested cross-validation and independent cohort testing; prediction is not presented as causal mechanism. Sample planning: Seek 1,500 discovery and 500 validation records plus 80 technical-replicate specimens.; outcomes verbatim: Primary: external functional-decline prediction. Secondary: calibration, assay cost, missing-layer degradation, and subgroup performance.; analysis verbatim inherited not authoritative if refined: Address age, disease, medication, and healthcare-selection confounding; prevent feature-selection leakage, control discovery multiplicity, and estimate cross-platform batch effects with bridge specimens.; falsification verbatim: Consistent large external gains from full-stack models would reject the sparse-equivalence hypothesis; absent gains over routine measures would challenge multiomic utility.; source ids unverified leads: multiomics-healthy-prevention-2025
serum-proteome-aging-cohort-2025
clinical-aging-clock-2025; prospectus precedence: No

L19 — Detecting individual aging acceleration without regression-to-the-mean traps

Rank: 129
Cluster: Epigenomics/multiomics/aging measurement
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: PRED_SURV; allocation or exposure unit: molecular history observed at frozen repeated prediction landmarks; independent sampling unit: participant with independent system testing; repeated measurement unit: landmark, assay and later mobility event; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: prediction, calibration or model transportability; prediction gain is not clinical efficacy, mechanistic necessity or causal treatment selection; specific constraint: Illness-driven assay timing is informative; landmarks must precede events; use inverse-censoring assumptions and competing-death rules without retrospective alert tuning.
Primary estimand: early-warning calibration and false-alert burden for future mobility decline at a fixed horizon
Candidate primary outcome: mobility-decline event after each eligible landmark
Primary finalized: No
Mathematical model: type: landmark/competing-risk prediction precision with event-free risk set, event counts, censoring and external calibration; formulas: At fixed landmark ell, use only H_i(ell) among alive event-free participants: F_1(t | H(ell))=Pr(T<=t,J=event | T>ell,H(ell)).
Evaluate calibrated cause-specific cumulative-incidence predictions and proper censoring-aware Brier/loss over (ell,tau]; do not treat death as independent censoring for the event risk.; variables: Y: mobility-decline event after each eligible landmark; i: participant with independent system testing; t: landmark, assay and later mobility event; the registered visit/time-window index, not another independent unit; A: molecular history observed at frozen repeated prediction landmarks; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: early-warning calibration and false-alert burden for future mobility decline at a fixed horizon; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; L: one proper or scientifically justified primary loss/scoring rule fixed for the actual outcome type; positive Gain favors new model; f base,f new: locked baseline/new prediction functions using only information available at the registered prediction time; test: untouched person/family/lab/site/time set as specified per thesis; p hat,y hat: frozen probability or continuous prediction; a cal,b cal: calibration intercept and slope, estimated on evaluation data for assessment; using them to update the model requires another evaluation set; ell: predeclared prediction landmark; H i(ell): history available no later than ell; T,J: first relevant event time and event type, including competing death; tau: fixed prediction horizon; F 1: conditional cumulative incidence for the primary event; early/prelandmark events reported separately; family code: PRED_SURV
Sample size design: family: landmark/competing-risk prediction precision with event-free risk set, event counts, censoring and external calibration; formula: Simulate or bootstrap the independent-unit test loss/calibration/decision criterion; plan cases/events, noncases and independent sites/families/programs separately. No universal events-per-variable or participant N is asserted.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Illness-driven assay timing is informative; landmarks must precede events; use inverse-censoring assumptions and competing-death rules without retrospective alert tuning.
Train/tune/test separation at the actual biological/participant/family/time level; no future, postoutcome or target-derived predictors.
Validation support must cover intended deployment; calibration and missing/failing measurements are reported with discrimination/loss.
A validation subset selected by culture success, assay viability, consent or data completeness is a different target unless the full strategy is explicitly evaluated.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: No gain in lead time at a matched false-alert rate would falsify the claimed advantage over ordinary clinical monitoring.; preanalysis gate: Illness-driven assay timing is informative; landmarks must precede events; use inverse-censoring assumptions and competing-death rules without retrospective alert tuning.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Illness-driven assay timing is informative; landmarks must precede events; use inverse-censoring assumptions and competing-death rules without retrospective alert tuning.
Repeat loss/calibration with independent-cluster resampling, prespecified subgroup/transport shifts and plausible missingness.
Compare fixed simple baselines, calibration and uncertainty coverage; report abstention/failure/turnaround rather than deleting hard cases.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Frozen prediction time/horizon, proper loss, baseline model, feature availability, independent split/holdout and target deployment population.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Illness-driven assay timing is informative; landmarks must precede events; use inverse-censoring assumptions and competing-death rules without retrospective alert tuning.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/128; question: Can repeated molecular trajectories detect genuine functional deterioration earlier than isolated high biological-age scores?; aims: Estimate ordinary within-person variation in repeated clinical and molecular aging measures.
Develop prespecified change-point alerts using training cohorts.
Prospectively validate alert timing, false alarms, and functional outcomes in held-out data.; design verbatim planning only: Population model: Existing repeated-measure aging biobanks linked to mobility and hospitalization records. Comparator: Single-score thresholds, linear trajectories, and routine clinical risk models. Causal replication: Temporal validation, negative-control pseudoalerts, and independent-system replication; no causal intervention claims. Sample planning: Seek 2,000 participants with at least four measurements; outcome prevalence determines precision.; outcomes verbatim: Primary: calibrated early warning of mobility decline. Secondary: false-alert burden, lead time, subgroup fairness, and sampling-cost sensitivity.; analysis verbatim inherited not authoritative if refined: Model illness-driven sampling, dropout, medication changes, and baseline disease; correct alert-selection multiplicity, separate model tuning from testing, and adjust laboratory batch drift.; falsification verbatim: No gain in lead time at a matched false-alert rate would falsify the claimed advantage over ordinary clinical monitoring.; source ids unverified leads: clinical-aging-clock-2025
serum-proteome-aging-cohort-2025
aging-clocks-critical-review-2026; prospectus precedence: No

L20 — Are clock-associated methylation loci drivers or passengers of cellular aging?

Rank: 117
Cluster: Epigenomics/multiomics/aging measurement
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: LAB; allocation or exposure unit: culture allocation to focused loci/control perturbation conditions; independent sampling unit: independent donor, with independently allocated culture units; repeated measurement unit: cell/assay/time replicate; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: controlled in-vitro/ex-vivo allocation contrast when culture allocation is randomized; no direct human therapeutic-efficacy claim; specific constraint: Clock change cannot define success; lineage identity/genomic stability are co-gates, not adjusted posttreatment confounders; locus-family multiplicity fixed in advance.
Primary estimand: condition effect on stress-recovery function independent of using the clock as the outcome
Candidate primary outcome: prespecified stress-recovery functional endpoint
Primary finalized: No
Mathematical model: type: multilevel donor-by-preparation/condition experimental contrast precision; independent-donor and independent-batch allocation simulation; formulas: Y_djbrt = alpha + z(A_djb)^T theta + gamma^T X_d + u_d + v_b + q_dj + epsilon_djbrt.
Delta_model = c^T theta standardized to sampled donors/conditions; r is a technical repeat, not another donor or manufacturing batch.; variables: Y: prespecified stress-recovery functional endpoint; i: independent donor, with independently allocated culture units; t: cell/assay/time replicate; the registered visit/time-window index, not another independent unit; A: culture allocation to focused loci/control perturbation conditions; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: condition effect on stress-recovery function independent of using the clock as the outcome; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; d: independent tissue/cell donor; j: independently allocated culture/explant/construct unit within donor; b: independent preparation/manufacturing/differentiation batch as appropriate; r: technical assay repeat or cell observation within allocated culture; z(A): indicators for the named controlled preparation/state/reference conditions; u d,v b,q dj: donor, batch and allocated-culture effects; crossed donor/preparation structure must follow actual allocation; X d: pretreatment donor/model attributes and baseline function only; family code: LAB
Sample size design: family: multilevel donor-by-preparation/condition experimental contrast precision; independent-donor and independent-batch allocation simulation; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Clock change cannot define success; lineage identity/genomic stability are co-gates, not adjusted posttreatment confounders; locus-family multiplicity fixed in advance.
Randomize independently allocated cultures/explants within blocking factors; donor and preparation/batch are the biological replication units for transport.
Technical replicates/cells/fields reduce measurement error but do not increase independent donor count.
Culture viability, identity loss, assay failure and lineage selection are outcomes/selection processes; do not silently condition treatment comparisons on successful posttreatment models.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Clock changes without functional effects support passenger status; uniform strong functional effects would reject the restricted-driver hypothesis.; preanalysis gate: Clock change cannot define success; lineage identity/genomic stability are co-gates, not adjusted posttreatment confounders; locus-family multiplicity fixed in advance.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Clock change cannot define success; lineage identity/genomic stability are co-gates, not adjusted posttreatment confounders; locus-family multiplicity fixed in advance.
Leave donors and manufacturing batches out; compare crossed/nested covariance structures supported by the design.
Report viable/failed model coverage, censored assays, donor-context heterogeneity and robustness to contamination/handling controls.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Confirmed donor/preparation access and allocated culture design; validated functional assay, dynamic range, technical error, detection limits and identity/viability gates.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Clock change cannot define success; lineage identity/genomic stability are co-gates, not adjusted posttreatment confounders; locus-family multiplicity fixed in advance.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/116; question: Do selected methylation loci driving age predictions directly alter cellular function, or merely report underlying state?; aims: Prioritize reproducible clock loci linked to accessible regulatory programs in human cells.
Test locus-specific perturbations against orthogonal functional and identity readouts.
Replicate causal effects across older donors and independent tissue models.; design verbatim planning only: Population model: Consented older-donor fibroblast and muscle-cell models in an institutional genomic laboratory. Comparator: Matched control loci, sham manipulations, and nonspecific state-change controls. Causal replication: Randomized culture allocation, orthogonal perturbation modalities, blinded assays, and independent donors. Sample planning: Plan 20–30 donors and a focused locus panel; pilot viability gates expansion.; outcomes verbatim: Primary: stress-recovery function independent of clock score. Secondary: identity preservation, chromatin state, proliferation, and genomic stability.; analysis verbatim inherited not authoritative if refined: Block donor age, sex, disease, and passage; preregister locus families, correct molecular multiplicity, and distribute donor replicates across experimental batches.; falsification verbatim: Clock changes without functional effects support passenger status; uniform strong functional effects would reject the restricted-driver hypothesis.; source ids unverified leads: reprogramming-genomic-tech-2026
epigenetic-interventions-review-2026
aging-clocks-critical-review-2026; prospectus precedence: No

L21 — Independent identity and biological-potency audit of marketed research peptides

Rank: 66
Cluster: Peptide translation/delivery/quality
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: POTENCY_CENSOR; allocation or exposure unit: legally acquired lot sampled within an explicit product-category frame; independent sampling unit: independent product lot with supplier clustering; repeated measurement unit: technical activity/chemistry replicate and laboratory; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: censored analytical distribution/change; a treatment effect requires the separately specified randomized assignment and correct sampling frame; specific constraint: Below-LOQ activity gives bounds rather than a positive substituted value; valid nonquantifiable activity differs from assay failure; chemistry-only lots have no invented potency endpoint.
Primary estimand: category-specific distribution of absolute log activity discrepancy versus reference-equivalent label amount, retaining censored lots
Candidate primary outcome: absolute log observed-activity/reference-expected-activity ratio
Primary finalized: No
Mathematical model: type: censored analytical/audit precision with independent lots/households, supplier/laboratory structure and lower-tail sensitivity; formulas: For latent positive concentration/activity Q with density f_Q, a quantified result contributes f_Q(q); left-censored q<LOQ contributes F_Q(LOQ); an interval result [L,U] contributes F_Q(U)-F_Q(L).
Technical assay failure is a separate missing-result state and cannot be treated as a biological below-LOQ result. Avoid LOQ/2 and arbitrary positive constants unless an explicitly labeled sensitivity model justifies them.
For expected reference activity E_l>0, signed R_l=log(Q_l/E_l) and discrepancy D_l=abs(R_l). If 0<=Q_l<LOQ_l, R_l<log(LOQ_l/E_l); D_l has lower bound max[0,log(E_l/LOQ_l)] and no finite upper bound without a positive lower activity bound.; variables: Y: absolute log observed-activity/reference-expected-activity ratio; i: independent product lot with supplier clustering; t: technical activity/chemistry replicate and laboratory; the registered visit/time-window index, not another independent unit; A: legally acquired lot sampled within an explicit product-category frame; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: category-specific distribution of absolute log activity discrepancy versus reference-equivalent label amount, retaining censored lots; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; Q,q: latent and measured per-thesis activity/concentration in assay-defined physical units; LOQ: validated, possibly batch-specific lower quantification limit; F Q,f Q: chosen distribution and density for latent quantity; below-limit shape may not be identified; L,U: reported analytical lower/upper bounds; l: independent product lot; E l: reference-equivalent activity implied by a quantity label under a validated category-specific assay; chemistry-only lots have no E_l potency endpoint; R l,D l: signed log ratio and absolute log discrepancy; Q=0 gives extended/infinite discrepancy rather than a finite invented log value; family code: POTENCY_CENSOR
Sample size design: family: censored analytical/audit precision with independent lots/households, supplier/laboratory structure and lower-tail sensitivity; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Below-LOQ activity gives bounds rather than a positive substituted value; valid nonquantifiable activity differs from assay failure; chemistry-only lots have no invented potency endpoint.
Validated assay range, blanks/recovery and category-specific activity units; never pool unlike biological assays as equivalent potency.
Censoring limits are known; unobserved lower-tail shape is assumption dependent; sampling frame/inclusion mechanism governs prevalence inference.
At true zero activity the ordinary finite log-ratio distribution is undefined; use extended-valued discrepancy/bounds and the nonquantifiable fraction; any quantifiable-only summary states its restricted population.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Near-complete identity–potency agreement with narrow uncertainty would weaken the hypothesis of substantial hidden functional variability.; preanalysis gate: Below-LOQ activity gives bounds rather than a positive substituted value; valid nonquantifiable activity differs from assay failure; chemistry-only lots have no invented potency endpoint.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Below-LOQ activity gives bounds rather than a positive substituted value; valid nonquantifiable activity differs from assay failure; chemistry-only lots have no invented potency endpoint.
Report nonquantifiable fraction and transparent interval/bounds alongside distribution-model estimates.
Vary lower-tail distributions, blank/recovery error, assay-failure MNAR and supplier/lab clustering.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Below-LOQ activity gives bounds rather than a positive substituted value; valid nonquantifiable activity differs from assay failure; chemistry-only lots have no invented potency endpoint.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/65; question: Do labeled peptide identity and quantity predict actual receptor activity across legally obtained research-only product lots?; aims: Build a blinded identity, purity, and content assessment across prespecified product categories.
Compare analytical results with validated receptor or cell-based potency assays.
Replicate discordant lots independently and develop a transparent research-quality reporting standard.; design verbatim planning only: Population model: Legally acquired laboratory-only samples plus authenticated institutional reference materials; no human use. Comparator: Reference-grade materials and label claims. Causal replication: Blinded coding, orthogonal analytical platforms, reference mixtures, and independent laboratory replication. Sample planning: Plan 60–90 lots across categories; sampling frame and pilot variance determine final coverage.; outcomes verbatim: Primary: lot-level discrepancy between assay-defined activity and reference-equivalent activity expected from the quantity label, with prespecified censoring or bounds for nonquantifiable activity. Report nonquantifiable lots and assay failures explicitly; do not discard absent-activity results or substitute arbitrary positive constants. Secondary: identity errors, impurity classes, analytical uncertainty and reporting completeness.; analysis verbatim inherited not authoritative if refined: Account for supplier clustering, acquisition date, and product class; prespecify analytical tolerances, control multiple product comparisons, and randomize lots across instrument batches.; falsification verbatim: Near-complete identity–potency agreement with narrow uncertainty would weaken the hypothesis of substantial hidden functional variability.; source ids unverified leads: fda-bpc157-briefing-2026
fda-pcac-peptides-july-2026
peptide-immunogenicity-safety-review-2025; prospectus precedence: Yes; prospectus file: thesis-prospectuses.json; prospectus design and estimand: The primary endpoint is the lot-level absolute log ratio of observed assay-defined activity to reference-equivalent activity expected from the labeled amount, measured at first controlled receipt and summarized within prespecified product categories. For activity below a validated quantification limit, retain the lot as a censored result and report a bound on discrepancy together with the fraction of nonquantifiable lots; do not drop it or substitute an arbitrary positive constant. Distinguish a valid finding of little or no activity from technical assay failure. Report signed ratios and analytical uncertainty as well. Category-specific models prevent unlike assays being pooled as equivalent biology. Products lacking quantity claims or validated assays enter separately labeled chemistry-only analyses, not an invented potency endpoint.; prospectus analysis plan: Estimate category-specific discrepancy distributions and uncertainty intervals with supplier clustering; technical replicates estimate measurement error rather than increase lot count. Prespecify censored-result methods for activity below quantification limits and report transparent bounds or sensitivity analyses when the lower tail cannot be identified. Report the nonquantifiable fraction separately; any continuous summary restricted to quantifiable lots must identify that restricted population. Compare blinded predictive models using label amount alone versus measured intact quantity, purity and impurity features, with cross-validation that holds suppliers or lots together. Prespecify analytical agreement limits from validation, not post hoc product rankings. Adjust replication analyses for deliberate oversampling of discordant lots and report all assay failures.; precedence note: Expanded prospectus refinements govern this scaffold; verbatim earlier brief material is retained for traceability, not as a contradictory analysis instruction. F15 is explicitly two arms; F30 has no unjustified symptom noninferiority claim.

L22 — When peptide impurities invalidate a nominally safe active molecule

Rank: 49
Cluster: Peptide translation/delivery/quality
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: LAB; allocation or exposure unit: donor culture randomized to characterized preparations/fractions/control; independent sampling unit: independent donor and preparation lot, crossed; repeated measurement unit: barrier assay and replicate; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: controlled in-vitro/ex-vivo allocation contrast when culture allocation is randomized; no direct human therapeutic-efficacy claim; specific constraint: Impurity attribution needs matched excipients/composition and orthogonal controls; observed chemical-feature associations alone do not identify a causal impurity effect.
Primary estimand: controlled preparation/fraction contrast in epithelial barrier integrity within donor
Candidate primary outcome: prespecified epithelial barrier-integrity measurement
Primary finalized: No
Mathematical model: type: multilevel donor-by-preparation/condition experimental contrast precision; independent-donor and independent-batch allocation simulation; formulas: Y_djbrt = alpha + z(A_djb)^T theta + gamma^T X_d + u_d + v_b + q_dj + epsilon_djbrt.
Delta_model = c^T theta standardized to sampled donors/conditions; r is a technical repeat, not another donor or manufacturing batch.; variables: Y: prespecified epithelial barrier-integrity measurement; i: independent donor and preparation lot, crossed; t: barrier assay and replicate; the registered visit/time-window index, not another independent unit; A: donor culture randomized to characterized preparations/fractions/control; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: controlled preparation/fraction contrast in epithelial barrier integrity within donor; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; d: independent tissue/cell donor; j: independently allocated culture/explant/construct unit within donor; b: independent preparation/manufacturing/differentiation batch as appropriate; r: technical assay repeat or cell observation within allocated culture; z(A): indicators for the named controlled preparation/state/reference conditions; u d,v b,q dj: donor, batch and allocated-culture effects; crossed donor/preparation structure must follow actual allocation; X d: pretreatment donor/model attributes and baseline function only; family code: LAB
Sample size design: family: multilevel donor-by-preparation/condition experimental contrast precision; independent-donor and independent-batch allocation simulation; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Impurity attribution needs matched excipients/composition and orthogonal controls; observed chemical-feature associations alone do not identify a causal impurity effect.
Randomize independently allocated cultures/explants within blocking factors; donor and preparation/batch are the biological replication units for transport.
Technical replicates/cells/fields reduce measurement error but do not increase independent donor count.
Culture viability, identity loss, assay failure and lineage selection are outcomes/selection processes; do not silently condition treatment comparisons on successful posttreatment models.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Equivalent toxicity after impurity removal, or no independent replication of implicated fractions, would reject impurity-driven attribution.; preanalysis gate: Impurity attribution needs matched excipients/composition and orthogonal controls; observed chemical-feature associations alone do not identify a causal impurity effect.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Impurity attribution needs matched excipients/composition and orthogonal controls; observed chemical-feature associations alone do not identify a causal impurity effect.
Leave donors and manufacturing batches out; compare crossed/nested covariance structures supported by the design.
Report viable/failed model coverage, censored assays, donor-context heterogeneity and robustness to contamination/handling controls.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Confirmed donor/preparation access and allocated culture design; validated functional assay, dynamic range, technical error, detection limits and identity/viability gates.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Impurity attribution needs matched excipients/composition and orthogonal controls; observed chemical-feature associations alone do not identify a causal impurity effect.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/48; question: Which impurity classes explain epithelial toxicity when nominally identical peptide preparations produce different biological responses?; aims: Characterize impurity profiles in authenticated development-stage peptide preparations.
Compare barrier integrity and stress responses using matched intestinal and skin models.
Validate causal attribution through orthogonal fractionation and independent preparation batches.; design verbatim planning only: Population model: Human intestinal and skin organotypic cultures with institutional reference preparations. Comparator: Purified active material, matched vehicle, and composition-characterized fractions. Causal replication: Randomized exposures, blinded barrier assays, orthogonal chemical attribution, and donor replication. Sample planning: Plan 12–18 donors and 20 characterized preparations; pilot effect variability guides expansion.; outcomes verbatim: Primary: epithelial barrier integrity. Secondary: viability, stress signaling, recovery after exposure, and attribution uncertainty.; analysis verbatim inherited not authoritative if refined: Block tissue donor, baseline barrier function, and formulation excipients; prespecify impurity families, control molecular multiplicity, and balance preparations across analytical batches.; falsification verbatim: Equivalent toxicity after impurity removal, or no independent replication of implicated fractions, would reject impurity-driven attribution.; source ids unverified leads: peptide-immunogenicity-safety-review-2025
fda-bpc157-briefing-2026
fda-compounding-questions-2026; prospectus precedence: No

L23 — Oral peptide delivery across heterogeneous intestinal physiology

Rank: 149
Cluster: Peptide translation/delivery/quality
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: LAB_FACTOR; allocation or exposure unit: organoid barrier unit allocated to formulation and physiology conditions; independent sampling unit: independent donor with independently allocated organoid cultures; repeated measurement unit: passage and activity/permeability replicate; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: controlled in-vitro/ex-vivo allocation contrast when culture allocation is randomized; no direct human therapeutic-efficacy claim; specific constraint: Passage and toxicity are separate outcomes; donor tissue context is not randomized and in-vitro passage cannot identify human oral bioavailability.
Primary estimand: formulation-by-physiology interaction on intact bioactive passage
Candidate primary outcome: intact bioactive peptide passage across the model barrier
Primary finalized: No
Mathematical model: type: multilevel donor-by-preparation/condition experimental contrast precision; independent-donor and independent-batch allocation simulation; formulas: Y_djbrt = alpha + z(A_djb)^T theta + gamma^T X_d + u_d + v_b + q_dj + epsilon_djbrt.
Delta_model = c^T theta standardized to sampled donors/conditions; r is a technical repeat, not another donor or manufacturing batch.
For two controlled factors B,C, add beta_B B + beta_C C + beta_BC B*C; estimate interaction as mu_11-mu_10-mu_01+mu_00 on the chosen additive scale. Donor age/disease attributes themselves are not randomized.; variables: Y: intact bioactive peptide passage across the model barrier; i: independent donor with independently allocated organoid cultures; t: passage and activity/permeability replicate; the registered visit/time-window index, not another independent unit; A: organoid barrier unit allocated to formulation and physiology conditions; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: formulation-by-physiology interaction on intact bioactive passage; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; d: independent tissue/cell donor; j: independently allocated culture/explant/construct unit within donor; b: independent preparation/manufacturing/differentiation batch as appropriate; r: technical assay repeat or cell observation within allocated culture; z(A): indicators for the named controlled preparation/state/reference conditions; u d,v b,q dj: donor, batch and allocated-culture effects; crossed donor/preparation structure must follow actual allocation; X d: pretreatment donor/model attributes and baseline function only; B,C: the actual two condition/context factors named in this thesis; fix levels and identify which factors are randomized; beta BC: interaction parameter; if context is observational it is not a causal effect of changing donor age/disease; family code: LAB_FACTOR
Sample size design: family: multilevel donor-by-preparation/condition experimental contrast precision; independent-donor and independent-batch allocation simulation; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Passage and toxicity are separate outcomes; donor tissue context is not randomized and in-vitro passage cannot identify human oral bioavailability.
Randomize independently allocated cultures/explants within blocking factors; donor and preparation/batch are the biological replication units for transport.
Technical replicates/cells/fields reduce measurement error but do not increase independent donor count.
Culture viability, identity loss, assay failure and lineage selection are outcomes/selection processes; do not silently condition treatment comparisons on successful posttreatment models.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Stable formulation rankings across conditions, or uptake coupled to persistent barrier injury, would defeat the proposed robust-delivery premise.; preanalysis gate: Passage and toxicity are separate outcomes; donor tissue context is not randomized and in-vitro passage cannot identify human oral bioavailability.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Passage and toxicity are separate outcomes; donor tissue context is not randomized and in-vitro passage cannot identify human oral bioavailability.
Leave donors and manufacturing batches out; compare crossed/nested covariance structures supported by the design.
Report viable/failed model coverage, censored assays, donor-context heterogeneity and robustness to contamination/handling controls.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Confirmed donor/preparation access and allocated culture design; validated functional assay, dynamic range, technical error, detection limits and identity/viability gates.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Passage and toxicity are separate outcomes; donor tissue context is not randomized and in-vitro passage cannot identify human oral bioavailability.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/148; question: Can delivery systems preserve peptide activity across the intestinal environments typical of aging and metabolic disease?; aims: Map stability and activity across clinically plausible intestinal conditions.
Compare delivery candidates in donor-derived intestinal barrier models.
Validate rank stability using independent donor models and pharmacokinetic simulations.; design verbatim planning only: Population model: Human intestinal organoid barriers representing older and metabolically impaired donors. Comparator: Unprotected reference peptide, matched carrier controls, and benchmark development formulations. Causal replication: Factorial condition randomization, blinded activity testing, and independent donor validation. Sample planning: Plan 20 donors and three candidate classes; pilot model reliability determines factorial scope.; outcomes verbatim: Primary: intact bioactive peptide passage. Secondary: barrier recovery, variability across donors, carrier toxicity, and simulated exposure uncertainty.; analysis verbatim inherited not authoritative if refined: Account for donor medication and metabolic status; prespecify condition interactions, correct formulation multiplicity, and block organoid passage and analytical batches.; falsification verbatim: Stable formulation rankings across conditions, or uptake coupled to persistent barrier injury, would defeat the proposed robust-delivery premise.; source ids unverified leads: nx210c-peptide-safety-2025
peptide-immunogenicity-safety-review-2025
rna-delivery-review-2025; prospectus precedence: No

L24 — Topical peptide exposure in aged and diabetic wound microenvironments

Rank: 50
Cluster: Peptide translation/delivery/quality
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: LAB; allocation or exposure unit: within-donor explant randomized to candidate/reference/vehicle condition; independent sampling unit: independent skin donor with allocated explant units; repeated measurement unit: histology, exposure and barrier time point; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: controlled in-vitro/ex-vivo allocation contrast when culture allocation is randomized; no direct human therapeutic-efficacy claim; specific constraint: Do not select only barrier-preserved treated samples; assess joint success/failure; diabetes/perfusion strata require replication and model transport remains provisional.
Primary estimand: within-donor candidate effect on re-epithelialization with preserved barrier under a fixed joint rule
Candidate primary outcome: re-epithelialization score paired with independent barrier function
Primary finalized: No
Mathematical model: type: multilevel donor-by-preparation/condition experimental contrast precision; independent-donor and independent-batch allocation simulation; formulas: Y_djbrt = alpha + z(A_djb)^T theta + gamma^T X_d + u_d + v_b + q_dj + epsilon_djbrt.
Delta_model = c^T theta standardized to sampled donors/conditions; r is a technical repeat, not another donor or manufacturing batch.
Define joint success J_dj=1{re-epithelialization meets registered criterion AND independent barrier function is preserved}; estimate randomized marginal joint-success contrasts and each component separately, or select a fully registered bivariate/co-primary rule before enrollment. Do not retain only successful posttreatment co-gates.; variables: Y: re-epithelialization score paired with independent barrier function; i: independent skin donor with allocated explant units; t: histology, exposure and barrier time point; the registered visit/time-window index, not another independent unit; A: within-donor explant randomized to candidate/reference/vehicle condition; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: within-donor candidate effect on re-epithelialization with preserved barrier under a fixed joint rule; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; d: independent tissue/cell donor; j: independently allocated culture/explant/construct unit within donor; b: independent preparation/manufacturing/differentiation batch as appropriate; r: technical assay repeat or cell observation within allocated culture; z(A): indicators for the named controlled preparation/state/reference conditions; u d,v b,q dj: donor, batch and allocated-culture effects; crossed donor/preparation structure must follow actual allocation; X d: pretreatment donor/model attributes and baseline function only; J dj: per-thesis joint-success indicator with thresholds externally justified before outcome analysis; family code: LAB
Sample size design: family: multilevel donor-by-preparation/condition experimental contrast precision; independent-donor and independent-batch allocation simulation; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Do not select only barrier-preserved treated samples; assess joint success/failure; diabetes/perfusion strata require replication and model transport remains provisional.
Randomize independently allocated cultures/explants within blocking factors; donor and preparation/batch are the biological replication units for transport.
Technical replicates/cells/fields reduce measurement error but do not increase independent donor count.
Culture viability, identity loss, assay failure and lineage selection are outcomes/selection processes; do not silently condition treatment comparisons on successful posttreatment models.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Equivalent intact exposure across wound environments, or no exposure–repair relationship, would reject the proposed delivery limitation.; preanalysis gate: Do not select only barrier-preserved treated samples; assess joint success/failure; diabetes/perfusion strata require replication and model transport remains provisional.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Do not select only barrier-preserved treated samples; assess joint success/failure; diabetes/perfusion strata require replication and model transport remains provisional.
Leave donors and manufacturing batches out; compare crossed/nested covariance structures supported by the design.
Report viable/failed model coverage, censored assays, donor-context heterogeneity and robustness to contamination/handling controls.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Confirmed donor/preparation access and allocated culture design; validated functional assay, dynamic range, technical error, detection limits and identity/viability gates.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Do not select only barrier-preserved treated samples; assess joint success/failure; diabetes/perfusion strata require replication and model transport remains provisional.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/49; question: Does a peptide tolerated on healthy experimental wounds retain exposure, activity, and repair compatibility in clinically impaired skin?; aims: Characterize peptide persistence in matched healthy and impaired wound-fluid matrices.
Relate local availability to repair in human skin explants.
Validate exposure–repair relationships in independent diabetic and older-donor samples.; design verbatim planning only: Population model: Consented discarded skin and wound-fluid biobank specimens; candidate development materials under institutional agreements. Comparator: Healthy-donor explants, vehicle, and intact reference peptide. Causal replication: Within-donor randomization, blinded histology, orthogonal exposure assays, and independent-donor replication. Sample planning: Plan 30–40 donors across impaired and control strata; pilot tissue availability determines balance.; outcomes verbatim: Primary: re-epithelialization with preserved barrier function. Secondary: intact local peptide, matrix remodeling, viability, and durability.; analysis verbatim inherited not authoritative if refined: Address diabetes severity, perfusion, donor age, and medications; prespecify exposure contrasts, correct secondary multiplicity, and block specimen-processing batches.; falsification verbatim: Equivalent intact exposure across wound environments, or no exposure–repair relationship, would reject the proposed delivery limitation.; source ids unverified leads: tcp25-first-human-peptide-2026
peptide-immunogenicity-safety-review-2025
skin-photoaging-systematic-review-2025; prospectus precedence: No

L25 — Dual-agonist weight loss and preservation of functional muscle

Rank: 82
Cluster: Peptide translation/delivery/quality
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: MIXED_EVIDENCE; allocation or exposure unit: original dual-agonist trial assignment; separate observational therapy validation; independent sampling unit: participant within trial/cohort; repeated measurement unit: strength, tissue imaging and treatment follow-up; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Avoid conditioning on achieved weight loss or lean mass; clinical cohort cannot inherit trial randomization; death/inability handling requires a registered rule.
Primary estimand: original-trial assignment effect on fixed lower-limb strength at the trial-aligned endpoint
Candidate primary outcome: objectively measured lower-limb strength endpoint
Primary finalized: No
Mathematical model: type: contrast precision/power by design-matrix or simulation planning; formulas: E(Y_it | A_i, X_i, Y_i0) = alpha_t + z(A_i)^T theta_t + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Delta = sum_t w_t c^T theta_t; estimate by baseline-adjusted marginal standardization, with unit/cluster-appropriate inference.; variables: Y: objectively measured lower-limb strength endpoint; i: participant within trial/cohort; t: strength, tissue imaging and treatment follow-up; the registered visit/time-window index, not another independent unit; A: original dual-agonist trial assignment; separate observational therapy validation; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: original-trial assignment effect on fixed lower-limb strength at the trial-aligned endpoint; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; z(A): prespecified indicators for the actual randomized arms; two arms reduce to one indicator; family code: MIXED_EVIDENCE
Sample size design: family: contrast precision/power by design-matrix or simulation planning; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Avoid conditioning on achieved weight loss or lean mass; clinical cohort cannot inherit trial randomization; death/inability handling requires a registered rule.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The formula applies to the original randomized comparison only; observational validation/cohort associations receive separate labels and cannot be pooled as randomized arms.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: No association between early reserve indicators and later function would reject the proposed vulnerability model.; preanalysis gate: Avoid conditioning on achieved weight loss or lean mass; clinical cohort cannot inherit trial randomization; death/inability handling requires a registered rule.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Avoid conditioning on achieved weight loss or lean mass; clinical cohort cannot inherit trial randomization; death/inability handling requires a registered rule.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Avoid conditioning on achieved weight loss or lean mass; clinical cohort cannot inherit trial randomization; death/inability handling requires a registered rule.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/81; question: Which early physiological changes distinguish beneficial fat loss from loss of muscle function during investigational dual-peptide agonism?; aims: Reanalyze available dual-agonist trial body-composition and activity trajectories.
Test early reserve indicators against subsequent strength and endurance.
Validate the indicator set in an independent supervised incretin-treatment cohort.; design verbatim planning only: Population model: Archived investigational dual-agonist trial data and clinically treated adults with overweight or obesity. Comparator: Trial placebo and matched approved-therapy trajectories. Causal replication: Randomized treatment contrasts, prespecified moderator analysis, and independent observational validation. Sample planning: Seek 150–250 trial participants and 100 validation participants; outcome availability is the access gate.; outcomes verbatim: Primary: change in a prespecified objectively measured lower-limb strength endpoint at the trial-aligned follow-up. Secondary: endurance, lean tissue change, fat distribution, physical activity, discontinuation, and gastrointestinal burden.; analysis verbatim inherited not authoritative if refined: Adjust baseline frailty, protein intake, disease, and activity; avoid conditioning on achieved weight loss, control moderator multiplicity, and harmonize imaging batches.; falsification verbatim: No association between early reserve indicators and later function would reject the proposed vulnerability model.; source ids unverified leads: amycretin-phase1b2a-2025
select-semaglutide-cardiovascular-2023
resistance-sarcopenia-meta-2026; prospectus precedence: No

L26 — Renal aging and the pharmacokinetic interpretation of peptide safety

Rank: 180
Cluster: Peptide translation/delivery/quality
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: PK; allocation or exposure unit: existing clinical dose/exposure profiles; no new treatment assigned; independent sampling unit: participant and independent compound/cohort; repeated measurement unit: concentration sample and residual assay; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: prediction, calibration or model transportability; prediction gain is not clinical efficacy, mechanistic necessity or causal treatment selection; specific constraint: Renal error, sparse sampling and concomitant drugs affect clearance models; exposure calibration is not identified causal toxicity.
Primary estimand: external exposure-prediction error of joint renal/physiological model versus age-only/renal-only models
Candidate primary outcome: measured concentration-time profile and derived exposure under the observed regimen
Primary finalized: No
Mathematical model: type: external/nested predictive-validation precision with target prevalence/variance, independent-unit holdout coverage and model complexity; formulas: Gain = mean_test[L(Y,f_base(X)) - L(Y,f_new(X))], with both predictions frozen before test outcomes; bootstrap independent held-out units.
For a binary target assess E[Y | p_hat=p] versus p and calibration intercept/slope logit Pr(Y=1)=a_cal+b_cal logit(p_hat); for a continuous target use Y=a_cal+b_cal y_hat+error plus scale-appropriate calibration/loss.
C_it=f_PK(t,dose_i;theta_i)+e_it; log(CL_i)=beta_0+beta_R log(Renal_i/Renal_ref)+beta_X^T X_i+u_i; validate concentration/exposure predictions on held-out compound/cohort.; variables: Y: measured concentration-time profile and derived exposure under the observed regimen; i: participant and independent compound/cohort; t: concentration sample and residual assay; the registered visit/time-window index, not another independent unit; A: existing clinical dose/exposure profiles; no new treatment assigned; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: external exposure-prediction error of joint renal/physiological model versus age-only/renal-only models; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; L: one proper or scientifically justified primary loss/scoring rule fixed for the actual outcome type; positive Gain favors new model; f base,f new: locked baseline/new prediction functions using only information available at the registered prediction time; test: untouched person/family/lab/site/time set as specified per thesis; p hat,y hat: frozen probability or continuous prediction; a cal,b cal: calibration intercept and slope, estimated on evaluation data for assessment; using them to update the model requires another evaluation set; C it: bioanalytically measured concentration at actual recorded time; f PK: prespecified pharmacokinetic model; structural form must be justified from data/biology; CL i: individual clearance, not directly observed when sparse samples do not identify it; Renal i,Renal ref: measured renal-function marker and fixed unit-matched reference scale; dose i: existing clinical regimen, no new dose recommendation; beta R,beta X: estimated covariate coefficients; family code: PK
Sample size design: family: external/nested predictive-validation precision with target prevalence/variance, independent-unit holdout coverage and model complexity; formula: Simulate or bootstrap the independent-unit test loss/calibration/decision criterion; plan cases/events, noncases and independent sites/families/programs separately. No universal events-per-variable or participant N is asserted.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Renal error, sparse sampling and concomitant drugs affect clearance models; exposure calibration is not identified causal toxicity.
Train/tune/test separation at the actual biological/participant/family/time level; no future, postoutcome or target-derived predictors.
Validation support must cover intended deployment; calibration and missing/failing measurements are reported with discrimination/loss.
A validation subset selected by culture success, assay viability, consent or data completeness is a different target unless the full strategy is explicitly evaluated.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: No improvement over age-only models, or failed external transport across compounds, would reject general physiological extrapolation.; preanalysis gate: Renal error, sparse sampling and concomitant drugs affect clearance models; exposure calibration is not identified causal toxicity.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Renal error, sparse sampling and concomitant drugs affect clearance models; exposure calibration is not identified causal toxicity.
Repeat loss/calibration with independent-cluster resampling, prespecified subgroup/transport shifts and plausible missingness.
Compare fixed simple baselines, calibration and uncertainty coverage; report abstention/failure/turnaround rather than deleting hard cases.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Frozen prediction time/horizon, proper loss, baseline model, feature availability, independent split/holdout and target deployment population.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Renal error, sparse sampling and concomitant drugs affect clearance models; exposure calibration is not identified causal toxicity.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/179; question: Can early healthy-volunteer peptide pharmacokinetics predict exposure and adverse-event risk in older adults with reduced renal reserve?; aims: Harmonize approved and investigational peptide pharmacokinetic datasets with renal measurements.
Develop mechanistic population models separating clearance from distribution changes.
Externally validate exposure predictions in archived older-adult clinical samples.; design verbatim planning only: Population model: Existing clinical pharmacology datasets and residual samples from supervised approved-therapy studies. Comparator: Age-only, renal-only, and joint physiological models. Causal replication: Prespecified external compound and cohort validation; mechanistic model predictions are separated from causal toxicity claims. Sample planning: Seek 300–500 pooled participants and approximately 80 validation profiles, conditional on data rights.; outcomes verbatim: Primary: external exposure-prediction error. Secondary: adverse-event calibration, subgroup uncertainty, and robustness to sparse clinical sampling.; analysis verbatim inherited not authoritative if refined: Address indication, concomitant drugs, renal measurement error, and body composition; control model-selection multiplicity and calibrate bioanalytical batches with shared references.; falsification verbatim: No improvement over age-only models, or failed external transport across compounds, would reject general physiological extrapolation.; source ids unverified leads: nx210c-peptide-safety-2025
peptide-immunogenicity-safety-review-2025
fda-off-label-use-2026; prospectus precedence: No

L27 — Bioactive collagen fragments versus ordinary amino-acid nutrition

Rank: 171
Cluster: Peptide translation/delivery/quality
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: LAB_TRIAL; allocation or exposure unit: original collagen trial assignment plus separately controlled donor tendon-model allocation; independent sampling unit: trial participant; independent tissue donor for model experiments; repeated measurement unit: fragment assay and mechanically loaded construct replicate; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: controlled in-vitro/ex-vivo allocation contrast when culture allocation is randomized; no direct human therapeutic-efficacy claim; specific constraint: Human fragment association and ex-vivo effects do not alone establish mediation of human tendon benefit; keep trial and donor-model evidence separate.
Primary estimand: controlled matrix-mechanical adaptation contrast versus nitrogen/amino-acid-matched reference; separate original-trial functional contrast
Candidate primary outcome: prespecified tendon-model matrix mechanical-adaptation endpoint
Primary finalized: No
Mathematical model: type: multilevel donor-by-preparation/condition experimental contrast precision; independent-donor and independent-batch allocation simulation; formulas: Y_djbrt = alpha + z(A_djb)^T theta + gamma^T X_d + u_d + v_b + q_dj + epsilon_djbrt.
Delta_model = c^T theta standardized to sampled donors/conditions; r is a technical repeat, not another donor or manufacturing batch.; variables: Y: prespecified tendon-model matrix mechanical-adaptation endpoint; i: trial participant; independent tissue donor for model experiments; t: fragment assay and mechanically loaded construct replicate; the registered visit/time-window index, not another independent unit; A: original collagen trial assignment plus separately controlled donor tendon-model allocation; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: controlled matrix-mechanical adaptation contrast versus nitrogen/amino-acid-matched reference; separate original-trial functional contrast; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; d: independent tissue/cell donor; j: independently allocated culture/explant/construct unit within donor; b: independent preparation/manufacturing/differentiation batch as appropriate; r: technical assay repeat or cell observation within allocated culture; z(A): indicators for the named controlled preparation/state/reference conditions; u d,v b,q dj: donor, batch and allocated-culture effects; crossed donor/preparation structure must follow actual allocation; X d: pretreatment donor/model attributes and baseline function only; family code: LAB_TRIAL
Sample size design: family: multilevel donor-by-preparation/condition experimental contrast precision; independent-donor and independent-batch allocation simulation; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Human fragment association and ex-vivo effects do not alone establish mediation of human tendon benefit; keep trial and donor-model evidence separate.
Randomize independently allocated cultures/explants within blocking factors; donor and preparation/batch are the biological replication units for transport.
Technical replicates/cells/fields reduce measurement error but do not increase independent donor count.
Culture viability, identity loss, assay failure and lineage selection are outcomes/selection processes; do not silently condition treatment comparisons on successful posttreatment models.
Original human-trial randomized effects and donor-model effects have separate units/outcomes; triangulation does not itself identify human mediation.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Equivalent responses under amino-acid matching or absent fragment–mechanics coupling would reject peptide-specific mediation.; preanalysis gate: Human fragment association and ex-vivo effects do not alone establish mediation of human tendon benefit; keep trial and donor-model evidence separate.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Human fragment association and ex-vivo effects do not alone establish mediation of human tendon benefit; keep trial and donor-model evidence separate.
Leave donors and manufacturing batches out; compare crossed/nested covariance structures supported by the design.
Report viable/failed model coverage, censored assays, donor-context heterogeneity and robustness to contamination/handling controls.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Confirmed donor/preparation access and allocated culture design; validated functional assay, dynamic range, technical error, detection limits and identity/viability gates.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Human fragment association and ex-vivo effects do not alone establish mediation of human tendon benefit; keep trial and donor-model evidence separate.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/170; question: Are tendon responses to collagen peptides mediated by absorbed signaling fragments or by their contribution to ordinary protein nutrition?; aims: Characterize circulating fragments in archived collagen-intervention samples.
Compare fragment-rich and amino-acid-matched exposures in loaded human tendon models.
Test whether fragment signatures mediate randomized changes in tendon mechanics.; design verbatim planning only: Population model: Archived supervised collagen trial specimens and donor-derived mechanically loaded tendon models. Comparator: Nitrogen- and amino-acid-matched controls, vehicle, and randomized trial controls. Causal replication: Blinded chemical profiling, controlled model exposure, and independent-donor replication linked to trial assignment. Sample planning: Seek 80–120 trial participants and 15–20 tissue donors; analytic detectability gates scope.; outcomes verbatim: Primary: matrix mechanical adaptation. Secondary: fragment exposure, collagen turnover, stiffness, and concordance with human functional outcomes.; analysis verbatim inherited not authoritative if refined: Address total protein, training load, baseline diet, and donor pathology; prespecify fragment families, correct metabolite multiplicity, and balance trial groups across analytical batches.; falsification verbatim: Equivalent responses under amino-acid matching or absent fragment–mechanics coupling would reject peptide-specific mediation.; source ids unverified leads: collagen-umbrella-review-2026
collagen-tendon-rct-2025
peptide-immunogenicity-safety-review-2025; prospectus precedence: No

L28 — Physical stability and reproducibility of peptide formulation potency

Rank: 85
Cluster: Peptide translation/delivery/quality
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: STABILITY; allocation or exposure unit: independent manufacturing batch aliquots randomized to stability conditions; independent sampling unit: manufacturing batch within formulation family; repeated measurement unit: time/condition potency and structure measurement; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: prediction, calibration or model transportability; prediction gain is not clinical efficacy, mechanistic necessity or causal treatment selection; specific constraint: Many time points do not replace manufacturing batches; all aliquots from a batch stay together in validation; nondetect potency remains censored.
Primary estimand: held-out batch prediction of potency loss relative to fresh reference from structural features
Candidate primary outcome: potency change over the registered storage interval
Primary finalized: No
Mathematical model: type: external/nested predictive-validation precision with target prevalence/variance, independent-unit holdout coverage and model complexity; formulas: Gain = mean_test[L(Y,f_base(X)) - L(Y,f_new(X))], with both predictions frozen before test outcomes; bootstrap independent held-out units.
For a binary target assess E[Y | p_hat=p] versus p and calibration intercept/slope logit Pr(Y=1)=a_cal+b_cal logit(p_hat); for a continuous target use Y=a_cal+b_cal y_hat+error plus scale-appropriate calibration/loss.
Below-LOQ potency contributes a censoring-aware likelihood or bounded proper-score interval; do not score an invented positive value or drop the low-potency batch. Distinguish true potency loss from a technical assay failure.; variables: Y: potency change over the registered storage interval; i: manufacturing batch within formulation family; t: time/condition potency and structure measurement; the registered visit/time-window index, not another independent unit; A: independent manufacturing batch aliquots randomized to stability conditions; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: held-out batch prediction of potency loss relative to fresh reference from structural features; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; L: one proper or scientifically justified primary loss/scoring rule fixed for the actual outcome type; positive Gain favors new model; f base,f new: locked baseline/new prediction functions using only information available at the registered prediction time; test: untouched person/family/lab/site/time set as specified per thesis; p hat,y hat: frozen probability or continuous prediction; a cal,b cal: calibration intercept and slope, estimated on evaluation data for assessment; using them to update the model requires another evaluation set; LOQ: validated batch/assay-specific potency quantification limit; family code: STABILITY
Sample size design: family: external/nested predictive-validation precision with target prevalence/variance, independent-unit holdout coverage and model complexity; formula: Simulate or bootstrap the independent-unit test loss/calibration/decision criterion; plan cases/events, noncases and independent sites/families/programs separately. No universal events-per-variable or participant N is asserted.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Many time points do not replace manufacturing batches; all aliquots from a batch stay together in validation; nondetect potency remains censored.
Train/tune/test separation at the actual biological/participant/family/time level; no future, postoutcome or target-derived predictors.
Validation support must cover intended deployment; calibration and missing/failing measurements are reported with discrimination/loss.
A validation subset selected by culture success, assay viability, consent or data completeness is a different target unless the full strategy is explicitly evaluated.
Hold manufacturing preparations/batches together across all aliquots/recipients/time points; time density cannot substitute for independent batches.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Content-only testing matching the orthogonal panel in external batches would refute the claimed added predictive value.; preanalysis gate: Many time points do not replace manufacturing batches; all aliquots from a batch stay together in validation; nondetect potency remains censored.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Many time points do not replace manufacturing batches; all aliquots from a batch stay together in validation; nondetect potency remains censored.
Repeat loss/calibration with independent-cluster resampling, prespecified subgroup/transport shifts and plausible missingness.
Compare fixed simple baselines, calibration and uncertainty coverage; report abstention/failure/turnaround rather than deleting hard cases.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Frozen prediction time/horizon, proper loss, baseline model, feature availability, independent split/holdout and target deployment population.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Many time points do not replace manufacturing batches; all aliquots from a batch stay together in validation; nondetect potency remains censored.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/84; question: Do storage-related structural changes alter peptide potency before conventional identity and content assays indicate a problem?; aims: Map structural, particle, and chemical stability in development-stage formulations.
Relate stability signatures to receptor potency and cellular stress.
Validate a reduced predictive assay panel across independently manufactured batches.; design verbatim planning only: Population model: Authenticated institutional peptide formulations under prespecified laboratory stability conditions. Comparator: Fresh reference material, chemically degraded controls, and content-only quality assessment. Causal replication: Randomized condition allocation, blinded potency testing, independent manufacturing batches, and external laboratory replication. Sample planning: Plan four formulation families and six batches per family; pilot assay variance determines sampling density.; outcomes verbatim: Primary: prediction of potency loss. Secondary: aggregation, particle burden, barrier-cell stress, and uncertainty of stability classification.; analysis verbatim inherited not authoritative if refined: Block formulation composition, container type, and production batch; preregister predictors, control stability-endpoint multiplicity, and include shared reference standards across assay runs.; falsification verbatim: Content-only testing matching the orthogonal panel in external batches would refute the claimed added predictive value.; source ids unverified leads: peptide-immunogenicity-safety-review-2025
fda-compounding-questions-2026
fda-unapproved-drugs-2026; prospectus precedence: No

L29 — Fibrotic matrix sequestration as a barrier to peptide target engagement

Rank: 89
Cluster: Peptide translation/delivery/quality
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: LAB_FACTOR; allocation or exposure unit: within-donor culture/explant assigned peptide and matrix-control condition; independent sampling unit: independent donor and allocated explant/culture; repeated measurement unit: spatial exposure image and receptor/function assay; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: controlled in-vitro/ex-vivo allocation contrast when culture allocation is randomized; no direct human therapeutic-efficacy claim; specific constraint: Bulk accumulation and free exposure are different quantities; fibrosis strata are observational; avoid postselection of visible high-uptake regions.
Primary estimand: peptide-by-matrix interaction in free target-site exposure relative to bulk accumulation
Candidate primary outcome: free target-site exposure in a prespecified tissue compartment
Primary finalized: No
Mathematical model: type: multilevel donor-by-preparation/condition experimental contrast precision; independent-donor and independent-batch allocation simulation; formulas: Y_djbrt = alpha + z(A_djb)^T theta + gamma^T X_d + u_d + v_b + q_dj + epsilon_djbrt.
Delta_model = c^T theta standardized to sampled donors/conditions; r is a technical repeat, not another donor or manufacturing batch.
For two controlled factors B,C, add beta_B B + beta_C C + beta_BC B*C; estimate interaction as mu_11-mu_10-mu_01+mu_00 on the chosen additive scale. Donor age/disease attributes themselves are not randomized.; variables: Y: free target-site exposure in a prespecified tissue compartment; i: independent donor and allocated explant/culture; t: spatial exposure image and receptor/function assay; the registered visit/time-window index, not another independent unit; A: within-donor culture/explant assigned peptide and matrix-control condition; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: peptide-by-matrix interaction in free target-site exposure relative to bulk accumulation; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; d: independent tissue/cell donor; j: independently allocated culture/explant/construct unit within donor; b: independent preparation/manufacturing/differentiation batch as appropriate; r: technical assay repeat or cell observation within allocated culture; z(A): indicators for the named controlled preparation/state/reference conditions; u d,v b,q dj: donor, batch and allocated-culture effects; crossed donor/preparation structure must follow actual allocation; X d: pretreatment donor/model attributes and baseline function only; B,C: the actual two condition/context factors named in this thesis; fix levels and identify which factors are randomized; beta BC: interaction parameter; if context is observational it is not a causal effect of changing donor age/disease; family code: LAB_FACTOR
Sample size design: family: multilevel donor-by-preparation/condition experimental contrast precision; independent-donor and independent-batch allocation simulation; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Bulk accumulation and free exposure are different quantities; fibrosis strata are observational; avoid postselection of visible high-uptake regions.
Randomize independently allocated cultures/explants within blocking factors; donor and preparation/batch are the biological replication units for transport.
Technical replicates/cells/fields reduce measurement error but do not increase independent donor count.
Culture viability, identity loss, assay failure and lineage selection are outcomes/selection processes; do not silently condition treatment comparisons on successful posttreatment models.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Equivalent target engagement at matched bulk concentration across fibrosis states would reject sequestration as a meaningful barrier.; preanalysis gate: Bulk accumulation and free exposure are different quantities; fibrosis strata are observational; avoid postselection of visible high-uptake regions.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Bulk accumulation and free exposure are different quantities; fibrosis strata are observational; avoid postselection of visible high-uptake regions.
Leave donors and manufacturing batches out; compare crossed/nested covariance structures supported by the design.
Report viable/failed model coverage, censored assays, donor-context heterogeneity and robustness to contamination/handling controls.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Confirmed donor/preparation access and allocated culture design; validated functional assay, dynamic range, technical error, detection limits and identity/viability gates.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Bulk accumulation and free exposure are different quantities; fibrosis strata are observational; avoid postselection of visible high-uptake regions.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/88; question: Does fibrosis reduce active peptide access to parenchymal cells even when bulk tissue concentration appears adequate?; aims: Quantify matrix association across healthy and fibrotic human tissue preparations.
Measure spatial exposure and receptor engagement in matched organotypic models.
Test whether matrix characteristics predict functional response across independent donors.; design verbatim planning only: Population model: Human renal and dermal fibrotic tissue biobanks with organotypic culture models. Comparator: Nonfibrotic matched tissue, reference peptide classes, and matrix-reduced analytical controls. Causal replication: Within-donor randomized comparisons, orthogonal exposure imaging, and independent-donor validation. Sample planning: Plan 25–35 donors and a focused peptide panel; detectability and tissue access govern expansion.; outcomes verbatim: Primary: free target-site exposure versus bulk accumulation. Secondary: receptor engagement, tissue function, spatial heterogeneity, and matrix-binding prediction.; analysis verbatim inherited not authoritative if refined: Adjust fibrosis severity, renal disease, donor age, and tissue handling; prespecify peptide–matrix interactions, control spatial multiplicity, and block analytical batches by donor.; falsification verbatim: Equivalent target engagement at matched bulk concentration across fibrosis states would reject sequestration as a meaningful barrier.; source ids unverified leads: rna-delivery-review-2025
nx210c-peptide-safety-2025
msc-diabetic-kidney-rct-2026; prospectus precedence: No

L30 — Exposure-aware surveillance of harms attributed to unapproved peptides

Rank: 183
Cluster: Peptide translation/delivery/quality
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: SIGNAL; allocation or exposure unit: naturally reported event/product record; exposure classification applied blinded; independent sampling unit: independent report/patient and center, with duplicate detection; repeated measurement unit: adjudicator label and retained-sample assay; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: reproducibility of pharmacovigilance signal classification; reports are selected and cannot identify causal incidence; specific constraint: No exposed-population denominator: neither disproportionality nor report frequency is incidence or causal harm; missing products and stimulated reporting remain limitations.
Primary estimand: independent-center reproducibility of product-specific signal classification versus name-only grouping
Candidate primary outcome: fixed product-event signal-classification category
Primary finalized: No
Mathematical model: type: held-out-center classification/agreement precision and sparse reporting-signal simulation; formulas: For locked signal categories, estimate center-held-out agreement matrix P(C_new=k,C_reference=l) and corrected/uncorrected agreement with clustered uncertainty.
A reporting odds/disproportionality ratio, if secondary, uses report counts only: ROR=(a/b)/(c/d); it has no exposed-population denominator and does not estimate incidence or causal risk.; variables: Y: fixed product-event signal-classification category; i: independent report/patient and center, with duplicate detection; t: adjudicator label and retained-sample assay; the registered visit/time-window index, not another independent unit; A: naturally reported event/product record; exposure classification applied blinded; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: independent-center reproducibility of product-specific signal classification versus name-only grouping; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; C new,C reference: blinded product-specific and independent adjudicated signal-classification labels; k,l: predeclared signal categories; a,b,c,d: deduplicated report-table cells for product/event versus comparators, with each denominator explicitly stated; family code: SIGNAL
Sample size design: family: held-out-center classification/agreement precision and sparse reporting-signal simulation; formula: Use design-specific simulation or independent-unit resampling for the stated estimand and target confidence width/power; preserve all allocation, censoring, clustering and measurement-error structure.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: No exposed-population denominator: neither disproportionality nor report frequency is incidence or causal harm; missing products and stimulated reporting remain limitations.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Duplicate patients/reports, indication, coexposure and stimulated reporting must be addressed; reference classification is imperfect.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: No improvement over name-only classification, or complete signal instability under plausible missingness, would reject the taxonomy’s utility.; preanalysis gate: No exposed-population denominator: neither disproportionality nor report frequency is incidence or causal harm; missing products and stimulated reporting remain limitations.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: No exposed-population denominator: neither disproportionality nor report frequency is incidence or causal harm; missing products and stimulated reporting remain limitations.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
No exposed-population denominator: neither disproportionality nor report frequency is incidence or causal harm; missing products and stimulated reporting remain limitations.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/182; question: Can quality-linked exposure classification improve adverse-event surveillance compared with treating every similarly named peptide product as equivalent?; aims: Develop a reproducible exposure taxonomy from regulator records and clinical toxicology cases.
Link de-identified adverse events to independently characterized retained samples where available.
Validate signal stability under missing-product and reporting-bias sensitivity analyses.; design verbatim planning only: Population model: De-identified poison-center, pharmacovigilance, and specialist clinical records with lawful access. Comparator: Name-only product grouping and approved-therapy reference categories. Causal replication: Blinded adjudication, negative-control events, independent-center replication; disproportionality is not interpreted as incidence or causation. Sample planning: Seek 300–600 eligible reports and 30 retained samples; actual availability defines scope.; outcomes verbatim: Primary: reproducibility of product-specific signal classification. Secondary: adjudication agreement, missingness burden, and clinically important event clusters.; analysis verbatim inherited not authoritative if refined: Address indication, coexposure, selective reporting, and denominator uncertainty; prespecify signal families, correct multiplicity, and model analytical batch effects for retained samples.; falsification verbatim: No improvement over name-only classification, or complete signal instability under plausible missingness, would reject the taxonomy’s utility.; source ids unverified leads: fda-unapproved-drugs-2026
fda-compounding-questions-2026
fda-bpc157-briefing-2026; prospectus precedence: No

L31 — Durable functional rejuvenation without loss of tissue identity

Rank: 90
Cluster: Regenerative biology/cellular therapies/tissue repair
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: LAB_JOINT; allocation or exposure unit: donor culture randomized to approved state/control conditions; independent sampling unit: independent donor with allocated culture units; repeated measurement unit: repair, identity and lineage observation over follow-up; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: controlled in-vitro/ex-vivo allocation contrast when culture allocation is randomized; no direct human therapeutic-efficacy claim; specific constraint: Do not condition primary repair analysis on posttreatment identity survivors; joint success includes identity loss/failure and functional gain must persist.
Primary estimand: condition effect on durable barrier-repair success jointly retaining lineage identity
Candidate primary outcome: durable barrier-repair endpoint with prespecified identity/abnormal-growth co-gates
Primary finalized: No
Mathematical model: type: multilevel donor-by-preparation/condition experimental contrast precision; independent-donor and independent-batch allocation simulation; formulas: Y_djbrt = alpha + z(A_djb)^T theta + gamma^T X_d + u_d + v_b + q_dj + epsilon_djbrt.
Delta_model = c^T theta standardized to sampled donors/conditions; r is a technical repeat, not another donor or manufacturing batch.
Define J_dj=1{durable repair criterion AND retained identity AND no prespecified abnormal-growth failure}; estimate condition-specific joint success and report each component separately.; variables: Y: durable barrier-repair endpoint with prespecified identity/abnormal-growth co-gates; i: independent donor with allocated culture units; t: repair, identity and lineage observation over follow-up; the registered visit/time-window index, not another independent unit; A: donor culture randomized to approved state/control conditions; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: condition effect on durable barrier-repair success jointly retaining lineage identity; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; d: independent tissue/cell donor; j: independently allocated culture/explant/construct unit within donor; b: independent preparation/manufacturing/differentiation batch as appropriate; r: technical assay repeat or cell observation within allocated culture; z(A): indicators for the named controlled preparation/state/reference conditions; u d,v b,q dj: donor, batch and allocated-culture effects; crossed donor/preparation structure must follow actual allocation; X d: pretreatment donor/model attributes and baseline function only; J dj: joint functional/identity success under a prespecified rule; do not restrict repair comparison to identity-preserving treated survivors; family code: LAB_JOINT
Sample size design: family: multilevel donor-by-preparation/condition experimental contrast precision; independent-donor and independent-batch allocation simulation; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Do not condition primary repair analysis on posttreatment identity survivors; joint success includes identity loss/failure and functional gain must persist.
Randomize independently allocated cultures/explants within blocking factors; donor and preparation/batch are the biological replication units for transport.
Technical replicates/cells/fields reduce measurement error but do not increase independent donor count.
Culture viability, identity loss, assay failure and lineage selection are outcomes/selection processes; do not silently condition treatment comparisons on successful posttreatment models.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Transient improvement only, persistent identity loss, or enrichment of abnormal clones would falsify functional rejuvenation despite younger clock scores.; preanalysis gate: Do not condition primary repair analysis on posttreatment identity survivors; joint success includes identity loss/failure and functional gain must persist.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Do not condition primary repair analysis on posttreatment identity survivors; joint success includes identity loss/failure and functional gain must persist.
Leave donors and manufacturing batches out; compare crossed/nested covariance structures supported by the design.
Report viable/failed model coverage, censored assays, donor-context heterogeneity and robustness to contamination/handling controls.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Confirmed donor/preparation access and allocated culture design; validated functional assay, dynamic range, technical error, detection limits and identity/viability gates.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Do not condition primary repair analysis on posttreatment identity survivors; joint success includes identity loss/failure and functional gain must persist.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/89; question: Can partial cellular reprogramming produce durable repair gains while preserving lineage identity and avoiding selective expansion of abnormal cells?; aims: Map identity, genomic stability, and repair trajectories in older-donor skin models.
Test persistence after withdrawal using longitudinal tissue-function and lineage measurements.
Replicate candidate stable states across donors and independent organotypic models.; design verbatim planning only: Population model: Older-donor human fibroblasts and organotypic skin in an institutional reprogramming laboratory. Comparator: Untreated tissue, nonspecific proliferation controls, and full-reprogramming reference states. Causal replication: Randomized culture allocation, blinded repair assays, orthogonal identity assessment, and independent donors. Sample planning: Plan 20–30 donors; pilot viability and identity retention gate later validation.; outcomes verbatim: Primary: durable barrier repair with retained identity. Secondary: clonal diversity, genomic instability, methylation changes, and abnormal growth.; analysis verbatim inherited not authoritative if refined: Block donor age, diabetes, passage, and baseline damage; preregister state-selection criteria, correct molecular multiplicity, and distribute replicates across culture and sequencing batches.; falsification verbatim: Transient improvement only, persistent identity loss, or enrichment of abnormal clones would falsify functional rejuvenation despite younger clock scores.; source ids unverified leads: reprogramming-genomic-tech-2026
epigenetic-interventions-review-2026
fda-unapproved-products-warning; prospectus precedence: No

L32 — Mechanistic potency tests for stromal-cell therapy in diabetic kidney disease

Rank: 160
Cluster: Regenerative biology/cellular therapies/tissue repair
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: LAB_PRED; allocation or exposure unit: authenticated preparation allocated to donor kidney models under randomized blinding; independent sampling unit: independent manufacturing preparation and recipient donor, crossed; repeated measurement unit: organoid transport assay and preparation-marker measurement; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: prediction, calibration or model transportability; prediction gain is not clinical efficacy, mechanistic necessity or causal treatment selection; specific constraint: Release markers versus mechanism markers require separate frozen models; hold both preparation batches and recipient donors out as appropriate; clinical correlates remain a distinct validation layer.
Primary estimand: held-out-preparation prediction of tubular transport recovery from preadministration potency components
Candidate primary outcome: blinded tubular transport recovery
Primary finalized: No
Mathematical model: type: external/nested predictive-validation precision with target prevalence/variance, independent-unit holdout coverage and model complexity; formulas: Gain = mean_test[L(Y,f_base(X)) - L(Y,f_new(X))], with both predictions frozen before test outcomes; bootstrap independent held-out units.
For a binary target assess E[Y | p_hat=p] versus p and calibration intercept/slope logit Pr(Y=1)=a_cal+b_cal logit(p_hat); for a continuous target use Y=a_cal+b_cal y_hat+error plus scale-appropriate calibration/loss.; variables: Y: blinded tubular transport recovery; i: independent manufacturing preparation and recipient donor, crossed; t: organoid transport assay and preparation-marker measurement; the registered visit/time-window index, not another independent unit; A: authenticated preparation allocated to donor kidney models under randomized blinding; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: held-out-preparation prediction of tubular transport recovery from preadministration potency components; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; L: one proper or scientifically justified primary loss/scoring rule fixed for the actual outcome type; positive Gain favors new model; f base,f new: locked baseline/new prediction functions using only information available at the registered prediction time; test: untouched person/family/lab/site/time set as specified per thesis; p hat,y hat: frozen probability or continuous prediction; a cal,b cal: calibration intercept and slope, estimated on evaluation data for assessment; using them to update the model requires another evaluation set; family code: LAB_PRED
Sample size design: family: external/nested predictive-validation precision with target prevalence/variance, independent-unit holdout coverage and model complexity; formula: Simulate or bootstrap the independent-unit test loss/calibration/decision criterion; plan cases/events, noncases and independent sites/families/programs separately. No universal events-per-variable or participant N is asserted.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Release markers versus mechanism markers require separate frozen models; hold both preparation batches and recipient donors out as appropriate; clinical correlates remain a distinct validation layer.
Train/tune/test separation at the actual biological/participant/family/time level; no future, postoutcome or target-derived predictors.
Validation support must cover intended deployment; calibration and missing/failing measurements are reported with discrimination/loss.
A validation subset selected by culture success, assay viability, consent or data completeness is a different target unless the full strategy is explicitly evaluated.
Hold manufacturing preparations/batches together across all aliquots/recipients/time points; time density cannot substitute for independent batches.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: No external gain over identity markers, or inconsistent preparation rankings, would reject the proposed potency panel.; preanalysis gate: Release markers versus mechanism markers require separate frozen models; hold both preparation batches and recipient donors out as appropriate; clinical correlates remain a distinct validation layer.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Release markers versus mechanism markers require separate frozen models; hold both preparation batches and recipient donors out as appropriate; clinical correlates remain a distinct validation layer.
Repeat loss/calibration with independent-cluster resampling, prespecified subgroup/transport shifts and plausible missingness.
Compare fixed simple baselines, calibration and uncertainty coverage; report abstention/failure/turnaround rather than deleting hard cases.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Frozen prediction time/horizon, proper loss, baseline model, feature availability, independent split/holdout and target deployment population.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Release markers versus mechanism markers require separate frozen models; hold both preparation batches and recipient donors out as appropriate; clinical correlates remain a distinct validation layer.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/159; question: Which pre-administration stromal-cell properties predict kidney-repair activity rather than merely satisfying conventional cell-identity markers?; aims: Compare preparation-level properties with repair in diabetic kidney tissue models.
Develop a compact potency assay linked to tubular and vascular function.
Validate blinded preparation rankings against independent models and available clinical trial correlates.; design verbatim planning only: Population model: Authenticated clinical-development stromal-cell preparations and diabetic human kidney organoid–vascular models. Comparator: Conventional release markers, inactive reference preparations, and vehicle. Causal replication: Randomized model allocation, blinded rankings, orthogonal function assays, and external preparation replication. Sample planning: Plan 20–30 preparations and 12–18 tissue donors; clinical sample access is conditional.; outcomes verbatim: Primary: blinded prediction of tubular transport recovery. Secondary: barrier function, fibrosis, preparation consistency, and concordance with clinical correlates.; analysis verbatim inherited not authoritative if refined: Account for donor metabolic status, passage, storage, and model maturity; prespecify potency components, correct assay-selection multiplicity, and block manufacturing and culture batches.; falsification verbatim: No external gain over identity markers, or inconsistent preparation rankings, would reject the proposed potency panel.; source ids unverified leads: msc-diabetic-kidney-rct-2026
fda-unapproved-products-warning
reprogramming-genomic-tech-2026; prospectus precedence: No

L33 — Extracellular-vesicle repair activity beyond particle counts

Rank: 91
Cluster: Regenerative biology/cellular therapies/tissue repair
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: LAB; allocation or exposure unit: recipient donor-model unit randomized to particle-matched/fraction/inactive/vehicle condition; independent sampling unit: independent manufacturing preparation and recipient donor; repeated measurement unit: transport/barrier/cargo assay replicate; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: controlled in-vitro/ex-vivo allocation contrast when culture allocation is randomized; no direct human therapeutic-efficacy claim; specific constraint: Particle counts and cargo are not independent biological replication; necessity controls need purity validation and soluble carryover sensitivity.
Primary estimand: specified vesicle/fraction condition effect on epithelial transport recovery
Candidate primary outcome: prespecified epithelial transport-recovery readout
Primary finalized: No
Mathematical model: type: multilevel donor-by-preparation/condition experimental contrast precision; independent-donor and independent-batch allocation simulation; formulas: Y_djbrt = alpha + z(A_djb)^T theta + gamma^T X_d + u_d + v_b + q_dj + epsilon_djbrt.
Delta_model = c^T theta standardized to sampled donors/conditions; r is a technical repeat, not another donor or manufacturing batch.; variables: Y: prespecified epithelial transport-recovery readout; i: independent manufacturing preparation and recipient donor; t: transport/barrier/cargo assay replicate; the registered visit/time-window index, not another independent unit; A: recipient donor-model unit randomized to particle-matched/fraction/inactive/vehicle condition; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: specified vesicle/fraction condition effect on epithelial transport recovery; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; d: independent tissue/cell donor; j: independently allocated culture/explant/construct unit within donor; b: independent preparation/manufacturing/differentiation batch as appropriate; r: technical assay repeat or cell observation within allocated culture; z(A): indicators for the named controlled preparation/state/reference conditions; u d,v b,q dj: donor, batch and allocated-culture effects; crossed donor/preparation structure must follow actual allocation; X d: pretreatment donor/model attributes and baseline function only; family code: LAB
Sample size design: family: multilevel donor-by-preparation/condition experimental contrast precision; independent-donor and independent-batch allocation simulation; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Particle counts and cargo are not independent biological replication; necessity controls need purity validation and soluble carryover sensitivity.
Randomize independently allocated cultures/explants within blocking factors; donor and preparation/batch are the biological replication units for transport.
Technical replicates/cells/fields reduce measurement error but do not increase independent donor count.
Culture viability, identity loss, assay failure and lineage selection are outcomes/selection processes; do not silently condition treatment comparisons on successful posttreatment models.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Equal repair at matched particle counts despite candidate-feature differences, or soluble-only activity, would reject vesicle-feature causality.; preanalysis gate: Particle counts and cargo are not independent biological replication; necessity controls need purity validation and soluble carryover sensitivity.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Particle counts and cargo are not independent biological replication; necessity controls need purity validation and soluble carryover sensitivity.
Leave donors and manufacturing batches out; compare crossed/nested covariance structures supported by the design.
Report viable/failed model coverage, censored assays, donor-context heterogeneity and robustness to contamination/handling controls.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Confirmed donor/preparation access and allocated culture design; validated functional assay, dynamic range, technical error, detection limits and identity/viability gates.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Particle counts and cargo are not independent biological replication; necessity controls need purity validation and soluble carryover sensitivity.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/90; question: Which extracellular-vesicle features are necessary for repair, and can they outperform particle number as a potency measure?; aims: Resolve vesicle-associated and soluble contributions to kidney epithelial repair.
Test necessity of candidate feature combinations using orthogonal loss-of-function comparisons.
Validate a feature-based potency metric across independently produced preparations.; design verbatim planning only: Population model: Authenticated stromal-cell vesicle preparations and donor-derived kidney epithelial injury models. Comparator: Particle-matched preparations, soluble fractions, inactive references, and vehicle. Causal replication: Randomized blinded exposures, orthogonal fraction characterization, necessity tests, and independent manufacturing batches. Sample planning: Plan 24 preparations and 15 donors; pilot fraction purity gates candidate selection.; outcomes verbatim: Primary: epithelial transport recovery. Secondary: barrier integrity, cargo retention, preparation identity, and potency reproducibility.; analysis verbatim inherited not authoritative if refined: Control producer-cell state, storage, soluble carryover, and recipient donor; preregister feature families, correct cargo multiplicity, and separate manufacturing from assay batch effects.; falsification verbatim: Equal repair at matched particle counts despite candidate-feature differences, or soluble-only activity, would reject vesicle-feature causality.; source ids unverified leads: extracellular-vesicle-therapy-meta-2024
msc-diabetic-kidney-rct-2026
fda-unapproved-products-warning; prospectus precedence: No

L34 — Aged extracellular matrix as a constraint on regenerative cell function

Rank: 51
Cluster: Regenerative biology/cellular therapies/tissue repair
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: LAB_FACTOR; allocation or exposure unit: model unit allocated to reciprocal cell-age/matrix-age combinations; independent sampling unit: independent cell donor and matrix donor, crossed; repeated measurement unit: mechanical/repair assay and follow-up; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: controlled in-vitro/ex-vivo allocation contrast when culture allocation is randomized; no direct human therapeutic-efficacy claim; specific constraint: Age group is a donor attribute, not randomized; randomized pairing identifies a model-context contrast conditional on sampled donors, not a population causal effect of aging.
Primary estimand: cell-age-by-matrix-age interaction in restored tissue mechanics under the fixed allocation scheme
Candidate primary outcome: prespecified tissue mechanical-restoration endpoint
Primary finalized: No
Mathematical model: type: multilevel donor-by-preparation/condition experimental contrast precision; independent-donor and independent-batch allocation simulation; formulas: Y_djbrt = alpha + z(A_djb)^T theta + gamma^T X_d + u_d + v_b + q_dj + epsilon_djbrt.
Delta_model = c^T theta standardized to sampled donors/conditions; r is a technical repeat, not another donor or manufacturing batch.
For two controlled factors B,C, add beta_B B + beta_C C + beta_BC B*C; estimate interaction as mu_11-mu_10-mu_01+mu_00 on the chosen additive scale. Donor age/disease attributes themselves are not randomized.; variables: Y: prespecified tissue mechanical-restoration endpoint; i: independent cell donor and matrix donor, crossed; t: mechanical/repair assay and follow-up; the registered visit/time-window index, not another independent unit; A: model unit allocated to reciprocal cell-age/matrix-age combinations; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: cell-age-by-matrix-age interaction in restored tissue mechanics under the fixed allocation scheme; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; d: independent tissue/cell donor; j: independently allocated culture/explant/construct unit within donor; b: independent preparation/manufacturing/differentiation batch as appropriate; r: technical assay repeat or cell observation within allocated culture; z(A): indicators for the named controlled preparation/state/reference conditions; u d,v b,q dj: donor, batch and allocated-culture effects; crossed donor/preparation structure must follow actual allocation; X d: pretreatment donor/model attributes and baseline function only; B,C: the actual two condition/context factors named in this thesis; fix levels and identify which factors are randomized; beta BC: interaction parameter; if context is observational it is not a causal effect of changing donor age/disease; family code: LAB_FACTOR
Sample size design: family: multilevel donor-by-preparation/condition experimental contrast precision; independent-donor and independent-batch allocation simulation; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Age group is a donor attribute, not randomized; randomized pairing identifies a model-context contrast conditional on sampled donors, not a population causal effect of aging.
Randomize independently allocated cultures/explants within blocking factors; donor and preparation/batch are the biological replication units for transport.
Technical replicates/cells/fields reduce measurement error but do not increase independent donor count.
Culture viability, identity loss, assay failure and lineage selection are outcomes/selection processes; do not silently condition treatment comparisons on successful posttreatment models.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Cell donor state fully determining function regardless of matrix, or no durable benefit from matrix normalization, would reject environmental dominance.; preanalysis gate: Age group is a donor attribute, not randomized; randomized pairing identifies a model-context contrast conditional on sampled donors, not a population causal effect of aging.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Age group is a donor attribute, not randomized; randomized pairing identifies a model-context contrast conditional on sampled donors, not a population causal effect of aging.
Leave donors and manufacturing batches out; compare crossed/nested covariance structures supported by the design.
Report viable/failed model coverage, censored assays, donor-context heterogeneity and robustness to contamination/handling controls.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Confirmed donor/preparation access and allocated culture design; validated functional assay, dynamic range, technical error, detection limits and identity/viability gates.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Age group is a donor attribute, not randomized; randomized pairing identifies a model-context contrast conditional on sampled donors, not a population causal effect of aging.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/50; question: Does an aged tissue matrix prevent otherwise competent donor cells from restoring mechanical function?; aims: Map matrix features and cell behavior in paired older and younger tissue specimens.
Separate cell-intrinsic and matrix-extrinsic effects using reciprocal model comparisons.
Test whether matrix normalization restores durable function across independent donors.; design verbatim planning only: Population model: Consented human dermal matrices with matched primary fibroblast organotypic models. Comparator: Reciprocal donor-age combinations, matrix-only controls, and unmodified tissue references. Causal replication: Factorial donor–matrix allocation, blinded mechanics, orthogonal matrix characterization, and independent replication. Sample planning: Plan 20 younger and 20 older donors; pilot matrix reproducibility sets pairing density.; outcomes verbatim: Primary: restored tissue mechanics. Secondary: barrier repair, cell identity, matrix turnover, and persistence after environmental withdrawal.; analysis verbatim inherited not authoritative if refined: Account for photoexposure, diabetes, anatomical site, and tissue processing; prespecify age-by-matrix interactions, correct feature multiplicity, and block donor and preparation batches.; falsification verbatim: Cell donor state fully determining function regardless of matrix, or no durable benefit from matrix normalization, would reject environmental dominance.; source ids unverified leads: skin-photoaging-systematic-review-2025
reprogramming-genomic-tech-2026
fda-unapproved-products-warning; prospectus precedence: No

L35 — Mechanical integration as the limiting endpoint for engineered tendon repair

Rank: 121
Cluster: Regenerative biology/cellular therapies/tissue repair
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: LAB_SURV; allocation or exposure unit: construct/explant unit randomized to maturation/control condition; independent sampling unit: independent tissue donor and fabrication batch; repeated measurement unit: repeated fatigue-loading cycle until interface failure; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: controlled in-vitro/ex-vivo allocation contrast when culture allocation is randomized; no direct human therapeutic-efficacy claim; specific constraint: Right-censored runouts, competing failure locations and construct size require explicit rules; choose survival versus fixed-cycle estimand before sample planning.
Primary estimand: fatigue-resistant interface load-transfer contrast, expressed as failure-time or fixed-cycle load endpoint
Candidate primary outcome: cycles to interface failure or retained interface load after fixed cycles, chosen before testing
Primary finalized: No
Mathematical model: type: donor/fabrication-clustered mechanical survival/runout precision or fixed-cycle-load contrast, chosen before confirmation; formulas: Y_djbrt = alpha + z(A_djb)^T theta + gamma^T X_d + u_d + v_b + q_dj + epsilon_djbrt.
Delta_model = c^T theta standardized to sampled donors/conditions; r is a technical repeat, not another donor or manufacturing batch.
If cycles-to-failure is selected: S_a(m)=Pr(M_a>m); compare restricted mean cycles integral_0^m_max S_a(m)dm, retaining right-censored runouts and competing failure locations explicitly.; variables: Y: cycles to interface failure or retained interface load after fixed cycles, chosen before testing; i: independent tissue donor and fabrication batch; t: repeated fatigue-loading cycle until interface failure; the registered visit/time-window index, not another independent unit; A: construct/explant unit randomized to maturation/control condition; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: fatigue-resistant interface load-transfer contrast, expressed as failure-time or fixed-cycle load endpoint; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; d: independent tissue/cell donor; j: independently allocated culture/explant/construct unit within donor; b: independent preparation/manufacturing/differentiation batch as appropriate; r: technical assay repeat or cell observation within allocated culture; z(A): indicators for the named controlled preparation/state/reference conditions; u d,v b,q dj: donor, batch and allocated-culture effects; crossed donor/preparation structure must follow actual allocation; X d: pretreatment donor/model attributes and baseline function only; M a: cycles to registered interface failure under condition a; m max: fixed maximum fatigue-test cycle budget; S a(m): interface-failure-free probability; competing failures need a registered definition/estimand; family code: LAB_SURV
Sample size design: family: donor/fabrication-clustered mechanical survival/runout precision or fixed-cycle-load contrast, chosen before confirmation; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Right-censored runouts, competing failure locations and construct size require explicit rules; choose survival versus fixed-cycle estimand before sample planning.
Randomize independently allocated cultures/explants within blocking factors; donor and preparation/batch are the biological replication units for transport.
Technical replicates/cells/fields reduce measurement error but do not increase independent donor count.
Culture viability, identity loss, assay failure and lineage selection are outcomes/selection processes; do not silently condition treatment comparisons on successful posttreatment models.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Strong markers without integration confirm marker insufficiency; no predictive advantage of interface measures would reject the proposed maturation framework.; preanalysis gate: Right-censored runouts, competing failure locations and construct size require explicit rules; choose survival versus fixed-cycle estimand before sample planning.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Right-censored runouts, competing failure locations and construct size require explicit rules; choose survival versus fixed-cycle estimand before sample planning.
Leave donors and manufacturing batches out; compare crossed/nested covariance structures supported by the design.
Report viable/failed model coverage, censored assays, donor-context heterogeneity and robustness to contamination/handling controls.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Confirmed donor/preparation access and allocated culture design; validated functional assay, dynamic range, technical error, detection limits and identity/viability gates.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Right-censored runouts, competing failure locations and construct size require explicit rules; choose survival versus fixed-cycle estimand before sample planning.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/120; question: Can engineered tendon constructs achieve stable mechanical integration rather than only expressing tendon-associated molecular markers?; aims: Define reproducible integration and fatigue endpoints in donor-derived tendon constructs.
Compare maturation conditions for durable host-interface load transfer.
Validate predictive interface signatures across independent donors and injury-model contexts.; design verbatim planning only: Population model: Human tendon explants and engineered constructs in ex-vivo mechanical injury models. Comparator: Unintegrated constructs, matrix-only controls, and native repaired-tissue references. Causal replication: Randomized construct allocation, blinded fatigue testing, orthogonal imaging, and independent donor validation. Sample planning: Plan 20–25 donors and three maturation conditions; pilot failure variability guides replication.; outcomes verbatim: Primary: fatigue-resistant interface load transfer. Secondary: matrix alignment, cell survival, failure location, and recovery after repeated loading.; analysis verbatim inherited not authoritative if refined: Block donor pathology, age, construct size, and baseline mechanics; prespecify maturation contrasts, correct secondary multiplicity, and distribute conditions across fabrication batches.; falsification verbatim: Strong markers without integration confirm marker insufficiency; no predictive advantage of interface measures would reject the proposed maturation framework.; source ids unverified leads: collagen-tendon-rct-2025
collagen-umbrella-review-2026
fda-unapproved-products-warning; prospectus precedence: No

L36 — Preparation-dependent platelet-rich plasma activity in hair-follicle repair

Rank: 192
Cluster: Regenerative biology/cellular therapies/tissue repair
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: LAB_TRIAL; allocation or exposure unit: follicle culture randomized to characterized PRP/plasma/reference; original trial allocation separate; independent sampling unit: preparation donor and follicle donor, crossed; independent clinical trial for transport; repeated measurement unit: follicle growth/viability time point; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: controlled in-vitro/ex-vivo allocation contrast when culture allocation is randomized; no direct human therapeutic-efficacy claim; specific constraint: Many follicles from one scalp are not independent donors; preparation-feature prediction must hold donors out; hair-density clinical concordance does not establish clinical efficacy alone.
Primary estimand: controlled preparation effect on reproducible follicle growth with viability; separate trial-linked validation
Candidate primary outcome: follicle growth under a prespecified viability co-gate
Primary finalized: No
Mathematical model: type: multilevel donor-by-preparation/condition experimental contrast precision; independent-donor and independent-batch allocation simulation; formulas: Y_djbrt = alpha + z(A_djb)^T theta + gamma^T X_d + u_d + v_b + q_dj + epsilon_djbrt.
Delta_model = c^T theta standardized to sampled donors/conditions; r is a technical repeat, not another donor or manufacturing batch.
Define joint success J_dj=1{follicle growth meets registered criterion AND viability is preserved}; estimate randomized marginal joint-success contrasts and each component separately, or select a fully registered bivariate/co-primary rule before enrollment. Do not retain only successful posttreatment co-gates.; variables: Y: follicle growth under a prespecified viability co-gate; i: preparation donor and follicle donor, crossed; independent clinical trial for transport; t: follicle growth/viability time point; the registered visit/time-window index, not another independent unit; A: follicle culture randomized to characterized PRP/plasma/reference; original trial allocation separate; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: controlled preparation effect on reproducible follicle growth with viability; separate trial-linked validation; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; d: independent tissue/cell donor; j: independently allocated culture/explant/construct unit within donor; b: independent preparation/manufacturing/differentiation batch as appropriate; r: technical assay repeat or cell observation within allocated culture; z(A): indicators for the named controlled preparation/state/reference conditions; u d,v b,q dj: donor, batch and allocated-culture effects; crossed donor/preparation structure must follow actual allocation; X d: pretreatment donor/model attributes and baseline function only; J dj: per-thesis joint-success indicator with thresholds externally justified before outcome analysis; family code: LAB_TRIAL
Sample size design: family: multilevel donor-by-preparation/condition experimental contrast precision; independent-donor and independent-batch allocation simulation; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Many follicles from one scalp are not independent donors; preparation-feature prediction must hold donors out; hair-density clinical concordance does not establish clinical efficacy alone.
Randomize independently allocated cultures/explants within blocking factors; donor and preparation/batch are the biological replication units for transport.
Technical replicates/cells/fields reduce measurement error but do not increase independent donor count.
Culture viability, identity loss, assay failure and lineage selection are outcomes/selection processes; do not silently condition treatment comparisons on successful posttreatment models.
Original human-trial randomized effects and donor-model effects have separate units/outcomes; triangulation does not itself identify human mediation.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: No replicated signature beyond platelet count, or opposite laboratory and clinical associations, would reject the proposed explanatory panel.; preanalysis gate: Many follicles from one scalp are not independent donors; preparation-feature prediction must hold donors out; hair-density clinical concordance does not establish clinical efficacy alone.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Many follicles from one scalp are not independent donors; preparation-feature prediction must hold donors out; hair-density clinical concordance does not establish clinical efficacy alone.
Leave donors and manufacturing batches out; compare crossed/nested covariance structures supported by the design.
Report viable/failed model coverage, censored assays, donor-context heterogeneity and robustness to contamination/handling controls.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Confirmed donor/preparation access and allocated culture design; validated functional assay, dynamic range, technical error, detection limits and identity/viability gates.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Many follicles from one scalp are not independent donors; preparation-feature prediction must hold donors out; hair-density clinical concordance does not establish clinical efficacy alone.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/191; question: Can standardized preparation descriptors explain variable platelet-rich plasma effects on human hair follicles and clinical hair-density outcomes?; aims: Audit preparation reporting and recover participant-level data from eligible trials.
Link characterized preparations to growth and viability in human follicle cultures.
Validate a compact preparation signature against independent trial and laboratory samples.; design verbatim planning only: Population model: Consented platelet preparations, discarded scalp follicles, and archived alopecia-trial data. Comparator: Matched plasma controls, inactive references, and standard-care trial comparators. Causal replication: Within-donor randomized cultures, blinded follicle assessment, and independent trial validation. Sample planning: Plan 40 preparation donors, 20 follicle donors, and accessible trial datasets; donor availability governs scope.; outcomes verbatim: Primary: reproducible follicle growth and viability. Secondary: preparation-signature transport, hair-density correlation, and reporting-standard completeness.; analysis verbatim inherited not authoritative if refined: Account for sex, alopecia severity, medication, and follicle donor; correct preparation-feature multiplicity, avoid trial-selection leakage, and balance culture and assay batches.; falsification verbatim: No replicated signature beyond platelet count, or opposite laboratory and clinical associations, would reject the proposed explanatory panel.; source ids unverified leads: alopecia-prp-meta-2025
hair-prp-vs-minoxidil-meta-2026
fda-unapproved-products-warning; prospectus precedence: No

L37 — Clonal selection masquerading as conditioned-media rejuvenation

Rank: 52
Cluster: Regenerative biology/cellular therapies/tissue repair
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: CLONE; allocation or exposure unit: donor culture randomized to authenticated media/control conditions; independent sampling unit: independent donor and media preparation; baseline clone nested in donor; repeated measurement unit: longitudinal lineage/clone, cell and repair assay; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: controlled in-vitro/ex-vivo allocation contrast when culture allocation is randomized; no direct human therapeutic-efficacy claim; specific constraint: Clone disappearance and selective expansion are outcomes; analyzing only surviving clones biases within-clone recovery; report missing lineage resolution explicitly.
Primary estimand: within-baseline-clone durable repair contrast versus shifts in clone abundance
Candidate primary outcome: clone-specific repair/function trajectory and population abundance
Primary finalized: No
Mathematical model: type: multilevel donor-by-preparation/condition experimental contrast precision; independent-donor and independent-batch allocation simulation; formulas: Y_djbrt = alpha + z(A_djb)^T theta + gamma^T X_d + u_d + v_b + q_dj + epsilon_djbrt.
Delta_model = c^T theta standardized to sampled donors/conditions; r is a technical repeat, not another donor or manufacturing batch.
Population mean at t = sum_k p_kt mu_kt; its change decomposes into sum_k p_k0(mu_kt-mu_k0) + sum_k(p_kt-p_k0)mu_k0 + sum_k(p_kt-p_k0)(mu_kt-mu_k0), only for baseline clones with defined tracked outcomes.
Report clone loss/appearance and unidentified missing lineage terms; surviving-clone contrasts alone cannot estimate whole-population rejuvenation.; variables: Y: clone-specific repair/function trajectory and population abundance; i: independent donor and media preparation; baseline clone nested in donor; t: longitudinal lineage/clone, cell and repair assay; the registered visit/time-window index, not another independent unit; A: donor culture randomized to authenticated media/control conditions; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: within-baseline-clone durable repair contrast versus shifts in clone abundance; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; d: independent tissue/cell donor; j: independently allocated culture/explant/construct unit within donor; b: independent preparation/manufacturing/differentiation batch as appropriate; r: technical assay repeat or cell observation within allocated culture; z(A): indicators for the named controlled preparation/state/reference conditions; u d,v b,q dj: donor, batch and allocated-culture effects; crossed donor/preparation structure must follow actual allocation; X d: pretreatment donor/model attributes and baseline function only; k: baseline clone nested in donor; p kt: clone proportion at time t; mu kt: clone-specific functional mean; disappearance may make it undefined; family code: CLONE
Sample size design: family: multilevel donor-by-preparation/condition experimental contrast precision; independent-donor and independent-batch allocation simulation; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Clone disappearance and selective expansion are outcomes; analyzing only surviving clones biases within-clone recovery; report missing lineage resolution explicitly.
Randomize independently allocated cultures/explants within blocking factors; donor and preparation/batch are the biological replication units for transport.
Technical replicates/cells/fields reduce measurement error but do not increase independent donor count.
Culture viability, identity loss, assay failure and lineage selection are outcomes/selection processes; do not silently condition treatment comparisons on successful posttreatment models.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Uniform within-clone recovery without altered diversity would reject clonal selection as the principal explanation.; preanalysis gate: Clone disappearance and selective expansion are outcomes; analyzing only surviving clones biases within-clone recovery; report missing lineage resolution explicitly.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Clone disappearance and selective expansion are outcomes; analyzing only surviving clones biases within-clone recovery; report missing lineage resolution explicitly.
Leave donors and manufacturing batches out; compare crossed/nested covariance structures supported by the design.
Report viable/failed model coverage, censored assays, donor-context heterogeneity and robustness to contamination/handling controls.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Confirmed donor/preparation access and allocated culture design; validated functional assay, dynamic range, technical error, detection limits and identity/viability gates.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Clone disappearance and selective expansion are outcomes; analyzing only surviving clones biases within-clone recovery; report missing lineage resolution explicitly.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/51; question: Do apparent repair and biological-age improvements after conditioned-media exposure reflect broad cellular recovery or selective outgrowth of preexisting clones?; aims: Map clonal diversity and function before and after conditioned-media exposure.
Distinguish within-clone recovery from changes in clone abundance.
Validate findings across independent media preparations and older-donor skin models.; design verbatim planning only: Population model: Consented older-donor primary cells and organotypic skin exposed to authenticated institutional conditioned media. Comparator: Matched unconditioned media, proliferation-matched controls, and vehicle. Causal replication: Randomized culture allocation, longitudinal lineage observation, blinded function, and independent donor replication. Sample planning: Plan 20 donors and 12 media preparations; pilot lineage resolution gates expansion.; outcomes verbatim: Primary: within-clone durable repair versus population replacement. Secondary: clonal diversity, barrier function, age scores, and abnormal growth.; analysis verbatim inherited not authoritative if refined: Account for starting clone abundance, donor disease, passage, and media composition; preregister clone-level contrasts, correct molecular multiplicity, and block sequencing and culture batches.; falsification verbatim: Uniform within-clone recovery without altered diversity would reject clonal selection as the principal explanation.; source ids unverified leads: multimodal-longevity-pilot-2026
reprogramming-genomic-tech-2026
epigenetic-interventions-review-2026; prospectus precedence: No

L38 — Functional mitochondrial transfer versus misleading donor-material uptake

Rank: 92
Cluster: Regenerative biology/cellular therapies/tissue repair
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: LAB; allocation or exposure unit: recipient coculture unit randomized to transfer-capable and matched multiarm control conditions; independent sampling unit: stromal donor and recipient donor, crossed when reused; disjoint donor pairs only are independent pairs; repeated measurement unit: energetic, transport and material-tracing replicate; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: controlled in-vitro/ex-vivo allocation contrast when culture allocation is randomized; no direct human therapeutic-efficacy claim; specific constraint: No two-factor factorial design is assumed; label carryover, fusion and donor-cell admixture can mimic transfer; identity controls co-gate interpretation; uptake alone is not functional rescue.
Primary estimand: controlled rescue contrast in persistent recipient energetic recovery
Candidate primary outcome: recipient energetic-recovery endpoint after the fixed persistence window
Primary finalized: No
Mathematical model: type: multilevel donor-by-preparation/condition experimental contrast precision; independent-donor and independent-batch allocation simulation; formulas: Y_djbrt = alpha + z(A_djb)^T theta + gamma^T X_d + u_d + v_b + q_dj + epsilon_djbrt.
Delta_model = c^T theta standardized to sampled donors/conditions; r is a technical repeat, not another donor or manufacturing batch.
If either donor is reused, include crossed stromal-donor and recipient-donor effects rather than treating all selected pairs as independent; predeclare rescue-versus-control contrasts.; variables: Y: recipient energetic-recovery endpoint after the fixed persistence window; i: stromal donor and recipient donor, crossed when reused; disjoint donor pairs only are independent pairs; t: energetic, transport and material-tracing replicate; the registered visit/time-window index, not another independent unit; A: recipient coculture unit randomized to transfer-capable and matched multiarm control conditions; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: controlled rescue contrast in persistent recipient energetic recovery; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; d: independent tissue/cell donor; j: independently allocated culture/explant/construct unit within donor; b: independent preparation/manufacturing/differentiation batch as appropriate; r: technical assay repeat or cell observation within allocated culture; z(A): indicators for the named controlled preparation/state/reference conditions; u d,v b,q dj: donor, batch and allocated-culture effects; crossed donor/preparation structure must follow actual allocation; X d: pretreatment donor/model attributes and baseline function only; u stromal,u recipient: crossed donor effects for reused donor pairings; family code: LAB
Sample size design: family: multilevel donor-by-preparation/condition experimental contrast precision; independent-donor and independent-batch allocation simulation; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: No two-factor factorial design is assumed; label carryover, fusion and donor-cell admixture can mimic transfer; identity controls co-gate interpretation; uptake alone is not functional rescue.
Randomize independently allocated cultures/explants within blocking factors; donor and preparation/batch are the biological replication units for transport.
Technical replicates/cells/fields reduce measurement error but do not increase independent donor count.
Culture viability, identity loss, assay failure and lineage selection are outcomes/selection processes; do not silently condition treatment comparisons on successful posttreatment models.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Rescue persisting without competent mitochondrial transfer, or uptake without improved energetics, would reject transfer as the necessary mechanism.; preanalysis gate: No two-factor factorial design is assumed; label carryover, fusion and donor-cell admixture can mimic transfer; identity controls co-gate interpretation; uptake alone is not functional rescue.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: No two-factor factorial design is assumed; label carryover, fusion and donor-cell admixture can mimic transfer; identity controls co-gate interpretation; uptake alone is not functional rescue.
Leave donors and manufacturing batches out; compare crossed/nested covariance structures supported by the design.
Report viable/failed model coverage, censored assays, donor-context heterogeneity and robustness to contamination/handling controls.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Confirmed donor/preparation access and allocated culture design; validated functional assay, dynamic range, technical error, detection limits and identity/viability gates.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
No two-factor factorial design is assumed; label carryover, fusion and donor-cell admixture can mimic transfer; identity controls co-gate interpretation; uptake alone is not functional rescue.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/91; question: Does stromal-cell-associated mitochondrial transfer genuinely restore recipient-cell energetics or merely create a misleading donor-material signal?; aims: Distinguish intact functional transfer from debris uptake in injured kidney-cell models.
Test necessity of donor mitochondrial competence for recipient recovery.
Validate durable rescue across donor pairs and independent measurement modalities.; design verbatim planning only: Population model: Authenticated stromal cells and donor-derived injured kidney epithelial models under institutional oversight. Comparator: Transfer-incompetent controls, nonfunctional donor material, conditioned medium, and vehicle. Causal replication: Randomized cocultures, orthogonal material tracing, blinded energetic assays, and independent donor-pair replication. Sample planning: Plan 15 stromal and 15 kidney donors with selected pairings; assay specificity gates scale.; outcomes verbatim: Primary: persistent recipient energetic recovery. Secondary: epithelial transport, transferred-material persistence, genomic identity, and proliferation.; analysis verbatim inherited not authoritative if refined: Control donor viability, injury severity, cell fusion, and label carryover; prespecify rescue contrasts, correct modality multiplicity, and block donor pairs across imaging batches.; falsification verbatim: Rescue persisting without competent mitochondrial transfer, or uptake without improved energetics, would reject transfer as the necessary mechanism.; source ids unverified leads: msc-diabetic-kidney-rct-2026
exercise-mimetics-review-2026
fda-unapproved-products-warning; prospectus precedence: No

L39 — Perfusion and maturation gates for functional kidney organoids

Rank: 76
Cluster: Regenerative biology/cellular therapies/tissue repair
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: LAB_FACTOR; allocation or exposure unit: organoid platform unit randomized to perfusion and maturation factors; independent sampling unit: donor-derived line with independent differentiation and device batches; repeated measurement unit: transport/exchange assay and time point; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: controlled in-vitro/ex-vivo allocation contrast when culture allocation is randomized; no direct human therapeutic-efficacy claim; specific constraint: Differentiation batch and device batch are separate; cell-type markers cannot replace function; three batches per line are not three donors.
Primary estimand: perfusion-by-maturation interaction in sustained transport and exchange
Candidate primary outcome: prespecified sustained kidney-organoid transport/exchange endpoint
Primary finalized: No
Mathematical model: type: multilevel donor-by-preparation/condition experimental contrast precision; independent-donor and independent-batch allocation simulation; formulas: Y_djbrt = alpha + z(A_djb)^T theta + gamma^T X_d + u_d + v_b + q_dj + epsilon_djbrt.
Delta_model = c^T theta standardized to sampled donors/conditions; r is a technical repeat, not another donor or manufacturing batch.
For two controlled factors B,C, add beta_B B + beta_C C + beta_BC B*C; estimate interaction as mu_11-mu_10-mu_01+mu_00 on the chosen additive scale. Donor age/disease attributes themselves are not randomized.; variables: Y: prespecified sustained kidney-organoid transport/exchange endpoint; i: donor-derived line with independent differentiation and device batches; t: transport/exchange assay and time point; the registered visit/time-window index, not another independent unit; A: organoid platform unit randomized to perfusion and maturation factors; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: perfusion-by-maturation interaction in sustained transport and exchange; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; d: independent tissue/cell donor; j: independently allocated culture/explant/construct unit within donor; b: independent preparation/manufacturing/differentiation batch as appropriate; r: technical assay repeat or cell observation within allocated culture; z(A): indicators for the named controlled preparation/state/reference conditions; u d,v b,q dj: donor, batch and allocated-culture effects; crossed donor/preparation structure must follow actual allocation; X d: pretreatment donor/model attributes and baseline function only; B,C: the actual two condition/context factors named in this thesis; fix levels and identify which factors are randomized; beta BC: interaction parameter; if context is observational it is not a causal effect of changing donor age/disease; family code: LAB_FACTOR
Sample size design: family: multilevel donor-by-preparation/condition experimental contrast precision; independent-donor and independent-batch allocation simulation; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Differentiation batch and device batch are separate; cell-type markers cannot replace function; three batches per line are not three donors.
Randomize independently allocated cultures/explants within blocking factors; donor and preparation/batch are the biological replication units for transport.
Technical replicates/cells/fields reduce measurement error but do not increase independent donor count.
Culture viability, identity loss, assay failure and lineage selection are outcomes/selection processes; do not silently condition treatment comparisons on successful posttreatment models.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: No functional advantage beyond marker abundance, or unstable qualification across lines, would reject the proposed threshold framework.; preanalysis gate: Differentiation batch and device batch are separate; cell-type markers cannot replace function; three batches per line are not three donors.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Differentiation batch and device batch are separate; cell-type markers cannot replace function; three batches per line are not three donors.
Leave donors and manufacturing batches out; compare crossed/nested covariance structures supported by the design.
Report viable/failed model coverage, censored assays, donor-context heterogeneity and robustness to contamination/handling controls.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Confirmed donor/preparation access and allocated culture design; validated functional assay, dynamic range, technical error, detection limits and identity/viability gates.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Differentiation batch and device batch are separate; cell-type markers cannot replace function; three batches per line are not three donors.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/75; question: Which maturation and perfusion measures predict durable kidney-organoid function better than cell-type markers alone?; aims: Benchmark transport and exchange functions against primary human kidney references.
Separate perfusion and maturation contributions using factorial model comparisons.
Validate qualification thresholds by predicting responses to standardized nonclinical injury and recovery.; design verbatim planning only: Population model: Established human kidney organoids, vascular coculture platforms, and primary-tissue reference specimens. Comparator: Static cultures, perfusion-only controls, maturation-only controls, and primary references. Causal replication: Randomized platform allocation, blinded transport assays, orthogonal imaging, and independent cell-line replication. Sample planning: Plan 12 donor-derived lines and three differentiation batches each; pilot platform reliability limits expansion.; outcomes verbatim: Primary: sustained transport and exchange function. Secondary: recovery after injury, cell maturity, fibrosis, and candidate-therapy ranking stability.; analysis verbatim inherited not authoritative if refined: Block donor, differentiation efficiency, size, and passage; prespecify factorial interactions, control functional-endpoint multiplicity, and explicitly model differentiation and device batches.; falsification verbatim: No functional advantage beyond marker abundance, or unstable qualification across lines, would reject the proposed threshold framework.; source ids unverified leads: reprogramming-genomic-tech-2026
msc-diabetic-kidney-rct-2026
fda-unapproved-products-warning; prospectus precedence: No

L40 — Preserving donor age and disease in regenerative therapy test beds

Rank: 78
Cluster: Regenerative biology/cellular therapies/tissue repair
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: MODEL_AGREE; allocation or exposure unit: matched donor material processed through existing model routes; independent sampling unit: independent donor with model-generation failure retained; repeated measurement unit: route, preparation response and primary-tissue reference; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: prediction, calibration or model transportability; prediction gain is not clinical efficacy, mechanistic necessity or causal treatment selection; specific constraint: Generation success is selected; report all attempted donors/routes and bounds for failure; within-donor matching does not guarantee exchangeable route construction.
Primary estimand: route-specific concordance of repair-response rankings with independent primary tissue
Candidate primary outcome: repair-response contrast under each model route and primary-tissue reference
Primary finalized: No
Mathematical model: type: external/nested predictive-validation precision with target prevalence/variance, independent-unit holdout coverage and model complexity; formulas: Gain = mean_test[L(Y,f_base(X)) - L(Y,f_new(X))], with both predictions frozen before test outcomes; bootstrap independent held-out units.
For a binary target assess E[Y | p_hat=p] versus p and calibration intercept/slope logit Pr(Y=1)=a_cal+b_cal logit(p_hat); for a continuous target use Y=a_cal+b_cal y_hat+error plus scale-appropriate calibration/loss.
Within context k, compute ranking R_k of the fixed condition-response vector; report rank concordance across held-out contexts with uncertainty propagated from independent donors/materials, not just point-rank correlations.; variables: Y: repair-response contrast under each model route and primary-tissue reference; i: independent donor with model-generation failure retained; t: route, preparation response and primary-tissue reference; the registered visit/time-window index, not another independent unit; A: matched donor material processed through existing model routes; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: route-specific concordance of repair-response rankings with independent primary tissue; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; L: one proper or scientifically justified primary loss/scoring rule fixed for the actual outcome type; positive Gain favors new model; f base,f new: locked baseline/new prediction functions using only information available at the registered prediction time; test: untouched person/family/lab/site/time set as specified per thesis; p hat,y hat: frozen probability or continuous prediction; a cal,b cal: calibration intercept and slope, estimated on evaluation data for assessment; using them to update the model requires another evaluation set; k: donor/tissue/laboratory/model-route context; R k: ranking of fixed reference systems/preparations under one independently specified functional readout; family code: MODEL_AGREE
Sample size design: family: external/nested predictive-validation precision with target prevalence/variance, independent-unit holdout coverage and model complexity; formula: Simulate or bootstrap the independent-unit test loss/calibration/decision criterion; plan cases/events, noncases and independent sites/families/programs separately. No universal events-per-variable or participant N is asserted.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Generation success is selected; report all attempted donors/routes and bounds for failure; within-donor matching does not guarantee exchangeable route construction.
Train/tune/test separation at the actual biological/participant/family/time level; no future, postoutcome or target-derived predictors.
Validation support must cover intended deployment; calibration and missing/failing measurements are reported with discrimination/loss.
A validation subset selected by culture success, assay viability, consent or data completeness is a different target unless the full strategy is explicitly evaluated.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Equivalent disease fidelity and treatment ranking across routes would reject the hypothesis that reprogramming obscures relevant recipient heterogeneity.; preanalysis gate: Generation success is selected; report all attempted donors/routes and bounds for failure; within-donor matching does not guarantee exchangeable route construction.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Generation success is selected; report all attempted donors/routes and bounds for failure; within-donor matching does not guarantee exchangeable route construction.
Repeat loss/calibration with independent-cluster resampling, prespecified subgroup/transport shifts and plausible missingness.
Compare fixed simple baselines, calibration and uncertainty coverage; report abstention/failure/turnaround rather than deleting hard cases.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Frozen prediction time/horizon, proper loss, baseline model, feature availability, independent split/holdout and target deployment population.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Generation success is selected; report all attempted donors/routes and bounds for failure; within-donor matching does not guarantee exchangeable route construction.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/77; question: Do reprogrammed patient-derived tissue models erase age and metabolic disease features needed to predict regenerative treatment response?; aims: Compare age, metabolic memory, and function across matched model-generation routes.
Test whether candidate stromal-cell preparations rank similarly across routes.
Validate model-specific predictions against independent primary-tissue repair measurements.; design verbatim planning only: Population model: Matched donor primary kidney cells, established direct-conversion models, and reprogrammed organoids spanning age and diabetes strata. Comparator: Within-donor model routes and independent primary-tissue references. Causal replication: Matched-donor design, blinded preparation ranking, prespecified route interactions, and external-donor replication. Sample planning: Plan 24–30 diverse donors; successful model generation and pilot reliability govern final coverage.; outcomes verbatim: Primary: concordance with primary-tissue repair response. Secondary: retained metabolic memory, functional heterogeneity, subgroup coverage, and model failure rates.; analysis verbatim inherited not authoritative if refined: Address donor disease severity, medications, sampling site, and selective model-generation success; correct route-by-treatment multiplicity and block donor comparisons across differentiation batches.; falsification verbatim: Equivalent disease fidelity and treatment ranking across routes would reject the hypothesis that reprogramming obscures relevant recipient heterogeneity.; source ids unverified leads: reprogramming-genomic-tech-2026
msc-diabetic-kidney-rct-2026
multiomics-healthy-prevention-2025; prospectus precedence: No

M01 — Circadian meal alignment under rotating work schedules

Rank: 13
Cluster: Nutrition/metabolism/meal timing
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT_LOG; allocation or exposure unit: participant randomized aligned meal-timing offer versus equally supported usual timing; independent sampling unit: participant with employer effects; repeated measurement unit: glucose-test and metabolic follow-up visit; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Matched meal provision is not proven equal consumption; postassignment intake, weight, phase and sleep are not primary confounders; six-month free-living effect is a separate strategy estimand.
Primary estimand: week-four assignment effect on log Matsuda index, reported as geometric-mean ratio
Candidate primary outcome: log oral-glucose-test-derived Matsuda insulin-sensitivity index
Primary finalized: No
Mathematical model: type: contrast precision/power by design-matrix or simulation planning; formulas: E(Y_it | A_i, X_i, Y_i0) = alpha_t + z(A_i)^T theta_t + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Delta = sum_t w_t c^T theta_t; estimate by baseline-adjusted marginal standardization, with unit/cluster-appropriate inference.
Y_i = log(Matsuda_i,week4); exp(c^T theta_week4) is the adjusted geometric-mean ratio, not an arithmetic insulin-sensitivity difference.; variables: Y: log oral-glucose-test-derived Matsuda insulin-sensitivity index; i: participant with employer effects; t: glucose-test and metabolic follow-up visit; the registered visit/time-window index, not another independent unit; A: participant randomized aligned meal-timing offer versus equally supported usual timing; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: week-four assignment effect on log Matsuda index, reported as geometric-mean ratio; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; z(A): prespecified indicators for the actual randomized arms; two arms reduce to one indicator; Matsuda i,week4: OGTT-derived index using an independently specified validated calculation; raw glucose/insulin units and sampling times must match it; family code: RCT_LOG
Sample size design: family: contrast precision/power by design-matrix or simulation planning; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Matched meal provision is not proven equal consumption; postassignment intake, weight, phase and sleep are not primary confounders; six-month free-living effect is a separate strategy estimand.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: No insulin-sensitivity advantage despite verified timing separation would reject the proposed timing mechanism; improved biomarkers without better functioning would limit clinical relevance.; preanalysis gate: Matched meal provision is not proven equal consumption; postassignment intake, weight, phase and sleep are not primary confounders; six-month free-living effect is a separate strategy estimand.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Matched meal provision is not proven equal consumption; postassignment intake, weight, phase and sleep are not primary confounders; six-month free-living effect is a separate strategy estimand.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Matched meal provision is not proven equal consumption; postassignment intake, weight, phase and sleep are not primary confounders; six-month free-living effect is a separate strategy estimand.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/12; question: Does a phase-informed meal-timing offer with matched food provision improve week-four insulin sensitivity compared with equally supported usual timing among rotating-shift workers?; aims: Map meals against activity and melatonin-estimated phase across rotations.
Randomize isocaloric timing alignment against equally supported unrestricted eating.
Test metabolic changes against six-month fatigue and work functioning.; design verbatim planning only: Randomize approximately 120 rotating-shift workers across two employers to aligned or unrestricted eating. Provide verified isocaloric meals for four weeks, then follow six months of free living; stratify shifts and include subset calorimetry.; outcomes verbatim: Primary: week-four log-transformed oral-glucose-test-derived Matsuda insulin-sensitivity index, adjusted for baseline. Secondary: six-month metabolic measures, actual energy intake and weight, sleep, fatigue, work function, adverse events and feasibility.; analysis verbatim inherited not authoritative if refined: Estimate the randomized assignment-package effect with baseline adjustment and prespecified employer/shift factors. Match food provision and support while measuring actual consumption; do not equate provision with isocaloric intake. Standardize assessment timing relative to circadian phase where feasible. Treat intake, weight and phase mediation as exploratory under explicit assumptions; six-month free-living ITT includes strategy-induced intake changes.; falsification verbatim: No insulin-sensitivity advantage despite verified timing separation would reject the proposed timing mechanism; improved biomarkers without better functioning would limit clinical relevance.; source ids unverified leads: tre-tread-rct-2026
metabolic-flexibility-t2d-meta-2025; prospectus precedence: Yes; prospectus file: thesis-prospectuses.json; prospectus design and estimand: Use an individually randomized parallel trial with allocation concealed and stratified by employer, shift pattern, and baseline metabolic risk. Both arms receive comparable food quantity and composition, dietitian contact, and logistics support; the comparator retains usual timing. The primary estimand is the assignment effect on the week-four log-transformed Matsuda insulin-sensitivity index, adjusted for baseline. The free-living estimand is the six-month effect of the offered strategy, including resulting changes in intake and weight. Randomization identifies the package effect, not each mechanism.; prospectus analysis plan: Estimate the primary baseline-adjusted assignment contrast with employer effects and report ratios and absolute descriptive changes. Model repeated functional outcomes separately with multiplicity controlled through a prespecified hierarchy. Describe timing separation and intake difference with uncertainty; lack of a significant calorie difference is not proof of equivalence. Do not adjust the main effect for post-randomization sleep, weight, or adherence. Mechanistic mediation and adherence analyses require explicit confounding assumptions and remain secondary. Account for phase-measurement error in sensitivity analyses.; precedence note: Expanded prospectus refinements govern this scaffold; verbatim earlier brief material is retained for traceability, not as a contradictory analysis instruction. F15 is explicitly two arms; F30 has no unjustified symptom noninferiority claim.

M02 — Does consumer glucose feedback improve food choices or intensify restriction?

Rank: 8
Cluster: Nutrition/metabolism/meal timing
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT3; allocation or exposure unit: participant randomized masked/visible/educated CGM feedback; independent sampling unit: participant; repeated measurement unit: blinded dietary recall and follow-up; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Choose primary comparison across three arms and fixed nine-month time summary; visible glucose excursions are not themselves a dietary-health endpoint.
Primary estimand: prespecified visible-feedback versus masked-sensing effect on sustained diet quality
Candidate primary outcome: validated diet-quality score from blinded repeated recalls
Primary finalized: No
Mathematical model: type: contrast precision/power by design-matrix or simulation planning; formulas: E(Y_it | A_i, X_i, Y_i0) = alpha_t + z(A_i)^T theta_t + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Delta = sum_t w_t c^T theta_t; estimate by baseline-adjusted marginal standardization, with unit/cluster-appropriate inference.; variables: Y: validated diet-quality score from blinded repeated recalls; i: participant; t: blinded dietary recall and follow-up; the registered visit/time-window index, not another independent unit; A: participant randomized masked/visible/educated CGM feedback; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: prespecified visible-feedback versus masked-sensing effect on sustained diet quality; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; z(A): prespecified indicators for the actual randomized arms; two arms reduce to one indicator; family code: RCT3
Sample size design: family: contrast precision/power by design-matrix or simulation planning; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Choose primary comparison across three arms and fixed nine-month time summary; visible glucose excursions are not themselves a dietary-health endpoint.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Primary contrast is fixed before observing outcomes; the third arm does not create another uncorrected primary test.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Failure of visible feedback to improve diet quality over masked sensing would leave feedback alone without demonstrated utility. Education plus feedback versus visible feedback tests the added education component; the combined arm versus masked sensing tests the bundle. Lower excursions with poorer diet would contradict an optimization claim.; preanalysis gate: Choose primary comparison across three arms and fixed nine-month time summary; visible glucose excursions are not themselves a dietary-health endpoint.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Choose primary comparison across three arms and fixed nine-month time summary; visible glucose excursions are not themselves a dietary-health endpoint.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Choose primary comparison across three arms and fixed nine-month time summary; visible glucose excursions are not themselves a dietary-health endpoint.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/7; question: Among adults without diabetes, does visible CGM feedback improve sustained diet quality and wellbeing compared with masked sensing?; aims: Compare perceived food quality, excursions and independently assessed dietary quality.
Randomize masked sensing, visible feedback and feedback with education.
Assess persistence after sensor removal and anxiety-mediated restriction.; design verbatim planning only: Randomize about 180 normoglycemic prospective CGM buyers to masked sensing, visible feedback or feedback with education. Match contact and follow participants for nine months, including six months without feedback; recruit across incomes and dietary traditions.; outcomes verbatim: Primary: validated diet-quality score assessed by blinded repeated recalls. Secondary: eating-related anxiety, nutrient adequacy, HbA1c, weight and quality of life.; analysis verbatim inherited not authoritative if refined: Prespecify anxiety-risk interactions and an outcome hierarchy. Address missing recalls, sensor artifacts and differential use; adjust precision for baseline diet and sensor literacy.; falsification verbatim: Failure of visible feedback to improve diet quality over masked sensing would leave feedback alone without demonstrated utility. Education plus feedback versus visible feedback tests the added education component; the combined arm versus masked sensing tests the bundle. Lower excursions with poorer diet would contradict an optimization claim.; source ids unverified leads: cgm-nondiabetic-review-2026
cgm-t2d-rct-2026
precision-nutrition-rct-2024; prospectus precedence: No

M03 — Fuel-switching tests as actionable prehabilitation markers

Rank: 46
Cluster: Nutrition/metabolism/meal timing
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RANK_INT; allocation or exposure unit: surgical candidate randomized enhanced prehabilitation versus usual care before scheduling outcomes; independent sampling unit: participant; repeated measurement unit: day-ninety vital/function status and subsequent surgery record; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Time starts at randomization, not completed surgery; cancelled/deferred surgery stays in primary denominator; operated-survivor recovery is secondary and selected.
Primary estimand: prehabilitation assignment effect on day-ninety joint survival/function ordering and baseline-flexibility interaction
Candidate primary outcome: day-ninety physical-function status with death below any living score
Primary finalized: No
Mathematical model: type: joint survival/function ordinal or probabilistic-index contrast and baseline-marker interaction simulation; formulas: Define Z^a as the day-ninety joint outcome with death below any living functional score; report F_a(z) and psi=Pr(Z_1>Z_0)+0.5Pr(Z_1=Z_0) for independent draws.
One candidate ordinal model is logit Pr(Z_i<=k | A_i,B_i,X_i)=alpha_k-tau A_i-eta B_i-kappa A_i B_i-gamma^T X_i; derive standardized distributions and assess proportional-odds sensitivity.; variables: Y: day-ninety physical-function status with death below any living score; i: participant; t: day-ninety vital/function status and subsequent surgery record; the registered visit/time-window index, not another independent unit; A: surgical candidate randomized enhanced prehabilitation versus usual care before scheduling outcomes; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: prehabilitation assignment effect on day-ninety joint survival/function ordering and baseline-flexibility interaction; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; Z^a: fixed day-ninety survival/function ordering from randomization, including cancelled/deferred surgery; Z 1,Z 0: independent draws from arm joint distributions; F a: assignment-specific joint distribution; psi: independent-draw probabilistic superiority; k: ordered outcome threshold, not a fabricated numerical death score; B i: baseline metabolic-flexibility marker measured before allocation; kappa: baseline-marker-by-assignment interaction on the declared ordinal scale; family code: RANK_INT
Sample size design: family: joint survival/function ordinal or probabilistic-index contrast and baseline-marker interaction simulation; formula: Use design-specific simulation or independent-unit resampling for the stated estimand and target confidence width/power; preserve all allocation, censoring, clustering and measurement-error structure.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Time starts at randomization, not completed surgery; cancelled/deferred surgery stays in primary denominator; operated-survivor recovery is secondary and selected.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Mortality reported separately and missing living scores distinguished from death; operated-survivor recovery remains conditional.
Ordinal proportional odds is a candidate model assumption, not a property guaranteed by the rank endpoint.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Poor repeatability, negligible incremental prediction, or absent treatment interaction would defeat the claim that flexibility testing guides care.; preanalysis gate: Time starts at randomization, not completed surgery; cancelled/deferred surgery stays in primary denominator; operated-survivor recovery is secondary and selected.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Time starts at randomization, not completed surgery; cancelled/deferred surgery stays in primary denominator; operated-survivor recovery is secondary and selected.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Time starts at randomization, not completed surgery; cancelled/deferred surgery stays in primary denominator; operated-survivor recovery is secondary and selected.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/45; question: Does measuring metabolic flexibility improve selection of prehabilitation for adults awaiting major elective surgery?; aims: Compare repeated meal-challenge flexibility with subset clamp reference measurements.
Assess incremental prediction of postoperative recovery beyond established variables.
Test prespecified flexibility-by-prehabilitation interaction through embedded randomization.; design verbatim planning only: Enroll approximately 220 surgical candidates at two hospitals; repeat physiology in 60. Embed randomization of enhanced prehabilitation versus usual care, collect baseline tests before allocation and follow recovery for ninety days.; outcomes verbatim: Primary: Day-90 physical-function status from randomization, with death ranked worse than any living functional score under a prespecified patient-informed ordering. Secondary: Patient-reported physical recovery thirty days after actual surgery, walking capacity, days at home, complications, test reliability and incremental prediction calibration.; analysis verbatim inherited not authoritative if refined: Use nested validation, shrinkage and calibration. Adjust baseline operation type, frailty, diabetes and scheduling; prespecify one flexibility-by-treatment interaction. Analyze the day-90 primary in all randomized participants, including cancelled or deferred surgery; report mortality and surgery occurrence separately. Treat postoperative recovery among operated survivors as a secondary conditional outcome, with explicit selection limits; assess missing living-participant scores separately from death.; falsification verbatim: Poor repeatability, negligible incremental prediction, or absent treatment interaction would defeat the claim that flexibility testing guides care.; source ids unverified leads: metabolic-flexibility-prehabilitation-2025
metabolic-flexibility-t2d-meta-2025
resistance-sarcopenia-meta-2026; prospectus precedence: No

M04 — Protein distribution versus protein sufficiency in later life

Rank: 60
Cluster: Nutrition/metabolism/meal timing
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT; allocation or exposure unit: participant randomized even versus dinner-heavy protein with common training; independent sampling unit: participant; repeated measurement unit: chair-rise and dietary assessment visit; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Intake matching must be checked against a justified equivalence criterion, not a nonsignificant difference; actual protein/training is postassignment.
Primary estimand: protein-distribution assignment effect on standardized chair-rise performance
Candidate primary outcome: standardized chair-rise endpoint at nine months
Primary finalized: No
Mathematical model: type: contrast precision/power by design-matrix or simulation planning; formulas: E(Y_it | A_i, X_i, Y_i0) = alpha_t + z(A_i)^T theta_t + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Delta = sum_t w_t c^T theta_t; estimate by baseline-adjusted marginal standardization, with unit/cluster-appropriate inference.; variables: Y: standardized chair-rise endpoint at nine months; i: participant; t: chair-rise and dietary assessment visit; the registered visit/time-window index, not another independent unit; A: participant randomized even versus dinner-heavy protein with common training; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: protein-distribution assignment effect on standardized chair-rise performance; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; z(A): prespecified indicators for the actual randomized arms; two arms reduce to one indicator; family code: RCT
Sample size design: family: contrast precision/power by design-matrix or simulation planning; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Intake matching must be checked against a justified equivalence criterion, not a nonsignificant difference; actual protein/training is postassignment.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Changed protein synthesis without superior chair-rise improvement would undermine translation from acute signaling to useful timing advice.; preanalysis gate: Intake matching must be checked against a justified equivalence criterion, not a nonsignificant difference; actual protein/training is postassignment.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Intake matching must be checked against a justified equivalence criterion, not a nonsignificant difference; actual protein/training is postassignment.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Intake matching must be checked against a justified equivalence criterion, not a nonsignificant difference; actual protein/training is postassignment.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/59; question: When total protein and exercise are held constant, does distributing protein across meals improve older adults’ everyday physical function?; aims: Characterize meal intake and subset anabolic responses.
Randomize protein distribution while matching total intake and training.
Assess sustained function, appetite burden and baseline-pattern effect modification.; design verbatim planning only: Randomize about 140 adults aged 65–80 with low strength to even or dinner-heavy protein distribution for nine months. Match total intake, food support and resistance training; measure stable-isotope synthesis in approximately 30 participants.; outcomes verbatim: Primary: standardized chair-rise performance. Secondary: walking speed, leg strength, lean mass, meal enjoyment and patient-reported activity limitation.; analysis verbatim inherited not authoritative if refined: Adjust baseline performance, prespecify meal-pattern interaction and verify total-intake equivalence. Record training attendance; control secondary multiplicity and keep lean mass separate from function.; falsification verbatim: Changed protein synthesis without superior chair-rise improvement would undermine translation from acute signaling to useful timing advice.; source ids unverified leads: protein-exercise-network-meta-2026
resistance-sarcopenia-meta-2026; prospectus precedence: No

M05 — Lipid discordance during plant-forward carbohydrate restriction

Rank: 72
Cluster: Nutrition/metabolism/meal timing
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT; allocation or exposure unit: participant randomized matched-carbohydrate food formulation; independent sampling unit: participant; repeated measurement unit: apolipoprotein and diet/medication visit; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Response assessment before randomization defines target population; ketones verify exposure, not cardiovascular benefit; medication and weight change are cointerventions/pathways.
Primary estimand: food-formulation assignment effect on apolipoprotein B at sixteen weeks
Candidate primary outcome: apolipoprotein B concentration
Primary finalized: No
Mathematical model: type: contrast precision/power by design-matrix or simulation planning; formulas: E(Y_it | A_i, X_i, Y_i0) = alpha_t + z(A_i)^T theta_t + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Delta = sum_t w_t c^T theta_t; estimate by baseline-adjusted marginal standardization, with unit/cluster-appropriate inference.; variables: Y: apolipoprotein B concentration; i: participant; t: apolipoprotein and diet/medication visit; the registered visit/time-window index, not another independent unit; A: participant randomized matched-carbohydrate food formulation; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: food-formulation assignment effect on apolipoprotein B at sixteen weeks; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; z(A): prespecified indicators for the actual randomized arms; two arms reduce to one indicator; family code: RCT
Sample size design: family: contrast precision/power by design-matrix or simulation planning; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Response assessment before randomization defines target population; ketones verify exposure, not cardiovascular benefit; medication and weight change are cointerventions/pathways.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Equivalent apolipoprotein B despite confirmed food substitution would reject the proposed lipid mechanism; glucose improvement alone cannot rescue that hypothesis.; preanalysis gate: Response assessment before randomization defines target population; ketones verify exposure, not cardiovascular benefit; medication and weight change are cointerventions/pathways.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Response assessment before randomization defines target population; ketones verify exposure, not cardiovascular benefit; medication and weight change are cointerventions/pathways.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Response assessment before randomization defines target population; ketones verify exposure, not cardiovascular benefit; medication and weight change are cointerventions/pathways.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/71; question: Can changes in food composition explain why some carbohydrate-restricted diets improve glucose measures while worsening atherogenic lipoproteins?; aims: Identify reproducible lipid responders during supervised dietary monitoring.
Randomize carbohydrate-matched formulations differing in fat and fiber sources.
Assess persistent lipid response, satiety, adequacy and adherence.; design verbatim planning only: Recruit approximately 150 carbohydrate-restricting adults with overweight. After monitored response assessment, randomize matched-carbohydrate formulations for sixteen weeks, then follow six-month maintenance. Initially maintain energy balance to separate composition from weight loss.; outcomes verbatim: Primary: apolipoprotein B change. Secondary: LDL particle measures, ambulatory glucose, blood pressure, gastrointestinal symptoms and dietary sustainability.; analysis verbatim inherited not authoritative if refined: Use baseline-adjusted intention-to-treat estimates and prespecified lipid strata. Track intake, weight and medications; ketones check exposure rather than establishing benefit.; falsification verbatim: Equivalent apolipoprotein B despite confirmed food substitution would reject the proposed lipid mechanism; glucose improvement alone cannot rescue that hypothesis.; source ids unverified leads: ketogenic-clinical-review-2026
cgm-nondiabetic-review-2026; prospectus precedence: No

M06 — Symptom-responsive fueling versus calendar-based cycle prescriptions

Rank: 42
Cluster: Nutrition/metabolism/meal timing
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT3; allocation or exposure unit: naturally cycling athlete randomized responsive/phase-calendar/standard nutrition; independent sampling unit: participant; repeated measurement unit: monthly symptom/training/cycle assessment; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Contraceptive cohort is descriptive and cannot be pooled as randomized; biochemical phase uncertainty and season require sensitivity checks.
Primary estimand: responsive versus phase-calendar assignment effect on low-energy-availability symptom burden
Candidate primary outcome: validated low-energy-availability symptom-burden score
Primary finalized: No
Mathematical model: type: contrast precision/power by design-matrix or simulation planning; formulas: E(Y_it | A_i, X_i, Y_i0) = alpha_t + z(A_i)^T theta_t + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Delta = sum_t w_t c^T theta_t; estimate by baseline-adjusted marginal standardization, with unit/cluster-appropriate inference.; variables: Y: validated low-energy-availability symptom-burden score; i: participant; t: monthly symptom/training/cycle assessment; the registered visit/time-window index, not another independent unit; A: naturally cycling athlete randomized responsive/phase-calendar/standard nutrition; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: responsive versus phase-calendar assignment effect on low-energy-availability symptom burden; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; z(A): prespecified indicators for the actual randomized arms; two arms reduce to one indicator; family code: RCT3
Sample size design: family: contrast precision/power by design-matrix or simulation planning; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Contraceptive cohort is descriptive and cannot be pooled as randomized; biochemical phase uncertainty and season require sensitivity checks.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Primary contrast is fixed before observing outcomes; the third arm does not create another uncorrected primary test.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: If responsive support adds no symptom or recovery benefit despite better intake matching, the hypothesized advantage over simpler advice is unsupported.; preanalysis gate: Contraceptive cohort is descriptive and cannot be pooled as randomized; biochemical phase uncertainty and season require sensitivity checks.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Contraceptive cohort is descriptive and cannot be pooled as randomized; biochemical phase uncertainty and season require sensitivity checks.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Contraceptive cohort is descriptive and cannot be pooled as randomized; biochemical phase uncertainty and season require sensitivity checks.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/41; question: Does adapting food availability to individual appetite and training demand prevent low energy availability better than fixed menstrual-phase rules?; aims: Measure individual appetite, verified cycles and training-load variation.
Compare responsive fueling, phase-calendar advice and standard nutrition.
Assess recovery and menstrual outcomes by baseline underfueling risk.; design verbatim planning only: Randomize approximately 120 naturally cycling endurance athletes to responsive fueling, phase-calendar advice or standard nutrition for six months. Add a descriptive contraceptive cohort; combine repeated interviews, training records and subset biochemical cycle verification.; outcomes verbatim: Primary: validated low-energy-availability symptom burden. Secondary: training completion, menstrual disturbance, injury-related lost days, fatigue and objective performance.; analysis verbatim inherited not authoritative if refined: Use participant-specific repeated-measure models. Prespecify underfueling interactions, account for illness and season, retain assignment-based comparisons and correct secondary multiplicity.; falsification verbatim: If responsive support adds no symptom or recovery benefit despite better intake matching, the hypothesized advantage over simpler advice is unsupported.; source ids unverified leads: menstrual-energy-intake-meta-2025
female-athlete-nutrition-review-2025; prospectus precedence: No

M07 — The incremental value of omics in personalized dietary advice

Rank: 25
Cluster: Nutrition/metabolism/meal timing
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT3; allocation or exposure unit: participant randomized conventional/omics-enhanced/generic personalized advice; independent sampling unit: participant with independent validation cohort; repeated measurement unit: fasting triglyceride and advice follow-up; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Lock advice algorithms and sequence contrasts; postrandomization recommendation concordance cannot prove omic mechanism or optimize subgroups.
Primary estimand: omics-enhanced versus conventional personalization assignment effect on fasting triglycerides
Candidate primary outcome: fasting triglyceride concentration under a prespecified skew-compatible scale
Primary finalized: No
Mathematical model: type: contrast precision/power by design-matrix or simulation planning; formulas: E(Y_it | A_i, X_i, Y_i0) = alpha_t + z(A_i)^T theta_t + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Delta = sum_t w_t c^T theta_t; estimate by baseline-adjusted marginal standardization, with unit/cluster-appropriate inference.; variables: Y: fasting triglyceride concentration under a prespecified skew-compatible scale; i: participant with independent validation cohort; t: fasting triglyceride and advice follow-up; the registered visit/time-window index, not another independent unit; A: participant randomized conventional/omics-enhanced/generic personalized advice; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: omics-enhanced versus conventional personalization assignment effect on fasting triglycerides; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; z(A): prespecified indicators for the actual randomized arms; two arms reduce to one indicator; family code: RCT3
Sample size design: family: contrast precision/power by design-matrix or simulation planning; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Lock advice algorithms and sequence contrasts; postrandomization recommendation concordance cannot prove omic mechanism or optimize subgroups.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Primary contrast is fixed before observing outcomes; the third arm does not create another uncorrected primary test.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: If omics rarely changes recommendations or fails to improve outcomes over conventional personalization, its added clinical value is not supported.; preanalysis gate: Lock advice algorithms and sequence contrasts; postrandomization recommendation concordance cannot prove omic mechanism or optimize subgroups.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Lock advice algorithms and sequence contrasts; postrandomization recommendation concordance cannot prove omic mechanism or optimize subgroups.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Lock advice algorithms and sequence contrasts; postrandomization recommendation concordance cannot prove omic mechanism or optimize subgroups.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/24; question: Do microbiome and postprandial data improve dietary outcomes beyond equally intensive personalization using preferences, budget and clinical history?; aims: Audit recommendation changes produced specifically by biological data.
Randomize conventional personalization, omics-enhanced personalization and matched generic advice.
Assess clinical utility, costs and transportability across dietary cultures.; design verbatim planning only: Randomize approximately 240 adults with elevated triglycerides to conventional personalization, omics-enhanced personalization or generic advice for nine months. Match contact, lock recommendation rules and laboratory pipelines, and retain an independent validation cohort.; outcomes verbatim: Primary: fasting triglycerides. Secondary: diet quality, blood pressure, wellbeing, recommendation concordance, out-of-pocket cost and adherence.; analysis verbatim inherited not authoritative if refined: Use ordered intention-to-treat contrasts and locked subgroups. Adjust baseline triglycerides and medications; test transportability rather than exploring unrestricted omic interactions.; falsification verbatim: If omics rarely changes recommendations or fails to improve outcomes over conventional personalization, its added clinical value is not supported.; source ids unverified leads: precision-nutrition-rct-2024
microbiome-fiber-prediabetes-rct-2025
multiomics-healthy-prevention-2025; prospectus precedence: No

M08 — Does blood-cell autophagic flux track useful adaptation to fasting?

Rank: 118
Cluster: Nutrition/metabolism/meal timing
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT_MECHANISM; allocation or exposure unit: participant randomized energy-matched eating schedule; predeclared assay subset selected independently of outcomes; independent sampling unit: participant for biological contrast; independent specimen/batch for reliability; repeated measurement unit: cell-specific flux time point and technical replicate; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Define assay-subset target and viability failures; cell mixture is a mechanism/measurement issue; do not adjust away treatment-induced weight or cell shifts without a distinct estimand.
Primary estimand: assignment effect and repeatability of one prespecified cell-specific autophagic-flux endpoint
Candidate primary outcome: validated cell-specific dynamic flux, not static marker abundance
Primary finalized: No
Mathematical model: type: contrast precision/power by design-matrix or simulation planning; formulas: E(Y_it | A_i, X_i, Y_i0) = alpha_t + z(A_i)^T theta_t + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Delta = sum_t w_t c^T theta_t; estimate by baseline-adjusted marginal standardization, with unit/cluster-appropriate inference.; variables: Y: validated cell-specific dynamic flux, not static marker abundance; i: participant for biological contrast; independent specimen/batch for reliability; t: cell-specific flux time point and technical replicate; the registered visit/time-window index, not another independent unit; A: participant randomized energy-matched eating schedule; predeclared assay subset selected independently of outcomes; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: assignment effect and repeatability of one prespecified cell-specific autophagic-flux endpoint; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; z(A): prespecified indicators for the actual randomized arms; two arms reduce to one indicator; family code: RCT_MECHANISM
Sample size design: family: contrast precision/power by design-matrix or simulation planning; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Define assay-subset target and viability failures; cell mixture is a mechanism/measurement issue; do not adjust away treatment-induced weight or cell shifts without a distinct estimand.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Effects disappearing after cell-composition adjustment or failing repeat sampling would invalidate a bulk-blood autophagy interpretation; biomarker change without function limits benefit claims.; preanalysis gate: Define assay-subset target and viability failures; cell mixture is a mechanism/measurement issue; do not adjust away treatment-induced weight or cell shifts without a distinct estimand.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Define assay-subset target and viability failures; cell mixture is a mechanism/measurement issue; do not adjust away treatment-induced weight or cell shifts without a distinct estimand.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Define assay-subset target and viability failures; cell mixture is a mechanism/measurement issue; do not adjust away treatment-induced weight or cell shifts without a distinct estimand.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/117; question: Is a fasting-related change in peripheral blood-cell autophagic flux reproducible and associated with improved metabolic function rather than simple cell redistribution?; aims: Validate within-person flux reliability across timing and processing conditions.
Randomize energy-matched time restriction versus conventional eating.
Test temporal relationships between cell-specific flux and metabolic function.; design verbatim planning only: Randomize approximately 90 adults with overweight to energy-matched time restriction or conventional eating for sixteen weeks. Intensively phenotype 40; standardize sampling times and processing and measure immune-cell composition. Infection observations remain exploratory.; outcomes verbatim: Primary: reproducibility and intervention effect on prespecified cell-specific flux. Secondary: insulin sensitivity, fatigue, immune-cell composition and adverse nutritional symptoms.; analysis verbatim inherited not authoritative if refined: Use blinded batches, technical controls and variance components. Account for infection, weight and cell mixture; prespecify one flux endpoint and adjust mechanistic multiplicity.; falsification verbatim: Effects disappearing after cell-composition adjustment or failing repeat sampling would invalidate a bulk-blood autophagy interpretation; biomarker change without function limits benefit claims.; source ids unverified leads: itre-autophagy-rct-2025
fmd-autophagy-rct-2025; prospectus precedence: No

M09 — Do dietary aging-clock changes predict preserved physical resilience?

Rank: 177
Cluster: Nutrition/metabolism/meal timing
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: ASSOC_EIV; allocation or exposure unit: dietary restriction trial/cohort score changes observed; comparator only where available; independent sampling unit: participant within independent trial/cohort; repeated measurement unit: aging score and walking-recovery visit; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: association or prospective prediction; temporal order and model adjustment alone do not identify a mechanism or treatment effect; specific constraint: Parent-trial randomization does not randomize clock change; weight and lean-mass mediation assumptions cannot be replaced by routine covariate adjustment.
Primary estimand: error-adjusted association of fixed score change with walking recovery
Candidate primary outcome: walking-recovery performance after exertion alongside score change
Primary finalized: No
Mathematical model: type: longitudinal association/reliability or held-out prediction precision simulation, not causal-efficacy power; formulas: Y_it = alpha + u_i + beta_W (B_it-Bbar_i) + beta_B Bbar_i + gamma^T X_it + f_time(t) + b_study + epsilon_it; u_i is a random intercept with a declared conditional mean/covariance model, not unrestricted fixed person effects.
B_observed,it = B_true,it + e_it; estimate measurement variance from replicates/validation or report sensitivity rather than assume it known.; variables: Y: walking-recovery performance after exertion alongside score change; i: participant within independent trial/cohort; t: aging score and walking-recovery visit; the registered visit/time-window index, not another independent unit; A: dietary restriction trial/cohort score changes observed; comparator only where available; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: error-adjusted association of fixed score change with walking recovery; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; B it: change in the independently prespecified aging score, paired with walking recovery; error estimated from repeat assays; Bbar i: person/donor mean over a prespecified observation schedule; beta W,beta B: within-unit and between-unit association coefficients, not causal effects; X it: explicitly time-ordered baseline/premeasurement covariates; posttreatment adjustment changes interpretation; f time(t): prespecified sampling/recovery/calendar trajectory; b study: study/site/batch effect where relevant; B true,e it: latent predictor and measurement error, requiring identification constraints; family code: ASSOC_EIV
Sample size design: family: longitudinal association/reliability or held-out prediction precision simulation, not causal-efficacy power; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Parent-trial randomization does not randomize clock change; weight and lean-mass mediation assumptions cannot be replaced by routine covariate adjustment.
Associations need a clear sampling/time frame; causal interpretation is not supplied by this model.
Measurement error, reverse causation, time-varying confounding and selected observation may remain despite adjustment.
Specify E[u_i | Bbar_i,X] and its covariance or a correlated-random-effects formulation; an unrestricted fixed person intercept would absorb the between-person term.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Scores that add no prediction beyond weight and fitness, or improve despite worsening function, would fail the proposed resilience interpretation.; preanalysis gate: Parent-trial randomization does not randomize clock change; weight and lean-mass mediation assumptions cannot be replaced by routine covariate adjustment.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Parent-trial randomization does not randomize clock change; weight and lean-mass mediation assumptions cannot be replaced by routine covariate adjustment.
Vary lag order/window and measurement-error reliability within a prespecified family; include negative-control windows when meaningful.
Separate within- and between-unit effects, missingness patterns and source-study replication; do not use cell/assay counts as independent donor counts.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Parent-trial randomization does not randomize clock change; weight and lean-mass mediation assumptions cannot be replaced by routine covariate adjustment.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/176; question: Do caloric-restriction-associated changes in organ-specific aging measures correspond to preserved function after accounting for weight loss and lean tissue changes?; aims: Replicate organ-score changes using accessible dietary-trial samples.
Relate trajectories to strength, exertional recovery and nutritional adequacy.
Validate locked associations prospectively in an independent restriction cohort.; design verbatim planning only: Seek existing restriction-trial data and biospecimens, then recruit approximately 100 adults for twelve-month supervised validation with a weight-maintenance comparator where feasible. Treat score-function associations as observational despite parent-trial randomization.; outcomes verbatim: Primary: association between prespecified score change and recovery of walking performance after exertion. Secondary: strength, fatigue, lean mass and score stability.; analysis verbatim inherited not authoritative if refined: Use causal diagrams, locked scores and external calibration. Separate baseline confounding from weight mediation, address regression to the mean and restrict the hypothesis family.; falsification verbatim: Scores that add no prediction beyond weight and fitness, or improve despite worsening function, would fail the proposed resilience interpretation.; source ids unverified leads: calerie-organ-aging-rct-2026
epigenetic-interventions-review-2026
dietary-aging-profiles-2026; prospectus precedence: No

M10 — Learned food cues, satiety and sustainable substitution

Rank: 21
Cluster: Nutrition/metabolism/meal timing
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: FACTOR_DUAL; allocation or exposure unit: participant randomized coaching and practical food support factorially; independent sampling unit: participant; repeated measurement unit: agency score and unplanned-eating diary; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Allocate familywise alpha or choose a primary hierarchy before enrollment; energy density and reward-task changes are mediators, not primary confounders.
Primary estimand: factorial coaching/food-support effects and one interaction on the registered dual primary family
Candidate primary outcome: eating-agency score and unplanned-eating frequency under a fixed hierarchy
Primary finalized: No
Mathematical model: type: factorial main-effect/interaction planning with actual cross-classified or cluster-period covariance; formulas: E(Y_ict | B_ict, C_ict, X_i) = alpha_t + beta_B B_ict + beta_C C_ict + beta_BC B_ict C_ict + gamma^T X_i + u_i + b_c.
I = mu_11 - mu_10 - mu_01 + mu_00 = beta_BC on the additive scale; main effects are standardized over the other randomized factor.; variables: Y: eating-agency score and unplanned-eating frequency under a fixed hierarchy; i: participant; t: agency score and unplanned-eating diary; the registered visit/time-window index, not another independent unit; A: participant randomized coaching and practical food support factorially; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: factorial coaching/food-support effects and one interaction on the registered dual primary family; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; B,C: the two assigned factors named in the per-thesis allocation specification; levels and contrasts must be fixed; mu bc: marginal mean outcome under the specified joint factor levels; I: additive interaction estimand; beta B,beta C,beta BC: factor coefficients; for nonidentity links derive marginal contrasts rather than interpreting coefficients as risk differences; family code: FACTOR_DUAL
Sample size design: family: factorial main-effect/interaction planning with actual cross-classified or cluster-period covariance; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Allocate familywise alpha or choose a primary hierarchy before enrollment; energy density and reward-task changes are mediators, not primary confounders.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Factor allocation and coding must match the design; interaction inference is separate from main-effect inference and needs its own precision.
Two primary outcomes plus factorial effects require a fixed familywise rule; the intervention-by-factor interaction is not automatically a third confirmatory claim.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Task changes without improved eating agency would refute their assumed practical importance; absent added coaching benefit would favor simpler food support.; preanalysis gate: Allocate familywise alpha or choose a primary hierarchy before enrollment; energy density and reward-task changes are mediators, not primary confounders.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Allocate familywise alpha or choose a primary hierarchy before enrollment; energy density and reward-task changes are mediators, not primary confounders.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Allocate familywise alpha or choose a primary hierarchy before enrollment; energy density and reward-task changes are mediators, not primary confounders.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/20; question: Can cue-aware coaching improve appetite regulation without rigid restriction beyond the effects of practical food support?; aims: Map daily cues, hunger and unplanned eating within individuals.
Randomize coaching and food support as separate factorial components.
Test reward-task prediction of sustained eating agency and functioning.; design verbatim planning only: Randomize approximately 160 adults with cue-triggered eating in a six-month factorial coaching/food-support trial, followed for six further months. Match contact; use grocery receipts and brief diaries instead of continuous calorie counting.; outcomes verbatim: Primary: validated eating-agency score and prespecified unplanned-eating frequency. Secondary: diet quality, distress, weight stability, triglycerides and reward-task reliability.; analysis verbatim inherited not authoritative if refined: Estimate assignment-based main effects and one interaction. Account for baseline food insecurity, sleep and medications; correct dual-primary testing. Measure energy-density changes as possible mediators rather than adjust them out of the primary contrast; interpret perceived-control and reward-task mediation under explicit confounding assumptions.; falsification verbatim: Task changes without improved eating agency would refute their assumed practical importance; absent added coaching benefit would favor simpler food support.; source ids unverified leads: dopamine-food-reward-review-2026
precision-nutrition-rct-2024
behavior-stress-sleep-rct-2026; prospectus precedence: No

M11 — Functional B12 surveillance during long-term metformin care

Rank: 5
Cluster: Micronutrients/supplement quality
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: CLUSTER; allocation or exposure unit: primary-care team randomized enhanced B12 pathway versus guideline-consistent care; independent sampling unit: primary-care team; participant nested within team; repeated measurement unit: blinded neurologic/function assessment over eighteen months; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Equal outcome ascertainment is essential; more diagnoses alone do not establish clinical benefit; few teams require small-cluster inference.
Primary estimand: care-pathway assignment effect on identically assessed neuropathy-related function
Candidate primary outcome: eighteen-month neuropathy-related functional-impairment score
Primary finalized: No
Mathematical model: type: contrast precision/power by design-matrix or simulation planning; formulas: E(Y_it | A_i, X_i, Y_i0) = alpha_t + z(A_i)^T theta_t + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Delta = sum_t w_t c^T theta_t; estimate by baseline-adjusted marginal standardization, with unit/cluster-appropriate inference.; variables: Y: eighteen-month neuropathy-related functional-impairment score; i: primary-care team; participant nested within team; t: blinded neurologic/function assessment over eighteen months; the registered visit/time-window index, not another independent unit; A: primary-care team randomized enhanced B12 pathway versus guideline-consistent care; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: care-pathway assignment effect on identically assessed neuropathy-related function; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; z(A): prespecified indicators for the actual randomized arms; two arms reduce to one indicator; family code: CLUSTER
Sample size design: family: contrast precision/power by design-matrix or simulation planning; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Equal outcome ascertainment is essential; more diagnoses alone do not establish clinical benefit; few teams require small-cluster inference.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Number of randomized clusters, cluster size imbalance and degrees of freedom govern inference; use small-sample correction/randomization inference when appropriate.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: More abnormal tests without earlier functional detection or improved symptoms would not support clinical utility of the enhanced pathway.; preanalysis gate: Equal outcome ascertainment is essential; more diagnoses alone do not establish clinical benefit; few teams require small-cluster inference.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Equal outcome ascertainment is essential; more diagnoses alone do not establish clinical benefit; few teams require small-cluster inference.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Equal outcome ascertainment is essential; more diagnoses alone do not establish clinical benefit; few teams require small-cluster inference.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/4; question: Does targeted functional B12 assessment improve recognition and recovery of neuropathy among people receiving long-term metformin?; aims: Compare serum B12, functional markers and standardized sensory findings.
Randomize clinically acceptable primary-care screening pathways.
Distinguish biochemical correction from subsequent sensory and walking recovery.; design verbatim planning only: Cluster-randomize twelve primary-care teams enrolling approximately 300 metformin-treated adults to enhanced symptom-triggered assessment or guideline-consistent care. Follow eighteen months with the same scheduled blinded neurologic and functional assessments in both arms and linked records; all clinically indicated testing and deficiency treatment remain available.; outcomes verbatim: Primary: Eighteen-month neuropathy-related functional impairment assessed identically in both arms. Secondary: Clinically adjudicated deficiency detection before substantial deterioration, neuropathy symptoms, gait confidence, unnecessary testing, treatment burden and walking performance.; analysis verbatim inherited not authoritative if refined: Account for clustering, diabetes duration, renal function, diet and other neuropathies. Use blinded functional assessment and deficiency adjudication; distinguish increased ascertainment from earlier clinically useful detection. Separate randomized screening-pathway effects from observational treatment-response associations and assess informative follow-up loss.; falsification verbatim: More abnormal tests without earlier functional detection or improved symptoms would not support clinical utility of the enhanced pathway.; source ids unverified leads: nih-ods-b12-2025
metformin-antiaging-uncertainty-2025; prospectus precedence: No

M12 — Hidden cumulative B6 exposure from supplement stacking

Rank: 17
Cluster: Micronutrients/supplement quality
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: EXPOSURE_RCT; allocation or exposure unit: participant randomized enhanced versus safety-screened usual reconciliation; independent sampling unit: participant; product lot contributes exposure calibration; repeated measurement unit: consumption day, product assay and follow-up; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Propagate consumption/assay uncertainty; product count is not exposure; urgent safety actions precede allocation; clinical neuropathy outcomes are exploratory unless event precision supports them.
Primary estimand: three-month assignment effect on independently calibrated average daily supplemental B6 exposure
Candidate primary outcome: average daily supplemental B6 exposure in a prespecified physical unit
Primary finalized: No
Mathematical model: type: randomized calibrated-exposure mean precision with skew, correlated lot/consumption measurement error and baseline covariance; formulas: Y_i=(1/T_i) sum_(d=1..T_i) sum_p q_idp c_p, where q is recorded servings and c is independently assayed B6 amount per serving for the actual lot.
Estimate the baseline-adjusted randomized mean difference at three months over repeated draws from validated q/c measurement-error distributions; retain shared lot calibration error across participants.; variables: Y: average daily supplemental B6 exposure in a prespecified physical unit; i: participant; product lot contributes exposure calibration; t: consumption day, product assay and follow-up; the registered visit/time-window index, not another independent unit; A: participant randomized enhanced versus safety-screened usual reconciliation; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: three-month assignment effect on independently calibrated average daily supplemental B6 exposure; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; d,p: consumption day and actual product lot; T i: registered exposure-assessment day count; q idp: consumed servings on day d of lot p with reporting uncertainty; c p: independently assayed B6 amount per serving in one declared physical unit, with correlated lot uncertainty; family code: EXPOSURE_RCT
Sample size design: family: randomized calibrated-exposure mean precision with skew, correlated lot/consumption measurement error and baseline covariance; formula: Use design-specific simulation or independent-unit resampling for the stated estimand and target confidence width/power; preserve all allocation, censoring, clustering and measurement-error structure.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Propagate consumption/assay uncertainty; product count is not exposure; urgent safety actions precede allocation; clinical neuropathy outcomes are exploratory unless event precision supports them.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Label amount/product count is not verified exposure; missing inventory and unassayed lots need explicit uncertainty/bounds, not automatic zero.
Clinical neuropathy associations remain observational/exploratory even when exposure-reduction strategy is randomized.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: If inventories reveal little excess or reconciliation fails to reduce verified exposure, the proposed prevention mechanism is weakened; temporal association alone cannot prove B6 causation.; preanalysis gate: Propagate consumption/assay uncertainty; product count is not exposure; urgent safety actions precede allocation; clinical neuropathy outcomes are exploratory unless event precision supports them.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Propagate consumption/assay uncertainty; product count is not exposure; urgent safety actions precede allocation; clinical neuropathy outcomes are exploratory unless event precision supports them.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Propagate consumption/assay uncertainty; product count is not exposure; urgent safety actions precede allocation; clinical neuropathy outcomes are exploratory unless event precision supports them.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/16; question: How often does supplement stacking produce excessive B6 exposure, and can a reconciliation tool reduce independently verified cumulative exposure?; aims: Link photographed inventories to measured product B6 content.
Relate cumulative exposure prospectively to standardized sensory outcomes.
Randomize safe educational reconciliation among nonurgent participants.; design verbatim planning only: Recruit a planning cohort of 250 adult supplement stackers for twelve months, with an embedded individual randomized comparison. Give everyone an immediate safety screen, actionable information and clinical referral as indicated; urgent cases bypass research allocation. Compare enhanced product-level reconciliation with safety-screened usual reconciliation. Repeated consumption logs and independent product assays calibrate cumulative exposure at baseline and three months, with later follow-up.; outcomes verbatim: Primary: three-month average daily supplemental B6 exposure, estimated from repeated consumption records with independently assayed product validation and propagated exposure uncertainty. Secondary: six- and twelve-month exposure, new sensory abnormalities, symptom trajectories, unnecessary stacking and referral completion; the study is not assumed powered for clinical neurologic harm.; analysis verbatim inherited not authoritative if refined: Model exposure uncertainty from assay and label variation. Estimate the randomized effect on verified cumulative exposure; treat new sensory abnormalities and symptom resolution as exploratory clinical outcomes unless event-based precision planning supports confirmatory testing. Adjust symptom associations for diabetes, alcohol and other neuropathies; prespecify endpoints and address differential product reporting.; falsification verbatim: If inventories reveal little excess or reconciliation fails to reduce verified exposure, the proposed prevention mechanism is weakened; temporal association alone cannot prove B6 causation.; source ids unverified leads: nih-ods-b6-2023
mvm-health-outcomes-review-2025; prospectus precedence: Yes; prospectus file: thesis-prospectuses.json; prospectus design and estimand: Use a prospective 12-month exposure cohort with an embedded individual randomized parallel trial. Both groups receive immediate safety screening, results requiring action, and access to clinicians. The enhanced group additionally receives a reconciled cumulative inventory and goal-based pharmacist discussion; usual care follows a documented standard pathway. The primary estimand is the three-month mean exposure difference under assignment to these offers, adjusted for baseline exposure. Report both absolute exposure and the distribution of changes; product count is not the endpoint.; prospectus analysis plan: Model the skewed exposure distribution with a prespecified suitable link or robust estimator, report an interpretable adjusted absolute difference, and propagate assay uncertainty through multiple draws or a measurement-error model. Preserve intention to treat even when clinicians change products. Use longitudinal models for neurological measures with baseline neuropathy, diabetes, alcohol use, renal status, and other relevant causes recorded before modeling. Do not interpret adjustment as eliminating all confounding. Correct or prioritize secondary endpoints and avoid data-derived toxic thresholds.; precedence note: Expanded prospectus refinements govern this scaffold; verbatim earlier brief material is retained for traceability, not as a contradictory analysis instruction. F15 is explicitly two arms; F30 has no unjustified symptom noninferiority claim.

M13 — When richer vitamin D testing changes clinical decisions

Rank: 114
Cluster: Micronutrients/supplement quality
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: PRED; allocation or exposure unit: expanded vitamin-D testing evaluated during clinician-directed bone-health care; independent sampling unit: participant with independent validation participants; repeated measurement unit: metabolite and lower-limb performance assessment; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: prediction, calibration or model transportability; prediction gain is not clinical efficacy, mechanistic necessity or causal treatment selection; specific constraint: Clinically indicated deficiency treatment is not withheld; observed management response is selected; threshold searching invalidates decision utility claims.
Primary estimand: incremental validated prediction of lower-limb function change beyond routine testing/clinical assessment
Candidate primary outcome: lower-limb physical-performance change at the fixed one-year horizon
Primary finalized: No
Mathematical model: type: external/nested predictive-validation precision with target prevalence/variance, independent-unit holdout coverage and model complexity; formulas: Gain = mean_test[L(Y,f_base(X)) - L(Y,f_new(X))], with both predictions frozen before test outcomes; bootstrap independent held-out units.
For a binary target assess E[Y | p_hat=p] versus p and calibration intercept/slope logit Pr(Y=1)=a_cal+b_cal logit(p_hat); for a continuous target use Y=a_cal+b_cal y_hat+error plus scale-appropriate calibration/loss.; variables: Y: lower-limb physical-performance change at the fixed one-year horizon; i: participant with independent validation participants; t: metabolite and lower-limb performance assessment; the registered visit/time-window index, not another independent unit; A: expanded vitamin-D testing evaluated during clinician-directed bone-health care; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: incremental validated prediction of lower-limb function change beyond routine testing/clinical assessment; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; L: one proper or scientifically justified primary loss/scoring rule fixed for the actual outcome type; positive Gain favors new model; f base,f new: locked baseline/new prediction functions using only information available at the registered prediction time; test: untouched person/family/lab/site/time set as specified per thesis; p hat,y hat: frozen probability or continuous prediction; a cal,b cal: calibration intercept and slope, estimated on evaluation data for assessment; using them to update the model requires another evaluation set; family code: PRED
Sample size design: family: external/nested predictive-validation precision with target prevalence/variance, independent-unit holdout coverage and model complexity; formula: Simulate or bootstrap the independent-unit test loss/calibration/decision criterion; plan cases/events, noncases and independent sites/families/programs separately. No universal events-per-variable or participant N is asserted.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Clinically indicated deficiency treatment is not withheld; observed management response is selected; threshold searching invalidates decision utility claims.
Train/tune/test separation at the actual biological/participant/family/time level; no future, postoutcome or target-derived predictors.
Validation support must cover intended deployment; calibration and missing/failing measurements are reported with discrimination/loss.
A validation subset selected by culture success, assay viability, consent or data completeness is a different target unless the full strategy is explicitly evaluated.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Improved biochemical reclassification without better calibrated functional prediction would reject the proposed added predictive value. Modeled decision benefit would remain hypothesis-generating; improved management from routine panel disclosure requires a separate implementation comparison.; preanalysis gate: Clinically indicated deficiency treatment is not withheld; observed management response is selected; threshold searching invalidates decision utility claims.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Clinically indicated deficiency treatment is not withheld; observed management response is selected; threshold searching invalidates decision utility claims.
Repeat loss/calibration with independent-cluster resampling, prespecified subgroup/transport shifts and plausible missingness.
Compare fixed simple baselines, calibration and uncertainty coverage; report abstention/failure/turnaround rather than deleting hard cases.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Frozen prediction time/horizon, proper loss, baseline model, feature availability, independent split/holdout and target deployment population.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Clinically indicated deficiency treatment is not withheld; observed management response is selected; threshold searching invalidates decision utility claims.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/113; question: Do additional vitamin D metabolite measures improve prediction of functional response beyond ordinary clinical assessment and standard testing?; aims: Assess reproducibility and reclassification from expanded metabolite panels.
Follow clinician-directed correction among patients with evaluation indications.
Compare added prediction and costs for meaningful functional response.; design verbatim planning only: Follow approximately 220 bone-health referrals and 80 independent validation participants for one year. Provide clinically indicated treatment throughout; blind expanded metabolite results unless actionable. Estimate response observationally rather than withholding deficiency treatment.; outcomes verbatim: Primary: Incremental prediction of change in lower-limb physical performance. Secondary: Model-based decision reclassification, documented management after any actionable disclosure, falls, biochemical normalization and testing cost.; analysis verbatim inherited not authoritative if refined: Adjust frailty, kidney function, season, adiposity and rehabilitation. Lock predictive models, assess calibration and decision utility, and account for treatment selection without threshold searching.; falsification verbatim: Improved biochemical reclassification without better calibrated functional prediction would reject the proposed added predictive value. Modeled decision benefit would remain hypothesis-generating; improved management from routine panel disclosure requires a separate implementation comparison.; source ids unverified leads: vital-vitamin-d-fractures-2022
endocrine-vitamin-d-guideline-2024
vitamin-d-status-rct-2024; prospectus precedence: No

M14 — Independent verification labels and safer supplement purchasing

Rank: 26
Cluster: Micronutrients/supplement quality
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT_AUDIT; allocation or exposure unit: buyer randomized report/icon/usual information; purchased lots sampled probabilistically in every arm; independent sampling unit: buyer for assignment; purchased lot with supplier/batch clustering for assay; repeated measurement unit: purchase, receipt linkage and technical assay; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Untested or unlinked lots have unknown quality, not brand-imputed passes; fix lot sampling inclusion probabilities and zero-purchase denominator rule.
Primary estimand: assignment effect on sampling-weighted proportion of purchased lots satisfying identity/label criteria
Candidate primary outcome: pass/fail analytical quality of an actual purchased lot
Primary finalized: No
Mathematical model: type: two-stage randomized buyer trial plus probability-sampled purchased-lot audit; survey ratio precision simulation; formulas: Q_hat_a = sum_(i:A_i=a) sum_(l in tested purchases_i) R_il q_il/pi_il / sum_(i:A_i=a) sum_(l in tested purchases_i) R_il/pi_il.
Delta_Q = Q_hat_1 - Q_hat_0; use buyer-randomization-aware, supplier/batch-aware survey estimation and sensitivity for unlinked/untested purchases.; variables: Y: pass/fail analytical quality of an actual purchased lot; i: buyer for assignment; purchased lot with supplier/batch clustering for assay; t: purchase, receipt linkage and technical assay; the registered visit/time-window index, not another independent unit; A: buyer randomized report/icon/usual information; purchased lots sampled probabilistically in every arm; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: assignment effect on sampling-weighted proportion of purchased lots satisfying identity/label criteria; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; q il: tested purchased lot meets all prespecified identity/label criteria (0/1); pi il: known inclusion probability of that actual purchase lot in the assay sample; R il: receipt/lot-linked, validly tested indicator; zero means unknown rather than quality failure or pass; Q a: proportion of purchased lots passing in arm a, a ratio target affected by purchase composition; z(A): prespecified indicators for the actual arms or randomized policy; family code: RCT_AUDIT
Sample size design: family: two-stage randomized buyer trial plus probability-sampled purchased-lot audit; survey ratio precision simulation; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Untested or unlinked lots have unknown quality, not brand-imputed passes; fix lot sampling inclusion probabilities and zero-purchase denominator rule.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Assay sampling probabilities must be known and positive; unlinked or technically failed lot results require missing-quality bounds, not design-weight fixes alone.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: If reports increase trust or spending without improving verified product selection, the proposed safety communication has failed.; preanalysis gate: Untested or unlinked lots have unknown quality, not brand-imputed passes; fix lot sampling inclusion probabilities and zero-purchase denominator rule.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Untested or unlinked lots have unknown quality, not brand-imputed passes; fix lot sampling inclusion probabilities and zero-purchase denominator rule.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Untested or unlinked lots have unknown quality, not brand-imputed passes; fix lot sampling inclusion probabilities and zero-purchase denominator rule.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/25; question: Does understandable independent product-verification information change supplement purchases and reduce exposure to mislabeled products?; aims: Audit prespecified label agreement and contaminants across retail channels.
Develop reports separating composition verification from evidence of benefit.
Randomize report formats and follow actual purchasing behavior.; design verbatim planning only: Blindly assay approximately 120 products across channels, then randomize about 300 buyers to interpretive reports, icons or usual information. Follow purchases for six months without supplying unverified products or assigning hazardous exposure. Reserve a prespecified laboratory budget for a probability sample of purchased lots from every arm; link receipts and lot identifiers to assays and retain untested or unlinked purchases as unknown quality.; outcomes verbatim: Primary: Sampling-weighted proportion of purchased product lots meeting prespecified identity and label-accuracy criteria, estimated from independently tested purchase-linked lots. Secondary: Misunderstanding of efficacy, total spending, unnecessary stacking and comprehension across literacy levels.; analysis verbatim inherited not authoritative if refined: Analyze assignment effects with baseline purchasing adjustment and purchase-lot sampling weights. Address vendor clustering, batch variation, missing receipts and unlinked lots; do not classify an untested purchased lot as passing solely from brand identity or an earlier lot. Preregister assay thresholds, the purchase-lot sampling frame and sensitivity analyses for unknown purchase quality.; falsification verbatim: If reports increase trust or spending without improving verified product selection, the proposed safety communication has failed.; source ids unverified leads: nih-ods-b6-2023
fda-heavy-metal-supplement-warning-2014
mvm-health-outcomes-review-2025; prospectus precedence: No

M15 — NAD target engagement versus fatigue and exertional recovery

Rank: 196
Cluster: Micronutrients/supplement quality
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT; allocation or exposure unit: participant randomized verified NAD precursor versus placebo; independent sampling unit: participant; repeated measurement unit: fatigue/interference and recovery assessment; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: A blood-metabolite rise is target engagement only; optional imaging selection limits mechanism inference; one baseline metabolic interaction must be frozen.
Primary estimand: precursor assignment effect on fatigue interference at twenty-four weeks
Candidate primary outcome: validated fatigue-interference score
Primary finalized: No
Mathematical model: type: contrast precision/power by design-matrix or simulation planning; formulas: E(Y_it | A_i, X_i, Y_i0) = alpha_t + z(A_i)^T theta_t + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Delta = sum_t w_t c^T theta_t; estimate by baseline-adjusted marginal standardization, with unit/cluster-appropriate inference.; variables: Y: validated fatigue-interference score; i: participant; t: fatigue/interference and recovery assessment; the registered visit/time-window index, not another independent unit; A: participant randomized verified NAD precursor versus placebo; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: precursor assignment effect on fatigue interference at twenty-four weeks; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; z(A): prespecified indicators for the actual randomized arms; two arms reduce to one indicator; family code: RCT
Sample size design: family: contrast precision/power by design-matrix or simulation planning; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: A blood-metabolite rise is target engagement only; optional imaging selection limits mechanism inference; one baseline metabolic interaction must be frozen.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Robust NAD elevation without improved fatigue or recovery would falsify the proposed functional pathway for this formulation and population.; preanalysis gate: A blood-metabolite rise is target engagement only; optional imaging selection limits mechanism inference; one baseline metabolic interaction must be frozen.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: A blood-metabolite rise is target engagement only; optional imaging selection limits mechanism inference; one baseline metabolic interaction must be frozen.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
A blood-metabolite rise is target engagement only; optional imaging selection limits mechanism inference; one baseline metabolic interaction must be frozen.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/195; question: Does increasing circulating NAD-related metabolites improve everyday fatigue or recovery in older adults with persistent low energy?; aims: Validate circulating NAD measures against subset muscle-energy assessments.
Compare one verified oral precursor with blinded placebo.
Assess functional mediation and persistence after study-product cessation.; design verbatim planning only: Randomize approximately 120 clinically evaluated adults aged 60–75 with fatigue to verified precursor or placebo for twenty-four weeks. Standardize exertion assessments; approximately 30 may undergo optional noninvasive muscle-energy imaging.; outcomes verbatim: Primary: validated fatigue interference score. Secondary: exercise recovery, walking capacity, daily activity, metabolite reproducibility and adverse events.; analysis verbatim inherited not authoritative if refined: Use assignment-based analysis, blinded batches and one metabolic-state interaction. Track supplements and fitness; test biochemical mediation only after demonstrating functional benefit.; falsification verbatim: Robust NAD elevation without improved fatigue or recovery would falsify the proposed functional pathway for this formulation and population.; source ids unverified leads: nad-wellness-review-2026
nmn-safety-meta-2026
nad-boosters-comparison-rct-2026; prospectus precedence: No

M16 — Dietary creatine status and meaningful cognitive response

Rank: 81
Cluster: Micronutrients/supplement quality
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT_INT; allocation or exposure unit: participant randomized creatine/placebo within baseline diet strata; independent sampling unit: participant; repeated measurement unit: repeated-baseline and prolonged-attention test; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Practice, sleep and time of day require standardized preassignment measurement; dietary strata must not be redefined from response.
Primary estimand: assignment effect and one baseline dietary-creatine interaction on sustained attention
Candidate primary outcome: prespecified sustained-attention performance during prolonged task
Primary finalized: No
Mathematical model: type: interaction-specific precision/power with modifier distribution and reliability; simulate the actual interaction; formulas: E(Y_it | A_i, B_i, X_i) = alpha_t + tau_t A_i + eta_t B_i + kappa_t A_i B_i + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Treatment contrast at baseline B=b is tau_t + kappa_t b; the interaction target is kappa_t, not a difference in within-subgroup p-values.; variables: Y: prespecified sustained-attention performance during prolonged task; i: participant; t: repeated-baseline and prolonged-attention test; the registered visit/time-window index, not another independent unit; A: participant randomized creatine/placebo within baseline diet strata; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: assignment effect and one baseline dietary-creatine interaction on sustained attention; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; B: baseline dietary-creatine status/stratum under the locked independent classification; centering/coding and any measurement-error model fixed before evaluation; eta: baseline modifier association; kappa: assignment-by-baseline-modifier interaction coefficient on the stated scale; family code: RCT_INT
Sample size design: family: interaction-specific precision/power with modifier distribution and reliability; simulate the actual interaction; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure
modifier prevalence/distribution and variance
modifier reliability and overlap
interaction size or precision target; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Practice, sleep and time of day require standardized preassignment measurement; dietary strata must not be redefined from response.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The modifier must be measured before assignment/initiation; interaction is scale dependent and requires support within modifier strata.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: No interaction with validated baseline status, or task changes without everyday transfer, would weaken personalized cognitive-enhancement claims.; preanalysis gate: Practice, sleep and time of day require standardized preassignment measurement; dietary strata must not be redefined from response.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Practice, sleep and time of day require standardized preassignment measurement; dietary strata must not be redefined from response.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Vary modifier measurement-error assumptions and retain a continuous modifier unless a cutoff was independently fixed.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Practice, sleep and time of day require standardized preassignment measurement; dietary strata must not be redefined from response.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/80; question: Does low dietary creatine identify adults who gain useful cognitive endurance from supplementation, beyond improvement on isolated memory tests?; aims: Validate dietary strata against biochemical and optional energy measures.
Randomize verified creatine versus placebo within dietary strata.
Test transfer to everyday prospective memory and work-error outcomes.; design verbatim planning only: Randomize approximately 160 adults across low- and higher-creatine dietary strata to verified creatine or placebo for sixteen weeks. Repeat baseline tasks to estimate practice effects; standardize sleep and testing time and include remote daily-function measures.; outcomes verbatim: Primary: sustained-attention performance under a standardized prolonged task. Secondary: prospective-memory failures, subjective cognitive functioning, sleep, memory and gastrointestinal tolerance.; analysis verbatim inherited not authoritative if refined: Prespecify one diet interaction and an outcome hierarchy. Include task reliability and sleep; report adherence analyses without replacing assignment-based estimates.; falsification verbatim: No interaction with validated baseline status, or task changes without everyday transfer, would weaken personalized cognitive-enhancement claims.; source ids unverified leads: creatine-cognition-meta-2024
mvm-health-outcomes-review-2025; prospectus precedence: No

M17 — Do sweat-guided electrolyte plans improve heat tolerance?

Rank: 61
Cluster: Micronutrients/supplement quality
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: CROSS; allocation or exposure unit: participant randomized personalized/standard hydration sequence; independent sampling unit: participant; repeated measurement unit: supervised heat-task period and repeated sweat assay; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Acclimatization and carryover affect periods; sweat classifications fixed from repeats; no outcome-selected electrolyte target.
Primary estimand: personalized advice effect on safe standardized heat-task performance
Candidate primary outcome: standardized task-performance endpoint under the common approved heat protocol
Primary finalized: No
Mathematical model: type: paired/crossover contrast precision with period, carryover and within-unit covariance; simulation for actual design; formulas: g(E[Y_ipt | A_ip,X_i]) = alpha_i + tau A_ip + pi_p + rho A_i,p-1 + gamma^T X_i + f_time(t) + b_cluster(i,p).
Target the registered within-unit assignment contrast, standardized over period/history; direct and lag effects are distinct parameters.; variables: Y: standardized task-performance endpoint under the common approved heat protocol; i: participant; t: supervised heat-task period and repeated sweat assay; the registered visit/time-window index, not another independent unit; A: participant randomized personalized/standard hydration sequence; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: personalized advice effect on safe standardized heat-task performance; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; p: randomized crossover period/session/episode, not the outcome probability; pi p: period effect; include calendar/season/sequence terms supported by the design; rho: prespecified lagged-assignment/carryover parameter; cannot in general identify arbitrary persistent carryover in a short two-period trial; f time(t): prespecified within-period recovery/day trend; g: identity for mean outcomes, log for positive/count means, or logit for binary/proportion outcomes as specified below; family code: CROSS
Sample size design: family: paired/crossover contrast precision with period, carryover and within-unit covariance; simulation for actual design; formula: For a simple two-condition paired mean only: Var(Delta_hat)=sigma_D^2/n_independent; actual cluster/period/lag models require design-specific simulation or c^T(D^T V^(-1)D)^(-1)c.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; sigma D^2: variance of independent-unit paired contrasts including measurement error; n independent: number of independent participants/teams/donors, not repeated samples; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Acclimatization and carryover affect periods; sweat classifications fixed from repeats; no outcome-selected electrolyte target.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Concealed randomized sequence, a scientifically justified recovery interval and no anticipation; period/carryover cannot be repaired solely by a nonsignificant test.
If carryover is plausible, report lag/period and first-period or otherwise valid design sensitivity; do not drop periods based only on a post hoc carryover p-value.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Better fluid-balance markers without performance or symptom improvement would reject the assumed practical benefit of personalization.; preanalysis gate: Acclimatization and carryover affect periods; sweat classifications fixed from repeats; no outcome-selected electrolyte target.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Acclimatization and carryover affect periods; sweat classifications fixed from repeats; no outcome-selected electrolyte target.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Compare planned lag/carryover structures and a valid first-period sensitivity where its sample supports inference; assess informative absent periods.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Acclimatization and carryover affect periods; sweat classifications fixed from repeats; no outcome-selected electrolyte target.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/60; question: Does tailoring rehydration advice to repeated sweat measurements improve heat-exercise performance beyond ordinary evidence-based hydration support?; aims: Quantify sweat and fluid-loss variability across heat conditions.
Compare repeated-measure personalization with standard advice by randomized crossover.
Assess field usability, symptoms and meaningful task performance.; design verbatim planning only: Enroll approximately 70 adults across sexes and activity levels in repeated laboratory heat sessions and supervised field testing. Randomize personalized versus standard hydration in crossover order; match palatability, energy and meal access with validated washouts.; outcomes verbatim: Primary: standardized task performance under a prespecified safe heat protocol. Secondary: thermal strain, symptoms, fluid balance, cost, acceptability and repeatability of sweat classifications.; analysis verbatim inherited not authoritative if refined: Model period, order, acclimatization and within-person variation. Prespecify high-loss classification, record relevant medications and menstrual factors, and avoid outcome-selected electrolyte targets.; falsification verbatim: Better fluid-balance markers without performance or symptom improvement would reject the assumed practical benefit of personalization.; source ids unverified leads: exercise-electrolyte-rehydration-review-2025
hyperhydration-heat-rct-2025
sweat-multianalyte-microfluidic-2025; prospectus precedence: No

M18 — Multivitamins, food insecurity and functional benefit

Rank: 15
Cluster: Micronutrients/supplement quality
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: FACTOR; allocation or exposure unit: participant randomized multivitamin/placebo and enhanced grocery support/existing services; independent sampling unit: participant with recruitment-site clustering; repeated measurement unit: practical cognitive-task and adequacy visit; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Confirmed deficiencies treated in all arms; assign multiplicity across factorial/interaction tests; food support uptake is postassignment.
Primary estimand: factorial multivitamin/food-support effects and baseline-inadequacy modification on practical cognition
Candidate primary outcome: externally scored practical cognitive-task composite
Primary finalized: No
Mathematical model: type: factorial main-effect/interaction planning with actual cross-classified or cluster-period covariance; formulas: E(Y_ict | B_ict, C_ict, X_i) = alpha_t + beta_B B_ict + beta_C C_ict + beta_BC B_ict C_ict + gamma^T X_i + u_i + b_c.
I = mu_11 - mu_10 - mu_01 + mu_00 = beta_BC on the additive scale; main effects are standardized over the other randomized factor.
Add baseline inadequacy Q_i and only the prespecified supplement-by-Q_i (and, if explicitly registered, food-support-by-Q_i) interaction; distinguish these from the supplement-by-food-support interaction.; variables: Y: externally scored practical cognitive-task composite; i: participant with recruitment-site clustering; t: practical cognitive-task and adequacy visit; the registered visit/time-window index, not another independent unit; A: participant randomized multivitamin/placebo and enhanced grocery support/existing services; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: factorial multivitamin/food-support effects and baseline-inadequacy modification on practical cognition; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; B,C: the two assigned factors named in the per-thesis allocation specification; levels and contrasts must be fixed; mu bc: marginal mean outcome under the specified joint factor levels; I: additive interaction estimand; beta B,beta C,beta BC: factor coefficients; for nonidentity links derive marginal contrasts rather than interpreting coefficients as risk differences; Q i: independently fixed baseline nutritional-inadequacy measure/category with validated reliability; distinct from randomized supplement factor B; family code: FACTOR
Sample size design: family: factorial main-effect/interaction planning with actual cross-classified or cluster-period covariance; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure
baseline-inadequacy modifier distribution/reliability and interaction precision target; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Confirmed deficiencies treated in all arms; assign multiplicity across factorial/interaction tests; food support uptake is postassignment.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Factor allocation and coding must match the design; interaction inference is separate from main-effect inference and needs its own precision.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Biochemical correction without functional gain would limit the supplementation hypothesis; equal benefit regardless of inadequacy would challenge the proposed responder mechanism.; preanalysis gate: Confirmed deficiencies treated in all arms; assign multiplicity across factorial/interaction tests; food support uptake is postassignment.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Confirmed deficiencies treated in all arms; assign multiplicity across factorial/interaction tests; food support uptake is postassignment.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Confirmed deficiencies treated in all arms; assign multiplicity across factorial/interaction tests; food support uptake is postassignment.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/14; question: Among older adults with food insecurity, does a multivitamin add cognitive or physical benefit beyond practical food support?; aims: Characterize nutrient inadequacy, food access and baseline function.
Separate supplement and food-support effects through factorial randomization.
Test inadequacy correction as a pathway to functional improvement.; design verbatim planning only: Randomize approximately 240 food-insecure adults over 65 to verified multivitamin/placebo and enhanced grocery support/existing services factorially for twelve months. Retain statutory benefits and clinically correct confirmed deficiencies in every group.; outcomes verbatim: Primary: prespecified composite of practical cognitive tasks. Secondary: physical performance, fatigue, diet adequacy, food security, quality of life and healthcare burden.; analysis verbatim inherited not authoritative if refined: Estimate factorial effects and baseline-inadequacy interaction. Account for recruitment-site clustering, attrition and cognitive multiplicity; record supplement crossover and support uptake.; falsification verbatim: Biochemical correction without functional gain would limit the supplementation hypothesis; equal benefit regardless of inadequacy would challenge the proposed responder mechanism.; source ids unverified leads: mvm-health-outcomes-review-2025
b-vitamin-cognition-meta-2025
nih-ods-b12-2025; prospectus precedence: No

M19 — Nonanemic iron deficiency and recovery in active women

Rank: 115
Cluster: Micronutrients/supplement quality
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: CLUSTER; allocation or exposure unit: care team randomized enhanced cause-assessment/follow-up versus guideline-consistent care; independent sampling unit: care team with participant outcomes; repeated measurement unit: fatigue/interference and iron-status visit; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Every confirmed deficiency is treated; iron-status change is not randomized; bleeding, energy availability and time-varying care complicate mediation.
Primary estimand: care-pathway assignment effect on fatigue interference at six months
Candidate primary outcome: fatigue interference with normal activities
Primary finalized: No
Mathematical model: type: contrast precision/power by design-matrix or simulation planning; formulas: E(Y_it | A_i, X_i, Y_i0) = alpha_t + z(A_i)^T theta_t + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Delta = sum_t w_t c^T theta_t; estimate by baseline-adjusted marginal standardization, with unit/cluster-appropriate inference.; variables: Y: fatigue interference with normal activities; i: care team with participant outcomes; t: fatigue/interference and iron-status visit; the registered visit/time-window index, not another independent unit; A: care team randomized enhanced cause-assessment/follow-up versus guideline-consistent care; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: care-pathway assignment effect on fatigue interference at six months; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; z(A): prespecified indicators for the actual randomized arms; two arms reduce to one indicator; family code: CLUSTER
Sample size design: family: contrast precision/power by design-matrix or simulation planning; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Every confirmed deficiency is treated; iron-status change is not randomized; bleeding, energy availability and time-varying care complicate mediation.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Number of randomized clusters, cluster size imbalance and degrees of freedom govern inference; use small-sample correction/randomization inference when appropriate.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: No functional advantage despite greater completion of cause assessment and follow-up would undermine the enhanced pathway claim. Improved laboratory values without improved fatigue or recovery would limit the clinical relevance of biochemical correction; neither pattern alone identifies an iron-specific causal effect.; preanalysis gate: Every confirmed deficiency is treated; iron-status change is not randomized; bleeding, energy availability and time-varying care complicate mediation.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Every confirmed deficiency is treated; iron-status change is not randomized; bleeding, energy availability and time-varying care complicate mediation.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Every confirmed deficiency is treated; iron-status change is not randomized; bleeding, energy availability and time-varying care complicate mediation.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/114; question: Does enhanced cause assessment and follow-up improve functional recovery beyond guideline-consistent care for active women with confirmed nonanemic iron deficiency?; aims: Confirm persistent deficiency beyond post-exercise biomarker fluctuation.
Compare enhanced cause assessment with guideline-consistent clinical pathways.
Separate functional recovery from concurrent intake and training changes.; design verbatim planning only: Enroll approximately 160 symptomatic active women with repeated baseline testing. Randomize care teams to enhanced cause assessment/follow-up or guideline-consistent care for six months; treat every confirmed deficiency and prospectively record bleeding, diet and training.; outcomes verbatim: Primary: fatigue interference with normal activities. Secondary: standardized submaximal exercise recovery, missed training days, gastrointestinal burden, iron status and recurrent deficiency.; analysis verbatim inherited not authoritative if refined: Estimate the assignment effect of the enhanced care pathway, accounting for care-team clustering. Describe iron-status and functional trajectories under explicit observational mediation assumptions; adjust baseline bleeding, inflammation, contraception, energy availability and sleep and handle time-varying cointerventions appropriately. Prespecify the functional primary and correct secondary comparisons.; falsification verbatim: No functional advantage despite greater completion of cause assessment and follow-up would undermine the enhanced pathway claim. Improved laboratory values without improved fatigue or recovery would limit the clinical relevance of biochemical correction; neither pattern alone identifies an iron-specific causal effect.; source ids unverified leads: female-athlete-nutrition-review-2025
menstrual-energy-intake-meta-2025
mvm-health-outcomes-review-2025; prospectus precedence: No

M20 — Biotin interference in self-ordered thyroid testing

Rank: 128
Cluster: Micronutrients/supplement quality
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: PAIRED_INTERFERENCE; allocation or exposure unit: measured biotin observed; coded aliquot randomized active versus sham removal; independent sampling unit: participant for exposure association; allocated aliquot for ex-vivo removal effect; repeated measurement unit: paired TSH/free-T4 platform and removal/control assay; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: observational biotin-discordance association plus distinct controlled aliquot-interference contrast; no randomized human biotin effect; specific constraint: Comparator must be biotin independent and removal recovery validated; controlled analytical reversal does not itself estimate clinical misdiagnosis incidence.
Primary estimand: biotin-associated excess classification discordance; separate randomized aliquot removal contrast
Candidate primary outcome: prespecified TSH classification-discordance indicator
Primary finalized: No
Mathematical model: type: paired classification/discordance precision with controlled aliquot comparison and exposure-error sensitivity; formulas: logit Pr(D_i=1 | B_i,X_i)=alpha+f(B_i)+gamma^T X_i, where D is locked paired-platform TSH classification discordance and B is measured biotin.
For split-aliquot randomization estimate Delta_removal=E[D_i(active removal)-D_i(sham removal)] with participant/specimen pairing and validated analyte recovery; this controlled ex-vivo effect differs from the exposure association.; variables: Y: prespecified TSH classification-discordance indicator; i: participant for exposure association; allocated aliquot for ex-vivo removal effect; t: paired TSH/free-T4 platform and removal/control assay; the registered visit/time-window index, not another independent unit; A: measured biotin observed; coded aliquot randomized active versus sham removal; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: biotin-associated excess classification discordance; separate randomized aliquot removal contrast; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; D i: prespecified TSH-platform classification-discordance indicator; B i: measured biotin in the declared sample/assay unit; f: prespecified biotin relationship, not an outcome-selected toxic/interference cutoff; Delta removal: paired randomized aliquot removal effect on analytical discordance; family code: PAIRED_INTERFERENCE
Sample size design: family: paired classification/discordance precision with controlled aliquot comparison and exposure-error sensitivity; formula: Use design-specific simulation or independent-unit resampling for the stated estimand and target confidence width/power; preserve all allocation, censoring, clustering and measurement-error structure.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Comparator must be biotin independent and removal recovery validated; controlled analytical reversal does not itself estimate clinical misdiagnosis incidence.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Comparator biotin independence, platform imprecision and removal/recovery controls require validation; inconclusive histories/results retained.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: No excess discordance, or failed sham-controlled reversibility with adequate precision, would weaken biotin-specific attribution; discrepancies alone never establish harm.; preanalysis gate: Comparator must be biotin independent and removal recovery validated; controlled analytical reversal does not itself estimate clinical misdiagnosis incidence.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Comparator must be biotin independent and removal recovery validated; controlled analytical reversal does not itself estimate clinical misdiagnosis incidence.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Comparator must be biotin independent and removal recovery validated; controlled analytical reversal does not itself estimate clinical misdiagnosis incidence.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/127; question: Does biotin exposure distort self-ordered thyroid-test classification, and can paired-platform checking distinguish interference from genuine endocrine change?; aims: Ascertain supplement products, ingestion timing and circulating biotin.
Quantify paired-platform TSH classification discordance against a validated biotin-independent comparator.
Test removal reversibility and orthogonal confirmation in discordant specimens.; design verbatim planning only: Prospectively recruit approximately 200 adults already ordering thyroid tests, without changing care. Record exposure before collection; blind laboratories to histories. Compare paired TSH/free-T4 platforms, verify comparator biotin independence, and test active versus sham removal with recovery controls in coded aliquots. Confirm selected free-T4 discrepancies using equilibrium-dialysis-based measurement.; outcomes verbatim: Primary: excess probability of prespecified TSH classification discordance associated with measured biotin. Secondary: free-T4 discordance, analytical reversibility, repeat-testing burden and adjudicated decision changes.; analysis verbatim inherited not authoritative if refined: Use paired models, accounting for thyroid disease, renal function and other interferences; distinguish observational exposure associations from controlled aliquot effects. Lock thresholds, quantify platform bias and imprecision, and assess missing histories and inconclusive results.; falsification verbatim: No excess discordance, or failed sham-controlled reversibility with adequate precision, would weaken biotin-specific attribution; discrepancies alone never establish harm.; source ids unverified leads: rwd-digital-platform-cohort-2026
testosterone-endocrine-statement-2026; prospectus precedence: No

M21 — Prospective validation of microbiome-based fiber response

Rank: 100
Cluster: Gut and oral microbiome
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT_INT; allocation or exposure unit: participant randomized fiber-rich foods versus matched support; independent sampling unit: participant; repeated measurement unit: HbA1c visit and repeated baseline stool assay; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Microbiome classifier and cutoffs fixed before outcomes; repeat stool samples estimate misclassification; interaction planning is not average-effect planning.
Primary estimand: locked baseline microbiome-by-treatment interaction on HbA1c
Candidate primary outcome: HbA1c at nine months, baseline adjusted
Primary finalized: No
Mathematical model: type: interaction-specific precision/power with modifier distribution and reliability; simulate the actual interaction; formulas: E(Y_it | A_i, B_i, X_i) = alpha_t + tau_t A_i + eta_t B_i + kappa_t A_i B_i + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Treatment contrast at baseline B=b is tau_t + kappa_t b; the interaction target is kappa_t, not a difference in within-subgroup p-values.; variables: Y: HbA1c at nine months, baseline adjusted; i: participant; t: HbA1c visit and repeated baseline stool assay; the registered visit/time-window index, not another independent unit; A: participant randomized fiber-rich foods versus matched support; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: locked baseline microbiome-by-treatment interaction on HbA1c; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; B: locked baseline microbial classifier/score estimated from repeat preallocation stool specimens; centering/coding and any measurement-error model fixed before evaluation; eta: baseline modifier association; kappa: assignment-by-baseline-modifier interaction coefficient on the stated scale; family code: RCT_INT
Sample size design: family: interaction-specific precision/power with modifier distribution and reliability; simulate the actual interaction; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure
modifier prevalence/distribution and variance
modifier reliability and overlap
interaction size or precision target; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Microbiome classifier and cutoffs fixed before outcomes; repeat stool samples estimate misclassification; interaction planning is not average-effect planning.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The modifier must be measured before assignment/initiation; interaction is scale dependent and requires support within modifier strata.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Failure to reproduce the interaction or added value beyond ordinary clinical variables would reject microbiome-guided allocation for this intervention.; preanalysis gate: Microbiome classifier and cutoffs fixed before outcomes; repeat stool samples estimate misclassification; interaction planning is not average-effect planning.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Microbiome classifier and cutoffs fixed before outcomes; repeat stool samples estimate misclassification; interaction planning is not average-effect planning.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Vary modifier measurement-error assumptions and retain a continuous modifier unless a cutoff was independently fixed.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Microbiome classifier and cutoffs fixed before outcomes; repeat stool samples estimate misclassification; interaction planning is not average-effect planning.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/99; question: Can a locked microbiome signature identify people with prediabetes who gain clinically meaningful benefit from a fiber-rich dietary intervention?; aims: Reconstruct a reproducible signature using permitted prior-trial data.
Lock classification before randomizing fiber support versus usual care.
Evaluate interaction, classification stability and added decision value.; design verbatim planning only: Randomize approximately 200 adults with prediabetes to supported fiber-rich foods or matched usual-care support for nine months. Lock microbial classification beforehand, repeat baseline stool sampling and blind outcomes; planning numbers require interaction-specific refinement.; outcomes verbatim: Primary: prespecified microbiome-by-treatment interaction for HbA1c. Secondary: glucose tolerance, bowel symptoms, diet quality, classification reliability and intervention burden.; analysis verbatim inherited not authoritative if refined: Use a short baseline-adjusted model, appropriate compositional methods and locked cutoffs. Control feature multiplicity and report uncertainty without optimizing the validation cohort.; falsification verbatim: Failure to reproduce the interaction or added value beyond ordinary clinical variables would reject microbiome-guided allocation for this intervention.; source ids unverified leads: microbiome-fiber-prediabetes-rct-2025
precision-nutrition-rct-2024; prospectus precedence: No

M22 — Plant diversity beyond nutrient composition

Rank: 139
Cluster: Gut and oral microbiome
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT; allocation or exposure unit: participant randomized nutrient-matched varied versus limited-variety foods; independent sampling unit: participant; repeated measurement unit: bowel symptom and stool collection visit; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Match actual fiber/energy/processing and report deviations; richness is exploratory rather than a clinical endpoint; antibiotic/laxative changes tracked.
Primary estimand: variety assignment effect on bowel-function symptom score
Candidate primary outcome: validated bowel-function symptom score at eight weeks
Primary finalized: No
Mathematical model: type: contrast precision/power by design-matrix or simulation planning; formulas: E(Y_it | A_i, X_i, Y_i0) = alpha_t + z(A_i)^T theta_t + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Delta = sum_t w_t c^T theta_t; estimate by baseline-adjusted marginal standardization, with unit/cluster-appropriate inference.; variables: Y: validated bowel-function symptom score at eight weeks; i: participant; t: bowel symptom and stool collection visit; the registered visit/time-window index, not another independent unit; A: participant randomized nutrient-matched varied versus limited-variety foods; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: variety assignment effect on bowel-function symptom score; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; z(A): prespecified indicators for the actual randomized arms; two arms reduce to one indicator; family code: RCT
Sample size design: family: contrast precision/power by design-matrix or simulation planning; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Match actual fiber/energy/processing and report deviations; richness is exploratory rather than a clinical endpoint; antibiotic/laxative changes tracked.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Microbiome change without superior bowel function would reject a functional-diversity claim; equivalent metabolite responses would weaken the substrate-variety mechanism.; preanalysis gate: Match actual fiber/energy/processing and report deviations; richness is exploratory rather than a clinical endpoint; antibiotic/laxative changes tracked.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Match actual fiber/energy/processing and report deviations; richness is exploratory rather than a clinical endpoint; antibiotic/laxative changes tracked.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Match actual fiber/energy/processing and report deviations; richness is exploratory rather than a clinical endpoint; antibiotic/laxative changes tracked.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/138; question: Does increasing whole-food plant variety improve bowel function independently of total fiber, energy and food-processing differences?; aims: Develop varied and limited-variety diets matched for major nutrients.
Randomize patterns with identical food provision and contact.
Test metabolite pathways, bowel function and post-support persistence.; design verbatim planning only: Randomize approximately 150 low-variety eaters, including a mild-constipation stratum, to nutrient-matched varied or limited-variety diets. Provide food for eight weeks, then observe six months; blind laboratories and assessors and repeat stool sampling.; outcomes verbatim: Primary: validated bowel-function symptom score. Secondary: spontaneous complete bowel movements, gastrointestinal quality of life, dietary adherence and prespecified microbial metabolic pathways.; analysis verbatim inherited not authoritative if refined: Adjust baseline symptoms, prespecify constipation interaction and pathway tests, and control multiplicity. Record laxatives, antibiotics and diet changes; keep richness exploratory.; falsification verbatim: Microbiome change without superior bowel function would reject a functional-diversity claim; equivalent metabolite responses would weaken the substrate-variety mechanism.; source ids unverified leads: biome-whole-food-rct-2026
microbiome-fiber-prediabetes-rct-2025; prospectus precedence: No

M23 — Postbiotic potency, stability and IBS symptom response

Rank: 93
Cluster: Gut and oral microbiome
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT_BATCH; allocation or exposure unit: participant randomized active postbiotic/placebo; independently characterized active batches allocated within active arm; independent sampling unit: participant for average trial effect; independent active batch for potency validation; repeated measurement unit: symptom visit and batch assay replicate; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Patients per batch do not replace independent batches; potency-marker validation stays exploratory until batch-level precision is justified.
Primary estimand: active-versus-placebo effect on IBS response; separate batch-level heterogeneity/potency association
Candidate primary outcome: validated IBS symptom-severity response under a fixed response definition
Primary finalized: No
Mathematical model: type: patient-level response trial plus separate independent-batch variance/potency-marker precision; formulas: logit Pr(Y_i=1 | A_i,X_i,b_batch)=alpha+tau A_i+gamma^T X_i+A_i b_batch(i), if IBS response is binary; if a continuous symptom change is selected, use its registered mean model.
Estimate the average randomized active/placebo effect first; potency-marker slope is a separate independent-active-batch analysis.; variables: Y: validated IBS symptom-severity response under a fixed response definition; i: participant for average trial effect; independent active batch for potency validation; t: symptom visit and batch assay replicate; the registered visit/time-window index, not another independent unit; A: participant randomized active postbiotic/placebo; independently characterized active batches allocated within active arm; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: active-versus-placebo effect on IBS response; separate batch-level heterogeneity/potency association; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; mu it: conditional mean of the actual per-thesis count; phi: overdispersion parameter, with Var(Y)=mu+mu^2/phi under this parameterization; z(A): prespecified indicators for the actual arms or randomized policy; b batch: active-batch effect, with placebo not spuriously assigned an active potency value; family code: RCT_BATCH
Sample size design: family: patient-level response trial plus separate independent-batch variance/potency-marker precision; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Patients per batch do not replace independent batches; potency-marker validation stays exploratory until batch-level precision is justified.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Finalize binary response versus continuous score before registration; independent batches, not patients or assay replicates, govern marker validation.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: No symptom advantage with sufficiently precise estimates despite confirmed product exposure would reject meaningful efficacy for this preparation. A potency marker would fail only if its prespecified prediction failed in an adequately replicated independent-batch evaluation; an imprecise null batch association is inconclusive.; preanalysis gate: Patients per batch do not replace independent batches; potency-marker validation stays exploratory until batch-level precision is justified.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Patients per batch do not replace independent batches; potency-marker validation stays exploratory until batch-level precision is justified.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Patients per batch do not replace independent batches; potency-marker validation stays exploratory until batch-level precision is justified.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/92; question: Does a defined postbiotic preparation improve IBS symptoms, and which batch-level characteristics warrant subsequent prospective validation?; aims: Validate independent batch identity, stability and candidate potency assays.
Compare standardized preparation against matched placebo in IBS.
Relate verified exposure and locked metabolites to symptom response.; design verbatim planning only: Randomize approximately 160 clinically diagnosed IBS patients to standardized postbiotic or placebo for twelve weeks, followed twelve more. Randomly allocate independently characterized active batches within the active arm, stratify subtype and maintain appropriate background care. Prespecify the number of independent batches and patients per batch after assay feasibility work; treat marker validation as exploratory unless batch-level precision is adequate.; outcomes verbatim: Primary: validated IBS symptom-severity response. Secondary: abdominal pain days, bowel-habit satisfaction, quality of life, adverse events and potency stability.; analysis verbatim inherited not authoritative if refined: Use subtype-adjusted assignment effects for the primary symptom endpoint. Estimate batch variation with patient clustering within independent batches; keep potency-marker associations exploratory unless independent-batch planning supports validation. Limit mechanistic tests, record diet/antibiotic changes and distinguish adherence from allocation.; falsification verbatim: No symptom advantage with sufficiently precise estimates despite confirmed product exposure would reject meaningful efficacy for this preparation. A potency marker would fail only if its prespecified prediction failed in an adequately replicated independent-batch evaluation; an imprecise null batch association is inconclusive.; source ids unverified leads: postbiotics-ibs-meta-2026
gut-probiotic-anxiety-rct-2026; prospectus precedence: No

M24 — Testing oral-to-gut microbial transmission after periodontal care

Rank: 178
Cluster: Gut and oral microbiome
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT_BINARY; allocation or exposure unit: participant randomized enhanced periodontal adherence support versus standard care; independent sampling unit: participant with prespecified strain-family structure; repeated measurement unit: paired oral/stool sample and strain detection visit; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Detection does not establish viable colonization or direction of transmission; treatment receipt and periodontal response are nonrandomized mediators.
Primary estimand: support-assignment effect on persistent detection of prespecified matched strains
Candidate primary outcome: persistent paired oral/stool strain-match detection indicator
Primary finalized: No
Mathematical model: type: binary risk-difference precision/power, simulated with actual cluster/trajectory structure; formulas: logit Pr(Y_i=1 | A_i,X_i) = alpha + z(A_i)^T theta + gamma^T X_i + b_cluster(i).
RD = E_X[p_hat(1,X) - p_hat(0,X)]; report absolute risk difference as well as any odds ratio.; variables: Y: persistent paired oral/stool strain-match detection indicator; i: participant with prespecified strain-family structure; t: paired oral/stool sample and strain detection visit; the registered visit/time-window index, not another independent unit; A: participant randomized enhanced periodontal adherence support versus standard care; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: support-assignment effect on persistent detection of prespecified matched strains; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; logit(p): log[p/(1-p)]; p hat(a,X): fitted outcome probability under arm a standardized to the prespecified population; RD: marginal risk difference; z(A): prespecified indicators for the actual arms or randomized policy; family code: RCT_BINARY
Sample size design: family: binary risk-difference precision/power, simulated with actual cluster/trajectory structure; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Detection does not establish viable colonization or direction of transmission; treatment receipt and periodontal response are nonrandomized mediators.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Lack of reliable strain matching or unchanged stool persistence despite oral improvement would weaken transmission claims; inflammation alone cannot establish the pathway.; preanalysis gate: Detection does not establish viable colonization or direction of transmission; treatment receipt and periodontal response are nonrandomized mediators.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Detection does not establish viable colonization or direction of transmission; treatment receipt and periodontal response are nonrandomized mediators.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Detection does not establish viable colonization or direction of transmission; treatment receipt and periodontal response are nonrandomized mediators.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/177; question: Does enhanced periodontal adherence support reduce persistent stool detection of strains matched to oral samples, and is that change associated with systemic inflammation?; aims: Establish repeated within-person strain matching across oral and stool sites.
Randomize clinically appropriate periodontal adherence support.
Test temporal transmission changes against dental recovery and inflammation.; design verbatim planning only: Randomize approximately 120 adults with periodontitis to enhanced adherence support plus standard care or standard care alone for nine months. Withhold no necessary treatment; repeatedly sample oral/stool sites and sequence a nested subset.; outcomes verbatim: Primary: Persistence of prespecified paired oral/stool strain matches. Secondary: Periodontal pocket improvement, bleeding, oral quality of life and one systemic inflammatory marker.; analysis verbatim inherited not authoritative if refined: Use blinded pipelines, contamination controls and uncertainty-aware source tracking. Estimate the randomized support effect, with treatment receipt and periodontal response analyzed as nonrandomized mediators. Adjust baseline smoking, diet and antibiotic exposure and model relevant time-varying changes under explicit assumptions. Call the molecular endpoint strain detection; infer viable colonization or direction only if additional independently validated viability and source-direction evidence supports it. Control strain-family multiplicity.; falsification verbatim: Lack of reliable strain matching or unchanged stool persistence despite oral improvement would weaken transmission claims; inflammation alone cannot establish the pathway.; source ids unverified leads: oral-gut-axis-review-2026
oral-microbiome-systemic-review-2026; prospectus precedence: No

M25 — Oral antisepsis, nitrate metabolism and functional tradeoffs

Rank: 140
Cluster: Gut and oral microbiome
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: CROSS; allocation or exposure unit: participant randomized dentist-supervised rinse/control sequence; independent sampling unit: participant; repeated measurement unit: saliva function assay and ambulatory pressure period; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Verify microbial/physiological recovery before periods; nitrate reduction is a mediator and mechanistic assay needs independent reliability.
Primary estimand: rinse assignment effect on ambulatory systolic blood pressure
Candidate primary outcome: ambulatory systolic pressure in mmHg
Primary finalized: No
Mathematical model: type: paired/crossover contrast precision with period, carryover and within-unit covariance; simulation for actual design; formulas: g(E[Y_ipt | A_ip,X_i]) = alpha_i + tau A_ip + pi_p + rho A_i,p-1 + gamma^T X_i + f_time(t) + b_cluster(i,p).
Target the registered within-unit assignment contrast, standardized over period/history; direct and lag effects are distinct parameters.; variables: Y: ambulatory systolic pressure in mmHg; i: participant; t: saliva function assay and ambulatory pressure period; the registered visit/time-window index, not another independent unit; A: participant randomized dentist-supervised rinse/control sequence; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: rinse assignment effect on ambulatory systolic blood pressure; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; p: randomized crossover period/session/episode, not the outcome probability; pi p: period effect; include calendar/season/sequence terms supported by the design; rho: prespecified lagged-assignment/carryover parameter; cannot in general identify arbitrary persistent carryover in a short two-period trial; f time(t): prespecified within-period recovery/day trend; g: identity for mean outcomes, log for positive/count means, or logit for binary/proportion outcomes as specified below; family code: CROSS
Sample size design: family: paired/crossover contrast precision with period, carryover and within-unit covariance; simulation for actual design; formula: For a simple two-condition paired mean only: Var(Delta_hat)=sigma_D^2/n_independent; actual cluster/period/lag models require design-specific simulation or c^T(D^T V^(-1)D)^(-1)c.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; sigma D^2: variance of independent-unit paired contrasts including measurement error; n independent: number of independent participants/teams/donors, not repeated samples; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Verify microbial/physiological recovery before periods; nitrate reduction is a mediator and mechanistic assay needs independent reliability.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Concealed randomized sequence, a scientifically justified recovery interval and no anticipation; period/carryover cannot be repaired solely by a nonsignificant test.
If carryover is plausible, report lag/period and first-period or otherwise valid design sensitivity; do not drop periods based only on a post hoc carryover p-value.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Reduced microbial nitrate conversion without a corresponding pressure effect would reject the proposed clinically relevant pathway in this setting.; preanalysis gate: Verify microbial/physiological recovery before periods; nitrate reduction is a mediator and mechanistic assay needs independent reliability.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Verify microbial/physiological recovery before periods; nitrate reduction is a mediator and mechanistic assay needs independent reliability.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Compare planned lag/carryover structures and a valid first-period sensitivity where its sample supports inference; assess informative absent periods.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Verify microbial/physiological recovery before periods; nitrate reduction is a mediator and mechanistic assay needs independent reliability.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/139; question: Do changes in oral nitrate-reducing activity after short-term antiseptic exposure alter ambulatory blood pressure while affecting oral health?; aims: Validate repeatable oral nitrate-reduction assays alongside composition.
Compare defined rinse and matched control under dental supervision.
Test reversibility and functional mediation of blood-pressure response.; design verbatim planning only: Enroll approximately 60 adults without urgent dental disease in dentist-supervised rinse/control crossover periods. Standardize relevant diet, document recovery between periods and repeat saliva and ambulatory pressure measurements; this mechanistic sample cannot establish cardiovascular-event effects.; outcomes verbatim: Primary: ambulatory systolic blood-pressure change. Secondary: oral nitrate-reduction capacity, gingival symptoms, taste effects and recovery of microbial function.; analysis verbatim inherited not authoritative if refined: Model period, order, carryover and baseline pressure. Prespecify one pathway, track nitrate intake and hygiene, and require functional-assay reliability before mediation analysis.; falsification verbatim: Reduced microbial nitrate conversion without a corresponding pressure effect would reject the proposed clinically relevant pathway in this setting.; source ids unverified leads: oral-microbiome-systemic-review-2026
oral-gut-axis-review-2026; prospectus precedence: No

M26 — Expectancy and microbial function in diet–mood interventions

Rank: 101
Cluster: Gut and oral microbiome
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: FACTOR; allocation or exposure unit: participant randomized dietary-support and blinded microbial/placebo components; independent sampling unit: participant; repeated measurement unit: depression/function and targeted-metabolite visit; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Psychiatric care remains stable as feasible but changes tracked; expectancy/microbial function do not become randomized mediators.
Primary estimand: factorial dietary/microbial assignment effects and registered interaction on depressive symptoms
Candidate primary outcome: validated depressive-symptom score at sixteen weeks
Primary finalized: No
Mathematical model: type: factorial main-effect/interaction planning with actual cross-classified or cluster-period covariance; formulas: E(Y_ict | B_ict, C_ict, X_i) = alpha_t + beta_B B_ict + beta_C C_ict + beta_BC B_ict C_ict + gamma^T X_i + u_i + b_c.
I = mu_11 - mu_10 - mu_01 + mu_00 = beta_BC on the additive scale; main effects are standardized over the other randomized factor.; variables: Y: validated depressive-symptom score at sixteen weeks; i: participant; t: depression/function and targeted-metabolite visit; the registered visit/time-window index, not another independent unit; A: participant randomized dietary-support and blinded microbial/placebo components; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: factorial dietary/microbial assignment effects and registered interaction on depressive symptoms; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; B,C: the two assigned factors named in the per-thesis allocation specification; levels and contrasts must be fixed; mu bc: marginal mean outcome under the specified joint factor levels; I: additive interaction estimand; beta B,beta C,beta BC: factor coefficients; for nonidentity links derive marginal contrasts rather than interpreting coefficients as risk differences; family code: FACTOR
Sample size design: family: factorial main-effect/interaction planning with actual cross-classified or cluster-period covariance; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Psychiatric care remains stable as feasible but changes tracked; expectancy/microbial function do not become randomized mediators.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Factor allocation and coding must match the design; interaction inference is separate from main-effect inference and needs its own precision.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Symptom gains without the proposed microbial changes, or strong expectancy effects without blinded-preparation benefit, would weaken a microbiome-specific mechanism.; preanalysis gate: Psychiatric care remains stable as feasible but changes tracked; expectancy/microbial function do not become randomized mediators.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Psychiatric care remains stable as feasible but changes tracked; expectancy/microbial function do not become randomized mediators.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Psychiatric care remains stable as feasible but changes tracked; expectancy/microbial function do not become randomized mediators.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/100; question: Do gut-directed interventions improve low mood through measurable microbial function, expectancy, or ordinary dietary improvement?; aims: Measure expectancy, diet and candidate microbial functions repeatedly.
Randomize diet and blinded microbial-preparation components factorially.
Test temporally ordered microbial mediation beyond expectancy and adequacy.; design verbatim planning only: Randomize approximately 180 adults with low mood receiving stable care to dietary support and blinded microbial-preparation/placebo components for sixteen weeks. Match dietary contact, follow six further months and measure a targeted metabolite panel.; outcomes verbatim: Primary: validated depressive-symptom score. Secondary: daily functioning, anxiety, gastrointestinal symptoms, treatment expectancy and prespecified microbial functional markers.; analysis verbatim inherited not authoritative if refined: Prespecify factorial contrasts and mechanistic hierarchy. Record psychiatric care, sleep, antibiotics and diet; use longitudinal mediation with unmeasured-confounding sensitivity analyses.; falsification verbatim: Symptom gains without the proposed microbial changes, or strong expectancy effects without blinded-preparation benefit, would weaken a microbiome-specific mechanism.; source ids unverified leads: gut-feelings-cognition-rct-2024
gut-probiotic-anxiety-rct-2026
nutritional-psychiatry-diet-meta-2025
gut-brain-dysbiosis-review-2026; prospectus precedence: No

M27 — Engraftment as a surrogate for recurrent infection prevention

Rank: 201
Cluster: Gut and oral microbiome
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: LANDMARK; allocation or exposure unit: original authorized trial allocation preserved; prediction conditional on prespecified event-free landmark; independent sampling unit: participant alive and recurrence-free at landmark; repeated measurement unit: prelandmark microbial sample and later recurrence/death event; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: prediction, calibration or model transportability; prediction gain is not clinical efficacy, mechanistic necessity or causal treatment selection; specific constraint: No future or postrecurrence samples; early events reported separately; conditional prediction cannot establish treatment-effect surrogacy.
Primary estimand: incremental prediction of recurrence after landmark through day ninety by functional markers versus strain counts
Candidate primary outcome: adjudicated recurrence after the fixed early landmark, with competing death
Primary finalized: No
Mathematical model: type: landmark/competing-risk prediction precision with event-free risk set, event counts, censoring and external calibration; formulas: At fixed landmark ell, use only H_i(ell) among alive event-free participants: F_1(t | H(ell))=Pr(T<=t,J=event | T>ell,H(ell)).
Evaluate calibrated cause-specific cumulative-incidence predictions and proper censoring-aware Brier/loss over (ell,tau]; do not treat death as independent censoring for the event risk.; variables: Y: adjudicated recurrence after the fixed early landmark, with competing death; i: participant alive and recurrence-free at landmark; t: prelandmark microbial sample and later recurrence/death event; the registered visit/time-window index, not another independent unit; A: original authorized trial allocation preserved; prediction conditional on prespecified event-free landmark; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: incremental prediction of recurrence after landmark through day ninety by functional markers versus strain counts; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; L: one proper or scientifically justified primary loss/scoring rule fixed for the actual outcome type; positive Gain favors new model; f base,f new: locked baseline/new prediction functions using only information available at the registered prediction time; test: untouched person/family/lab/site/time set as specified per thesis; p hat,y hat: frozen probability or continuous prediction; a cal,b cal: calibration intercept and slope, estimated on evaluation data for assessment; using them to update the model requires another evaluation set; ell: predeclared prediction landmark; H i(ell): history available no later than ell; T,J: first relevant event time and event type, including competing death; tau: fixed prediction horizon; F 1: conditional cumulative incidence for the primary event; early/prelandmark events reported separately; family code: LANDMARK
Sample size design: family: landmark/competing-risk prediction precision with event-free risk set, event counts, censoring and external calibration; formula: Simulate or bootstrap the independent-unit test loss/calibration/decision criterion; plan cases/events, noncases and independent sites/families/programs separately. No universal events-per-variable or participant N is asserted.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: No future or postrecurrence samples; early events reported separately; conditional prediction cannot establish treatment-effect surrogacy.
Train/tune/test separation at the actual biological/participant/family/time level; no future, postoutcome or target-derived predictors.
Validation support must cover intended deployment; calibration and missing/failing measurements are reported with discrimination/loss.
A validation subset selected by culture success, assay viability, consent or data completeness is a different target unless the full strategy is explicitly evaluated.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Failure of prespecified strain or functional markers to predict subsequent recurrence in independent data would defeat their individual prognostic use. Even successful prediction would not establish that a treatment-induced marker change is a valid surrogate for the treatment effect on recurrence.; preanalysis gate: No future or postrecurrence samples; early events reported separately; conditional prediction cannot establish treatment-effect surrogacy.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: No future or postrecurrence samples; early events reported separately; conditional prediction cannot establish treatment-effect surrogacy.
Repeat loss/calibration with independent-cluster resampling, prespecified subgroup/transport shifts and plausible missingness.
Compare fixed simple baselines, calibration and uncertainty coverage; report abstention/failure/turnaround rather than deleting hard cases.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Frozen prediction time/horizon, proper loss, baseline model, feature availability, independent split/holdout and target deployment population.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
No future or postrecurrence samples; early events reported separately; conditional prediction cannot establish treatment-effect surrogacy.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/200; question: Does durable engraftment of a defined microbial consortium predict protection against recurrent C. difficile infection beyond early community recovery?; aims: Standardize strain detection and functional assays in authorized-trial samples.
Relate longitudinal patterns to adjudicated infection recurrence.
Externally validate restricted functional prediction against strain-count models.; design verbatim planning only: Obtain authorized consortium-trial samples and outcomes for approximately 100–150 participants, plus an independent validation cohort if available. Track ninety-day adjudicated recurrence and preserve parent-trial allocation. Lock an early prediction landmark and, if feasible, additional sequential landmarks; use only samples obtained before each prediction time. Actual sample access and timing must be confirmed first.; outcomes verbatim: Primary: Incremental prediction of adjudicated recurrence after a prespecified early landmark through day ninety from functional markers versus strain counts measured by that landmark. Secondary: Overall ninety-day recurrence, landmark-specific time to recurrence, symptom burden, adverse infections and assay reproducibility.; analysis verbatim inherited not authoritative if refined: Use penalized landmark survival models and external calibration among participants alive and recurrence-free at each prediction time. Report early events separately and use no post-recurrence or future samples as predictors. Account for antibiotics, recurrence history, immunocompromise and competing death; distinguish conditional individual prediction from randomized treatment effects and formal treatment-effect surrogacy.; falsification verbatim: Failure of prespecified strain or functional markers to predict subsequent recurrence in independent data would defeat their individual prognostic use. Even successful prediction would not establish that a treatment-induced marker change is a valid surrogate for the treatment effect on recurrence.; source ids unverified leads: defined-microbiome-consortium-rct-2026
oral-gut-axis-review-2026; prospectus precedence: No

M28 — Separating pretreatment effects from microbiota-associated sleep changes

Rank: 237
Cluster: Gut and oral microbiome
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: OBS_TRAJECTORY; allocation or exposure unit: original trial allocation; antibiotic/engraftment protocol phases not separately randomized; independent sampling unit: participant; repeated measurement unit: serial microbial function, phase and sleep measurement; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: association or prospective prediction; temporal order and model adjustment alone do not identify a mechanism or treatment effect; specific constraint: Antibiotic, expectancy and engraftment components remain confounded; before-after trajectories cannot isolate a component effect.
Primary estimand: prespecified lagged relationship between microbial functional change and objective sleep efficiency, allocation aware
Candidate primary outcome: objectively measured sleep efficiency paired with earlier microbial change
Primary finalized: No
Mathematical model: type: longitudinal association/reliability or held-out prediction precision simulation, not causal-efficacy power; formulas: Y_it = alpha + u_i + beta_W (B_it-Bbar_i) + beta_B Bbar_i + gamma^T X_it + f_time(t) + b_study + epsilon_it; u_i is a random intercept with a declared conditional mean/covariance model, not unrestricted fixed person effects.
B_observed,it = B_true,it + e_it; estimate measurement variance from replicates/validation or report sensitivity rather than assume it known.
Use the registered lagged predictor B_i,t-lag rather than a future/post-sleep measurement; include original trial allocation and protocol-phase indicators without calling their unrandomized components causal.; variables: Y: objectively measured sleep efficiency paired with earlier microbial change; i: participant; t: serial microbial function, phase and sleep measurement; the registered visit/time-window index, not another independent unit; A: original trial allocation; antibiotic/engraftment protocol phases not separately randomized; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: prespecified lagged relationship between microbial functional change and objective sleep efficiency, allocation aware; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; B it: prespecified microbial functional-change measure before the fixed lagged sleep window; Bbar i: person/donor mean over a prespecified observation schedule; beta W,beta B: within-unit and between-unit association coefficients, not causal effects; X it: explicitly time-ordered baseline/premeasurement covariates; posttreatment adjustment changes interpretation; f time(t): prespecified sampling/recovery/calendar trajectory; b study: study/site/batch effect where relevant; B true,e it: latent predictor and measurement error, requiring identification constraints; lag: fixed mechanistically justified microbial-to-sleep lag window, not optimized on the same outcome data; family code: OBS_TRAJECTORY
Sample size design: family: longitudinal association/reliability or held-out prediction precision simulation, not causal-efficacy power; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Antibiotic, expectancy and engraftment components remain confounded; before-after trajectories cannot isolate a component effect.
Associations need a clear sampling/time frame; causal interpretation is not supplied by this model.
Measurement error, reverse causation, time-varying confounding and selected observation may remain despite adjustment.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Sleep gains occurring before engraftment or only in subjective ratings would weaken the proposed microbial sleep mechanism.; preanalysis gate: Antibiotic, expectancy and engraftment components remain confounded; before-after trajectories cannot isolate a component effect.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Antibiotic, expectancy and engraftment components remain confounded; before-after trajectories cannot isolate a component effect.
Vary lag order/window and measurement-error reliability within a prespecified family; include negative-control windows when meaningful.
Separate within- and between-unit effects, missingness patterns and source-study replication; do not use cell/assay counts as independent donor counts.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Antibiotic, expectancy and engraftment components remain confounded; before-after trajectories cannot isolate a component effect.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/236; question: In authorized microbiota-based insomnia research, which sleep changes follow antibiotic pretreatment, expectancy, or subsequent microbial engraftment?; aims: Collect serial sleep and samples around distinct protocol components.
Compare pretreatment, engraftment and subjective/objective sleep trajectories.
Replicate locked temporal signatures before proposing component randomization.; design verbatim planning only: Access an existing authorized insomnia trial and add dense serial sampling in approximately 60 participants. Compare randomized controls and within-person protocol phases, with key-point polysomnography; component effects remain unrandomized and cannot establish definitive causal attribution.; outcomes verbatim: Primary: temporal relationship between prespecified microbial functional change and objectively measured sleep efficiency. Secondary: insomnia severity, daytime function, antibiotic adverse effects and expectation scores.; analysis verbatim inherited not authoritative if refined: Use allocation-aware interrupted trajectories, prespecified lag windows and cointervention sensitivity analyses. Correct pathway multiplicity and avoid attributing all post-treatment change to engraftment.; falsification verbatim: Sleep gains occurring before engraftment or only in subjective ratings would weaken the proposed microbial sleep mechanism.; source ids unverified leads: fmt-insomnia-rct-2026
gut-brain-dysbiosis-review-2026
defined-microbiome-consortium-rct-2026; prospectus precedence: No

M29 — Interpretation support and reproducibility of consumer microbiome reports

Rank: 27
Cluster: Gut and oral microbiome
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT; allocation or exposure unit: intended test user randomized contextual counseling versus ordinary report; independent sampling unit: participant with service effects; repeated measurement unit: informed-decision assessment and repeated/split stool report; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Report repeatability, biological validity and clinical utility are different families; service-selected missing receipts and restrictive eating need independent follow-up.
Primary estimand: counseling-offer effect on validated informed-decision quality
Candidate primary outcome: prespecified informed-decision-quality score
Primary finalized: No
Mathematical model: type: contrast precision/power by design-matrix or simulation planning; formulas: E(Y_it | A_i, X_i, Y_i0) = alpha_t + z(A_i)^T theta_t + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Delta = sum_t w_t c^T theta_t; estimate by baseline-adjusted marginal standardization, with unit/cluster-appropriate inference.; variables: Y: prespecified informed-decision-quality score; i: participant with service effects; t: informed-decision assessment and repeated/split stool report; the registered visit/time-window index, not another independent unit; A: intended test user randomized contextual counseling versus ordinary report; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: counseling-offer effect on validated informed-decision quality; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; z(A): prespecified indicators for the actual randomized arms; two arms reduce to one indicator; family code: RCT
Sample size design: family: contrast precision/power by design-matrix or simulation planning; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Report repeatability, biological validity and clinical utility are different families; service-selected missing receipts and restrictive eating need independent follow-up.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Reproducible recommendations would contradict the proposed instability. Contextual counseling that improves neither informed decisions nor symptoms, or increases restriction or spending without benefit, would fail its utility claim. This two-arm comparison cannot establish the incremental utility of stool testing over ordinary symptom-and-diet assessment.; preanalysis gate: Report repeatability, biological validity and clinical utility are different families; service-selected missing receipts and restrictive eating need independent follow-up.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Report repeatability, biological validity and clinical utility are different families; service-selected missing receipts and restrictive eating need independent follow-up.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Report repeatability, biological validity and clinical utility are different families; service-selected missing receipts and restrictive eating need independent follow-up.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/26; question: Among intended consumer-test users, does contextual counseling improve informed decisions and symptoms compared with ordinary report delivery, and how stable are repeated recommendations?; aims: Assess repeated- and split-sample agreement across selected services.
Blindly classify recommendations for evidence, cost and restriction burden.
Randomize interpretation support and follow decisions and symptoms.; design verbatim planning only: Recruit approximately 180 intended consumer-test users, including a mild-symptom stratum. Assess repeated and split samples through research agreements; randomize contextual counseling versus ordinary report delivery and follow decisions and symptoms for six months.; outcomes verbatim: Primary: validated informed-decision quality. Secondary: gastrointestinal quality of life, unnecessary spending, restrictive eating, care-seeking and report reproducibility.; analysis verbatim inherited not authoritative if refined: Use assignment effects, service-level random effects and locked symptom strata. Record concurrent care and missing receipts; separate reproducibility, validity and utility outcome families.; falsification verbatim: Reproducible recommendations would contradict the proposed instability. Contextual counseling that improves neither informed decisions nor symptoms, or increases restriction or spending without benefit, would fail its utility claim. This two-arm comparison cannot establish the incremental utility of stool testing over ordinary symptom-and-diet assessment.; source ids unverified leads: gut-brain-dysbiosis-review-2026
precision-nutrition-rct-2024
microbiome-fiber-prediabetes-rct-2025; prospectus precedence: No

M30 — Salivary function, microbial ecology and chewing in polypharmacy

Rank: 68
Cluster: Gut and oral microbiome
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT; allocation or exposure unit: participant randomized integrated dry-mouth care versus usual dental support; independent sampling unit: participant with site effects; repeated measurement unit: oral function, saliva and microbial visit; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Prescriber medication decisions retained; saliva and microbiome changes may be mediators; time-varying medication changes require explicit mediation assumptions.
Primary estimand: integrated-care assignment effect on oral-health-related quality of life at twelve months
Candidate primary outcome: validated oral-health-related quality-of-life score
Primary finalized: No
Mathematical model: type: contrast precision/power by design-matrix or simulation planning; formulas: E(Y_it | A_i, X_i, Y_i0) = alpha_t + z(A_i)^T theta_t + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Delta = sum_t w_t c^T theta_t; estimate by baseline-adjusted marginal standardization, with unit/cluster-appropriate inference.; variables: Y: validated oral-health-related quality-of-life score; i: participant with site effects; t: oral function, saliva and microbial visit; the registered visit/time-window index, not another independent unit; A: participant randomized integrated dry-mouth care versus usual dental support; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: integrated-care assignment effect on oral-health-related quality of life at twelve months; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; z(A): prespecified indicators for the actual randomized arms; two arms reduce to one indicator; family code: RCT
Sample size design: family: contrast precision/power by design-matrix or simulation planning; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Prescriber medication decisions retained; saliva and microbiome changes may be mediators; time-varying medication changes require explicit mediation assumptions.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Functional improvement without preceding microbial change would weaken a microbiome-primary explanation; biomarkers adding no prediction would argue against routine profiling.; preanalysis gate: Prescriber medication decisions retained; saliva and microbiome changes may be mediators; time-varying medication changes require explicit mediation assumptions.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Prescriber medication decisions retained; saliva and microbiome changes may be mediators; time-varying medication changes require explicit mediation assumptions.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Prescriber medication decisions retained; saliva and microbiome changes may be mediators; time-varying medication changes require explicit mediation assumptions.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/67; question: Does improving clinically managed dry mouth restore oral function through salivary changes, microbial changes, or both in older adults using multiple medicines?; aims: Map repeated saliva, microbial function and oral symptoms.
Randomize enhanced dry-mouth pathways against usual dental support.
Test temporal mediation and incremental microbial prediction of function.; design verbatim planning only: Randomize approximately 160 older adults with dry mouth to enhanced integrated care or usual dental support for twelve months. Prescribers retain medication decisions; all receive essential care. Repeat saliva, chewing and targeted microbial assessments.; outcomes verbatim: Primary: validated oral-health-related quality of life. Secondary: chewing ability, salivary flow, incident dental lesions, nutrition-related difficulty and targeted microbial function.; analysis verbatim inherited not authoritative if refined: Adjust dental disease, medication burden, diabetes and smoking. Separate pathway effects from mediation; account for site clustering, medication changes and microbial multiplicity.; falsification verbatim: Functional improvement without preceding microbial change would weaken a microbiome-primary explanation; biomarkers adding no prediction would argue against routine profiling.; source ids unverified leads: oral-microbiome-systemic-review-2026
oral-gut-axis-review-2026; prospectus precedence: No

M31 — Preserving functional muscle during incretin-associated weight loss

Rank: 4
Cluster: Obesity/metabolic medicines
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT_DUAL; allocation or exposure unit: incretin-treated initiator randomized resistance support versus contact-matched education; independent sampling unit: participant; repeated measurement unit: chair-rise and lower-limb strength assessment; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Fix co-primary success rule and familywise alpha; achieved weight loss/attendance are not baseline adjustment; inability/death rules need prespecification.
Primary estimand: resistance-support assignment effects on the fixed chair-rise and lower-limb-strength primary family
Candidate primary outcome: chair-rise performance and prespecified lower-limb strength endpoint
Primary finalized: No
Mathematical model: type: contrast precision/power by design-matrix or simulation planning; formulas: E(Y_it | A_i, X_i, Y_i0) = alpha_t + z(A_i)^T theta_t + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Delta = sum_t w_t c^T theta_t; estimate by baseline-adjusted marginal standardization, with unit/cluster-appropriate inference.; variables: Y: chair-rise performance and prespecified lower-limb strength endpoint; i: participant; t: chair-rise and lower-limb strength assessment; the registered visit/time-window index, not another independent unit; A: incretin-treated initiator randomized resistance support versus contact-matched education; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: resistance-support assignment effects on the fixed chair-rise and lower-limb-strength primary family; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; z(A): prespecified indicators for the actual randomized arms; two arms reduce to one indicator; family code: RCT_DUAL
Sample size design: family: contrast precision/power by design-matrix or simulation planning; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Fix co-primary success rule and familywise alpha; achieved weight loss/attendance are not baseline adjustment; inability/death rules need prespecification.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The listed primary endpoints need a conjunction/hierarchy or familywise alpha rule; a favorable result chosen after looking is not confirmation.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Lean-mass preservation without better strength or activity would not establish functional protection; absent functional advantage despite training adherence would reject the support strategy.; preanalysis gate: Fix co-primary success rule and familywise alpha; achieved weight loss/attendance are not baseline adjustment; inability/death rules need prespecification.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Fix co-primary success rule and familywise alpha; achieved weight loss/attendance are not baseline adjustment; inability/death rules need prespecification.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Fix co-primary success rule and familywise alpha; achieved weight loss/attendance are not baseline adjustment; inability/death rules need prespecification.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/3; question: Can tailored resistance training preserve strength and mobility during clinician-managed incretin therapy beyond the effect of nutrition advice alone?; aims: Map early strength, lean-mass and activity trajectories.
Randomize resistance support against contact-matched mobility/nutrition education.
Test frailty and early strength decline as rehabilitation indicators.; design verbatim planning only: Randomize approximately 180 clinically treated incretin initiators to resistance support or contact-matched mobility/nutrition education for twelve months. Prescribers manage medicines; both groups receive nutrition support. Offer home options and blind functional assessors.; outcomes verbatim: Primary: chair-rise performance and prespecified lower-limb strength endpoint. Secondary: walking ability, daily activity limitation, lean mass, falls and treatment persistence.; analysis verbatim inherited not authoritative if refined: Control dual-primary testing; track drug exposure, attendance and nutrition. Prespecify frailty interaction and avoid adjusting away treatment-mediated weight loss in main analyses.; falsification verbatim: Lean-mass preservation without better strength or activity would not establish functional protection; absent functional advantage despite training adherence would reject the support strategy.; source ids unverified leads: glp1-weight-loss-rct-review-2025
resistance-sarcopenia-meta-2026
protein-exercise-network-meta-2026; prospectus precedence: No

M32 — A structured care bridge after medically necessary treatment cessation

Rank: 1
Cluster: Obesity/metabolic medicines
Status: DRAFT — primary death estimand unresolved; not ready for registration
Design: model family: DEATH_PENDING; allocation or exposure unit: participant randomized postcessation care bridge versus routine care with equal research visits; independent sampling unit: participant within clinic; repeated measurement unit: quality-of-life and clinical restart/switching visit; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: primary estimand unresolved pending an explicit preregistration decision; ordinary survivor mean alone is not an overall treatment-policy effect; specific constraint: Do not report observed-survivor mean difference as an overall causal treatment-policy effect; no ordinary score after death; restart/switching/incomplete attendance remain strategy consequences.
Primary estimand: twelve-month bridge-assignment effect on patient-important status after choosing an explicit death strategy
Candidate primary outcome: twelve-month obesity-related quality of life when alive; primary outcome undefined until death strategy chosen
Primary finalized: No
Mathematical model: type: unresolved: joint survival/function rank/composite simulation OR explicitly identified/bounded survivor-estimand precision; mean-score power cannot be finalized; formulas: No unconditional mean-score primary estimand exists until death handling is selected: Y^a is undefined when survival S^a=0.
Candidate 1: define ordered joint Z^a=(death status, living function/quality of life) with a patient-informed ordering; estimate the full F_a(z) and independent-draw superiority probability psi=Pr(Z_1>Z_0)+0.5Pr(Z_1=Z_0).
Candidate 2: SACE=E[Y^1-Y^0 | S^1=S^0=1]; this always-survivor causal estimand is not identified by comparing observed survivors without additional assumptions/bounds. Observed-survivor summaries must be labeled conditional.; variables: Y: twelve-month obesity-related quality of life when alive; primary outcome undefined until death strategy chosen; i: participant within clinic; t: quality-of-life and clinical restart/switching visit; the registered visit/time-window index, not another independent unit; A: participant randomized postcessation care bridge versus routine care with equal research visits; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: twelve-month bridge-assignment effect on patient-important status after choosing an explicit death strategy; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; a: assigned strategy, including clinician-directed changes per its treatment policy; S^a: potential survival indicator at the primary horizon under assignment a; Y^a: living-participant endpoint under a, undefined after death; Z^a: registered joint survival/functional or quality-of-life outcome if that option is chosen; Z 1,Z 0: independent draws from the arm-specific joint distributions; not paired individual benefit; F a: arm-specific joint-outcome distribution; psi: probabilistic superiority for independent draws, if selected; SACE: survivor average causal effect among the unobserved always-survivor principal stratum; family code: DEATH_PENDING
Sample size design: family: unresolved: joint survival/function rank/composite simulation OR explicitly identified/bounded survivor-estimand precision; mean-score power cannot be finalized; formula: Use design-specific simulation or independent-unit resampling for the stated estimand and target confidence width/power; preserve all allocation, censoring, clustering and measurement-error structure.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Do not report observed-survivor mean difference as an overall causal treatment-policy effect; no ordinary score after death; restart/switching/incomplete attendance remain strategy consequences.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Do not impute a conventional living score after death or combine death with a missing living assessment.
If a rank/composite is selected, update question, effect measure, estimator and power together; if a survivor estimand is selected, specify its untestable assumptions/sensitivity.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Improved attendance without better quality of life or functional outcomes would fail the bridge’s central claim; weight stability alone would not establish wellbeing benefit.; preanalysis gate: Do not report observed-survivor mean difference as an overall causal treatment-policy effect; no ordinary score after death; restart/switching/incomplete attendance remain strategy consequences.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Do not report observed-survivor mean difference as an overall causal treatment-policy effect; no ordinary score after death; restart/switching/incomplete attendance remain strategy consequences.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose and justify the death/intercurrent-event estimand with patients/clinicians/statistician before registration; align score, validated inability rule, primary question and sample plan.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Do not report observed-survivor mean difference as an overall causal treatment-policy effect; no ordinary score after death; restart/switching/incomplete attendance remain strategy consequences.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/0; question: Among adults referred after independently determined medically necessary treatment cessation, does offering a structured care bridge improve obesity-related quality of life at twelve months compared with routine care and equal research assessment?; aims: Map cessation reasons and early symptoms, access and distress.
Randomize integrated follow-up versus routine post-cessation care.
Identify early signals predicting additional clinical support needs.; design verbatim planning only: Enroll adults after cessation independently chosen by their clinician for a medical indication; research never induces or delays cessation. A pragmatic individual randomized trial compares an offered structured bridge with routine care plus the same research outcome schedule. Retain clinician-directed restart, switching and incomplete attendance in the assigned-strategy estimand; confirm the outcome instrument and intercurrent-event handling before enrollment.; outcomes verbatim: Primary: twelve-month obesity-related quality of life under the prespecified intercurrent-event strategy. Secondary: weight trajectory, eating distress, glycemic measures, physical function, care continuity, treatment restart/switching and financial strain.; analysis verbatim inherited not authoritative if refined: The candidate primary is twelve-month quality of life under randomized strategy assignment. Before registration, define death handling and choose either a survival-incorporated outcome or an explicit survivor estimand with stated assumptions, then align analysis and power. Never impute ordinary quality-of-life scores after death or describe an observed-survivor contrast as the overall causal effect. Retain clinician-led restart/switching in the relevant strategy estimand; handle missing living-participant scores separately.; falsification verbatim: Improved attendance without better quality of life or functional outcomes would fail the bridge’s central claim; weight stability alone would not establish wellbeing benefit.; source ids unverified leads: glp1-weight-loss-rct-review-2025
select-semaglutide-cardiovascular-2023
behavior-stress-sleep-rct-2026; prospectus precedence: Yes; prospectus file: thesis-prospectuses.json; prospectus design and estimand: Use a pragmatic parallel trial with the patient as randomization unit and routine care as comparator. Both arms retain ordinary clinical access and urgent escalation. The candidate primary contrast is the assignment effect on 12-month obesity-related quality of life, including clinician-led restart, switching and incomplete attendance. Quality of life is undefined after death, so the ordinary mean-score formulation remains provisional: before trial registration, choose a survival-incorporated primary outcome or an explicitly defined survivor estimand with its identifying assumptions, then align the question, analysis and power calculation. Comparing observed survivors alone must not be described as an overall causal treatment-policy effect. Define the validated instrument, licensing, scoring direction and minimum important difference before enrollment. The intervention is a package; the trial does not isolate counseling from additional attention.; prospectus analysis plan: For the candidate mean-score outcome, use a baseline-adjusted repeated-measures model with treatment-by-visit terms and clinic effects, targeting the 12-month contrast. Finalize or replace this estimator before enrollment to match the prespecified primary death strategy; a ranked composite, if selected, requires its own effect measure and power plan. Report absolute score differences among living respondents with the appropriate estimand limitations. Restrict covariate adjustment to baseline prognostic factors. Do not adjust the primary estimate for post-randomization weight, attendance or restart. Secondary longitudinal weight and function estimates use a prespecified hierarchy; process mediation remains exploratory with explicit untestable confounding assumptions.; precedence note: Expanded prospectus refinements govern this scaffold; verbatim earlier brief material is retained for traceability, not as a contradictory analysis instruction. F15 is explicitly two arms; F30 has no unjustified symptom noninferiority claim.

M33 — Weight-independent cardiovascular pathways in incretin treatment

Rank: 127
Cluster: Obesity/metabolic medicines
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: MEDIATION; allocation or exposure unit: original outcomes-trial treatment randomization; weight trajectory not randomized; independent sampling unit: participant within independent outcomes trial; repeated measurement unit: longitudinal weight/confounder trajectory and cardiovascular/death event; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: assumption-dependent longitudinal interventional mediation, distinct from the randomized total effect; specific constraint: Sequential mediator-outcome exchangeability and supported interventions are additional assumptions; no conditioning on achieved weight loss; transport cohort separate.
Primary estimand: interventional indirect effect through a prespecified stochastic weight-trajectory distribution
Candidate primary outcome: fixed-horizon cardiovascular outcome with explicit competing-death or joint-outcome definition
Primary finalized: No
Mathematical model: type: longitudinal mediation/g-formula uncertainty and bias simulation with event/mediator history support; formulas: IIE=E[Y^(1,G_1)]-E[Y^(1,G_0)], where G_a draws the prespecified supported weight trajectory distribution under strategy a with the registered history conditioning.
Estimate via longitudinal g-computation integrating treatment-affected confounder and mediator histories; randomized treatment total effects are estimated separately. Choose fixed-horizon event risk/joint outcome before this contrast, including competing death.; variables: Y: fixed-horizon cardiovascular outcome with explicit competing-death or joint-outcome definition; i: participant within independent outcomes trial; t: longitudinal weight/confounder trajectory and cardiovascular/death event; the registered visit/time-window index, not another independent unit; A: original outcomes-trial treatment randomization; weight trajectory not randomized; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: interventional indirect effect through a prespecified stochastic weight-trajectory distribution; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; Y^(a,G b): fixed-horizon potential cardiovascular/joint outcome under treatment a and stochastic mediator intervention G_b; G a: fully specified longitudinal stochastic weight-trajectory intervention, not ordinary adjustment for achieved weight loss; IIE: interventional indirect effect on the fixed additive outcome scale; direction interpreted relative to outcome coding; family code: MEDIATION
Sample size design: family: longitudinal mediation/g-formula uncertainty and bias simulation with event/mediator history support; formula: Use design-specific simulation or independent-unit resampling for the stated estimand and target confidence width/power; preserve all allocation, censoring, clustering and measurement-error structure.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Sequential mediator-outcome exchangeability and supported interventions are additional assumptions; no conditioning on achieved weight loss; transport cohort separate.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Sequential mediator-outcome exchangeability, positivity/support, consistency and correct longitudinal models; randomization alone does not identify mediation.
Define treatment-induced confounder handling, informative mediator missingness and terminal events explicitly.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: If residual benefit vanishes under credible confounding sensitivity analyses, a weight-independent interpretation is unsupported.; preanalysis gate: Sequential mediator-outcome exchangeability and supported interventions are additional assumptions; no conditioning on achieved weight loss; transport cohort separate.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Sequential mediator-outcome exchangeability and supported interventions are additional assumptions; no conditioning on achieved weight loss; transport cohort separate.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Sequential mediator-outcome exchangeability and supported interventions are additional assumptions; no conditioning on achieved weight loss; transport cohort separate.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/126; question: How much of cardiovascular benefit associated with indicated incretin therapy is compatible with pathways beyond achieved weight loss?; aims: Specify causal pathways linking treatment, weight and cardiovascular events.
Estimate longitudinal mediation in accessible randomized participant data.
Test alternative mediators and transportability in an independent cohort.; design verbatim planning only: Seek completed outcomes-trial participant data and repeated mediators, preserving randomized total-effect comparisons. Add an eligible new-user cohort for transportability. Sample size and event availability remain contingent on access; mediation is not independently randomized.; outcomes verbatim: Primary: interventional indirect effects through prespecified weight trajectories. Secondary: effects through blood pressure, glycemia and inflammation, plus subgroup transportability and sensitivity bounds.; analysis verbatim inherited not authoritative if refined: Use longitudinal g-computation or weighting with explicit confounding assumptions. Address discontinuation, missing mediators and competing death; restrict mediator families and report sensitivity bounds.; falsification verbatim: If residual benefit vanishes under credible confounding sensitivity analyses, a weight-independent interpretation is unsupported.; source ids unverified leads: select-semaglutide-cardiovascular-2023
glp1-weight-loss-rct-review-2025; prospectus precedence: No

M34 — Decision quality before obesity pharmacotherapy initiation

Rank: 9
Cluster: Obesity/metabolic medicines
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT_DUAL; allocation or exposure unit: participant randomized balanced medication decision aid versus time-matched information; independent sampling unit: participant with clinic effects; repeated measurement unit: informed-choice/values-concordance assessment and follow-up; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Specify whether conjunction, hierarchy or alpha-split defines success; later treatment choice is not an adjustment covariate and persistence is not universal benefit.
Primary estimand: aid-assignment effect on the registered informed-choice/values-concordance family
Candidate primary outcome: validated informed-choice and values-concordance measures
Primary finalized: No
Mathematical model: type: contrast precision/power by design-matrix or simulation planning; formulas: E(Y_it | A_i, X_i, Y_i0) = alpha_t + z(A_i)^T theta_t + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Delta = sum_t w_t c^T theta_t; estimate by baseline-adjusted marginal standardization, with unit/cluster-appropriate inference.; variables: Y: validated informed-choice and values-concordance measures; i: participant with clinic effects; t: informed-choice/values-concordance assessment and follow-up; the registered visit/time-window index, not another independent unit; A: participant randomized balanced medication decision aid versus time-matched information; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: aid-assignment effect on the registered informed-choice/values-concordance family; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; z(A): prespecified indicators for the actual randomized arms; two arms reduce to one indicator; family code: RCT_DUAL
Sample size design: family: contrast precision/power by design-matrix or simulation planning; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Specify whether conjunction, hierarchy or alpha-split defines success; later treatment choice is not an adjustment covariate and persistence is not universal benefit.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The listed primary endpoints need a conjunction/hierarchy or familywise alpha rule; a favorable result chosen after looking is not confirmation.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Higher uptake without better understanding or values concordance would fail the aid’s purpose; increased regret would signal harmful framing.; preanalysis gate: Specify whether conjunction, hierarchy or alpha-split defines success; later treatment choice is not an adjustment covariate and persistence is not universal benefit.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Specify whether conjunction, hierarchy or alpha-split defines success; later treatment choice is not an adjustment covariate and persistence is not universal benefit.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Specify whether conjunction, hierarchy or alpha-split defines success; later treatment choice is not an adjustment covariate and persistence is not universal benefit.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/8; question: Can a balanced decision aid improve informed treatment choice, realistic expectations and sustained wellbeing when adults consider obesity medication?; aims: Identify expectation mismatches through qualitative patient and clinician interviews.
Randomize a co-designed decision aid against usual counseling.
Assess sustained values concordance, regret and wellbeing regardless of uptake.; design verbatim planning only: Randomize approximately 240 initial-consultation patients across primary/specialty care to a decision aid or time-matched health information. Clinicians prescribe independently; follow everyone for twelve months, oversampling lower-literacy and underinsured participants.; outcomes verbatim: Primary: validated informed-choice and values-concordance measures. Secondary: decisional regret, quality of life, understanding of adverse effects, financial burden and treatment persistence.; analysis verbatim inherited not authoritative if refined: Use clinic-adjusted assignment effects and baseline knowledge. Prespecify literacy/insurance interactions, correct primary multiplicity and measure clinician-mediated contamination.; falsification verbatim: Higher uptake without better understanding or values concordance would fail the aid’s purpose; increased regret would signal harmful framing.; source ids unverified leads: glp1-weight-loss-rct-review-2025
select-semaglutide-cardiovascular-2023
fda-fdalabel-drug-labeling-2026; prospectus precedence: No

M35 — Net functional benefit in older adults with obesity and frailty

Rank: 41
Cluster: Obesity/metabolic medicines
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: OBS_RANK; allocation or exposure unit: clinician-selected new incretin versus eligible active alternative care; independent sampling unit: participant; repeated measurement unit: twelve-month survival/function status; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: observational comparison/association; causal interpretation is conditional on explicit exchangeability, positivity, consistency and observation assumptions; specific constraint: Separate mortality and living-function distribution; no imputation of living function after death; survivor-specific change remains conditional and indication confounding unresolved.
Primary estimand: standardized joint survival/function distribution contrast under the baseline care strategies, conditional on exchangeability
Candidate primary outcome: instrumental daily-living status ordered with death below any living score
Primary finalized: No
Mathematical model: type: observational overlap-weighted/g-method precision simulation with effective independent units, outcome/event frequency and residual-bias scenarios; formulas: W_i(t) = product_(k<=t) P(A_ik | A_i,<k,X_i0)/P(A_ik | H_ik), optionally multiplied by separately justified observation weights.
m_a = E_X[m(a,X)] for a baseline strategy, or the sequential g-formula for sustained strategies; Delta=m_1-m_0. Fit a weighted, baseline-aligned mean/risk model matching Y rather than assume the treatment coefficient is causal.
Let Z^a be the ordered twelve-month survival/function status, with death below any living state. Report F_a(z)=Pr(Z^a<=z) and an independent-draw superiority contrast Pr(Z_1>Z_0)+0.5Pr(Z_1=Z_0); it is not the probability that a particular person benefits.; variables: Y: instrumental daily-living status ordered with death below any living score; i: participant; t: twelve-month survival/function status; the registered visit/time-window index, not another independent unit; A: clinician-selected new incretin versus eligible active alternative care; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: standardized joint survival/function distribution contrast under the baseline care strategies, conditional on exchangeability; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; H ik: measured predecision history including prior treatment and time-varying covariates; m(a,X): outcome mean/risk model under specified treatment strategy and observed covariates; m a: standardized mean/risk under a, identified only under the listed assumptions; k: ordered treatment decision time; Z: joint death/living-function outcome with fixed patient-informed ordering; Z 1,Z 0: independent draws from standardized strategy distributions, not paired person-specific counterfactuals; F a: standardized cumulative distribution; family code: OBS_RANK
Sample size design: family: observational overlap-weighted/g-method precision simulation with effective independent units, outcome/event frequency and residual-bias scenarios; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Separate mortality and living-function distribution; no imputation of living function after death; survivor-specific change remains conditional and indication confounding unresolved.
Common eligibility/time zero and a defensible active comparator; treatment choice is not randomized.
Conditional exchangeability (including time-varying feedback where relevant), positivity and correct observation/treatment/outcome models are untestable or only partly testable; observed overlap does not prove them.
Do not adjust for posttreatment variables as baseline confounders; define a separate mediation/per-protocol estimand before using g-methods.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: If functional reserve does not modify outcomes or early strength adds no predictive value, the proposed selection framework fails.; preanalysis gate: Separate mortality and living-function distribution; no imputation of living function after death; survivor-specific change remains conditional and indication confounding unresolved.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Separate mortality and living-function distribution; no imputation of living function after death; survivor-specific change remains conditional and indication confounding unresolved.
Overlap/effective-sample-size and weight-tail diagnostics with declared truncation sensitivity; report population changes caused by trimming.
Quantitative unmeasured-confounding and informative-loss sensitivity; negative controls/falsification windows where scientifically defensible.
Compare g-computation/weighting/doubly robust estimates on the same target population and time zero, without interpreting agreement as proof.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Separate mortality and living-function distribution; no imputation of living function after death; survivor-specific change remains conditional and indication confounding unresolved.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/40; question: Which older adults experience improved daily function rather than increased vulnerability after clinically indicated incretin treatment?; aims: Map frailty, nutrition and daily function around initiation.
Emulate an eligibility-aligned active-comparator target trial.
Externally validate early warnings of functional deterioration.; design verbatim planning only: Enroll approximately 250 adults over 70 through geriatric/obesity services, supplementing linked records where available. Compare clinically selected new users with eligible alternative-care initiators; measure function before treatment and at three, six and twelve months.; outcomes verbatim: Primary: Twelve-month instrumental daily-living status, with death ranked worse than any living functional score under a prespecified patient-informed ordering. Secondary: Survivor-specific change in daily-living function, mobility, malnutrition, falls, hospital-free days, strength and patient-defined treatment goals.; analysis verbatim inherited not authoritative if refined: Align eligibility and time zero; use active comparators and weighting. Estimate the twelve-month joint survival/function distribution, report mortality separately and do not impute living function after death. Address indication bias and informative missingness among living participants; label survivor-specific change as conditional. Prespecify reserve strata and report residual confounding.; falsification verbatim: If functional reserve does not modify outcomes or early strength adds no predictive value, the proposed selection framework fails.; source ids unverified leads: glp1-weight-loss-rct-review-2025
resistance-sarcopenia-meta-2026
select-semaglutide-cardiovascular-2023; prospectus precedence: No

M36 — Product provenance and preventable harm in online obesity medicines

Rank: 142
Cluster: Obesity/metabolic medicines
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: AUDIT; allocation or exposure unit: product sampled through defined risk-stratified provenance/channel frame; independent sampling unit: independent product lot with vendor/batch clustering; repeated measurement unit: technical analytical replicate and provenance record; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: analytical quality prevalence within the defined sampling frame, not clinical harm incidence or a randomized treatment effect; specific constraint: Risk-stratified convenience samples cannot estimate all-market prevalence without inclusion probabilities; quality failure is not harm incidence.
Primary estimand: sampling-frame-weighted prevalence of prespecified analytical quality failure
Candidate primary outcome: lot-level analytical failure indicator
Primary finalized: No
Mathematical model: type: stratified clustered prevalence/audit precision with analytical-error and sampling-selection sensitivity; formulas: p_hat = sum_l w_l F_l / sum_l w_l; w_l=1/pi_l when inclusion probabilities are known. Cluster uncertainty by supplier/batch and propagate validated assay classification error.
If sampling is purposive with unknown pi_l, report the finite acquired-sample failure proportion and bounds; no probability-sample all-market prevalence claim.; variables: Y: lot-level analytical failure indicator; i: independent product lot with vendor/batch clustering; t: technical analytical replicate and provenance record; the registered visit/time-window index, not another independent unit; A: product sampled through defined risk-stratified provenance/channel frame; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: sampling-frame-weighted prevalence of prespecified analytical quality failure; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; l: independent lot within the explicit acquisition frame; F l: prespecified lot-level analytical failure indicator; pi l: lot inclusion probability in that frame; not assumed known for convenience purchases; w l: design weight if justified; family code: AUDIT
Sample size design: family: stratified clustered prevalence/audit precision with analytical-error and sampling-selection sensitivity; formula: Use design-specific simulation or independent-unit resampling for the stated estimand and target confidence width/power; preserve all allocation, censoring, clustering and measurement-error structure.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Risk-stratified convenience samples cannot estimate all-market prevalence without inclusion probabilities; quality failure is not harm incidence.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Explicit sampling frame and positive known inclusion probabilities for population weighting; vendor/batch dependence and assay uncertainty retained.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Equivalent verified quality across provenance groups would weaken the proposed discriminator; anecdotal harm alone cannot confirm contamination.; preanalysis gate: Risk-stratified convenience samples cannot estimate all-market prevalence without inclusion probabilities; quality failure is not harm incidence.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Risk-stratified convenience samples cannot estimate all-market prevalence without inclusion probabilities; quality failure is not harm incidence.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Risk-stratified convenience samples cannot estimate all-market prevalence without inclusion probabilities; quality failure is not harm incidence.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/141; question: Which product and distribution characteristics identify preventable safety problems among obesity medicines obtained outside ordinary licensed dispensing?; aims: Construct a reproducible advertised-product and dispensing-channel sampling frame.
Independently test lawfully obtained products for quality failures.
Combine surveillance and interviews to identify detection and access gaps.; design verbatim planning only: Legally obtain approximately 80 risk-stratified products through a regulatory/laboratory partnership and interview around 60 patients/clinicians. Compare explicit provenance categories without administering products. Report failure prevalence only within the defined sampling frame, never unsupported harm incidence.; outcomes verbatim: Primary: prevalence of prespecified analytical quality failures within the defined sampling frame. Secondary: provenance completeness, labeling discordance, access barriers and adverse-event reporting completeness.; analysis verbatim inherited not authoritative if refined: Weight the sampling frame and quantify vendor/batch clustering and selection uncertainty. Separate reporting signals from causal attribution; preregister analytical thresholds and disclosure rules.; falsification verbatim: Equivalent verified quality across provenance groups would weaken the proposed discriminator; anecdotal harm alone cannot confirm contamination.; source ids unverified leads: fda-compounding-questions-2026
fda-unapproved-drugs-2026
peptide-immunogenicity-safety-review-2025; prospectus precedence: No

M37 — Metformin, exercise adaptation and baseline metabolic state

Rank: 164
Cluster: Obesity/metabolic medicines
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT_INT; allocation or exposure unit: original authorized exercise-metformin trial assignment or future approved placebo trial; independent sampling unit: participant; repeated measurement unit: aerobic-capacity and metabolic assessment; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Prefer existing trials; no causal effect modification by a posttreatment insulin-sensitivity measure; clinical equipoise/access remain unconfirmed.
Primary estimand: metformin assignment effect and one baseline insulin-resistance interaction on aerobic adaptation
Candidate primary outcome: objectively measured aerobic capacity at six months
Primary finalized: No
Mathematical model: type: interaction-specific precision/power with modifier distribution and reliability; simulate the actual interaction; formulas: E(Y_it | A_i, B_i, X_i) = alpha_t + tau_t A_i + eta_t B_i + kappa_t A_i B_i + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Treatment contrast at baseline B=b is tau_t + kappa_t b; the interaction target is kappa_t, not a difference in within-subgroup p-values.; variables: Y: objectively measured aerobic capacity at six months; i: participant; t: aerobic-capacity and metabolic assessment; the registered visit/time-window index, not another independent unit; A: original authorized exercise-metformin trial assignment or future approved placebo trial; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: metformin assignment effect and one baseline insulin-resistance interaction on aerobic adaptation; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; B: baseline insulin-resistance measure under independently fixed scale/cutoff; centering/coding and any measurement-error model fixed before evaluation; eta: baseline modifier association; kappa: assignment-by-baseline-modifier interaction coefficient on the stated scale; family code: RCT_INT
Sample size design: family: interaction-specific precision/power with modifier distribution and reliability; simulate the actual interaction; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure
modifier prevalence/distribution and variance
modifier reliability and overlap
interaction size or precision target; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Prefer existing trials; no causal effect modification by a posttreatment insulin-sensitivity measure; clinical equipoise/access remain unconfirmed.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The modifier must be measured before assignment/initiation; interaction is scale dependent and requires support within modifier strata.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: No reproducible interaction would reject metabolic-state personalization; favorable signaling without functional or metabolic benefit would not establish geroprotection.; preanalysis gate: Prefer existing trials; no causal effect modification by a posttreatment insulin-sensitivity measure; clinical equipoise/access remain unconfirmed.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Prefer existing trials; no causal effect modification by a posttreatment insulin-sensitivity measure; clinical equipoise/access remain unconfirmed.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Vary modifier measurement-error assumptions and retain a continuous modifier unless a cutoff was independently fixed.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Prefer existing trials; no causal effect modification by a posttreatment insulin-sensitivity measure; clinical equipoise/access remain unconfirmed.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/163; question: Does baseline insulin resistance modify the effect of metformin on functional adaptation to exercise in adults without diabetes?; aims: Characterize baseline insulin sensitivity and energy-related function.
Compare supervised metformin/placebo alongside matched exercise where ethical.
Test one metabolic-state interaction for fitness and glucose outcomes.; design verbatim planning only: Prefer existing authorized exercise–metformin trial data. Alternatively, approximately 120 adults could enter an ethics-approved six-month placebo-controlled exercise trial where genuine equipoise exists and indicated care is preserved. Regulatory approval and clinical oversight precede recruitment.; outcomes verbatim: Primary: change in objectively measured aerobic capacity. Secondary: insulin sensitivity, strength, fatigue, training adherence, gastrointestinal effects and B12 status.; analysis verbatim inherited not authoritative if refined: Prespecify one insulin-resistance interaction and a restricted outcome hierarchy. Use assignment effects, track attendance and missing tests, and avoid equating mediation with proof.; falsification verbatim: No reproducible interaction would reject metabolic-state personalization; favorable signaling without functional or metabolic benefit would not establish geroprotection.; source ids unverified leads: metformin-antiaging-uncertainty-2025
metabolic-flexibility-t2d-meta-2025
nih-ods-b12-2025; prospectus precedence: No

M38 — Appetite relief, food reward and eating agency during incretin treatment

Rank: 110
Cluster: Obesity/metabolic medicines
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: OBS_LONG; allocation or exposure unit: clinician-selected new incretin treatment versus active lifestyle comparator; independent sampling unit: participant; repeated measurement unit: early cue task/mobile report and later agency visit; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: observational comparison/association; causal interpretation is conditional on explicit exchangeability, positivity, consistency and observation assumptions; specific constraint: Nausea, cue reactivity and weight are possible mediators; their adjustment does not prove agency benefit independent of them.
Primary estimand: standardized eating-agency change contrast under comparable baseline strategies, conditional on exchangeability
Candidate primary outcome: validated eating-agency score at nine months
Primary finalized: No
Mathematical model: type: observational overlap-weighted/g-method precision simulation with effective independent units, outcome/event frequency and residual-bias scenarios; formulas: W_i(t) = product_(k<=t) P(A_ik | A_i,<k,X_i0)/P(A_ik | H_ik), optionally multiplied by separately justified observation weights.
m_a = E_X[m(a,X)] for a baseline strategy, or the sequential g-formula for sustained strategies; Delta=m_1-m_0. Fit a weighted, baseline-aligned mean/risk model matching Y rather than assume the treatment coefficient is causal.; variables: Y: validated eating-agency score at nine months; i: participant; t: early cue task/mobile report and later agency visit; the registered visit/time-window index, not another independent unit; A: clinician-selected new incretin treatment versus active lifestyle comparator; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: standardized eating-agency change contrast under comparable baseline strategies, conditional on exchangeability; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; H ik: measured predecision history including prior treatment and time-varying covariates; m(a,X): outcome mean/risk model under specified treatment strategy and observed covariates; m a: standardized mean/risk under a, identified only under the listed assumptions; k: ordered treatment decision time; family code: OBS_LONG
Sample size design: family: observational overlap-weighted/g-method precision simulation with effective independent units, outcome/event frequency and residual-bias scenarios; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Nausea, cue reactivity and weight are possible mediators; their adjustment does not prove agency benefit independent of them.
Common eligibility/time zero and a defensible active comparator; treatment choice is not randomized.
Conditional exchangeability (including time-varying feedback where relevant), positivity and correct observation/treatment/outcome models are untestable or only partly testable; observed overlap does not prove them.
Do not adjust for posttreatment variables as baseline confounders; define a separate mediation/per-protocol estimand before using g-methods.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Cue reduction that tracks only nausea or fails to improve agency would contradict the proposed functional reward pathway.; preanalysis gate: Nausea, cue reactivity and weight are possible mediators; their adjustment does not prove agency benefit independent of them.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Nausea, cue reactivity and weight are possible mediators; their adjustment does not prove agency benefit independent of them.
Overlap/effective-sample-size and weight-tail diagnostics with declared truncation sensitivity; report population changes caused by trimming.
Quantitative unmeasured-confounding and informative-loss sensitivity; negative controls/falsification windows where scientifically defensible.
Compare g-computation/weighting/doubly robust estimates on the same target population and time zero, without interpreting agreement as proof.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Nausea, cue reactivity and weight are possible mediators; their adjustment does not prove agency benefit independent of them.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/109; question: Does reduced food-cue reactivity during indicated incretin treatment translate into improved eating agency independently of nausea and weight change?; aims: Validate distinctions among intrusive thoughts, hunger, aversion and agency.
Compare eligible medication initiators with active lifestyle comparators.
Test temporal cue-response relationships with functioning and wellbeing.; design verbatim planning only: Follow approximately 150 treatment-naive eligible adults across clinician-selected medication and active lifestyle-comparator cohorts for nine months. Match eligibility and history; collect dense early mobile reports and behavioral cue tasks. Prescribing remains independent of research.; outcomes verbatim: Primary: validated eating-agency change. Secondary: food enjoyment, intrusive thoughts, cue-driven eating frequency, nausea, nutritional adequacy and quality of life.; analysis verbatim inherited not authoritative if refined: Align time zero and use propensity methods. Adjust baseline distress, diet, sleep and socioeconomic context; model nausea/weight as mediators and report residual confounding.; falsification verbatim: Cue reduction that tracks only nausea or fails to improve agency would contradict the proposed functional reward pathway.; source ids unverified leads: glp1-weight-loss-rct-review-2025
dopamine-food-reward-review-2026; prospectus precedence: No

M39 — Dietary support for gastrointestinal tolerance during obesity treatment

Rank: 2
Cluster: Obesity/metabolic medicines
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT_LONG; allocation or exposure unit: clinically treated initiator randomized proactive dietitian support versus routine information; independent sampling unit: participant stratified by clinic/medicine; repeated measurement unit: weekly early and monthly later symptom assessment; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Define time weights/horizon before analysis; clinician-directed dose/switch/stop retained under treatment policy; persistence alone does not imply benefit.
Primary estimand: dietitian-support offer effect on prespecified gastrointestinal-interference time summary
Candidate primary outcome: validated gastrointestinal symptom-interference score
Primary finalized: No
Mathematical model: type: contrast precision/power by design-matrix or simulation planning; formulas: E(Y_it | A_i, X_i, Y_i0) = alpha_t + z(A_i)^T theta_t + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Delta = sum_t w_t c^T theta_t; estimate by baseline-adjusted marginal standardization, with unit/cluster-appropriate inference.; variables: Y: validated gastrointestinal symptom-interference score; i: participant stratified by clinic/medicine; t: weekly early and monthly later symptom assessment; the registered visit/time-window index, not another independent unit; A: clinically treated initiator randomized proactive dietitian support versus routine information; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: dietitian-support offer effect on prespecified gastrointestinal-interference time summary; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; z(A): prespecified indicators for the actual randomized arms; two arms reduce to one indicator; family code: RCT_LONG
Sample size design: family: contrast precision/power by design-matrix or simulation planning; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Define time weights/horizon before analysis; clinician-directed dose/switch/stop retained under treatment policy; persistence alone does not imply benefit.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: More contacts without less symptom interference or better nutritional adequacy would fail the support strategy; persistence despite worsening wellbeing would be an adverse result.; preanalysis gate: Define time weights/horizon before analysis; clinician-directed dose/switch/stop retained under treatment policy; persistence alone does not imply benefit.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Define time weights/horizon before analysis; clinician-directed dose/switch/stop retained under treatment policy; persistence alone does not imply benefit.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Define time weights/horizon before analysis; clinician-directed dose/switch/stop retained under treatment policy; persistence alone does not imply benefit.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/1; question: Can early structured dietary support reduce gastrointestinal burden and preserve nutritional adequacy during clinician-managed incretin therapy?; aims: Map early symptoms against intake and daily functioning.
Randomize proactive dietitian support versus routine clinic information.
Assess nutrition, wellbeing and clinically appropriate treatment continuity.; design verbatim planning only: Randomize approximately 200 clinically treated initiators to proactive dietitian support or routine information for six months. Maintain identical prescriber access; record clinician-directed medication changes and collect weekly early symptoms followed by monthly assessments.; outcomes verbatim: Primary: validated gastrointestinal symptom interference. Secondary: nutrient adequacy, dehydration-related clinical contacts, quality of life, clinically appropriate persistence and healthcare burden.; analysis verbatim inherited not authoritative if refined: Stratify by clinic and medicine. Use a treatment-policy estimand, assess informative missingness and correct secondary multiplicity; persistence alone never defines benefit.; falsification verbatim: More contacts without less symptom interference or better nutritional adequacy would fail the support strategy; persistence despite worsening wellbeing would be an adverse result.; source ids unverified leads: glp1-weight-loss-rct-review-2025
amycretin-phase1b2a-2025
fda-fdalabel-drug-labeling-2026; prospectus precedence: No

M40 — Coverage expansion, treatment access and real-world metabolic benefit

Rank: 64
Cluster: Obesity/metabolic medicines
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: DID; allocation or exposure unit: eligible population exposed to staggered insurance-coverage policy expansion; independent sampling unit: independent policy jurisdiction/insurer and adoption cohort; repeated measurement unit: person laboratory record by policy/calendar period; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: quasi-experimental policy/update association with conditional causal interpretation; observational timing is not randomization; specific constraint: No naive two-way fixed-effect average under heterogeneous timing; not-yet/never-treated comparators, parallel-trend plausibility and laboratory-observation selection required.
Primary estimand: group-time average coverage-expansion effect on glycemic control among eligible high-risk adults
Candidate primary outcome: prespecified glycemic-control endpoint in a defined eligible population
Primary finalized: No
Mathematical model: type: staggered adoption/event-study precision with independent adoption units and bias/parallel-trend sensitivity; formulas: ATT(g,t)=E[Y_t-Y_(g-1) | G=g] - E_w[Y_t-Y_(g-1) | G>t or G=never], with predeclared eligibility/comparator weights.
Aggregate only identified group-time effects with explicit nonnegative target weights. Estimate pre-adoption leads and placebo dates; no favorable pretrend test proves parallel untreated trends.; variables: Y: prespecified glycemic-control endpoint in a defined eligible population; i: independent policy jurisdiction/insurer and adoption cohort; t: person laboratory record by policy/calendar period; the registered visit/time-window index, not another independent unit; A: eligible population exposed to staggered insurance-coverage policy expansion; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: group-time average coverage-expansion effect on glycemic control among eligible high-risk adults; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; G: first policy-adoption cohort (M40) or documented device/version update cohort (E04); g,t: adoption cohort and outcome calendar period; E w: comparison-group expectation after justified pre-exposure covariate weighting; ATT(g,t): group-time average effect, causally interpreted only under conditional parallel untreated trends/no anticipation/appropriate interference assumptions; family code: DID
Sample size design: family: staggered adoption/event-study precision with independent adoption units and bias/parallel-trend sensitivity; formula: Use design-specific simulation or independent-unit resampling for the stated estimand and target confidence width/power; preserve all allocation, censoring, clustering and measurement-error structure.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: No naive two-way fixed-effect average under heterogeneous timing; not-yet/never-treated comparators, parallel-trend plausibility and laboratory-observation selection required.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Common eligibility/time zero, no anticipation, credible untreated-trend comparability, stable composition and no unaddressed simultaneous policies/updates.
Use independent adoption/update units for uncertainty; people and readings do not create new independent adoption events.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Absent credible parallel pretrends would invalidate causal claims; increased dispensing without improved health or reduced burden would weaken the access-benefit hypothesis.; preanalysis gate: No naive two-way fixed-effect average under heterogeneous timing; not-yet/never-treated comparators, parallel-trend plausibility and laboratory-observation selection required.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: No naive two-way fixed-effect average under heterogeneous timing; not-yet/never-treated comparators, parallel-trend plausibility and laboratory-observation selection required.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Vary never-treated versus not-yet-treated comparisons, admissible pretrends, event windows, concurrent-policy exclusions and missing-laboratory observation weights.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
No naive two-way fixed-effect average under heterogeneous timing; not-yet/never-treated comparators, parallel-trend plausibility and laboratory-observation selection required.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/63; question: When insurance coverage for obesity medicines expands, who gains meaningful health benefit and who remains excluded?; aims: Map policy timing, eligibility and out-of-pocket requirements.
Estimate coverage effects using aligned populations and policy comparisons.
Test rurality, deprivation and continuity as benefit modifiers.; design verbatim planning only: Obtain linked claims, prescribing and laboratory records around staggered coverage expansions, plus an approximately 200-person outcomes substudy. Compare eligible populations with similar pretrends and documented concurrent policies; actual cohort size depends on approved access.; outcomes verbatim: Primary: policy effect on glycemic control among eligible high-risk adults. Secondary: treatment continuity, obesity-related quality of life, avoidable clinical events, financial burden and distributional effects.; analysis verbatim inherited not authoritative if refined: Use staggered difference-in-differences, event-study diagnostics and falsification dates. Address composition, spillovers, missing laboratories and concurrent policies; control subgroup multiplicity and report uncertainty.; falsification verbatim: Absent credible parallel pretrends would invalidate causal claims; increased dispensing without improved health or reduced burden would weaken the access-benefit hypothesis.; source ids unverified leads: select-semaglutide-cardiovascular-2023
glp1-weight-loss-rct-review-2025; prospectus precedence: No

N01 — Circadian phase rather than clock time for personalized light guidance

Rank: 150
Cluster: Sleep/circadian/breathing
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT_INT; allocation or exposure unit: participant randomized phase-anchored versus clock-time guidance; independent sampling unit: participant; repeated measurement unit: daytime-impairment visit and phase specimen; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Adequate sleep-opportunity advice is common; achieved phase shift is postassignment, and phase measurement error weakens interaction estimation.
Primary estimand: guidance assignment effect and one baseline phase-mismatch interaction on daytime impairment
Candidate primary outcome: prespecified daytime-impairment scale at eight weeks
Primary finalized: No
Mathematical model: type: interaction-specific precision/power with modifier distribution and reliability; simulate the actual interaction; formulas: E(Y_it | A_i, B_i, X_i) = alpha_t + tau_t A_i + eta_t B_i + kappa_t A_i B_i + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Treatment contrast at baseline B=b is tau_t + kappa_t b; the interaction target is kappa_t, not a difference in within-subgroup p-values.; variables: Y: prespecified daytime-impairment scale at eight weeks; i: participant; t: daytime-impairment visit and phase specimen; the registered visit/time-window index, not another independent unit; A: participant randomized phase-anchored versus clock-time guidance; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: guidance assignment effect and one baseline phase-mismatch interaction on daytime impairment; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; B: baseline phase-mismatch estimate with stated circadian-phase measurement uncertainty; centering/coding and any measurement-error model fixed before evaluation; eta: baseline modifier association; kappa: assignment-by-baseline-modifier interaction coefficient on the stated scale; family code: RCT_INT
Sample size design: family: interaction-specific precision/power with modifier distribution and reliability; simulate the actual interaction; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure
modifier prevalence/distribution and variance
modifier reliability and overlap
interaction size or precision target; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Adequate sleep-opportunity advice is common; achieved phase shift is postassignment, and phase measurement error weakens interaction estimation.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The modifier must be measured before assignment/initiation; interaction is scale dependent and requires support within modifier strata.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Equivalent function despite demonstrably different phase alignment would undermine the proposed precision advantage.; preanalysis gate: Adequate sleep-opportunity advice is common; achieved phase shift is postassignment, and phase measurement error weakens interaction estimation.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Adequate sleep-opportunity advice is common; achieved phase shift is postassignment, and phase measurement error weakens interaction estimation.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Vary modifier measurement-error assumptions and retain a continuous modifier unless a cutoff was independently fixed.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Adequate sleep-opportunity advice is common; achieved phase shift is postassignment, and phase measurement error weakens interaction estimation.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/149; question: Does phase-anchored light guidance improve daytime function more than identical guidance anchored to preferred clock times?; aims: Quantify disagreement between reported chronotype and measured dim-light melatonin onset.
Randomize timing rules while matching guidance intensity and contact.
Test whether achieved phase alignment mediates functional improvement.; design verbatim planning only: Planning assumption: 80 adults with delayed, irregular sleep and no untreated sleep disorder; two-week baseline followed by eight-week parallel randomization. Both arms receive adequate sleep-opportunity advice and professionally screened light guidance.; outcomes verbatim: Primary: prespecified daytime impairment scale. Secondary: actigraphic sleep midpoint, melatonin phase, adherence, mood and psychomotor vigilance.; analysis verbatim inherited not authoritative if refined: Intention-to-treat models adjust baseline impairment and season; test one phase-mismatch interaction. Treat mediation as assumption-dependent and control secondary-outcome false discovery.; falsification verbatim: Equivalent function despite demonstrably different phase alignment would undermine the proposed precision advantage.; source ids unverified leads: light-circadian-review
dim-light-insomnia-rct-2024; prospectus precedence: No

N02 — When sleep dashboards worsen sleep: causal testing of score disclosure

Rank: 48
Cluster: Sleep/circadian/breathing
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT; allocation or exposure unit: participant randomized truthful sleep-score disclosure format; independent sampling unit: participant; repeated measurement unit: worry, sleep diary and independent actigraphy visit; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Unauthorized checking dilutes an offer effect; objective sleep and anxiety-related attrition provide independent benefit/harm checks.
Primary estimand: hiding detailed nightly scores versus full disclosure effect on sleep-related preoccupation
Candidate primary outcome: sleep-related preoccupation score at ten weeks
Primary finalized: No
Mathematical model: type: contrast precision/power by design-matrix or simulation planning; formulas: E(Y_it | A_i, X_i, Y_i0) = alpha_t + z(A_i)^T theta_t + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Delta = sum_t w_t c^T theta_t; estimate by baseline-adjusted marginal standardization, with unit/cluster-appropriate inference.; variables: Y: sleep-related preoccupation score at ten weeks; i: participant; t: worry, sleep diary and independent actigraphy visit; the registered visit/time-window index, not another independent unit; A: participant randomized truthful sleep-score disclosure format; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: hiding detailed nightly scores versus full disclosure effect on sleep-related preoccupation; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; z(A): prespecified indicators for the actual randomized arms; two arms reduce to one indicator; family code: RCT
Sample size design: family: contrast precision/power by design-matrix or simulation planning; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Unauthorized checking dilutes an offer effect; objective sleep and anxiety-related attrition provide independent benefit/harm checks.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: No reduction in worry, or improved worry accompanied by worse independently measured sleep, would challenge the proposed benefit.; preanalysis gate: Unauthorized checking dilutes an offer effect; objective sleep and anxiety-related attrition provide independent benefit/harm checks.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Unauthorized checking dilutes an offer effect; objective sleep and anxiety-related attrition provide independent benefit/harm checks.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Unauthorized checking dilutes an offer effect; objective sleep and anxiety-related attrition provide independent benefit/harm checks.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/47; question: Does hiding nightly sleep scores reduce insomnia-related worry without reducing useful self-management?; aims: Identify which dashboard interpretations predict next-day sleep worry.
Randomize granular versus summary-only feedback using identical devices.
Estimate whether worry changes precede altered behavior or symptoms.; design verbatim planning only: Planning assumption: 100 habitual wearable users with subclinical insomnia; ten-week parallel trial after baseline. Interfaces display truthful device outputs with different detail; participants retain access to clinical evaluation.; outcomes verbatim: Primary: sleep-related preoccupation score. Secondary: insomnia severity, diary sleep efficiency, actigraphy, time spent checking and daytime functioning.; analysis verbatim inherited not authoritative if refined: Use baseline-adjusted longitudinal models, prespecified baseline-worry moderation and sensitivity analyses for unauthorized app checking. Assess attrition by anxiety and digital literacy.; falsification verbatim: No reduction in worry, or improved worry accompanied by worse independently measured sleep, would challenge the proposed benefit.; source ids unverified leads: wearables-living-review
belief-effects-health-review-2026
va-dod-insomnia-guideline-2025; prospectus precedence: No

N03 — Equitable sleep-apnea screening and postdiagnosis treatment linkage

Rank: 47
Cluster: Sleep/circadian/breathing
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: DIAG_RCT; allocation or exposure unit: screening validation in referred adults, followed by separate individual navigation randomization among clinically diagnosed eligible patients; independent sampling unit: participant within each stage; nested stages are not interchangeable denominators; repeated measurement unit: paired oximetry/PSG record, then eight-week treatment-initiation assessment; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: diagnostic/benchmark agreement or classification validity; reader/interface/navigation effects are causal only in their separately randomized component; specific constraint: Consumer result never determines diagnosis or eligibility; stage-two effect is conditional on clinical eligibility and cannot be generalized to all screened adults.
Primary estimand: screening calibration/sensitivity and, separately, navigation assignment effect on recommended-treatment initiation
Candidate primary outcome: PSG-defined apnea status; documented clinician-recommended treatment initiation by eight weeks
Primary finalized: No
Mathematical model: type: two distinct stages: diagnostic cases/noncases precision, then randomized initiation risk-difference precision among eligible patients; formulas: Sensitivity = TP/(TP+FN); false-alert burden = FP/(registered nonevent time/opportunities); report participant/family-clustered uncertainty and failed/indeterminate cases.
For probability scores, report calibration and a fixed proper score in participant/source-disjoint evaluation, with uncertain reference labels modeled or bounded.
Stage two: RD_nav=Pr(initiation by 8 weeks | navigation offer)-Pr(initiation by 8 weeks | usual follow-up) among independently clinically diagnosed eligible randomized patients.; variables: Y: PSG-defined apnea status; documented clinician-recommended treatment initiation by eight weeks; i: participant within each stage; nested stages are not interchangeable denominators; t: paired oximetry/PSG record, then eight-week treatment-initiation assessment; the registered visit/time-window index, not another independent unit; A: screening validation in referred adults, followed by separate individual navigation randomization among clinically diagnosed eligible patients; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: screening calibration/sensitivity and, separately, navigation assignment effect on recommended-treatment initiation; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; TP,FN,FP: true-positive, false-negative and false-positive counts relative to the independently defined reference and matching window, not an assumed perfect gold standard; reference: per-thesis independent reference, including adjudication uncertainty; nonevent time/opportunities: fixed at-risk denominator; event specificity and false alerts per hour are distinct targets; RD nav: navigation-assignment risk difference in the stage-two eligible population, not the screening population; family code: DIAG_RCT
Sample size design: family: two distinct stages: diagnostic cases/noncases precision, then randomized initiation risk-difference precision among eligible patients; formula: An independent-case rough starting point is n_case ~= z_(1-alpha*/2)^2 Se(1-Se)/h_Se^2; replace with interval-based/clustered simulation for the actual design. Plan noncases/observation time and reference failures separately.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; Se: plausible sensitivity from prior evidence or pilot, not an asserted study result; h Se: desired sensitivity interval half-width; n case: independent positive cases; repeated events within a person require correlation adjustment; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Consumer result never determines diagnosis or eligibility; stage-two effect is conditional on clinical eligibility and cannot be generalized to all screened adults.
Reference independence, fixed thresholds/window and an explicit denominator; incorporation/verification bias can invalidate apparent accuracy.
Participant/family-level test splits and paired comparisons; report abstention, assay failure and indeterminate truth rather than force every case into a binary label.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: No improvement in indicated treatment initiation would undermine navigation benefit; persistent subgroup errors would undermine equitable screening, independently.; preanalysis gate: Consumer result never determines diagnosis or eligibility; stage-two effect is conditional on clinical eligibility and cannot be generalized to all screened adults.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Consumer result never determines diagnosis or eligibility; stage-two effect is conditional on clinical eligibility and cannot be generalized to all screened adults.
Vary plausible reference misclassification, verification-selection and event-matching window under a prespecified sensitivity family.
Report subgroup usable-data coverage and calibration; avoid interpreting an imprecise null difference as equivalence.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Consumer result never determines diagnosis or eligibility; stage-two effect is conditional on clinical eligibility and cannot be generalized to all screened adults.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/46; question: Does consumer oximetry perform equitably, and can navigation after confirmed apnea diagnosis improve initiation of clinician-recommended treatment?; aims: Measure screening accuracy against polysomnography across skin-pigmentation and perfusion strata.
Quantify subgroup calibration and false-positive and false-negative classification.
Randomize postdiagnosis navigation to test improvement in treatment initiation.; design verbatim planning only: Planning assumption: 150 referred adults receive concurrent consumer oximetry and clinical polysomnography. After clinical diagnosis, approximately 80 participants offered treatment enter nested randomization to navigation or usual follow-up. Research screening results never determine diagnosis, treatment eligibility or clinical recommendations.; outcomes verbatim: Co-primary stages: screening calibration against polysomnography, then documented initiation of clinician-recommended treatment within eight weeks. Secondary: classification errors, costs, symptoms, acceptability and informed treatment choice.; analysis verbatim inherited not authoritative if refined: Estimate diagnostic accuracy with uncertainty intervals and prespecified subgroup contrasts. Separately estimate intention-to-treat navigation effects among diagnosed, treatment-eligible participants; account for treatment preference and report diagnostic selection limits.; falsification verbatim: No improvement in indicated treatment initiation would undermine navigation benefit; persistent subgroup errors would undermine equitable screening, independently.; source ids unverified leads: sleep-apnea-oximetry-meta-2025
wearables-living-review
va-dod-insomnia-guideline-2025; prospectus precedence: No

N04 — Separating sleep-stage explanations from effective insomnia care

Rank: 108
Cluster: Sleep/circadian/breathing
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT_PROP; allocation or exposure unit: participant randomized add-on uncertainty education before otherwise identical CBT-I; independent sampling unit: participant with therapist clustering; repeated measurement unit: therapy session/practice opportunity and follow-up; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Corrected beliefs are process outcomes; therapist contamination and unequal opportunities need explicit handling; sleep-stage claims remain exploratory.
Primary estimand: education-offer effect on completed CBT-I sessions/practice under the registered primary adherence rule
Candidate primary outcome: completed sessions and assigned practice under a fixed denominator/hierarchy
Primary finalized: No
Mathematical model: type: proportion/binomial design with opportunity denominators, overdispersion and independent-person precision; formulas: K_it | n_it, p_it ~ Binomial(n_it,p_it) with beta-binomial or clustered dependence if needed; logit(p_it)=alpha_t+z(A_i)^T theta_t+gamma^T X_i+u_i.
Report the chosen participant-average or opportunity-average probability difference; define n_it=0 before analysis.; variables: Y: completed sessions and assigned practice under a fixed denominator/hierarchy; i: participant with therapist clustering; t: therapy session/practice opportunity and follow-up; the registered visit/time-window index, not another independent unit; A: participant randomized add-on uncertainty education before otherwise identical CBT-I; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: education-offer effect on completed CBT-I sessions/practice under the registered primary adherence rule; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; K it: number of primary successes/events within the registered opportunity set; n it: number of eligible opportunities, defined independently of observed success; p it: success/event probability under that explicitly defined opportunity set; z(A): prespecified indicators for the actual arms or randomized policy; family code: RCT_PROP
Sample size design: family: proportion/binomial design with opportunity denominators, overdispersion and independent-person precision; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Corrected beliefs are process outcomes; therapist contamination and unequal opportunities need explicit handling; sleep-stage claims remain exploratory.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
If the intervention changes opportunity counts, participant-average and pooled-opportunity targets differ; choose and report the intended target.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Equal adherence despite corrected beliefs, or increased distress after education, would challenge the proposed explanatory mechanism.; preanalysis gate: Corrected beliefs are process outcomes; therapist contamination and unequal opportunities need explicit handling; sleep-stage claims remain exploratory.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Corrected beliefs are process outcomes; therapist contamination and unequal opportunities need explicit handling; sleep-stage claims remain exploratory.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Corrected beliefs are process outcomes; therapist contamination and unequal opportunities need explicit handling; sleep-stage claims remain exploratory.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/107; question: Does teaching the uncertainty of wearable sleep stages improve engagement with cognitive behavioral therapy for insomnia?; aims: Characterize mismatch among stage estimates, subjective quality and clinical sleep measures.
Randomize a brief stage-uncertainty module versus matched general sleep education.
Test whether changed beliefs mediate CBT-I engagement.; design verbatim planning only: Planning assumption: 90 adults beginning clinician-supervised CBT-I, stratified by wearable use. All receive the same therapy; add-on education is randomized before treatment and followed for twelve weeks.; outcomes verbatim: Primary: completed CBT-I sessions and assigned behavioral practice. Secondary: insomnia severity, daytime function, checking behavior and objective continuity in a polysomnography subset.; analysis verbatim inherited not authoritative if refined: Use intention-to-treat risk differences for adherence and baseline-adjusted symptom models. Account for therapist clustering, expectations and missing device data; separate exploratory stage analyses.; falsification verbatim: Equal adherence despite corrected beliefs, or increased distress after education, would challenge the proposed explanatory mechanism.; source ids unverified leads: va-dod-insomnia-guideline-2025
wearables-living-review
nocebo-communication-review-2026; prospectus precedence: No

N05 — Roster-aware circadian guidance for rotating healthcare workers

Rank: 83
Cluster: Sleep/circadian/breathing
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: CROSS_COUNT; allocation or exposure unit: worker roster-transition block randomized to guidance sequence; independent sampling unit: worker; repeated measurement unit: transition-day off-duty vigilance test across roster blocks; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Six transitions are repeated within worker, not six workers; changing roster history and incomplete recovery create carryover; do not interpret unavailable tests as adherence failure.
Primary estimand: roster-aware versus generic advice effect on vigilance lapses at safe common transition assessments
Candidate primary outcome: psychomotor-vigilance lapse count
Primary finalized: No
Mathematical model: type: paired/crossover contrast precision with period, carryover and within-unit covariance; simulation for actual design; formulas: Y_ipt ~ NegativeBinomial(mu_ipt,phi); log(mu_ipt)=alpha_i+tau A_ip+pi_p+rho A_i,p-1+gamma^T X_i+b_cluster(i,p).
Target the registered within-unit assignment contrast, standardized over period/history; direct and lag effects are distinct parameters.; variables: Y: psychomotor-vigilance lapse count; i: worker; t: transition-day off-duty vigilance test across roster blocks; the registered visit/time-window index, not another independent unit; A: worker roster-transition block randomized to guidance sequence; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: roster-aware versus generic advice effect on vigilance lapses at safe common transition assessments; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; p: randomized crossover period/session/episode, not the outcome probability; pi p: period effect; include calendar/season/sequence terms supported by the design; rho: prespecified lagged-assignment/carryover parameter; cannot in general identify arbitrary persistent carryover in a short two-period trial; f time(t): prespecified within-period recovery/day trend; g: identity for mean outcomes, log for positive/count means, or logit for binary/proportion outcomes as specified below; mu ipt: mean vigilance lapse/error count; phi: count overdispersion parameter; family code: CROSS_COUNT
Sample size design: family: paired/crossover contrast precision with period, carryover and within-unit covariance; simulation for actual design; formula: For a simple two-condition paired mean only: Var(Delta_hat)=sigma_D^2/n_independent; actual cluster/period/lag models require design-specific simulation or c^T(D^T V^(-1)D)^(-1)c.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; sigma D^2: variance of independent-unit paired contrasts including measurement error; n independent: number of independent participants/teams/donors, not repeated samples; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Six transitions are repeated within worker, not six workers; changing roster history and incomplete recovery create carryover; do not interpret unavailable tests as adherence failure.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Concealed randomized sequence, a scientifically justified recovery interval and no anticipation; period/carryover cannot be repaired solely by a nonsignificant test.
If carryover is plausible, report lag/period and first-period or otherwise valid design sensitivity; do not drop periods based only on a post hoc carryover p-value.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Benefits limited to immediate subjective alertness with unchanged vigilance or shortened sleep opportunity would weaken the hypothesis.; preanalysis gate: Six transitions are repeated within worker, not six workers; changing roster history and incomplete recovery create carryover; do not interpret unavailable tests as adherence failure.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Six transitions are repeated within worker, not six workers; changing roster history and incomplete recovery create carryover; do not interpret unavailable tests as adherence failure.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Compare planned lag/carryover structures and a valid first-period sensitivity where its sample supports inference; assess informative absent periods.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Six transitions are repeated within worker, not six workers; changing roster history and incomplete recovery create carryover; do not interpret unavailable tests as adherence failure.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/82; question: Can roster-aware light and sleep planning reduce shift-transition impairment more than generic shift-work advice?; aims: Map within-person phase and alertness changes across naturally occurring roster transitions.
Compare roster-aware and generic guidance in randomized transition blocks.
Identify which household constraints limit implementation and benefit.; design verbatim planning only: Planning assumption: 60 rotating healthcare workers across six roster transitions each. Randomized crossover blocks have prespecified recovery intervals; no shift assignment or intentional sleep deprivation is introduced.; outcomes verbatim: Primary: transition-day psychomotor vigilance lapses during safe off-duty testing. Secondary: diary sleep opportunity, actigraphic regularity, phase estimates and work-family burden.; analysis verbatim inherited not authoritative if refined: Hierarchical models include worker, roster sequence, prior shifts and season. Estimate carryover explicitly; exclude unsafe on-duty testing and distinguish missingness from intervention nonadherence.; falsification verbatim: Benefits limited to immediate subjective alertness with unchanged vigilance or shortened sleep opportunity would weaken the hypothesis.; source ids unverified leads: light-circadian-review
circadian-ultradian-clinical-review-2024
behavior-stress-sleep-rct-2026; prospectus precedence: No

N06 — Phenotype-guided behavioral sleep support during menopause

Rank: 151
Cluster: Sleep/circadian/breathing
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT_INT; allocation or exposure unit: participant randomized phenotype-matched versus standard module sequence; independent sampling unit: participant; repeated measurement unit: daytime-impairment visit; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Strategy-by-phenotype interaction cannot identify which module is best for that phenotype without within-phenotype randomized module comparisons.
Primary estimand: assignment effect of the matched strategy and prespecified baseline phenotype contrasts on impairment
Candidate primary outcome: sleep-related daytime-impairment score at twelve weeks
Primary finalized: No
Mathematical model: type: interaction-specific precision/power with modifier distribution and reliability; simulate the actual interaction; formulas: E(Y_it | A_i, B_i, X_i) = alpha_t + tau_t A_i + eta_t B_i + kappa_t A_i B_i + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Treatment contrast at baseline B=b is tau_t + kappa_t b; the interaction target is kappa_t, not a difference in within-subgroup p-values.; variables: Y: sleep-related daytime-impairment score at twelve weeks; i: participant; t: daytime-impairment visit; the registered visit/time-window index, not another independent unit; A: participant randomized phenotype-matched versus standard module sequence; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: assignment effect of the matched strategy and prespecified baseline phenotype contrasts on impairment; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; B: one of the two prespecified pretreatment menopause sleep-phenotype strata; centering/coding and any measurement-error model fixed before evaluation; eta: baseline modifier association; kappa: assignment-by-baseline-modifier interaction coefficient on the stated scale; family code: RCT_INT
Sample size design: family: interaction-specific precision/power with modifier distribution and reliability; simulate the actual interaction; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure
modifier prevalence/distribution and variance
modifier reliability and overlap
interaction size or precision target; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Strategy-by-phenotype interaction cannot identify which module is best for that phenotype without within-phenotype randomized module comparisons.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The modifier must be measured before assignment/initiation; interaction is scale dependent and requires support within modifier strata.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Unstable phenotypes, or no functional advantage of matching despite stable phenotypes and module-specific target engagement, would undermine the practical matching strategy. Establishing a predictive phenotype for individual modules would require randomized module comparisons within phenotype strata.; preanalysis gate: Strategy-by-phenotype interaction cannot identify which module is best for that phenotype without within-phenotype randomized module comparisons.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Strategy-by-phenotype interaction cannot identify which module is best for that phenotype without within-phenotype randomized module comparisons.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Vary modifier measurement-error assumptions and retain a continuous modifier unless a cutoff was independently fixed.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Strategy-by-phenotype interaction cannot identify which module is best for that phenotype without within-phenotype randomized module comparisons.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/150; question: Does phenotype-matched sequencing of behavioral sleep modules improve daytime function more than an equally intensive standard sequence during menopause?; aims: Define reproducible nighttime symptom phenotypes using diaries and unobtrusive temperature sensing.
Randomize phenotype-matched versus standard-sequence behavioral modules.
Test whether symptom-specific improvement explains daytime benefit.; design verbatim planning only: Planning assumption: 90 perimenopausal or early postmenopausal adults with sleep complaints; parallel twelve-week trial after two-week characterization. Both groups receive established behavioral sleep support and appropriate clinical assessment.; outcomes verbatim: Primary: sleep-related daytime impairment. Secondary: insomnia severity, awakening timing, vasomotor burden, cognitive complaints and treatment acceptability.; analysis verbatim inherited not authoritative if refined: Estimate the intention-to-treat effect of the matched strategy on daytime impairment, adjusting baseline severity, stable hormone treatment and caregiving load. Prespecify two phenotype strata and report their uncertain strategy effects; a strategy-by-phenotype contrast does not establish which module works best for each phenotype. Treat symptom mediation as exploratory and assumption-dependent.; falsification verbatim: Unstable phenotypes, or no functional advantage of matching despite stable phenotypes and module-specific target engagement, would undermine the practical matching strategy. Establishing a predictive phenotype for individual modules would require randomized module comparisons within phenotype strata.; source ids unverified leads: menopause-cognition-sleep-review-2025
menopause-behavioral-symptoms-review-2026
va-dod-insomnia-guideline-2025; prospectus precedence: No

N07 — Bedtime breathing: sleep benefit beyond respiratory-driven HRV changes

Rank: 136
Cluster: Sleep/circadian/breathing
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT_LOG; allocation or exposure unit: participant randomized comfortable breathing versus matched relaxation; independent sampling unit: participant; repeated measurement unit: nightly sleep-onset diary and physiology visit; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Postassignment respiratory rate and adherence should not enter the primary total-effect adjustment; within-session HRV is not sleep benefit.
Primary estimand: breathing-program assignment effect on diary sleep-onset latency
Candidate primary outcome: diary sleep-onset latency in minutes under a fixed zero/rounding rule
Primary finalized: No
Mathematical model: type: contrast precision/power by design-matrix or simulation planning; formulas: E(Y_it | A_i, X_i, Y_i0) = alpha_t + tau_t A_i + beta_t Y_i0 + gamma_t^T X_i + u_i, targeting mean latency in minutes; use robust inference.
Delta = sum_t w_t c^T theta_t; estimate by baseline-adjusted marginal standardization, with unit/cluster-appropriate inference.; variables: Y: diary sleep-onset latency in minutes under a fixed zero/rounding rule; i: participant; t: nightly sleep-onset diary and physiology visit; the registered visit/time-window index, not another independent unit; A: participant randomized comfortable breathing versus matched relaxation; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: breathing-program assignment effect on diary sleep-onset latency; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; z(A): prespecified indicators for the actual randomized arms; two arms reduce to one indicator; family code: RCT_LOG
Sample size design: family: contrast precision/power by design-matrix or simulation planning; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Postassignment respiratory rate and adherence should not enter the primary total-effect adjustment; within-session HRV is not sleep benefit.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
A log-scale sensitivity analysis changes the scale of summary; define zeros/rounding without arbitrary offsets and do not silently replace the mean-minute target.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Only within-session HRV improvement, with no sleep or arousal change, would undermine the proposed sleep mechanism.; preanalysis gate: Postassignment respiratory rate and adherence should not enter the primary total-effect adjustment; within-session HRV is not sleep benefit.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Postassignment respiratory rate and adherence should not enter the primary total-effect adjustment; within-session HRV is not sleep benefit.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Postassignment respiratory rate and adherence should not enter the primary total-effect adjustment; within-session HRV is not sleep benefit.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/135; question: Does comfortable slow breathing improve insomnia symptoms beyond a matched attentional relaxation exercise?; aims: Validate simultaneous respiration and ECG measures before and after practice.
Randomize bedtime breathing or matched attention training for eight weeks.
Test whether reduced cognitive arousal predicts improved sleep independently of paced-session HRV.; design verbatim planning only: Planning assumption: 80 adults with elevated pre-sleep arousal, with clinically significant disorders assessed first; parallel groups receive equal instructor contact and optional usual care.; outcomes verbatim: Primary: diary sleep-onset latency. Secondary: insomnia severity, spontaneous-breathing HRV, respiratory rate, actigraphy and next-day alertness.; analysis verbatim inherited not authoritative if refined: Baseline-adjusted models incorporate respiratory rate and adherence; compare mechanisms using temporally ordered observations. Register one primary outcome and correct secondary physiology comparisons.; falsification verbatim: Only within-session HRV improvement, with no sleep or arousal change, would undermine the proposed sleep mechanism.; source ids unverified leads: slow-breathing-hrv-meta-2022
breathwork-paramedic-rct-2026
breathing-stress-review-2025; prospectus precedence: No

N08 — Context-aware acoustic sleep support in noisy housing

Rank: 195
Cluster: Sleep/circadian/breathing
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: CROSS_BINARY; allocation or exposure unit: participant-night block randomized acoustic masking versus comparator; independent sampling unit: participant in prespecified EEG subcohort; repeated measurement unit: independently detected environmental-noise event and scored EEG window; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: EEG subcohort defines primary population; detect environmental events independently of masking; carryover, alarm recognition and missing EEG nights are separate gates.
Primary estimand: within-person assignment effect on probability of an EEG awakening after an independently defined noise event
Candidate primary outcome: EEG-awakening indicator in the fixed postnoise window
Primary finalized: No
Mathematical model: type: paired/crossover contrast precision with period, carryover and within-unit covariance; simulation for actual design; formulas: logit Pr(Y_ipt=1)=alpha_i+tau_i A_ip+pi_p+rho A_i,p-1+gamma^T X_i; report marginal probability contrasts, with tau_i pooled only under stated hierarchical assumptions.
Target the registered within-unit assignment contrast, standardized over period/history; direct and lag effects are distinct parameters.; variables: Y: EEG-awakening indicator in the fixed postnoise window; i: participant in prespecified EEG subcohort; t: independently detected environmental-noise event and scored EEG window; the registered visit/time-window index, not another independent unit; A: participant-night block randomized acoustic masking versus comparator; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: within-person assignment effect on probability of an EEG awakening after an independently defined noise event; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; p: randomized crossover period/session/episode, not the outcome probability; pi p: period effect; include calendar/season/sequence terms supported by the design; rho: prespecified lagged-assignment/carryover parameter; cannot in general identify arbitrary persistent carryover in a short two-period trial; f time(t): prespecified within-period recovery/day trend; g: identity for mean outcomes, log for positive/count means, or logit for binary/proportion outcomes as specified below; tau i: person-specific randomized log-odds contrast; repeated task trials are dependent within person/block; family code: CROSS_BINARY
Sample size design: family: paired/crossover contrast precision with period, carryover and within-unit covariance; simulation for actual design; formula: For a simple two-condition paired mean only: Var(Delta_hat)=sigma_D^2/n_independent; actual cluster/period/lag models require design-specific simulation or c^T(D^T V^(-1)D)^(-1)c.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; sigma D^2: variance of independent-unit paired contrasts including measurement error; n independent: number of independent participants/teams/donors, not repeated samples; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: EEG subcohort defines primary population; detect environmental events independently of masking; carryover, alarm recognition and missing EEG nights are separate gates.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Concealed randomized sequence, a scientifically justified recovery interval and no anticipation; period/carryover cannot be repaired solely by a nonsignificant test.
If carryover is plausible, report lag/period and first-period or otherwise valid design sensitivity; do not drop periods based only on a post hoc carryover p-value.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Reduced perceived disturbance without fewer independent awakenings, or poorer alarm recognition, would challenge the practical benefit.; preanalysis gate: EEG subcohort defines primary population; detect environmental events independently of masking; carryover, alarm recognition and missing EEG nights are separate gates.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: EEG subcohort defines primary population; detect environmental events independently of masking; carryover, alarm recognition and missing EEG nights are separate gates.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Compare planned lag/carryover structures and a valid first-period sensitivity where its sample supports inference; assess informative absent periods.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
EEG subcohort defines primary population; detect environmental events independently of masking; carryover, alarm recognition and missing EEG nights are separate gates.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/194; question: Can adaptable acoustic masking reduce noise-linked awakenings without impairing alarm recognition or worsening perceived control?; aims: Link privacy-preserving sound-event features to independently scored awakenings.
Randomize masking versus acoustically credible control across home nights.
Measure alarm recognition, autonomy and differential benefit by housing conditions.; design verbatim planning only: Planning assumption: 50 adults in noisy housing complete randomized crossover night blocks over eight weeks. Prespecified portable-EEG nights in a subcohort provide the primary independently scored awakening endpoint; the whole cohort provides diary and daytime outcomes. Validate privacy-preserving environmental-noise detection separately from the assigned masking sound. No raw conversation audio is stored; every participant can disable sound.; outcomes verbatim: Primary: probability of an independently scored EEG awakening after a prespecified environmental-noise event during the portable-EEG assessment nights. Secondary: total independently scored awakenings in that subcohort, whole-cohort diary disturbance and daytime sleepiness, alarm-detection performance and household acceptability.; analysis verbatim inherited not authoritative if refined: Estimate randomized within-person effects in the prespecified EEG assessment population, using environmental-noise events detected independently of the masking output. Include weekday, co-sleeping, prior-night sleep, period and carryover; predefine exposure windows. Report subcohort selection and missing EEG nights, and keep whole-cohort subjective outcomes separate from independently scored awakening effects.; falsification verbatim: Reduced perceived disturbance without fewer independent awakenings, or poorer alarm recognition, would challenge the practical benefit.; source ids unverified leads: wearables-living-review
va-dod-insomnia-guideline-2025; prospectus precedence: No

N09 — Positive-airway-pressure feedback that prioritizes function over scores

Rank: 38
Cluster: Sleep/circadian/breathing
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT_BINARY; allocation or exposure unit: participant randomized PAP feedback format; independent sampling unit: participant; repeated measurement unit: device upload and six-month clinical persistence assessment; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Missing uploads are not necessarily nonuse; usage is a mediator; switching or clinician-directed stopping require strategy-specific endpoint rules.
Primary estimand: function-oriented feedback versus usage-dashboard offer effect on six-month PAP persistence
Candidate primary outcome: persistence with prescribed PAP at six months under a fixed clinical definition
Primary finalized: No
Mathematical model: type: binary risk-difference precision/power, simulated with actual cluster/trajectory structure; formulas: logit Pr(Y_i=1 | A_i,X_i) = alpha + z(A_i)^T theta + gamma^T X_i + b_cluster(i).
RD = E_X[p_hat(1,X) - p_hat(0,X)]; report absolute risk difference as well as any odds ratio.; variables: Y: persistence with prescribed PAP at six months under a fixed clinical definition; i: participant; t: device upload and six-month clinical persistence assessment; the registered visit/time-window index, not another independent unit; A: participant randomized PAP feedback format; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: function-oriented feedback versus usage-dashboard offer effect on six-month PAP persistence; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; logit(p): log[p/(1-p)]; p hat(a,X): fitted outcome probability under arm a standardized to the prespecified population; RD: marginal risk difference; z(A): prespecified indicators for the actual arms or randomized policy; family code: RCT_BINARY
Sample size design: family: binary risk-difference precision/power, simulated with actual cluster/trajectory structure; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Missing uploads are not necessarily nonuse; usage is a mediator; switching or clinician-directed stopping require strategy-specific endpoint rules.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Better understanding without sustained use or daytime benefit would weaken the proposed feedback mechanism.; preanalysis gate: Missing uploads are not necessarily nonuse; usage is a mediator; switching or clinician-directed stopping require strategy-specific endpoint rules.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Missing uploads are not necessarily nonuse; usage is a mediator; switching or clinician-directed stopping require strategy-specific endpoint rules.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Missing uploads are not necessarily nonuse; usage is a mediator; switching or clinician-directed stopping require strategy-specific endpoint rules.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/37; question: Does feedback emphasizing symptoms and confirmed respiratory control improve PAP persistence more than usage-hour dashboards alone?; aims: Describe discordance among device usage, residual events and perceived recovery.
Randomize clinically interpretable integrated feedback versus standard adherence displays.
Test whether improved treatment understanding precedes sustained use.; design verbatim planning only: Planning assumption: 100 adults initiating prescribed PAP at one sleep center; twelve-week parallel trial with usual troubleshooting in both arms and six-month follow-up. Clinicians retain access to all treatment data.; outcomes verbatim: Primary: persistence with prescribed PAP at six months. Secondary: daytime sleepiness, functional outcomes, objectively logged use, mask burden and inappropriate score checking.; analysis verbatim inherited not authoritative if refined: Intention-to-treat models adjust severity, initial mask comfort and insurance constraints; treat usage as a mediator rather than a baseline confounder. Examine missing uploads without assuming nonuse.; falsification verbatim: Better understanding without sustained use or daytime benefit would weaken the proposed feedback mechanism.; source ids unverified leads: va-dod-insomnia-guideline-2025
sleep-apnea-oximetry-meta-2025
wearables-remote-monitoring-2026; prospectus precedence: No

N10 — Weekend regularity without sacrificing sleep opportunity

Rank: 158
Cluster: Sleep/circadian/breathing
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT; allocation or exposure unit: participant randomized regularity coaching versus matched education; independent sampling unit: participant; repeated measurement unit: Monday vigilance assessment over ten weeks; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Timing and sleep duration may change together; postrandomization duration adjustment does not isolate a pure timing effect.
Primary estimand: added-regularity coaching effect on mean Monday vigilance with common duration support
Candidate primary outcome: prespecified Monday vigilance score
Primary finalized: No
Mathematical model: type: contrast precision/power by design-matrix or simulation planning; formulas: E(Y_it | A_i, X_i, Y_i0) = alpha_t + z(A_i)^T theta_t + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Delta = sum_t w_t c^T theta_t; estimate by baseline-adjusted marginal standardization, with unit/cluster-appropriate inference.; variables: Y: prespecified Monday vigilance score; i: participant; t: Monday vigilance assessment over ten weeks; the registered visit/time-window index, not another independent unit; A: participant randomized regularity coaching versus matched education; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: added-regularity coaching effect on mean Monday vigilance with common duration support; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; z(A): prespecified indicators for the actual randomized arms; two arms reduce to one indicator; family code: RCT
Sample size design: family: contrast precision/power by design-matrix or simulation planning; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Timing and sleep duration may change together; postrandomization duration adjustment does not isolate a pure timing effect.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: No Monday benefit despite reduced timing variability and comparable sleep opportunity would weaken the practical regularity rationale. If duration also changes, the trial estimates the coaching strategy effect and cannot by itself separate timing from duration mechanisms.; preanalysis gate: Timing and sleep duration may change together; postrandomization duration adjustment does not isolate a pure timing effect.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Timing and sleep duration may change together; postrandomization duration adjustment does not isolate a pure timing effect.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Timing and sleep duration may change together; postrandomization duration adjustment does not isolate a pure timing effect.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/157; question: Does reducing weekend timing variability improve Monday functioning when total sleep opportunity remains adequate?; aims: Measure weekday-weekend phase, opportunity and functional variation prospectively.
Randomize added regularity coaching while matching sleep-opportunity support and contact.
Test baseline opportunity constraints as a prespecified effect modifier, and describe achieved timing and duration changes separately.; design verbatim planning only: Planning assumption: 90 adults with large weekday-weekend timing shifts, excluding rotating workers and untreated sleep disorders; ten-week parallel trial spanning comparable seasons. Both arms receive the same duration-focused support for adequate sleep opportunity; regularity coaching versus matched general sleep education is added at random. Participants negotiate achievable schedules rather than receiving enforced bedtimes.; outcomes verbatim: Primary: repeated Monday vigilance performance. Secondary: daytime impairment, actigraphic regularity, diary opportunity, phase estimates and social participation.; analysis verbatim inherited not authoritative if refined: Estimate the intention-to-treat effect of added regularity coaching before adherence analyses. Include baseline timing shifts, work start time and childcare constraints; use baseline opportunity constraints for moderation. Report timing and duration changes jointly. Post-randomization duration adjustment does not by itself isolate a causal timing effect; any mediation analysis requires explicit additional assumptions.; falsification verbatim: No Monday benefit despite reduced timing variability and comparable sleep opportunity would weaken the practical regularity rationale. If duration also changes, the trial estimates the coaching strategy effect and cannot by itself separate timing from duration mechanisms.; source ids unverified leads: light-circadian-review
wearable-mci-variability-2026
behavior-stress-sleep-rct-2026; prospectus precedence: No

N11 — Respiration-aware HRV: separating measurement artifacts from recovery

Rank: 22
Cluster: Autonomic regulation/stress/HRV
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: CROSS; allocation or exposure unit: participant randomized guided-breathing/quiet-attention sequence; independent sampling unit: participant; repeated measurement unit: ECG window within standardized post-session visit; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Baseline, period and sequence adjustment only in primary total-effect model; respiration and allocation guesses are postassignment; optical agreement is a separate measurement estimand.
Primary estimand: within-person assignment effect on log RMSSD ten to fifteen minutes after session in spontaneous breathing
Candidate primary outcome: ECG-derived log RMSSD under fixed posture/timing
Primary finalized: No
Mathematical model: type: paired/crossover contrast precision with period, carryover and within-unit covariance; simulation for actual design; formulas: g(E[Y_ipt | A_ip,X_i]) = alpha_i + tau A_ip + pi_p + rho A_i,p-1 + gamma^T X_i + f_time(t) + b_cluster(i,p).
Target the registered within-unit assignment contrast, standardized over period/history; direct and lag effects are distinct parameters.; variables: Y: ECG-derived log RMSSD under fixed posture/timing; i: participant; t: ECG window within standardized post-session visit; the registered visit/time-window index, not another independent unit; A: participant randomized guided-breathing/quiet-attention sequence; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: within-person assignment effect on log RMSSD ten to fifteen minutes after session in spontaneous breathing; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; p: randomized crossover period/session/episode, not the outcome probability; pi p: period effect; include calendar/season/sequence terms supported by the design; rho: prespecified lagged-assignment/carryover parameter; cannot in general identify arbitrary persistent carryover in a short two-period trial; f time(t): prespecified within-period recovery/day trend; g: identity for mean outcomes, log for positive/count means, or logit for binary/proportion outcomes as specified below; family code: CROSS
Sample size design: family: paired/crossover contrast precision with period, carryover and within-unit covariance; simulation for actual design; formula: For a simple two-condition paired mean only: Var(Delta_hat)=sigma_D^2/n_independent; actual cluster/period/lag models require design-specific simulation or c^T(D^T V^(-1)D)^(-1)c.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; sigma D^2: variance of independent-unit paired contrasts including measurement error; n independent: number of independent participants/teams/donors, not repeated samples; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Baseline, period and sequence adjustment only in primary total-effect model; respiration and allocation guesses are postassignment; optical agreement is a separate measurement estimand.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Concealed randomized sequence, a scientifically justified recovery interval and no anticipation; period/carryover cannot be repaired solely by a nonsignificant test.
If carryover is plausible, report lag/period and first-period or otherwise valid design sensitivity; do not drop periods based only on a post hoc carryover p-value.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: No reproducible post-session ECG contrast with adequate precision would challenge the short-term total-effect hypothesis. Optical changes absent in ECG would support device-specific error. Persistence of an ECG contrast alone would not establish respiration-independent or lasting autonomic recovery.; preanalysis gate: Baseline, period and sequence adjustment only in primary total-effect model; respiration and allocation guesses are postassignment; optical agreement is a separate measurement estimand.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Baseline, period and sequence adjustment only in primary total-effect model; respiration and allocation guesses are postassignment; optical agreement is a separate measurement estimand.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Compare planned lag/carryover structures and a valid first-period sensitivity where its sample supports inference; assess informative absent periods.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Baseline, period and sequence adjustment only in primary total-effect model; respiration and allocation guesses are postassignment; optical agreement is a separate measurement estimand.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/21; question: Does guided breathing versus quiet attention change post-session ECG-derived HRV under standardized posture and measurement timing, and how do optical-device error and measured respiration affect interpretation?; aims: Validate optical HRV against ECG across respiration and posture conditions.
Randomize comfortable paced-breathing versus quiet-attention sessions within participants.
Test which standardized changes predict recovery from a separate mild laboratory stressor.; design verbatim planning only: Planning assumption: 70 adults spanning age and skin-pigmentation ranges complete counterbalanced laboratory visits; a four-week home substudy tests repeatability. Clinical screening precedes stress tasks.; outcomes verbatim: Primary: within-person assignment effect on ECG-derived log RMSSD during a standardized spontaneous-breathing assessment 10–15 minutes after the session. Secondary: synchronized optical-versus-ECG error, respiratory dynamics, independent stress-response associations, distress and measurement acceptability.; analysis verbatim inherited not authoritative if refined: Use the randomized crossover contrast with baseline physiology, period and sequence; prespecify carryover and signal-quality handling. Do not adjust the primary total-effect estimate for post-treatment respiration. Analyze respiratory contributions descriptively and optical agreement with a reference-error framework; independent stress-response prediction is associational and does not validate durable recovery.; falsification verbatim: No reproducible post-session ECG contrast with adequate precision would challenge the short-term total-effect hypothesis. Optical changes absent in ECG would support device-specific error. Persistence of an ECG contrast alone would not establish respiration-independent or lasting autonomic recovery.; source ids unverified leads: slow-breathing-hrv-meta-2022
wearable-stress-measurement-review-2026
wearables-living-review; prospectus precedence: Yes; prospectus file: thesis-prospectuses.json; prospectus design and estimand: The participant is the analysis unit. The primary estimand is the mean within-person difference in post-session log RMSSD under assignment to guided breathing versus quiet attention, measured in the same seated spontaneous-breathing condition. Concealed randomized sequence and standardized contact support identification. The primary total-effect model adjusts baseline values and period, not post-treatment respiration; an explicitly descriptive mechanistic model examines respiratory contributions.; prospectus analysis plan: Fit mixed models with participant effects, randomized condition, period, sequence and baseline log RMSSD. Report paired differences with confidence intervals, not only significance. Use hierarchical agreement analyses that account for multiple windows within people; repeated windows are not independent sample size. Prespecify motion and pigmentation contrasts as measurement questions with uncertainty. Explore carryover using design information and sensitivity analyses; analyze recovery slopes separately with multiplicity limits and no diagnostic claims.; precedence note: Expanded prospectus refinements govern this scaffold; verbatim earlier brief material is retained for traceability, not as a contradictory analysis instruction. F15 is explicitly two arms; F30 has no unjustified symptom noninferiority claim.

N12 — Does personalized HRV biofeedback outperform a fixed comfortable program?

Rank: 172
Cluster: Autonomic regulation/stress/HRV
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT_LONG; allocation or exposure unit: participant randomized frozen adaptive versus fixed comfortable biofeedback program; independent sampling unit: participant; repeated measurement unit: weekly stress measurement over eight weeks; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Calibration repeatability and the stability interaction must be prespecified; do not select responders from observed outcome improvement.
Primary estimand: adaptive-program assignment effect on a fixed weighted average of weekly perceived stress
Candidate primary outcome: weekly perceived-stress score with weights fixed before analysis
Primary finalized: No
Mathematical model: type: contrast precision/power by design-matrix or simulation planning; formulas: E(Y_it | A_i, X_i, Y_i0) = alpha_t + z(A_i)^T theta_t + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Delta = sum_t w_t c^T theta_t; estimate by baseline-adjusted marginal standardization, with unit/cluster-appropriate inference.; variables: Y: weekly perceived-stress score with weights fixed before analysis; i: participant; t: weekly stress measurement over eight weeks; the registered visit/time-window index, not another independent unit; A: participant randomized frozen adaptive versus fixed comfortable biofeedback program; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: adaptive-program assignment effect on a fixed weighted average of weekly perceived stress; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; z(A): prespecified indicators for the actual randomized arms; two arms reduce to one indicator; family code: RCT_LONG
Sample size design: family: contrast precision/power by design-matrix or simulation planning; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Calibration repeatability and the stability interaction must be prespecified; do not select responders from observed outcome improvement.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Unstable individual rankings or equal benefit across arms despite successful adaptation would undermine the precision claim.; preanalysis gate: Calibration repeatability and the stability interaction must be prespecified; do not select responders from observed outcome improvement.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Calibration repeatability and the stability interaction must be prespecified; do not select responders from observed outcome improvement.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Calibration repeatability and the stability interaction must be prespecified; do not select responders from observed outcome improvement.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/171; question: Does adapting feedback to a participant's reproducible physiological response improve everyday stress regulation beyond a fixed breathing program?; aims: Estimate calibration repeatability across independent visits.
Randomize adaptive versus fixed programs with identical feedback contact.
Test whether calibration stability prospectively identifies durable symptom benefit.; design verbatim planning only: Planning assumption: 90 stressed adults without unstable cardiovascular illness complete repeated calibration then eight-week parallel training. Adaptive rules are frozen before recruitment; both arms use comfortable supervised slow breathing.; outcomes verbatim: Primary: weekly perceived stress over follow-up. Secondary: spontaneous-rest HRV, stressor recovery, adherence and perceived control, assessed beyond training sessions.; analysis verbatim inherited not authoritative if refined: Model intention-to-treat change and one prespecified stability interaction; avoid selecting responders using outcomes. Track expectancy, respiratory rate and differential practice, with sensitivity analyses for informative dropout.; falsification verbatim: Unstable individual rankings or equal benefit across arms despite successful adaptation would undermine the precision claim.; source ids unverified leads: slow-breathing-hrv-review-2025
n-of-1-methods-review-2025
breathing-stress-review-2025; prospectus precedence: No

N13 — Just-in-time breathing prompts: benefit, interruption and alarm fatigue

Rank: 73
Cluster: Autonomic regulation/stress/HRV
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: MRT_POLICY; allocation or exposure unit: participant randomized trigger policy, then eligible moment randomized prompt/no prompt; independent sampling unit: participant; moment is the within-person randomization unit; repeated measurement unit: eligible prompt event and proximal anxiety window; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Availability is history-dependent; compare longer-term policy outcomes using participant randomization and common scheduled visits, not event-selected means.
Primary estimand: proximal prompt excursion effect among available moments within policy; common-support policy moderation only
Candidate primary outcome: change in momentary anxiety over the fixed postdecision window
Primary finalized: No
Mathematical model: type: micro-randomized-trial power/precision simulation for availability, randomization probabilities, proximal correlation and lagged effects; formulas: E[Y_i,t+Delta(1)-Y_i,t+Delta(0) | I_it=1,S_it] = f(S_it)^T beta, a proximal excursion effect under the registered future-treatment policy.
Estimate beta using centered A_it-p_it with known p_it=P(A_it=1 | H_it,I_it=1); a basic unmoderated moment is sum I_it(A_it-p_it)Y_i,t+Delta / sum I_it p_it(1-p_it), with prespecified weighting/augmentation as needed.
For participant-randomized trigger policy Z, estimate common-schedule outcomes by E[Q | Z=1]-E[Q | Z=0]; event-selected proximal moderator comparisons require eligibility common support.; variables: Y: change in momentary anxiety over the fixed postdecision window; i: participant; moment is the within-person randomization unit; t: eligible prompt event and proximal anxiety window; the registered visit/time-window index, not another independent unit; A: participant randomized trigger policy, then eligible moment randomized prompt/no prompt; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: proximal prompt excursion effect among available moments within policy; common-support policy moderation only; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; I it: availability/eligibility indicator decided before current randomization; H it: observed predecision history including prior prompts/breaks; p it: known, positive randomized prompt/show probabilities among available moments; S it: prespecified predecision moderator vector; Delta: fixed proximal outcome delay/window, not a mean difference in this formula; f(S)^T beta: registered moderation model; beta is estimated with participant/coach dependence; Z: baseline randomized trigger policy; Q: common scheduled weekly symptom/burden/abandonment outcome; family code: MRT_POLICY
Sample size design: family: micro-randomized-trial power/precision simulation for availability, randomization probabilities, proximal correlation and lagged effects; formula: Simulate the actual availability/history/randomization process and estimate Var(beta_hat) with participant/coach dependence; moment count alone is not the independent sample size.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Availability is history-dependent; compare longer-term policy outcomes using participant randomization and common scheduled visits, not event-selected means.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Sequential randomization at eligible moments, positivity of prompt/no prompt, consistency and prespecified availability; proximal effect does not automatically identify a long-term strategy effect.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: No proximal benefit at receptive moments, or greater burden without sustained engagement, would challenge the adaptive policy.; preanalysis gate: Availability is history-dependent; compare longer-term policy outcomes using participant randomization and common scheduled visits, not event-selected means.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Availability is history-dependent; compare longer-term policy outcomes using participant randomization and common scheduled visits, not event-selected means.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Vary availability models, lagged effects, proximal windows and informative postprompt missingness without searching a favorable window.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Availability is history-dependent; compare longer-term policy outcomes using participant randomization and common scheduled visits, not event-selected means.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/72; question: When do context-sensitive breathing prompts improve anxiety, and when do they become another source of stress?; aims: Map prompt receptivity and false stress alerts prospectively.
Micro-randomize eligible moments to prompt or no prompt under two trigger policies.
Estimate whether accumulated burden attenuates benefit across weeks.; design verbatim planning only: Planning assumption: 80 adults with elevated anxiety use a study app for eight weeks. Randomly assign a receptive-window or physiology-only trigger policy, then micro-randomize prompt versus no prompt at eligible moments within each policy. Collect common scheduled symptom and burden assessments in both groups. Participant-defined quiet hours and safety exclusions precede randomization; breathing is an adjunct to existing care.; outcomes verbatim: Primary: change in momentary anxiety after eligible events. Secondary: weekly symptoms, prompt dismissal, abandonment, perceived surveillance and spontaneous-breathing recovery.; analysis verbatim inherited not authoritative if refined: Use weighted micro-randomized estimators for proximal prompt effects, accounting for availability and prior prompts. Compare effect moderation only where policy eligibility has common support. Use the randomized policy assignment and common scheduled assessments for exploratory policy comparisons of symptoms, burden and abandonment; event-selected proximal comparisons alone do not establish policy superiority. Prespecify time-varying moderation and distinguish proximal effects from longer-term effects.; falsification verbatim: No proximal benefit at receptive moments, or greater burden without sustained engagement, would challenge the adaptive policy.; source ids unverified leads: slow-breathing-jit-rct-2026
wearable-stress-ml-review-2025
digital-cbt-wearable-rct-2025; prospectus precedence: No

N14 — Stress detection that generalizes beyond the laboratory

Rank: 116
Cluster: Autonomic regulation/stress/HRV
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: PRED; allocation or exposure unit: stress model evaluated before any separate display randomization; independent sampling unit: participant and held-out context; repeated measurement unit: sensor/diary episode; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: prediction, calibration or model transportability; prediction gain is not clinical efficacy, mechanistic necessity or causal treatment selection; specific constraint: Physical arousal is not distress truth; medication/movement confounding limits interpretation; display effects require their separate randomization.
Primary estimand: context-held-out calibration and false-alert burden for self-reported distress episodes
Candidate primary outcome: self-reported distress-episode indicator under a prespecified definition
Primary finalized: No
Mathematical model: type: external/nested predictive-validation precision with target prevalence/variance, independent-unit holdout coverage and model complexity; formulas: Gain = mean_test[L(Y,f_base(X)) - L(Y,f_new(X))], with both predictions frozen before test outcomes; bootstrap independent held-out units.
For a binary target assess E[Y | p_hat=p] versus p and calibration intercept/slope logit Pr(Y=1)=a_cal+b_cal logit(p_hat); for a continuous target use Y=a_cal+b_cal y_hat+error plus scale-appropriate calibration/loss.; variables: Y: self-reported distress-episode indicator under a prespecified definition; i: participant and held-out context; t: sensor/diary episode; the registered visit/time-window index, not another independent unit; A: stress model evaluated before any separate display randomization; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: context-held-out calibration and false-alert burden for self-reported distress episodes; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; L: one proper or scientifically justified primary loss/scoring rule fixed for the actual outcome type; positive Gain favors new model; f base,f new: locked baseline/new prediction functions using only information available at the registered prediction time; test: untouched person/family/lab/site/time set as specified per thesis; p hat,y hat: frozen probability or continuous prediction; a cal,b cal: calibration intercept and slope, estimated on evaluation data for assessment; using them to update the model requires another evaluation set; family code: PRED
Sample size design: family: external/nested predictive-validation precision with target prevalence/variance, independent-unit holdout coverage and model complexity; formula: Simulate or bootstrap the independent-unit test loss/calibration/decision criterion; plan cases/events, noncases and independent sites/families/programs separately. No universal events-per-variable or participant N is asserted.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Physical arousal is not distress truth; medication/movement confounding limits interpretation; display effects require their separate randomization.
Train/tune/test separation at the actual biological/participant/family/time level; no future, postoutcome or target-derived predictors.
Validation support must cover intended deployment; calibration and missing/failing measurements are reported with discrimination/loss.
A validation subset selected by culture success, assay viability, consent or data completeness is a different target unless the full strategy is explicitly evaluated.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: No improvement under context-held-out evaluation, or interpretation unchanged by uncertainty, would undermine the proposed translation strategy.; preanalysis gate: Physical arousal is not distress truth; medication/movement confounding limits interpretation; display effects require their separate randomization.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Physical arousal is not distress truth; medication/movement confounding limits interpretation; display effects require their separate randomization.
Repeat loss/calibration with independent-cluster resampling, prespecified subgroup/transport shifts and plausible missingness.
Compare fixed simple baselines, calibration and uncertainty coverage; report abstention/failure/turnaround rather than deleting hard cases.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Frozen prediction time/horizon, proper loss, baseline model, feature availability, independent split/holdout and target deployment population.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Physical arousal is not distress truth; medication/movement confounding limits interpretation; display effects require their separate randomization.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/115; question: Can wearable stress models retain useful calibration across home, service work and caregiving without equating physical arousal with distress?; aims: Collect matched laboratory and naturalistic episodes with contextual labels.
Compare frozen models on held-out occupations and participants.
Randomize truthful uncertainty displays to test whether calibration improves interpretation.; design verbatim planning only: Planning assumption: 120 adults from three contrasting work or caregiving contexts provide four weeks of sensing and brief diaries; a separate session supplies laboratory references. Display randomization occurs after external validation.; outcomes verbatim: Primary: external calibration for self-reported distress episodes. Secondary: false-alert burden, subgroup error, interpretation accuracy and willingness to use feedback.; analysis verbatim inherited not authoritative if refined: Prevent participant and temporal leakage; preregister context-level holdouts. Adjust observational comparisons for movement and medication; causal claims apply only to randomized display effects.; falsification verbatim: No improvement under context-held-out evaluation, or interpretation unchanged by uncertainty, would undermine the proposed translation strategy.; source ids unverified leads: wearable-stress-ml-review-2025
wearable-stress-measurement-review-2026
wearables-living-review; prospectus precedence: No

N15 — Interoceptive attention as a moderator of mindfulness benefit and harm

Rank: 152
Cluster: Autonomic regulation/stress/HRV
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT_INT; allocation or exposure unit: participant randomized body-focused versus externally anchored attention; independent sampling unit: participant with instructor clustering; repeated measurement unit: stress and adverse-experience visit; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Baseline threat sensitivity must precede allocation; intolerance-related dropout and panic-like experiences require parallel harm analysis.
Primary estimand: assignment effect and one baseline threat-sensitivity interaction on perceived stress
Candidate primary outcome: validated perceived-stress score at six weeks
Primary finalized: No
Mathematical model: type: interaction-specific precision/power with modifier distribution and reliability; simulate the actual interaction; formulas: E(Y_it | A_i, B_i, X_i) = alpha_t + tau_t A_i + eta_t B_i + kappa_t A_i B_i + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Treatment contrast at baseline B=b is tau_t + kappa_t b; the interaction target is kappa_t, not a difference in within-subgroup p-values.; variables: Y: validated perceived-stress score at six weeks; i: participant with instructor clustering; t: stress and adverse-experience visit; the registered visit/time-window index, not another independent unit; A: participant randomized body-focused versus externally anchored attention; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: assignment effect and one baseline threat-sensitivity interaction on perceived stress; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; B: baseline threat/interoceptive-anxiety sensitivity measured before allocation; centering/coding and any measurement-error model fixed before evaluation; eta: baseline modifier association; kappa: assignment-by-baseline-modifier interaction coefficient on the stated scale; family code: RCT_INT
Sample size design: family: interaction-specific precision/power with modifier distribution and reliability; simulate the actual interaction; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure
modifier prevalence/distribution and variance
modifier reliability and overlap
interaction size or precision target; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Baseline threat sensitivity must precede allocation; intolerance-related dropout and panic-like experiences require parallel harm analysis.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The modifier must be measured before assignment/initiation; interaction is scale dependent and requires support within modifier strata.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: No differential response by threat sensitivity, or equally increased distress in both programs, would weaken the matching hypothesis.; preanalysis gate: Baseline threat sensitivity must precede allocation; intolerance-related dropout and panic-like experiences require parallel harm analysis.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Baseline threat sensitivity must precede allocation; intolerance-related dropout and panic-like experiences require parallel harm analysis.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Vary modifier measurement-error assumptions and retain a continuous modifier unless a cutoff was independently fixed.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Baseline threat sensitivity must precede allocation; intolerance-related dropout and panic-like experiences require parallel harm analysis.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/151; question: Who benefits from body-focused mindfulness, and who responds better to externally anchored attention when monitoring bodily sensations increases anxiety?; aims: Measure interoceptive accuracy, confidence and threat interpretation separately.
Randomize body-focused or externally anchored mindfulness with matched instructors.
Test whether changes in threat appraisal explain differential stress recovery.; design verbatim planning only: Planning assumption: 90 nonclinical adults with a range of anxiety sensitivity complete six-week programs and a follow-up laboratory challenge; severe psychiatric symptoms trigger referral before enrollment.; outcomes verbatim: Primary: perceived stress. Secondary: distress during practice, panic-like symptoms, discontinuation, recovery physiology and daily functioning.; analysis verbatim inherited not authoritative if refined: Predefine one treatment-by-threat-sensitivity interaction; account for instructor clustering and baseline expectations. Model adverse experiences alongside benefits and report missingness by intolerance.; falsification verbatim: No differential response by threat sensitivity, or equally increased distress in both programs, would weaken the matching hypothesis.; source ids unverified leads: mindfulness-stress-meta-2026
belief-effects-health-review-2026
breathing-stress-review-2025; prospectus precedence: No

N16 — Does positive writing accelerate recovery or merely change appraisal?

Rank: 165
Cluster: Autonomic regulation/stress/HRV
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: EPISODE; allocation or exposure unit: ordinary-stress episode writing assignment randomized within participant; laboratory sequence counterbalanced; independent sampling unit: participant; repeated measurement unit: episode rumination time point and lab visit; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Episode severity measured before assignment; previous writing may carry over; improved mood does not establish physiological recovery.
Primary estimand: randomized writing effect on the fixed postepisode rumination trajectory summary
Candidate primary outcome: rumination score over the registered postepisode recovery interval
Primary finalized: No
Mathematical model: type: paired/crossover contrast precision with period, carryover and within-unit covariance; simulation for actual design; formulas: g(E[Y_ipt | A_ip,X_i]) = alpha_i + tau A_ip + pi_p + rho A_i,p-1 + gamma^T X_i + f_time(t) + b_cluster(i,p).
Target the registered within-unit assignment contrast, standardized over period/history; direct and lag effects are distinct parameters.; variables: Y: rumination score over the registered postepisode recovery interval; i: participant; t: episode rumination time point and lab visit; the registered visit/time-window index, not another independent unit; A: ordinary-stress episode writing assignment randomized within participant; laboratory sequence counterbalanced; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: randomized writing effect on the fixed postepisode rumination trajectory summary; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; p: randomized crossover period/session/episode, not the outcome probability; pi p: period effect; include calendar/season/sequence terms supported by the design; rho: prespecified lagged-assignment/carryover parameter; cannot in general identify arbitrary persistent carryover in a short two-period trial; f time(t): prespecified within-period recovery/day trend; g: identity for mean outcomes, log for positive/count means, or logit for binary/proportion outcomes as specified below; family code: EPISODE
Sample size design: family: paired/crossover contrast precision with period, carryover and within-unit covariance; simulation for actual design; formula: For a simple two-condition paired mean only: Var(Delta_hat)=sigma_D^2/n_independent; actual cluster/period/lag models require design-specific simulation or c^T(D^T V^(-1)D)^(-1)c.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; sigma D^2: variance of independent-unit paired contrasts including measurement error; n independent: number of independent participants/teams/donors, not repeated samples; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Episode severity measured before assignment; previous writing may carry over; improved mood does not establish physiological recovery.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Concealed randomized sequence, a scientifically justified recovery interval and no anticipation; period/carryover cannot be repaired solely by a nonsignificant test.
If carryover is plausible, report lag/period and first-period or otherwise valid design sensitivity; do not drop periods based only on a post hoc carryover p-value.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Equivalent rumination and physiological recovery despite mood improvement would refute the proposed recovery mechanism while preserving a narrower affect benefit.; preanalysis gate: Episode severity measured before assignment; previous writing may carry over; improved mood does not establish physiological recovery.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Episode severity measured before assignment; previous writing may carry over; improved mood does not establish physiological recovery.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Compare planned lag/carryover structures and a valid first-period sensitivity where its sample supports inference; assess informative absent periods.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Episode severity measured before assignment; previous writing may carry over; improved mood does not establish physiological recovery.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/164; question: Does positive expressive writing alter physiological recovery after everyday stress independently of improved reported mood?; aims: Characterize within-person stress episodes and subsequent rumination.
Randomize writing versus neutral factual description after eligible episodes.
Compare changes in appraisal, rumination and standardized recovery physiology.; design verbatim planning only: Planning assumption: 70 adults complete six weeks of event-triggered diary assessments with randomized writing assignments and two counterbalanced laboratory sessions. Analyses remain within ordinary stress contexts.; outcomes verbatim: Primary: post-episode rumination trajectory. Secondary: positive affect, ECG recovery, spontaneous-rest HRV, sleep-related worry and intervention burden.; analysis verbatim inherited not authoritative if refined: Use multilevel randomized-episode contrasts with participant fixed effects, episode severity and prior assignments. Assess carryover across episodes; separate registered mechanism tests from exploratory subjective-physiology concordance.; falsification verbatim: Equivalent rumination and physiological recovery despite mood improvement would refute the proposed recovery mechanism while preserving a narrower affect benefit.; source ids unverified leads: expressive-writing-review-2025
allostatic-consensus-2023
wearable-stress-measurement-review-2026; prospectus precedence: No

N17 — Can everyday HRV meaningfully track multisystem stress burden?

Rank: 212
Cluster: Autonomic regulation/stress/HRV
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: WITHIN_PRED; allocation or exposure unit: natural HRV/allostatic-load trajectories, no assigned intervention; independent sampling unit: participant; repeated measurement unit: standardized ECG and allostatic panel visit; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: association or prospective prediction; temporal order and model adjustment alone do not identify a mechanism or treatment effect; specific constraint: Composite construction and measurement error fixed in advance; association is not causal stress damage and multiple panel variants cannot be searched freely.
Primary estimand: incremental out-of-participant prediction of allostatic-load change from HRV beyond fixed clinical/social predictors
Candidate primary outcome: prespecified allostatic-load change
Primary finalized: No
Mathematical model: type: independent-person/site prospective prediction-gain and calibration precision, with within-person dependence; formulas: Y_it = alpha + u_i + beta_W (B_it-Bbar_i) + beta_B Bbar_i + gamma^T X_it + f_time(t) + b_study + epsilon_it; u_i is a random intercept with a declared conditional mean/covariance model, not unrestricted fixed person effects.
B_observed,it = B_true,it + e_it; estimate measurement variance from replicates/validation or report sensitivity rather than assume it known.
Primary prediction target: Gain=mean_test[L(Y,f_base(X))-L(Y,f_new(X,B_history))], with locked models and proper outcome-specific loss; report calibration as well as gain.
For new held-out persons/sites, predict using the training-estimated population random-effect distribution or only genuinely available preprediction calibration; never fit a person intercept using that test person's future outcome.; variables: Y: prespecified allostatic-load change; i: participant; t: standardized ECG and allostatic panel visit; the registered visit/time-window index, not another independent unit; A: natural HRV/allostatic-load trajectories, no assigned intervention; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: incremental out-of-participant prediction of allostatic-load change from HRV beyond fixed clinical/social predictors; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; B it: standardized ECG HRV history available before the allostatic-load prediction window; Bbar i: person/donor mean over a prespecified observation schedule; beta W,beta B: within-unit and between-unit association coefficients, not causal effects; X it: explicitly time-ordered baseline/premeasurement covariates; posttreatment adjustment changes interpretation; f time(t): prespecified sampling/recovery/calendar trajectory; b study: study/site/batch effect where relevant; B true,e it: latent predictor and measurement error, requiring identification constraints; Gain: held-out improvement over the complete prespecified routine/sleep/autonomic or clinical baseline; L: registered proper/validated outcome-specific loss; f base,f new: models frozen in development, with all feature selection internal to training; B history: predictor history available before the registered prediction time; family code: WITHIN_PRED
Sample size design: family: independent-person/site prospective prediction-gain and calibration precision, with within-person dependence; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Composite construction and measurement error fixed in advance; association is not causal stress damage and multiple panel variants cannot be searched freely.
Associations need a clear sampling/time frame; causal interpretation is not supplied by this model.
Measurement error, reverse causation, time-varying confounding and selected observation may remain despite adjustment.
All feature selection and time normalization occur inside training splits; later outcome windows are untouched and only information available at prediction is used.
For the secondary within/between association model, specify the conditional distribution of random intercept u_i; fixed person effects cannot separately estimate beta_B for time-invariant person means.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: No incremental predictive value or poor within-person concordance would undermine HRV as a broad stress-burden monitor.; preanalysis gate: Composite construction and measurement error fixed in advance; association is not causal stress damage and multiple panel variants cannot be searched freely.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Composite construction and measurement error fixed in advance; association is not causal stress damage and multiple panel variants cannot be searched freely.
Vary lag order/window and measurement-error reliability within a prespecified family; include negative-control windows when meaningful.
Separate within- and between-unit effects, missingness patterns and source-study replication; do not use cell/assay counts as independent donor counts.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Composite construction and measurement error fixed in advance; association is not causal stress damage and multiple panel variants cannot be searched freely.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/211; question: Does longitudinal standardized HRV add useful information about change in multisystem allostatic load beyond symptoms and social conditions?; aims: Construct a prespecified clinically accessible multisystem panel.
Measure HRV repeatability and longitudinal concordance with panel changes.
Test incremental prediction against symptom-only and context-aware baselines.; design verbatim planning only: Planning assumption: 140 adults assessed at baseline, six and twelve months, with repeated standardized ECG and home wearable periods. No intervention or medication change is assigned.; outcomes verbatim: Primary: incremental out-of-sample prediction of prespecified allostatic-load change. Secondary: measurement reliability, calibration, subgroup error and participant interpretation.; analysis verbatim inherited not authoritative if refined: Use repeated-measures measurement-error models and participant-level validation. Adjust confounding using a causal diagram; report associations without causal stress-damage claims. Limit panel variants and correct exploratory comparisons.; falsification verbatim: No incremental predictive value or poor within-person concordance would undermine HRV as a broad stress-burden monitor.; source ids unverified leads: allostatic-consensus-2023
wearable-hrv-inflammation-2026
allostatic-cvd-review-2025; prospectus precedence: No

N18 — Accessible biofeedback across sensory and digital barriers

Rank: 95
Cluster: Autonomic regulation/stress/HRV
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT_COUNT; allocation or exposure unit: participant randomized accessible modality choice versus standard adaptable interface; independent sampling unit: participant; repeated measurement unit: usable practice session; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Selected modality is not randomized; no modality equivalence/efficacy claim from choices; low-vision/hearing strata need uncertainty-aware interaction reporting.
Primary estimand: accessible-choice offer effect on completed usable practice sessions
Candidate primary outcome: count of completed usable practice sessions over six weeks
Primary finalized: No
Mathematical model: type: overdispersed count contrast planning by simulation, with follow-up/cluster/zero inflation assumptions; formulas: Y_it ~ NegativeBinomial(mu_it,phi); log(mu_it)=alpha_t+z(A_i)^T theta_t+beta Y_i0+gamma^T X_i+b_cluster(i)+u_i.
Delta = standardized E[Y | do(A=1)] - E[Y | do(A=0)] under randomization; report ratio exp(c^T theta) only when the log-mean model supports it.; variables: Y: count of completed usable practice sessions over six weeks; i: participant; t: usable practice session; the registered visit/time-window index, not another independent unit; A: participant randomized accessible modality choice versus standard adaptable interface; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: accessible-choice offer effect on completed usable practice sessions; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; mu it: conditional mean of the actual per-thesis count; phi: overdispersion parameter, with Var(Y)=mu+mu^2/phi under this parameterization; z(A): prespecified indicators for the actual arms or randomized policy; family code: RCT_COUNT
Sample size design: family: overdispersed count contrast planning by simulation, with follow-up/cluster/zero inflation assumptions; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Selected modality is not randomized; no modality equivalence/efficacy claim from choices; low-vision/hearing strata need uncertainty-aware interaction reporting.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: No adherence gain after verified accessibility improvement, or increased cognitive burden, would weaken the proposed interface pathway.; preanalysis gate: Selected modality is not randomized; no modality equivalence/efficacy claim from choices; low-vision/hearing strata need uncertainty-aware interaction reporting.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Selected modality is not randomized; no modality equivalence/efficacy claim from choices; low-vision/hearing strata need uncertainty-aware interaction reporting.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Selected modality is not randomized; no modality equivalence/efficacy claim from choices; low-vision/hearing strata need uncertainty-aware interaction reporting.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/94; question: Does offering a choice among accessible visual, auditory and haptic HRV-feedback interfaces improve usable practice and stress outcomes compared with a standard adaptable visual interface?; aims: Co-design interfaces with participants who have sensory or digital-access barriers.
Validate equivalent physiological information across interfaces.
Randomize accessible choice versus a standard adaptable visual interface.; design verbatim planning only: Planning assumption: 80 adults, deliberately including low-vision and hearing-impaired participants, complete six weeks of comfortable HRV biofeedback. All receive reasonable accommodations and loaned equipment; no essential access is withheld.; outcomes verbatim: Primary: completed usable practice sessions. Secondary: perceived stress, comprehension, spontaneous-rest HRV, frustration and setup burden.; analysis verbatim inherited not authoritative if refined: Estimate intention-to-treat effects of offering accessible choice, with baseline access-needs strata; analyze usability and physiology separately. Report selected modalities descriptively rather than treating their nonrandomized outcome differences as modality efficacy. Prespecify interaction limits and report uncertainty where subgroup numbers remain small; this is not a modality-equivalence trial.; falsification verbatim: No adherence gain after verified accessibility improvement, or increased cognitive burden, would weaken the proposed interface pathway.; source ids unverified leads: slow-breathing-hrv-review-2025
wearable-stress-measurement-review-2026
wearables-remote-monitoring-2026; prospectus precedence: No

N19 — Dyadic stress regulation in caregiving households

Rank: 122
Cluster: Autonomic regulation/stress/HRV
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: DYAD; allocation or exposure unit: household/dyad assigned shared practice versus equal-contact individual support; independent sampling unit: caregiver-supporter dyad; repeated measurement unit: caregiver/supporter repeated reports and episode physiology; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Confirm randomized household allocation; supporter data are correlated secondary outcomes; synchrony is not inherently beneficial and recipients need not participate.
Primary estimand: household-strategy effect on caregiver perceived stress
Candidate primary outcome: caregiver perceived-stress score at eight weeks
Primary finalized: No
Mathematical model: type: contrast precision/power by design-matrix or simulation planning; formulas: E(Y_it | A_i, X_i, Y_i0) = alpha_t + z(A_i)^T theta_t + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Delta = sum_t w_t c^T theta_t; estimate by baseline-adjusted marginal standardization, with unit/cluster-appropriate inference.; variables: Y: caregiver perceived-stress score at eight weeks; i: caregiver-supporter dyad; t: caregiver/supporter repeated reports and episode physiology; the registered visit/time-window index, not another independent unit; A: household/dyad assigned shared practice versus equal-contact individual support; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: household-strategy effect on caregiver perceived stress; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; z(A): prespecified indicators for the actual randomized arms; two arms reduce to one indicator; family code: DYAD
Sample size design: family: contrast precision/power by design-matrix or simulation planning; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Confirm randomized household allocation; supporter data are correlated secondary outcomes; synchrony is not inherently beneficial and recipients need not participate.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Randomization and inference occur by dyad/household; caregiver and supporter outcomes are not independent observations.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: No stress improvement, increased relationship strain, or benefit explained solely by additional contact would challenge shared-practice specificity.; preanalysis gate: Confirm randomized household allocation; supporter data are correlated secondary outcomes; synchrony is not inherently beneficial and recipients need not participate.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Confirm randomized household allocation; supporter data are correlated secondary outcomes; synchrony is not inherently beneficial and recipients need not participate.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Confirm randomized household allocation; supporter data are correlated secondary outcomes; synchrony is not inherently beneficial and recipients need not participate.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/121; question: Does shared low-burden regulation practice improve caregiver recovery more than individually delivered support with equal contact?; aims: Map within-household timing of stress and support using paired diaries.
Randomize households to optional shared or individual comfortable breathing support.
Test whether perceived support or coordination burden explains differences.; design verbatim planning only: Planning assumption: 50 adult caregiver-supporter dyads complete an eight-week parallel household trial; both arms receive equivalent practical caregiver resources. Care recipients are not required to participate.; outcomes verbatim: Primary: caregiver perceived stress. Secondary: recovery after caregiving episodes, sleep disturbance, relationship strain, adherence and supporter burden.; analysis verbatim inherited not authoritative if refined: Dyadic models account for nonindependence, baseline caregiving intensity and illness fluctuations. Prespecify support moderation; avoid treating physiological synchrony as inherently beneficial.; falsification verbatim: No stress improvement, increased relationship strain, or benefit explained solely by additional contact would challenge shared-practice specificity.; source ids unverified leads: breathing-stress-review-2025
mindfulness-stress-meta-2026
behavior-stress-sleep-rct-2026; prospectus precedence: No

N20 — Clinically meaningful autonomic biofeedback in cardiac rehabilitation

Rank: 188
Cluster: Autonomic regulation/stress/HRV
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT; allocation or exposure unit: participant randomized adjunct biofeedback versus attention-matched rehabilitation adjunct; independent sampling unit: participant; repeated measurement unit: anxiety-interference and ECG visit; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Rehabilitation attendance after assignment is a pathway, not a baseline confounder; arrhythmia-induced missing HRV must not exclude clinical outcomes.
Primary estimand: biofeedback-offer effect on anxiety-related daily interference
Candidate primary outcome: validated anxiety-related daily-interference score
Primary finalized: No
Mathematical model: type: contrast precision/power by design-matrix or simulation planning; formulas: E(Y_it | A_i, X_i, Y_i0) = alpha_t + z(A_i)^T theta_t + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Delta = sum_t w_t c^T theta_t; estimate by baseline-adjusted marginal standardization, with unit/cluster-appropriate inference.; variables: Y: validated anxiety-related daily-interference score; i: participant; t: anxiety-interference and ECG visit; the registered visit/time-window index, not another independent unit; A: participant randomized adjunct biofeedback versus attention-matched rehabilitation adjunct; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: biofeedback-offer effect on anxiety-related daily interference; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; z(A): prespecified indicators for the actual randomized arms; two arms reduce to one indicator; family code: RCT
Sample size design: family: contrast precision/power by design-matrix or simulation planning; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Rehabilitation attendance after assignment is a pathway, not a baseline confounder; arrhythmia-induced missing HRV must not exclude clinical outcomes.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: HRV improvement without functional benefit, or equal benefit from matched education, would weaken biofeedback-specific clinical utility.; preanalysis gate: Rehabilitation attendance after assignment is a pathway, not a baseline confounder; arrhythmia-induced missing HRV must not exclude clinical outcomes.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Rehabilitation attendance after assignment is a pathway, not a baseline confounder; arrhythmia-induced missing HRV must not exclude clinical outcomes.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Rehabilitation attendance after assignment is a pathway, not a baseline confounder; arrhythmia-induced missing HRV must not exclude clinical outcomes.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/187; question: Does adding low-burden HRV biofeedback to existing cardiac rehabilitation improve daily anxiety-related functioning beyond an attention-matched program?; aims: Establish interpretable measurements under stable medication and rhythm conditions.
Randomize adjunctive biofeedback versus matched stress education.
Determine whether functional benefit persists after supervised contact ends.; design verbatim planning only: Planning assumption: 80 clinically stable rehabilitation patients complete an eight-week parallel adjunct trial with three-month follow-up. Existing rehabilitation and prescribed treatment continue in both groups; exercise is not manipulated.; outcomes verbatim: Primary: anxiety-related daily interference. Secondary: quality of life, perceived control, standardized ECG HRV, adherence and adverse symptoms.; analysis verbatim inherited not authoritative if refined: Stratify randomization by relevant medication class; adjust baseline interference and rehabilitation attendance. Analyze arrhythmia-related missing physiology explicitly and register one clinical primary endpoint.; falsification verbatim: HRV improvement without functional benefit, or equal benefit from matched education, would weaken biofeedback-specific clinical utility.; source ids unverified leads: hrv-biofeedback-cardiac-meta-2026
slow-breathing-hrv-review-2025
belief-effects-health-review-2026; prospectus precedence: No

N21 — Durable learning rather than training-game improvement

Rank: 166
Cluster: Cognitive performance/learning
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT; allocation or exposure unit: participant assigned spaced retrieval versus matched repeated exposure; independent sampling unit: participant; repeated measurement unit: practical transfer task and delayed follow-up; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Confirm allocation sequence; alternate forms and held-out procedures distinguish practical transfer from practiced-item gains.
Primary estimand: retrieval-practice assignment effect on delayed untrained practical application
Candidate primary outcome: blinded practical-application score at three months
Primary finalized: No
Mathematical model: type: contrast precision/power by design-matrix or simulation planning; formulas: E(Y_it | A_i, X_i, Y_i0) = alpha_t + z(A_i)^T theta_t + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Delta = sum_t w_t c^T theta_t; estimate by baseline-adjusted marginal standardization, with unit/cluster-appropriate inference.; variables: Y: blinded practical-application score at three months; i: participant; t: practical transfer task and delayed follow-up; the registered visit/time-window index, not another independent unit; A: participant assigned spaced retrieval versus matched repeated exposure; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: retrieval-practice assignment effect on delayed untrained practical application; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; z(A): prespecified indicators for the actual randomized arms; two arms reduce to one indicator; family code: RCT
Sample size design: family: contrast precision/power by design-matrix or simulation planning; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Confirm allocation sequence; alternate forms and held-out procedures distinguish practical transfer from practiced-item gains.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: No delayed application advantage despite better trained-item recall would undermine useful transfer; broad enhancement requires independently replicated untrained-domain gains.; preanalysis gate: Confirm allocation sequence; alternate forms and held-out procedures distinguish practical transfer from practiced-item gains.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Confirm allocation sequence; alternate forms and held-out procedures distinguish practical transfer from practiced-item gains.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Confirm allocation sequence; alternate forms and held-out procedures distinguish practical transfer from practiced-item gains.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/165; question: Does spaced retrieval practice improve delayed use of newly learned information in realistic tasks beyond matched repeated exposure?; aims: Estimate immediate learning under matched exposure and time.
Randomize retrieval versus rereading schedules for unfamiliar practical material.
Measure delayed application and untrained cognitive outcomes separately.; design verbatim planning only: Planning assumption: 120 adults from varied educational backgrounds complete six weeks of online training with supervised baseline and three-month follow-up. Content concerns unfamiliar everyday procedures rather than privileged academic knowledge.; outcomes verbatim: Primary: blinded scoring of delayed practical application. Secondary: retention, confidence calibration, unrelated working-memory tasks, dropout and time cost.; analysis verbatim inherited not authoritative if refined: Preregister one transfer outcome and use alternate assessment forms. Adjust baseline knowledge and literacy, correct secondary-domain multiplicity, and analyze differential adherence without replacing intention-to-treat estimates.; falsification verbatim: No delayed application advantage despite better trained-item recall would undermine useful transfer; broad enhancement requires independently replicated untrained-domain gains.; source ids unverified leads: multisensory-learning-neuroplasticity-2025
vr-cognitive-remediation-rct-2026; prospectus precedence: No

N22 — Testing personal focus rhythms without assuming a universal cycle

Rank: 227
Cluster: Cognitive performance/learning
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: CROSS; allocation or exposure unit: candidate participant scheduling block randomized/counterbalanced after frozen baseline rhythm estimation; independent sampling unit: participant; repeated measurement unit: standardized work simulation within scheduling block; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Counterbalancing alone is not randomization; confirm random sequence; frequency search and schedule rules frozen using separate baseline data.
Primary estimand: rhythm-informed versus equally frequent uninformed break schedule effect on error-adjusted output
Candidate primary outcome: error-adjusted work-simulation output
Primary finalized: No
Mathematical model: type: paired/crossover contrast precision with period, carryover and within-unit covariance; simulation for actual design; formulas: g(E[Y_ipt | A_ip,X_i]) = alpha_i + tau A_ip + pi_p + rho A_i,p-1 + gamma^T X_i + f_time(t) + b_cluster(i,p).
Target the registered within-unit assignment contrast, standardized over period/history; direct and lag effects are distinct parameters.; variables: Y: error-adjusted work-simulation output; i: participant; t: standardized work simulation within scheduling block; the registered visit/time-window index, not another independent unit; A: candidate participant scheduling block randomized/counterbalanced after frozen baseline rhythm estimation; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: rhythm-informed versus equally frequent uninformed break schedule effect on error-adjusted output; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; p: randomized crossover period/session/episode, not the outcome probability; pi p: period effect; include calendar/season/sequence terms supported by the design; rho: prespecified lagged-assignment/carryover parameter; cannot in general identify arbitrary persistent carryover in a short two-period trial; f time(t): prespecified within-period recovery/day trend; g: identity for mean outcomes, log for positive/count means, or logit for binary/proportion outcomes as specified below; family code: CROSS
Sample size design: family: paired/crossover contrast precision with period, carryover and within-unit covariance; simulation for actual design; formula: For a simple two-condition paired mean only: Var(Delta_hat)=sigma_D^2/n_independent; actual cluster/period/lag models require design-specific simulation or c^T(D^T V^(-1)D)^(-1)c.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; sigma D^2: variance of independent-unit paired contrasts including measurement error; n independent: number of independent participants/teams/donors, not repeated samples; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Counterbalancing alone is not randomization; confirm random sequence; frequency search and schedule rules frozen using separate baseline data.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Concealed randomized sequence, a scientifically justified recovery interval and no anticipation; period/carryover cannot be repaired solely by a nonsignificant test.
If carryover is plausible, report lag/period and first-period or otherwise valid design sensitivity; do not drop periods based only on a post hoc carryover p-value.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Unreliable rhythms or no advantage over shuffled schedules would undermine personalized ultradian scheduling despite plausible rhythm biology.; preanalysis gate: Counterbalancing alone is not randomization; confirm random sequence; frequency search and schedule rules frozen using separate baseline data.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Counterbalancing alone is not randomization; confirm random sequence; frequency search and schedule rules frozen using separate baseline data.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Compare planned lag/carryover structures and a valid first-period sensitivity where its sample supports inference; assess informative absent periods.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Counterbalancing alone is not randomization; confirm random sequence; frequency search and schedule rules frozen using separate baseline data.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/226; question: Do reproducible within-day attention patterns improve break scheduling beyond equally frequent breaks placed without rhythm information?; aims: Estimate attention-pattern reliability across independent baseline weeks.
Compare individualized timing with shuffled timing while matching break duration.
Determine whether schedule benefits replicate on unfamiliar tasks.; design verbatim planning only: Planning assumption: 70 adults complete three baseline weeks followed by counterbalanced two-week scheduling blocks. Brief cognitive probes and self-selected work tasks are collected without changing sleep opportunity.; outcomes verbatim: Primary: error-adjusted output on standardized work simulations. Secondary: fatigue, perceived focus, schedule burden and stability of estimated phase.; analysis verbatim inherited not authoritative if refined: Use held-out baseline data to select rhythms and frozen scheduling rules. Model circadian time, task practice, period and carryover; correct exploratory frequency searches to prevent invented cycles.; falsification verbatim: Unreliable rhythms or no advantage over shuffled schedules would undermine personalized ultradian scheduling despite plausible rhythm biology.; source ids unverified leads: circadian-ultradian-clinical-review-2024
digital-n-of-1-physiology-2026
light-circadian-review; prospectus precedence: No

N23 — When multisensory learning helps and when it overloads

Rank: 167
Cluster: Cognitive performance/learning
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: FACTOR; allocation or exposure unit: participant randomized coordinated cues and learning-load conditions; independent sampling unit: participant; repeated measurement unit: novel practical task item and one-month retention test; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Confirm exact factorial levels and assignment unit; customary sensory aids remain; age averages do not identify sensory-access mechanisms.
Primary estimand: cue-by-load interaction on retained practical performance
Candidate primary outcome: retained novel assembly/navigation-task score
Primary finalized: No
Mathematical model: type: factorial main-effect/interaction planning with actual cross-classified or cluster-period covariance; formulas: E(Y_i | C_i,L_i,X_i)=alpha+z(C_i)^T beta_C+beta_L L_i+L_i z(C_i)^T beta_CL+gamma^T X_i.
C has three cue levels (complementary, redundant, single-channel) and L has the two prespecified load levels; test the registered cue-by-load contrasts or omnibus interaction with an explicit primary multiplicity rule.; variables: Y: retained novel assembly/navigation-task score; i: participant; t: novel practical task item and one-month retention test; the registered visit/time-window index, not another independent unit; A: participant randomized coordinated cues and learning-load conditions; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: cue-by-load interaction on retained practical performance; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; B,C: the two assigned factors named in the per-thesis allocation specification; levels and contrasts must be fixed; mu bc: marginal mean outcome under the specified joint factor levels; I: additive interaction estimand; beta B,beta C,beta BC: factor coefficients; for nonidentity links derive marginal contrasts rather than interpreting coefficients as risk differences; C i: three-level randomized cue factor, not a binary cue indicator; z(C i): two prespecified cue contrasts for the three levels; L i: randomized two-level load factor; beta CL: two-component cue-by-load interaction vector; family code: FACTOR
Sample size design: family: factorial main-effect/interaction planning with actual cross-classified or cluster-period covariance; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Confirm exact factorial levels and assignment unit; customary sensory aids remain; age averages do not identify sensory-access mechanisms.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Factor allocation and coding must match the design; interaction inference is separate from main-effect inference and needs its own precision.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: No retained advantage for complementary cues, or benefits explained only by additional information exposure, would weaken the proposed integration mechanism.; preanalysis gate: Confirm exact factorial levels and assignment unit; customary sensory aids remain; age averages do not identify sensory-access mechanisms.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Confirm exact factorial levels and assignment unit; customary sensory aids remain; age averages do not identify sensory-access mechanisms.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Confirm exact factorial levels and assignment unit; customary sensory aids remain; age averages do not identify sensory-access mechanisms.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/166; question: Does coordinated audiovisual learning improve durable procedural memory in older adults, or merely increase engagement and cognitive load?; aims: Validate matched instructional materials across sensory conditions.
Randomize complementary, redundant and single-channel instruction across low and high complexity.
Test retained practical performance and sensory-access moderation.; design verbatim planning only: Planning assumption: 100 community-dwelling adults aged fifty-five and older complete a factorial laboratory learning study and one-month follow-up. Participants use customary sensory aids; content and exposure time are matched.; outcomes verbatim: Primary: retained performance on a novel practical assembly or navigation task. Secondary: immediate accuracy, perceived load, confidence and accessibility burden.; analysis verbatim inherited not authoritative if refined: Register the cue-by-load interaction and a small set of sensory-function moderators. Use blinded scoring and alternate transfer tasks; distinguish impairment-related access effects from age averages.; falsification verbatim: No retained advantage for complementary cues, or benefits explained only by additional information exposure, would weaken the proposed integration mechanism.; source ids unverified leads: multisensory-learning-neuroplasticity-2025
vr-cognitive-remediation-rct-2026; prospectus precedence: No

N24 — Reward schedules, sustained effort and the limits of dopamine narratives

Rank: 168
Cluster: Cognitive performance/learning
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT3; allocation or exposure unit: participant randomized fixed/performance-linked/autonomy-supportive reward schedule; independent sampling unit: participant; repeated measurement unit: learning-task and delayed unrewarded session; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Equal compensation floors protect interpretation; persistence and accuracy must be reported jointly; no dopamine mechanism is identified.
Primary estimand: prespecified reward-schedule contrast in voluntary persistence after reward removal
Candidate primary outcome: time/trials voluntarily persisted in delayed unrewarded task, with chosen scoring rule
Primary finalized: No
Mathematical model: type: contrast precision/power by design-matrix or simulation planning; formulas: E(Y_it | A_i, X_i, Y_i0) = alpha_t + z(A_i)^T theta_t + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Delta = sum_t w_t c^T theta_t; estimate by baseline-adjusted marginal standardization, with unit/cluster-appropriate inference.; variables: Y: time/trials voluntarily persisted in delayed unrewarded task, with chosen scoring rule; i: participant; t: learning-task and delayed unrewarded session; the registered visit/time-window index, not another independent unit; A: participant randomized fixed/performance-linked/autonomy-supportive reward schedule; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: prespecified reward-schedule contrast in voluntary persistence after reward removal; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; z(A): prespecified indicators for the actual randomized arms; two arms reduce to one indicator; family code: RCT3
Sample size design: family: contrast precision/power by design-matrix or simulation planning; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Equal compensation floors protect interpretation; persistence and accuracy must be reported jointly; no dopamine mechanism is identified.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Primary contrast is fixed before observing outcomes; the third arm does not create another uncorrected primary test.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Increased rewarded activity followed by lower unrewarded persistence would challenge a durable motivation benefit, even if immediate productivity rises.; preanalysis gate: Equal compensation floors protect interpretation; persistence and accuracy must be reported jointly; no dopamine mechanism is identified.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Equal compensation floors protect interpretation; persistence and accuracy must be reported jointly; no dopamine mechanism is identified.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Equal compensation floors protect interpretation; persistence and accuracy must be reported jointly; no dopamine mechanism is identified.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/167; question: Can transparent reward scheduling improve sustained task engagement without reducing later intrinsic motivation?; aims: Disentangle effort allocation, reward prediction and subjective enjoyment in matched tasks.
Randomize transparent reward schedules during four weeks of voluntary learning.
Assess unrewarded persistence after rewards are removed.; design verbatim planning only: Planning assumption: 110 adults complete equivalent learning content under fixed, performance-linked or autonomy-supportive rewards, with compensation floors equal across groups and a delayed unrewarded session.; outcomes verbatim: Primary: voluntary persistence after reward removal. Secondary: learning retention, effort allocation, enjoyment, perceived autonomy and reported pressure.; analysis verbatim inherited not authoritative if refined: Use intention-to-treat comparisons with preregistered schedule contrasts; adjust baseline motivation and financial strain. Analyze task accuracy alongside effort to distinguish useful engagement from hurried responding.; falsification verbatim: Increased rewarded activity followed by lower unrewarded persistence would challenge a durable motivation benefit, even if immediate productivity rises.; source ids unverified leads: dopamine-reward-strategy-2026
dopamine-reward-human-meta-2025
dopamine-timescales-review-2025; prospectus precedence: No

N25 — Enhancement expectations and metacognitive miscalibration

Rank: 104
Cluster: Cognitive performance/learning
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT; allocation or exposure unit: participant randomized truthful enhancement framing with identical evidence; independent sampling unit: participant; repeated measurement unit: confidence and accuracy judgment on untrained task; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Define confidence scale and proper scoring rule; guessing beliefs after randomization does not isolate expectancy causally.
Primary estimand: framing assignment effect on prespecified confidence-accuracy calibration loss
Candidate primary outcome: fixed calibration loss on untrained tasks
Primary finalized: No
Mathematical model: type: contrast precision/power by design-matrix or simulation planning; formulas: E(Y_it | A_i, X_i, Y_i0) = alpha_t + z(A_i)^T theta_t + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Delta = sum_t w_t c^T theta_t; estimate by baseline-adjusted marginal standardization, with unit/cluster-appropriate inference.; variables: Y: fixed calibration loss on untrained tasks; i: participant; t: confidence and accuracy judgment on untrained task; the registered visit/time-window index, not another independent unit; A: participant randomized truthful enhancement framing with identical evidence; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: framing assignment effect on prespecified confidence-accuracy calibration loss; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; z(A): prespecified indicators for the actual randomized arms; two arms reduce to one indicator; family code: RCT
Sample size design: family: contrast precision/power by design-matrix or simulation planning; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Define confidence scale and proper scoring rule; guessing beliefs after randomization does not isolate expectancy causally.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Equivalent confidence and objective gains across framings would undermine expectancy-driven miscalibration; accurate self-assessment despite framing would also challenge the mechanism.; preanalysis gate: Define confidence scale and proper scoring rule; guessing beliefs after randomization does not isolate expectancy causally.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Define confidence scale and proper scoring rule; guessing beliefs after randomization does not isolate expectancy causally.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Define confidence scale and proper scoring rule; guessing beliefs after randomization does not isolate expectancy causally.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/103; question: Does enhancement-oriented framing change confidence more than objective cognitive performance during an otherwise identical attention program?; aims: Measure baseline enhancement beliefs and confidence calibration.
Randomize equally accurate benefit-focused or uncertainty-focused descriptions of the same training.
Track whether confidence changes predict continued use despite absent objective gain.; design verbatim planning only: Planning assumption: 120 adults complete a four-week attention program with assessor blinding. Information conditions differ in emphasis while disclosing identical evidence and uncertainty; informed consent explains communication research.; outcomes verbatim: Primary: confidence-accuracy calibration on untrained tasks. Secondary: sustained attention, perceived focus, willingness to pay and continuation decisions.; analysis verbatim inherited not authoritative if refined: Prespecify calibration scoring and one belief interaction. Control practice effects with alternate forms, model attrition by expectations and separate subjective from objective outcomes under a multiplicity plan.; falsification verbatim: Equivalent confidence and objective gains across framings would undermine expectancy-driven miscalibration; accurate self-assessment despite framing would also challenge the mechanism.; source ids unverified leads: belief-effects-health-review-2026
nocebo-communication-review-2026
neurofeedback-adhd-meta-2025; prospectus precedence: No

N26 — Predicting cognitive fatigue before recommending a break

Rank: 173
Cluster: Cognitive performance/learning
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: MRT; allocation or exposure unit: eligible fatigue window randomized to brief-break offer versus no offer; independent sampling unit: participant; repeated measurement unit: decision moment and subsequent task window; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Freeze fatigue threshold and models before trial outcomes; prior breaks influence availability, and prediction quality alone is not causal break benefit.
Primary estimand: proximal assignment excursion effect on error-adjusted performance among available moments
Candidate primary outcome: postdecision change in error-adjusted task performance
Primary finalized: No
Mathematical model: type: micro-randomized-trial power/precision simulation for availability, randomization probabilities, proximal correlation and lagged effects; formulas: E[Y_i,t+Delta(1)-Y_i,t+Delta(0) | I_it=1,S_it] = f(S_it)^T beta, a proximal excursion effect under the registered future-treatment policy.
Estimate beta using centered A_it-p_it with known p_it=P(A_it=1 | H_it,I_it=1); a basic unmoderated moment is sum I_it(A_it-p_it)Y_i,t+Delta / sum I_it p_it(1-p_it), with prespecified weighting/augmentation as needed.; variables: Y: postdecision change in error-adjusted task performance; i: participant; t: decision moment and subsequent task window; the registered visit/time-window index, not another independent unit; A: eligible fatigue window randomized to brief-break offer versus no offer; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: proximal assignment excursion effect on error-adjusted performance among available moments; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; I it: availability/eligibility indicator decided before current randomization; H it: observed predecision history including prior prompts/breaks; p it: known, positive randomized prompt/show probabilities among available moments; S it: prespecified predecision moderator vector; Delta: fixed proximal outcome delay/window, not a mean difference in this formula; f(S)^T beta: registered moderation model; beta is estimated with participant/coach dependence; family code: MRT
Sample size design: family: micro-randomized-trial power/precision simulation for availability, randomization probabilities, proximal correlation and lagged effects; formula: Simulate the actual availability/history/randomization process and estimate Var(beta_hat) with participant/coach dependence; moment count alone is not the independent sample size.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Freeze fatigue threshold and models before trial outcomes; prior breaks influence availability, and prediction quality alone is not causal break benefit.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Sequential randomization at eligible moments, positivity of prompt/no prompt, consistency and prespecified availability; proximal effect does not automatically identify a long-term strategy effect.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Accurate fatigue prediction with equal break effects at random moments would undermine targeted intervention value.; preanalysis gate: Freeze fatigue threshold and models before trial outcomes; prior breaks influence availability, and prediction quality alone is not causal break benefit.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Freeze fatigue threshold and models before trial outcomes; prior breaks influence availability, and prediction quality alone is not causal break benefit.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Vary availability models, lagged effects, proximal windows and informative postprompt missingness without searching a favorable window.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Freeze fatigue threshold and models before trial outcomes; prior breaks influence availability, and prediction quality alone is not causal break benefit.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/172; question: Do person-adapted fatigue models identify moments when a brief break causally restores performance rather than merely predicting low scores?; aims: Train fatigue models using baseline task and self-report data.
Validate on held-out people, weeks and unfamiliar tasks.
Micro-randomize breaks at predicted and randomly selected moments.; design verbatim planning only: Planning assumption: 80 adults provide six weeks of brief remote cognitive tasks and optional noninvasive wearable features. Frozen models determine eligible windows; all participants retain the right to rest at any time.; outcomes verbatim: Primary: post-randomization change in error-adjusted task performance. Secondary: calibration, subjective fatigue, false alerts and interruption burden.; analysis verbatim inherited not authoritative if refined: Use time-respecting validation and micro-randomized causal estimators; account for prior breaks, availability and task difficulty. Register one fatigue threshold before trial analysis and report subgroup uncertainty.; falsification verbatim: Accurate fatigue prediction with equal break effects at random moments would undermine targeted intervention value.; source ids unverified leads: digital-phenotyping-relapse-review-2026
wearable-stress-ml-review-2025
digital-n-of-1-physiology-2026; prospectus precedence: No

N27 — Practical cognitive compensation for menopausal brain fog

Rank: 123
Cluster: Cognitive performance/learning
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT_PROP; allocation or exposure unit: participant randomized prospective-memory support versus comparator; independent sampling unit: participant; repeated measurement unit: prospectively declared intention and follow-up; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Intervention could change which intentions are recorded; fix denominator and assess goal difficulty; no claim of reversing a biological deficit.
Primary estimand: support-offer effect on completion of declared intentions under a fixed scoring rule
Candidate primary outcome: completed intentions divided by eligible prospectively recorded intentions
Primary finalized: No
Mathematical model: type: proportion/binomial design with opportunity denominators, overdispersion and independent-person precision; formulas: K_it | n_it, p_it ~ Binomial(n_it,p_it) with beta-binomial or clustered dependence if needed; logit(p_it)=alpha_t+z(A_i)^T theta_t+gamma^T X_i+u_i.
Report the chosen participant-average or opportunity-average probability difference; define n_it=0 before analysis.; variables: Y: completed intentions divided by eligible prospectively recorded intentions; i: participant; t: prospectively declared intention and follow-up; the registered visit/time-window index, not another independent unit; A: participant randomized prospective-memory support versus comparator; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: support-offer effect on completion of declared intentions under a fixed scoring rule; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; K it: number of primary successes/events within the registered opportunity set; n it: number of eligible opportunities, defined independently of observed success; p it: success/event probability under that explicitly defined opportunity set; z(A): prespecified indicators for the actual arms or randomized policy; family code: RCT_PROP
Sample size design: family: proportion/binomial design with opportunity denominators, overdispersion and independent-person precision; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Intervention could change which intentions are recorded; fix denominator and assess goal difficulty; no claim of reversing a biological deficit.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
If the intervention changes opportunity counts, participant-average and pooled-opportunity targets differ; choose and report the intended target.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Improved self-confidence without fewer missed intentions, or unsustainable reminder burden, would weaken practical effectiveness.; preanalysis gate: Intervention could change which intentions are recorded; fix denominator and assess goal difficulty; no claim of reversing a biological deficit.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Intervention could change which intentions are recorded; fix denominator and assess goal difficulty; no claim of reversing a biological deficit.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Intervention could change which intentions are recorded; fix denominator and assess goal difficulty; no claim of reversing a biological deficit.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/122; question: Can prospective-memory support improve everyday functioning during menopause without claiming to reverse a biological cognitive deficit?; aims: Characterize discordance among cognitive complaints, task performance and daily missed intentions.
Randomize contextual memory support versus matched general information.
Test whether functional benefit varies with symptom burden rather than laboratory impairment.; design verbatim planning only: Planning assumption: 90 perimenopausal or early postmenopausal adults reporting brain fog complete a ten-week parallel trial and three-month follow-up. Clinical assessment and usual symptom care continue independently.; outcomes verbatim: Primary: prospectively recorded completion of participant-defined intentions, corroborated where feasible. Secondary: cognitive interference, confidence, untrained memory tasks and reminder burden.; analysis verbatim inherited not authoritative if refined: Prespecify functional scoring before randomization; adjust baseline sleep, mood and stable treatment. Separate support-mediated improvements from exploratory symptom associations and avoid causal hormone interpretations.; falsification verbatim: Improved self-confidence without fewer missed intentions, or unsustainable reminder burden, would weaken practical effectiveness.; source ids unverified leads: menopause-cognition-sleep-review-2025
menopause-behavioral-symptoms-review-2026
behavior-stress-sleep-rct-2026; prospectus precedence: No

N28 — Does immersive cognitive remediation transfer outside virtual reality?

Rank: 153
Cluster: Cognitive performance/learning
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: FACTOR; allocation or exposure unit: participant randomized immersion and context-variation training components factorially; independent sampling unit: participant; repeated measurement unit: untrained functional-capacity scenario at follow-up; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Context variation is a randomized training component, not merely a follow-up scenario; use independent untrained scenarios and scale invariance; cybersickness and medication changes affect retention.
Primary estimand: factorial immersion/context-variation effects and registered immersion-by-context-variation interaction on functional transfer
Candidate primary outcome: validated untrained functional-capacity task score
Primary finalized: No
Mathematical model: type: factorial main-effect/interaction planning with actual cross-classified or cluster-period covariance; formulas: E(Y_ict | B_ict, C_ict, X_i) = alpha_t + beta_B B_ict + beta_C C_ict + beta_BC B_ict C_ict + gamma^T X_i + u_i + b_c.
I = mu_11 - mu_10 - mu_01 + mu_00 = beta_BC on the additive scale; main effects are standardized over the other randomized factor.; variables: Y: validated untrained functional-capacity task score; i: participant; t: untrained functional-capacity scenario at follow-up; the registered visit/time-window index, not another independent unit; A: participant randomized immersion and context-variation training components factorially; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: factorial immersion/context-variation effects and registered immersion-by-context-variation interaction on functional transfer; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; B,C: the two assigned factors named in the per-thesis allocation specification; levels and contrasts must be fixed; mu bc: marginal mean outcome under the specified joint factor levels; I: additive interaction estimand; beta B,beta C,beta BC: factor coefficients; for nonidentity links derive marginal contrasts rather than interpreting coefficients as risk differences; family code: FACTOR
Sample size design: family: factorial main-effect/interaction planning with actual cross-classified or cluster-period covariance; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Context variation is a randomized training component, not merely a follow-up scenario; use independent untrained scenarios and scale invariance; cybersickness and medication changes affect retention.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Factor allocation and coding must match the design; interaction inference is separate from main-effect inference and needs its own precision.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Presence or trained-task gains without functional transfer would undermine immersion-specific utility; no interaction challenges the proposed mechanism.; preanalysis gate: Context variation is a randomized training component, not merely a follow-up scenario; use independent untrained scenarios and scale invariance; cybersickness and medication changes affect retention.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Context variation is a randomized training component, not merely a follow-up scenario; use independent untrained scenarios and scale invariance; cybersickness and medication changes affect retention.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Context variation is a randomized training component, not merely a follow-up scenario; use independent untrained scenarios and scale invariance; cybersickness and medication changes affect retention.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/152; question: Does immersion add functional transfer to otherwise matched cognitive remediation in clinically stable mood or psychosis-spectrum conditions?; aims: Develop equivalent immersive and desktop tasks with matched cognitive demands.
Randomize immersion and context-variation components factorially.
Measure blinded everyday functional performance after training and follow-up.; design verbatim planning only: Planning assumption: 80 clinically stable adults recruited through an established mental-health service complete an eight-week adjunctive program. Existing treatment is maintained; allocation is concealed and assessors are blinded.; outcomes verbatim: Primary: validated functional-capacity task using untrained scenarios. Secondary: cognition, symptoms, participation, presence, cybersickness and retention at three months.; analysis verbatim inherited not authoritative if refined: Register the immersion-by-context interaction and assess practice effects with alternate scenarios. Adjust baseline functioning and medication stability; compare dropout and adverse experiences across conditions.; falsification verbatim: Presence or trained-task gains without functional transfer would undermine immersion-specific utility; no interaction challenges the proposed mechanism.; source ids unverified leads: vr-cognitive-remediation-rct-2026
multisensory-learning-neuroplasticity-2025; prospectus precedence: No

N29 — Notification batching and restoration of attention after interruption

Rank: 169
Cluster: Cognitive performance/learning
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: CROSS; allocation or exposure unit: participant-controlled notification block; randomized sequence must be confirmed; independent sampling unit: participant; repeated measurement unit: interruption/resumption simulation within block; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Emergency bypass remains common; carryover and workload differ by block; observational screen-time association is not the randomized estimand.
Primary estimand: batching versus usual notification assignment effect on accuracy-adjusted resumption time
Candidate primary outcome: accuracy-adjusted resumption time on standardized complex tasks
Primary finalized: No
Mathematical model: type: paired/crossover contrast precision with period, carryover and within-unit covariance; simulation for actual design; formulas: g(E[Y_ipt | A_ip,X_i]) = alpha_i + tau A_ip + pi_p + rho A_i,p-1 + gamma^T X_i + f_time(t) + b_cluster(i,p).
Target the registered within-unit assignment contrast, standardized over period/history; direct and lag effects are distinct parameters.; variables: Y: accuracy-adjusted resumption time on standardized complex tasks; i: participant; t: interruption/resumption simulation within block; the registered visit/time-window index, not another independent unit; A: participant-controlled notification block; randomized sequence must be confirmed; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: batching versus usual notification assignment effect on accuracy-adjusted resumption time; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; p: randomized crossover period/session/episode, not the outcome probability; pi p: period effect; include calendar/season/sequence terms supported by the design; rho: prespecified lagged-assignment/carryover parameter; cannot in general identify arbitrary persistent carryover in a short two-period trial; f time(t): prespecified within-period recovery/day trend; g: identity for mean outcomes, log for positive/count means, or logit for binary/proportion outcomes as specified below; family code: CROSS
Sample size design: family: paired/crossover contrast precision with period, carryover and within-unit covariance; simulation for actual design; formula: For a simple two-condition paired mean only: Var(Delta_hat)=sigma_D^2/n_independent; actual cluster/period/lag models require design-specific simulation or c^T(D^T V^(-1)D)^(-1)c.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; sigma D^2: variance of independent-unit paired contrasts including measurement error; n independent: number of independent participants/teams/donors, not repeated samples; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Emergency bypass remains common; carryover and workload differ by block; observational screen-time association is not the randomized estimand.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Concealed randomized sequence, a scientifically justified recovery interval and no anticipation; period/carryover cannot be repaired solely by a nonsignificant test.
If carryover is plausible, report lag/period and first-period or otherwise valid design sensitivity; do not drop periods based only on a post hoc carryover p-value.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Fewer notifications without reduced resumption costs, or communication harm outweighing performance benefit, would undermine the proposed intervention.; preanalysis gate: Emergency bypass remains common; carryover and workload differ by block; observational screen-time association is not the randomized estimand.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Emergency bypass remains common; carryover and workload differ by block; observational screen-time association is not the randomized estimand.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Compare planned lag/carryover structures and a valid first-period sensitivity where its sample supports inference; assess informative absent periods.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Emergency bypass remains common; carryover and workload differ by block; observational screen-time association is not the randomized estimand.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/168; question: Does batching nonurgent smartphone notifications improve recovery of complex-task performance without reducing necessary social responsiveness?; aims: Estimate individual resumption costs under controlled interruptions.
Randomize batching and usual delivery in counterbalanced real-world blocks.
Test whether resumption improvements predict meaningful work completion and communication satisfaction.; design verbatim planning only: Planning assumption: 90 adults complete six weeks of participant-controlled phone settings plus standardized work simulations; emergency contacts always bypass batching. Employer-required rapid-response roles are analyzed separately or excluded.; outcomes verbatim: Primary: accuracy-adjusted resumption time after interruption. Secondary: completed planned tasks, perceived autonomy, missed important messages and social strain.; analysis verbatim inherited not authoritative if refined: Within-person models include workload, urgency, sequence and carryover; register baseline control as one moderator. Avoid interpreting screen-time reduction alone as benefit or inferring dopamine changes.; falsification verbatim: Fewer notifications without reduced resumption costs, or communication harm outweighing performance benefit, would undermine the proposed intervention.; source ids unverified leads: dopamine-timescales-review-2025
dopamine-reward-strategy-2026
belief-effects-health-review-2026; prospectus precedence: No

N30 — How much cognitive self-tracking improvement is repeated testing?

Rank: 35
Cluster: Cognitive performance/learning
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: TRANSFER; allocation or exposure unit: participant randomized frequent/weekly/sparse testing schedule; independent sampling unit: participant; repeated measurement unit: practiced-domain and unfamiliar-domain endpoint task; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Use a single difference-of-effects contrast; external scales fixed, no within-arm standardization; p-value significance differences do not test transfer.
Primary estimand: frequent-minus-sparse assignment effect on practiced standardized outcome minus corresponding unfamiliar-task effect
Candidate primary outcome: externally scaled practiced performance and unfamiliar-task composite at twelve weeks
Primary finalized: No
Mathematical model: type: multiarm paired-outcome difference-of-effects contrast precision using cross-domain covariance; formulas: Delta_practiced=E[Y_P|frequent]-E[Y_P|sparse]; Delta_unfamiliar=E[Y_U|frequent]-E[Y_U|sparse].
Primary D=Delta_practiced-Delta_unfamiliar; Var(D_hat)=Var(Delta_P_hat)+Var(Delta_U_hat)-2Cov(Delta_P_hat,Delta_U_hat), using participant-level joint inference and baseline adjustment.; variables: Y: externally scaled practiced performance and unfamiliar-task composite at twelve weeks; i: participant; t: practiced-domain and unfamiliar-domain endpoint task; the registered visit/time-window index, not another independent unit; A: participant randomized frequent/weekly/sparse testing schedule; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: frequent-minus-sparse assignment effect on practiced standardized outcome minus corresponding unfamiliar-task effect; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; Y P,Y U: practiced-domain and unfamiliar-task scores on externally fixed compatible standard scales; D: single difference-of-assignment-effects contrast; Delta P hat,Delta U hat: correlated randomized contrasts from the same participants; family code: TRANSFER
Sample size design: family: multiarm paired-outcome difference-of-effects contrast precision using cross-domain covariance; formula: Use design-specific simulation or independent-unit resampling for the stated estimand and target confidence width/power; preserve all allocation, censoring, clustering and measurement-error structure.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Use a single difference-of-effects contrast; external scales fixed, no within-arm standardization; p-value significance differences do not test transfer.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
External scales/form equivalence must be fixed; frequent versus sparse is primary and the weekly arm is secondary; do not infer transfer from differing p-values.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Equivalent practice-related and unfamiliar-task gains across frequencies would weaken the measurement-artifact hypothesis.; preanalysis gate: Use a single difference-of-effects contrast; external scales fixed, no within-arm standardization; p-value significance differences do not test transfer.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Use a single difference-of-effects contrast; external scales fixed, no within-arm standardization; p-value significance differences do not test transfer.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Use a single difference-of-effects contrast; external scales fixed, no within-arm standardization; p-value significance differences do not test transfer.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/34; question: Can alternate forms and sparse assessment distinguish genuine change from practice effects in consumer-style cognitive monitoring?; aims: Estimate reliability and alternate-form equivalence before longitudinal use.
Randomize assessment frequency without assigning any enhancement intervention.
Compare score trajectories with blinded unfamiliar tasks and everyday functioning.; design verbatim planning only: Planning assumption: 120 adults complete twelve weeks of frequent, weekly or baseline-follow-up testing, with a common endpoint assessment. Randomization separates testing exposure from underlying change; participants report routine changes prospectively.; outcomes verbatim: Primary: frequency-related difference in endpoint unfamiliar-task performance versus practiced-task gains. Secondary: confidence in enhancement, regression-to-mean artifacts, burden and retention.; analysis verbatim inherited not authoritative if refined: Use item-level measurement models, randomized frequency contrasts and baseline-adjusted estimates. Audit form equivalence across education and language background; register domain families and correct multiplicity.; falsification verbatim: Equivalent practice-related and unfamiliar-task gains across frequencies would weaken the measurement-artifact hypothesis.; source ids unverified leads: n-of-1-methods-review-2025
digital-n-of-1-physiology-2026
neurofeedback-adhd-meta-2025; prospectus precedence: Yes; prospectus file: thesis-prospectuses.json; prospectus design and estimand: The participant is the analysis unit. The primary estimand is the frequent-versus-sparse assignment effect on practiced-domain standardized endpoint performance minus the corresponding effect on the unfamiliar-task composite at twelve weeks. This single difference-of-effects contrast uses externally fixed scales, not within-arm standardization. Weekly testing is a secondary exposure–response contrast. Randomization identifies effects of being offered a testing schedule, including burden and expectancy, rather than a pure biological practice mechanism.; prospectus analysis plan: Fit prespecified baseline-adjusted endpoint models and estimate the primary interaction contrast with participant-level uncertainty. Use item-level models only if pilot fit and invariance justify them. Assess form and device effects, control secondary domain multiplicity and analyze the weekly arm without multiplying primary hypotheses. Examine regression-to-the-mean patterns descriptively; avoid adjusting the randomized estimate for post-randomization routine changes. Report unfamiliar-task effects directly, not merely whether their p-value differs from practiced-task significance.; precedence note: Expanded prospectus refinements govern this scaffold; verbatim earlier brief material is retained for traceability, not as a contradictory analysis instruction. F15 is explicitly two arms; F30 has no unjustified symptom noninferiority claim.

N31 — Contingent EEG neurofeedback beyond expectancy and task practice

Rank: 154
Cluster: Neurotechnology/neurostimulation
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT; allocation or exposure unit: participant randomized contingent versus credible yoked EEG feedback; independent sampling unit: participant; repeated measurement unit: functional visit and independent neural-target task; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Usual ADHD care preserved; postassignment allocation guesses are not routine adjustment; independent regulation outside feedback is secondary.
Primary estimand: contingent-feedback assignment effect on daily executive-function impairment
Candidate primary outcome: validated daily executive-function impairment score
Primary finalized: No
Mathematical model: type: contrast precision/power by design-matrix or simulation planning; formulas: E(Y_it | A_i, X_i, Y_i0) = alpha_t + z(A_i)^T theta_t + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Delta = sum_t w_t c^T theta_t; estimate by baseline-adjusted marginal standardization, with unit/cluster-appropriate inference.; variables: Y: validated daily executive-function impairment score; i: participant; t: functional visit and independent neural-target task; the registered visit/time-window index, not another independent unit; A: participant randomized contingent versus credible yoked EEG feedback; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: contingent-feedback assignment effect on daily executive-function impairment; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; z(A): prespecified indicators for the actual randomized arms; two arms reduce to one indicator; family code: RCT
Sample size design: family: contrast precision/power by design-matrix or simulation planning; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Usual ADHD care preserved; postassignment allocation guesses are not routine adjustment; independent regulation outside feedback is secondary.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: No between-arm regulation difference undermines target engagement; regulation without functional transfer weakens enhancement claims.; preanalysis gate: Usual ADHD care preserved; postassignment allocation guesses are not routine adjustment; independent regulation outside feedback is secondary.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Usual ADHD care preserved; postassignment allocation guesses are not routine adjustment; independent regulation outside feedback is secondary.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Usual ADHD care preserved; postassignment allocation guesses are not routine adjustment; independent regulation outside feedback is secondary.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/153; question: Does genuinely contingent portable EEG feedback improve untrained functioning in adults with ADHD beyond credible yoked feedback?; aims: Validate artifact-controlled target measurement and contingency fidelity.
Randomize active contingent or yoked feedback with equal coaching.
Test delayed untrained attention and daily functioning alongside blinding success.; design verbatim planning only: Planning assumption: 70 adults with clinically confirmed ADHD complete an eight-week adjunctive trial with three-month follow-up. Stable usual treatment continues; allocation is concealed and assessors are blinded.; outcomes verbatim: Primary: validated daily executive-function impairment. Secondary: untrained attention, target regulation outside feedback, adverse experiences and participant guesses about allocation.; analysis verbatim inherited not authoritative if refined: Preregister one neural target and one functional endpoint; model baseline impairment and stable medication strata. Assess blinding without conditioning the primary causal estimate on post-randomization guesses.; falsification verbatim: No between-arm regulation difference undermines target engagement; regulation without functional transfer weakens enhancement claims.; source ids unverified leads: neurofeedback-adhd-meta-2025
neurofeedback-regulation-review-2026
belief-effects-health-review-2026; prospectus precedence: No

N32 — Neurofeedback process and tolerability in trauma-related symptoms

Rank: 222
Cluster: Neurotechnology/neurostimulation
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT_BINARY; allocation or exposure unit: participant randomized preparation/process support before identical neurofeedback; independent sampling unit: participant; repeated measurement unit: session-level target regulation and deterioration check; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Threshold and artifact rules fixed before outcomes; dissociation association remains confounded; randomized preparation does not test neurofeedback efficacy.
Primary estimand: preparation assignment effect on reliable acquisition of the frozen neural-regulation target
Candidate primary outcome: reliable target-acquisition indicator under a prespecified artifact-controlled threshold
Primary finalized: No
Mathematical model: type: binary risk-difference precision/power, simulated with actual cluster/trajectory structure; formulas: logit Pr(Y_i=1 | A_i,X_i) = alpha + z(A_i)^T theta + gamma^T X_i + b_cluster(i).
RD = E_X[p_hat(1,X) - p_hat(0,X)]; report absolute risk difference as well as any odds ratio.; variables: Y: reliable target-acquisition indicator under a prespecified artifact-controlled threshold; i: participant; t: session-level target regulation and deterioration check; the registered visit/time-window index, not another independent unit; A: participant randomized preparation/process support before identical neurofeedback; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: preparation assignment effect on reliable acquisition of the frozen neural-regulation target; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; logit(p): log[p/(1-p)]; p hat(a,X): fitted outcome probability under arm a standardized to the prespecified population; RD: marginal risk difference; z(A): prespecified indicators for the actual arms or randomized policy; family code: RCT_BINARY
Sample size design: family: binary risk-difference precision/power, simulated with actual cluster/trajectory structure; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Threshold and artifact rules fixed before outcomes; dissociation association remains confounded; randomized preparation does not test neurofeedback efficacy.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Improved comprehension without improved learning would undermine preparation efficacy; absent dissociation association challenges the proposed tolerability marker.; preanalysis gate: Threshold and artifact rules fixed before outcomes; dissociation association remains confounded; randomized preparation does not test neurofeedback efficacy.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Threshold and artifact rules fixed before outcomes; dissociation association remains confounded; randomized preparation does not test neurofeedback efficacy.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Threshold and artifact rules fixed before outcomes; dissociation association remains confounded; randomized preparation does not test neurofeedback efficacy.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/221; question: Can a standardized process-support protocol improve neurofeedback learnability while identifying dissociation-related intolerance in PTSD?; aims: Develop reproducible target-learning and adverse-experience definitions.
Randomize additional process support versus matched neutral preparation during an existing supervised program.
Relate prespecified learning trajectories and dissociation to tolerability and function.; design verbatim planning only: Planning assumption: 60 clinically stable adults with PTSD complete a six-week adjunct program in a specialist service. Both arms receive identical neurofeedback and existing treatment; only preparation is randomized.; outcomes verbatim: Primary: reliable acquisition of the preregistered neural-regulation target. Secondary: dissociation, symptom worsening, retention, daily function and comprehension.; analysis verbatim inherited not authoritative if refined: Estimate the preparation effect causally; analyze dissociation associations with explicit confounding limits. Use blinded target scoring, artifact rejection and prespecified deterioration thresholds rather than retrospective responder labels.; falsification verbatim: Improved comprehension without improved learning would undermine preparation efficacy; absent dissociation association challenges the proposed tolerability marker.; source ids unverified leads: neurofeedback-regulation-review-2026
belief-effects-health-review-2026
nocebo-communication-review-2026; prospectus precedence: No

N33 — Auricular vagus stimulation: reproducible physiology beyond HRV

Rank: 217
Cluster: Neurotechnology/neurostimulation
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: CROSS; allocation or exposure unit: participant randomized active/sham stimulation visit sequence; independent sampling unit: participant; repeated measurement unit: independent challenge-autonomic response within visit; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: No autonomic endpoint alone establishes vagal specificity; posttreatment respiration/guesses excluded from primary total-effect model; carryover and sham activity evaluated separately.
Primary estimand: active-versus-sensory-matched-sham assignment effect on fixed independent autonomic response
Candidate primary outcome: prespecified independent autonomic challenge-response endpoint
Primary finalized: No
Mathematical model: type: paired/crossover contrast precision with period, carryover and within-unit covariance; simulation for actual design; formulas: g(E[Y_ipt | A_ip,X_i]) = alpha_i + tau A_ip + pi_p + rho A_i,p-1 + gamma^T X_i + f_time(t) + b_cluster(i,p).
Target the registered within-unit assignment contrast, standardized over period/history; direct and lag effects are distinct parameters.; variables: Y: prespecified independent autonomic challenge-response endpoint; i: participant; t: independent challenge-autonomic response within visit; the registered visit/time-window index, not another independent unit; A: participant randomized active/sham stimulation visit sequence; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: active-versus-sensory-matched-sham assignment effect on fixed independent autonomic response; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; p: randomized crossover period/session/episode, not the outcome probability; pi p: period effect; include calendar/season/sequence terms supported by the design; rho: prespecified lagged-assignment/carryover parameter; cannot in general identify arbitrary persistent carryover in a short two-period trial; f time(t): prespecified within-period recovery/day trend; g: identity for mean outcomes, log for positive/count means, or logit for binary/proportion outcomes as specified below; family code: CROSS
Sample size design: family: paired/crossover contrast precision with period, carryover and within-unit covariance; simulation for actual design; formula: For a simple two-condition paired mean only: Var(Delta_hat)=sigma_D^2/n_independent; actual cluster/period/lag models require design-specific simulation or c^T(D^T V^(-1)D)^(-1)c.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; sigma D^2: variance of independent-unit paired contrasts including measurement error; n independent: number of independent participants/teams/donors, not repeated samples; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: No autonomic endpoint alone establishes vagal specificity; posttreatment respiration/guesses excluded from primary total-effect model; carryover and sham activity evaluated separately.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Concealed randomized sequence, a scientifically justified recovery interval and no anticipation; period/carryover cannot be repaired solely by a nonsignificant test.
If carryover is plausible, report lag/period and first-period or otherwise valid design sensitivity; do not drop periods based only on a post hoc carryover p-value.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Failure to replicate an active-versus-sham response under acceptable sensory matching would undermine reliable physiological efficacy. Responses reproduced by the sensory control would weaken specificity. Respiration and allocation guesses inform alternative explanations but cannot establish or exclude neural target engagement by themselves.; preanalysis gate: No autonomic endpoint alone establishes vagal specificity; posttreatment respiration/guesses excluded from primary total-effect model; carryover and sham activity evaluated separately.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: No autonomic endpoint alone establishes vagal specificity; posttreatment respiration/guesses excluded from primary total-effect model; carryover and sham activity evaluated separately.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Compare planned lag/carryover structures and a valid first-period sensitivity where its sample supports inference; assess informative absent periods.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
No autonomic endpoint alone establishes vagal specificity; posttreatment respiration/guesses excluded from primary total-effect model; carryover and sham activity evaluated separately.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/216; question: Does supervised auricular vagus stimulation produce reproducible autonomic or stress-recovery changes beyond sensory-matched sham, with respiration measured concurrently?; aims: Validate sensory matching and allocation concealment with a credible sham.
Compare active and sham sessions using concurrent respiration and ECG.
Test reproducibility and independent stress-recovery endpoints.; design verbatim planning only: Planning assumption: 50 screened adults complete counterbalanced active and sham visits with protocol-defined washout under an established clinical-device research protocol. Staff applying the intervention are separate from blinded assessors.; outcomes verbatim: Primary: prespecified independent autonomic response during a standardized challenge. Secondary: respiration-adjusted HRV, distress recovery, blinding and tolerability.; analysis verbatim inherited not authoritative if refined: Estimate the randomized active-versus-sham effect with period, sequence and baseline physiology; test carryover and correct multiple autonomic outcomes. Use respiration-standardized measurements and sensory-matching checks to assess robustness, while keeping the primary treatment estimate separate from post-treatment adjustment. Analyze allocation guesses descriptively and with explicitly assumption-dependent sensitivity analyses, rather than treating adjustment for guesses as proof that expectancy explains an effect. No proposed autonomic endpoint alone establishes vagal pathway specificity.; falsification verbatim: Failure to replicate an active-versus-sham response under acceptable sensory matching would undermine reliable physiological efficacy. Responses reproduced by the sensory control would weaken specificity. Respiration and allocation guesses inform alternative explanations but cannot establish or exclude neural target engagement by themselves.; source ids unverified leads: tvns-affective-stress-rct-2025
tavns-hemostasis-human-2026
slow-breathing-hrv-meta-2022; prospectus precedence: No

N34 — Transcranial photobiomodulation in MCI: function and durability

Rank: 191
Cluster: Neurotechnology/neurostimulation
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT; allocation or exposure unit: participant randomized supervised home photobiomodulation versus credible sham; independent sampling unit: participant with informant clustering; repeated measurement unit: instrumental-function visit and informant report; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Progression/caregiver burden create informative missingness; define death/inability strategy before registration rather than imputing ordinary function after death.
Primary estimand: assignment effect on validated instrumental function at the fixed durability endpoint
Candidate primary outcome: blinded informant-assessed instrumental-function score
Primary finalized: No
Mathematical model: type: contrast precision/power by design-matrix or simulation planning; formulas: E(Y_it | A_i, X_i, Y_i0) = alpha_t + z(A_i)^T theta_t + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Delta = sum_t w_t c^T theta_t; estimate by baseline-adjusted marginal standardization, with unit/cluster-appropriate inference.; variables: Y: blinded informant-assessed instrumental-function score; i: participant with informant clustering; t: instrumental-function visit and informant report; the registered visit/time-window index, not another independent unit; A: participant randomized supervised home photobiomodulation versus credible sham; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: assignment effect on validated instrumental function at the fixed durability endpoint; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; z(A): prespecified indicators for the actual randomized arms; two arms reduce to one indicator; family code: RCT
Sample size design: family: contrast precision/power by design-matrix or simulation planning; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Progression/caregiver burden create informative missingness; define death/inability strategy before registration rather than imputing ordinary function after death.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Practice-task gains without functional or delayed benefit would weaken clinical utility; compromised sham credibility would limit causal interpretation.; preanalysis gate: Progression/caregiver burden create informative missingness; define death/inability strategy before registration rather than imputing ordinary function after death.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Progression/caregiver burden create informative missingness; define death/inability strategy before registration rather than imputing ordinary function after death.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Progression/caregiver burden create informative missingness; define death/inability strategy before registration rather than imputing ordinary function after death.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/190; question: Does supervised home transcranial photobiomodulation improve everyday functioning in clinically characterized MCI beyond credible sham and training effects?; aims: Validate sham credibility and objective adherence in the selected device.
Randomize active or sham adjunctive treatment under specialist monitoring.
Assess delayed function, caregiver burden and untrained cognition.; design verbatim planning only: Planning assumption: 60 adults with MCI due to Alzheimer disease recruited through an established memory clinic; twelve-week parallel treatment and three-month follow-up use unchanged usual care.; outcomes verbatim: Primary: validated instrumental-function measure with blinded informant assessment. Secondary: alternate-form cognition, daily participation, adherence, caregiver setup time and adverse effects.; analysis verbatim inherited not authoritative if refined: Use baseline-adjusted intention-to-treat models with diagnosis severity strata. Register one cognitive composite, account for informant clustering and examine missingness related to progression or caregiver burden.; falsification verbatim: Practice-task gains without functional or delayed benefit would weaken clinical utility; compromised sham credibility would limit causal interpretation.; source ids unverified leads: home-tpbm-mci-rct-2026
wearable-mci-variability-2026
belief-effects-health-review-2026; prospectus precedence: No

N35 — Sleep improvement as a pathway in depression photobiomodulation trials

Rank: 225
Cluster: Neurotechnology/neurostimulation
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT_MECHANISM; allocation or exposure unit: participant randomized photobiomodulation versus comparator under stable depression care; independent sampling unit: participant; repeated measurement unit: daily-function, sleep and mood visit; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Temporal order is insufficient for mediation; treatment-induced mood/sleep confounding and expectancy prevent naive path interpretation.
Primary estimand: assignment effect on daily function; any sleep/mood mediation is separate and assumption-dependent
Candidate primary outcome: validated daily-function score at follow-up
Primary finalized: No
Mathematical model: type: contrast precision/power by design-matrix or simulation planning; formulas: E(Y_it | A_i, X_i, Y_i0) = alpha_t + z(A_i)^T theta_t + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Delta = sum_t w_t c^T theta_t; estimate by baseline-adjusted marginal standardization, with unit/cluster-appropriate inference.; variables: Y: validated daily-function score at follow-up; i: participant; t: daily-function, sleep and mood visit; the registered visit/time-window index, not another independent unit; A: participant randomized photobiomodulation versus comparator under stable depression care; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: assignment effect on daily function; any sleep/mood mediation is separate and assumption-dependent; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; z(A): prespecified indicators for the actual randomized arms; two arms reduce to one indicator; family code: RCT_MECHANISM
Sample size design: family: contrast precision/power by design-matrix or simulation planning; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Temporal order is insufficient for mediation; treatment-induced mood/sleep confounding and expectancy prevent naive path interpretation.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: No independent sleep improvement or absent delayed functional benefit would undermine a sleep-mediated pathway; unblinding would weaken specificity.; preanalysis gate: Temporal order is insufficient for mediation; treatment-induced mood/sleep confounding and expectancy prevent naive path interpretation.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Temporal order is insufficient for mediation; treatment-induced mood/sleep confounding and expectancy prevent naive path interpretation.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Temporal order is insufficient for mediation; treatment-induced mood/sleep confounding and expectancy prevent naive path interpretation.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/224; question: Are functional improvements during transcranial photobiomodulation for depression explained by improved sleep, direct mood change or expectancy?; aims: Establish temporal reliability of ambulatory sleep and mood measures.
Randomize active versus credible sham adjunctive treatment.
Test prespecified sleep-first and mood-first longitudinal pathways.; design verbatim planning only: Planning assumption: 60 clinically stable outpatients with major depression and sleep complaints enter a supervised eight-week parallel trial, with stable usual treatment and a polysomnography subset. The research uses an established investigational device protocol.; outcomes verbatim: Primary: validated daily functioning at follow-up. Secondary: sleep continuity, depression symptoms, daytime sleepiness, expectations, adverse events and blinding.; analysis verbatim inherited not authoritative if refined: Estimate treatment effects by intention to treat, then compare longitudinal mediation models with explicit no-unmeasured-mediator-confounding assumptions. Correct pathway multiplicity and avoid interpreting temporal order as proof of mechanism.; falsification verbatim: No independent sleep improvement or absent delayed functional benefit would undermine a sleep-mediated pathway; unblinding would weaken specificity.; source ids unverified leads: tpbm-mdd-sleep-rct-2025
tpbm-insomnia-rct-2025
belief-effects-health-review-2026; prospectus precedence: No

N36 — Focused ultrasound sham quality and peripheral sensory confounding

Rank: 236
Cluster: Neurotechnology/neurostimulation
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: CROSS_IMAGING; allocation or exposure unit: participant randomized/counterbalanced session condition under existing MRI-guided protocol; independent sampling unit: participant; repeated measurement unit: target-network imaging contrast, voxel and session; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Confirm random sequence; whole-brain voxels are not independent participants; sensory/expectancy adjustment alone cannot identify target-specific efficacy.
Primary estimand: condition contrast in predefined target-network imaging response after sensory-control comparisons
Candidate primary outcome: registered target-network response summary
Primary finalized: No
Mathematical model: type: paired/crossover contrast precision with period, carryover and within-unit covariance; simulation for actual design; formulas: g(E[Y_ipt | A_ip,X_i]) = alpha_i + tau A_ip + pi_p + rho A_i,p-1 + gamma^T X_i + f_time(t) + b_cluster(i,p).
Target the registered within-unit assignment contrast, standardized over period/history; direct and lag effects are distinct parameters.; variables: Y: registered target-network response summary; i: participant; t: target-network imaging contrast, voxel and session; the registered visit/time-window index, not another independent unit; A: participant randomized/counterbalanced session condition under existing MRI-guided protocol; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: condition contrast in predefined target-network imaging response after sensory-control comparisons; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; p: randomized crossover period/session/episode, not the outcome probability; pi p: period effect; include calendar/season/sequence terms supported by the design; rho: prespecified lagged-assignment/carryover parameter; cannot in general identify arbitrary persistent carryover in a short two-period trial; f time(t): prespecified within-period recovery/day trend; g: identity for mean outcomes, log for positive/count means, or logit for binary/proportion outcomes as specified below; family code: CROSS_IMAGING
Sample size design: family: paired/crossover contrast precision with period, carryover and within-unit covariance; simulation for actual design; formula: For a simple two-condition paired mean only: Var(Delta_hat)=sigma_D^2/n_independent; actual cluster/period/lag models require design-specific simulation or c^T(D^T V^(-1)D)^(-1)c.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; sigma D^2: variance of independent-unit paired contrasts including measurement error; n independent: number of independent participants/teams/donors, not repeated samples; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Confirm random sequence; whole-brain voxels are not independent participants; sensory/expectancy adjustment alone cannot identify target-specific efficacy.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Concealed randomized sequence, a scientifically justified recovery interval and no anticipation; period/carryover cannot be repaired solely by a nonsignificant test.
If carryover is plausible, report lag/period and first-period or otherwise valid design sensitivity; do not drop periods based only on a post hoc carryover p-value.
Predefined network summary is primary; voxelwise exploration requires spatial multiplicity correction, not independent-observation counting.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Target responses reproduced by sensory control, or disappearance under credible masking, would undermine direct-target specificity.; preanalysis gate: Confirm random sequence; whole-brain voxels are not independent participants; sensory/expectancy adjustment alone cannot identify target-specific efficacy.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Confirm random sequence; whole-brain voxels are not independent participants; sensory/expectancy adjustment alone cannot identify target-specific efficacy.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Compare planned lag/carryover structures and a valid first-period sensitivity where its sample supports inference; assess informative absent periods.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Confirm random sequence; whole-brain voxels are not independent participants; sensory/expectancy adjustment alone cannot identify target-specific efficacy.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/235; question: How much apparent focused-ultrasound target engagement remains after accounting for auditory, somatic and expectancy responses?; aims: Audit sensory signatures and blinding in existing approved-session data.
Prospectively compare active and matched-control sessions under an established protocol.
Test reproducibility of a preregistered target-linked imaging endpoint.; design verbatim planning only: Planning assumption: existing data access plus 24 screened participants at an MRI-guided neuromodulation center, with counterbalanced sessions and protocol-defined washout. This is a mechanistic substudy, not an efficacy trial.; outcomes verbatim: Primary: predefined target-network imaging response. Secondary: peripheral evoked responses, perceived sensations, allocation guesses, symptoms and adverse events.; analysis verbatim inherited not authoritative if refined: Use blinded imaging pipelines, anatomical target registration and session-level mixed models. Correct voxelwise inference and distinguish exploratory symptom changes; model period and carryover.; falsification verbatim: Target responses reproduced by sensory control, or disappearance under credible masking, would undermine direct-target specificity.; source ids unverified leads: tfus-amygdala-target-engagement-2025
belief-effects-health-review-2026
nocebo-communication-review-2026; prospectus precedence: No

N37 — BCI recalibration burden as a determinant of useful independence

Rank: 184
Cluster: Neurotechnology/neurostimulation
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: NOF1_RATIO; allocation or exposure unit: task/session block randomized within each existing implanted BCI participant; independent sampling unit: individual participant for person-specific inference; independent participant for cautious pooling; repeated measurement unit: communication/setup session and failed day; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Ratio can be undefined on zero setup; predefine aggregate ratio/zero rule and failed-day scoring; adaptive recalibration produces carryover and few persons limit population claims.
Primary estimand: person-specific decoding-policy effect on useful communication time per setup hour
Candidate primary outcome: useful independently completed communication time and setup time, each recorded separately
Primary finalized: No
Mathematical model: type: person-specific randomization/precision planning with carryover and small independent-participant replication; no large-population powered-N claim; formulas: g(E[Y_ipt | A_ip,X_i]) = alpha_i + tau A_ip + pi_p + rho A_i,p-1 + gamma^T X_i + f_time(t) + b_cluster(i,p).
Target the registered within-unit assignment contrast, standardized over period/history; direct and lag effects are distinct parameters.
For person i and policy a, R_i(a)=sum_sessions U_is(a)/sum_sessions H_is(a); Delta_i=R_i(1)-R_i(0). Randomization inference permutes only allowed block assignments.; variables: Y: useful independently completed communication time and setup time, each recorded separately; i: individual participant for person-specific inference; independent participant for cautious pooling; t: communication/setup session and failed day; the registered visit/time-window index, not another independent unit; A: task/session block randomized within each existing implanted BCI participant; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: person-specific decoding-policy effect on useful communication time per setup hour; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; p: randomized crossover period/session/episode, not the outcome probability; pi p: period effect; include calendar/season/sequence terms supported by the design; rho: prespecified lagged-assignment/carryover parameter; cannot in general identify arbitrary persistent carryover in a short two-period trial; f time(t): prespecified within-period recovery/day trend; g: identity for mean outcomes, log for positive/count means, or logit for binary/proportion outcomes as specified below; U is: independently completed useful communication minutes in session s, including zero useful time on failed days; H is: setup hours under the same policy/session schedule; a positive-denominator aggregation rule must be registered; s: communication/setup session; R i(a): person-specific ratio of totals, distinct from mean of session ratios; family code: NOF1_RATIO
Sample size design: family: person-specific randomization/precision planning with carryover and small independent-participant replication; no large-population powered-N claim; formula: For a simple two-condition paired mean only: Var(Delta_hat)=sigma_D^2/n_independent; actual cluster/period/lag models require design-specific simulation or c^T(D^T V^(-1)D)^(-1)c.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; sigma D^2: variance of independent-unit paired contrasts including measurement error; n independent: number of independent participants/teams/donors, not repeated samples; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Ratio can be undefined on zero setup; predefine aggregate ratio/zero rule and failed-day scoring; adaptive recalibration produces carryover and few persons limit population claims.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Concealed randomized sequence, a scientifically justified recovery interval and no anticipation; period/carryover cannot be repaired solely by a nonsignificant test.
If carryover is plausible, report lag/period and first-period or otherwise valid design sensitivity; do not drop periods based only on a post hoc carryover p-value.
Person-specific randomization inference needs the actual permissible assignment schedule; partial pooling does not make a handful of participants representative.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Reduced calibration time accompanied by communication errors or additional caregiver rescue would undermine useful independence.; preanalysis gate: Ratio can be undefined on zero setup; predefine aggregate ratio/zero rule and failed-day scoring; adaptive recalibration produces carryover and few persons limit population claims.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Ratio can be undefined on zero setup; predefine aggregate ratio/zero rule and failed-day scoring; adaptive recalibration produces carryover and few persons limit population claims.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Compare planned lag/carryover structures and a valid first-period sensitivity where its sample supports inference; assess informative absent periods.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Ratio can be undefined on zero setup; predefine aggregate ratio/zero rule and failed-day scoring; adaptive recalibration produces carryover and few persons limit population claims.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/183; question: Can bounded adaptive decoding reduce daily recalibration time without sacrificing reliable communication for existing implantable-BCI users?; aims: Quantify day-to-day drift and user effort in existing approved BCI studies.
Compare frozen and bounded adaptive decoding in randomized within-user blocks.
Estimate sustained communication utility and caregiver time outside benchmark sessions.; design verbatim planning only: Planning assumption: six to ten already implanted participants through a multisite clinical-research consortium, with repeated task blocks and remote-use observations. No new implants or surgical changes are proposed.; outcomes verbatim: Primary: independently completed useful communication time per setup hour. Secondary: corrected message accuracy, failed sessions, fatigue, rollback frequency and caregiver assistance.; analysis verbatim inherited not authoritative if refined: Use participant-specific randomization inference with hierarchical descriptive pooling; model period, task difficulty and adaptation carryover. Report all failed days and do not generalize small-sample averages to all disabilities.; falsification verbatim: Reduced calibration time accompanied by communication errors or additional caregiver rescue would undermine useful independence.; source ids unverified leads: ibci-outcome-measures-2025
bci-clinical-translation-2026; prospectus precedence: No

N38 — Restored tactile feedback: meaningful utility beyond sensation reports

Rank: 207
Cluster: Neurotechnology/neurostimulation
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: NOF1_BINARY; allocation or exposure unit: feedback task block randomized within each existing-interface participant; independent sampling unit: individual participant with independent participants as replication; repeated measurement unit: object interaction/trial within block; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Repeated trials improve each person's precision without increasing participant count; task interaction and order effects need fixed tasks and valid within-person randomization.
Primary estimand: person-specific tactile-plus-visual versus visual-only effect on safe independent interaction probability
Candidate primary outcome: safe successful object-interaction indicator
Primary finalized: No
Mathematical model: type: person-specific randomization/precision planning with carryover and small independent-participant replication; no large-population powered-N claim; formulas: logit Pr(Y_ipt=1)=alpha_i+tau_i A_ip+pi_p+rho A_i,p-1+gamma^T X_i; report marginal probability contrasts, with tau_i pooled only under stated hierarchical assumptions.
Target the registered within-unit assignment contrast, standardized over period/history; direct and lag effects are distinct parameters.; variables: Y: safe successful object-interaction indicator; i: individual participant with independent participants as replication; t: object interaction/trial within block; the registered visit/time-window index, not another independent unit; A: feedback task block randomized within each existing-interface participant; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: person-specific tactile-plus-visual versus visual-only effect on safe independent interaction probability; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; p: randomized crossover period/session/episode, not the outcome probability; pi p: period effect; include calendar/season/sequence terms supported by the design; rho: prespecified lagged-assignment/carryover parameter; cannot in general identify arbitrary persistent carryover in a short two-period trial; f time(t): prespecified within-period recovery/day trend; g: identity for mean outcomes, log for positive/count means, or logit for binary/proportion outcomes as specified below; tau i: person-specific randomized log-odds contrast; repeated task trials are dependent within person/block; family code: NOF1_BINARY
Sample size design: family: person-specific randomization/precision planning with carryover and small independent-participant replication; no large-population powered-N claim; formula: For a simple two-condition paired mean only: Var(Delta_hat)=sigma_D^2/n_independent; actual cluster/period/lag models require design-specific simulation or c^T(D^T V^(-1)D)^(-1)c.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; sigma D^2: variance of independent-unit paired contrasts including measurement error; n independent: number of independent participants/teams/donors, not repeated samples; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Repeated trials improve each person's precision without increasing participant count; task interaction and order effects need fixed tasks and valid within-person randomization.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Concealed randomized sequence, a scientifically justified recovery interval and no anticipation; period/carryover cannot be repaired solely by a nonsignificant test.
If carryover is plausible, report lag/period and first-period or otherwise valid design sensitivity; do not drop periods based only on a post hoc carryover p-value.
Person-specific randomization inference needs the actual permissible assignment schedule; partial pooling does not make a handful of participants representative.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Sensation without improved interaction, or increased workload offsetting accuracy, would weaken functional-restoration claims.; preanalysis gate: Repeated trials improve each person's precision without increasing participant count; task interaction and order effects need fixed tasks and valid within-person randomization.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Repeated trials improve each person's precision without increasing participant count; task interaction and order effects need fixed tasks and valid within-person randomization.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Compare planned lag/carryover structures and a valid first-period sensitivity where its sample supports inference; assess informative absent periods.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Repeated trials improve each person's precision without increasing participant count; task interaction and order effects need fixed tasks and valid within-person randomization.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/206; question: Does clinically delivered tactile feedback improve functional object interaction beyond visual feedback alone in already implanted neuroprosthesis participants?; aims: Establish within-person percept reliability using existing approved settings.
Randomize feedback availability during matched functional tasks.
Measure sustained utility, mental workload and user preference over repeated sessions.; design verbatim planning only: Planning assumption: four to eight participants with existing sensory interfaces complete counterbalanced task blocks under current specialist protocols. No new implants, surgery or novel stimulation settings are introduced.; outcomes verbatim: Primary: successful safe object interactions with prespecified independence criteria. Secondary: completion time, grasp errors, visual attention, mental workload and preferred feedback use.; analysis verbatim inherited not authoritative if refined: Use within-person randomized contrasts, task-by-feedback interactions and explicit order effects. Present individual effects with uncertainty; repeated trials do not substitute for independent participant counts.; falsification verbatim: Sensation without improved interaction, or increased workload offsetting accuracy, would weaken functional-restoration claims.; source ids unverified leads: icms-human-decade-safety-2025
neuroprosthesis-sensation-tetraplegia-2026
ibci-outcome-measures-2025; prospectus precedence: No

N39 — Consumer EEG attention signals across artifacts and access barriers

Rank: 106
Cluster: Neurotechnology/neurostimulation
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: BENCH_DIAG; allocation or exposure unit: task/artifact condition randomized; EEG devices compared on paired recordings; independent sampling unit: participant, held out across all trials; repeated measurement unit: task epoch, device and artifact condition; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: diagnostic/benchmark agreement or classification validity; reader/interface/navigation effects are causal only in their separately randomized component; specific constraint: Hair/fit differences are observational; any artifact filtering must be trained without test participants; data usability is a separate endpoint.
Primary estimand: out-of-participant discrimination of randomized task conditions after frozen artifact control
Candidate primary outcome: known randomized task-condition label and fixed EEG model score
Primary finalized: No
Mathematical model: type: diagnostic/benchmark sensitivity, false-alert and calibration precision with independent cases/noncases and cluster/reference uncertainty; formulas: Sensitivity = TP/(TP+FN); false-alert burden = FP/(registered nonevent time/opportunities); report participant/family-clustered uncertainty and failed/indeterminate cases.
For probability scores, report calibration and a fixed proper score in participant/source-disjoint evaluation, with uncertain reference labels modeled or bounded.; variables: Y: known randomized task-condition label and fixed EEG model score; i: participant, held out across all trials; t: task epoch, device and artifact condition; the registered visit/time-window index, not another independent unit; A: task/artifact condition randomized; EEG devices compared on paired recordings; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: out-of-participant discrimination of randomized task conditions after frozen artifact control; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; TP,FN,FP: true-positive, false-negative and false-positive counts relative to the independently defined reference and matching window, not an assumed perfect gold standard; reference: per-thesis independent reference, including adjudication uncertainty; nonevent time/opportunities: fixed at-risk denominator; event specificity and false alerts per hour are distinct targets; family code: BENCH_DIAG
Sample size design: family: diagnostic/benchmark sensitivity, false-alert and calibration precision with independent cases/noncases and cluster/reference uncertainty; formula: An independent-case rough starting point is n_case ~= z_(1-alpha*/2)^2 Se(1-Se)/h_Se^2; replace with interval-based/clustered simulation for the actual design. Plan noncases/observation time and reference failures separately.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; Se: plausible sensitivity from prior evidence or pilot, not an asserted study result; h Se: desired sensitivity interval half-width; n case: independent positive cases; repeated events within a person require correlation adjustment; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Hair/fit differences are observational; any artifact filtering must be trained without test participants; data usability is a separate endpoint.
Reference independence, fixed thresholds/window and an explicit denominator; incorporation/verification bias can invalidate apparent accuracy.
Participant/family-level test splits and paired comparisons; report abstention, assay failure and indeterminate truth rather than force every case into a binary label.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Indices primarily tracking artifacts, or loss of performance on new participants, would undermine cognitive-state interpretation.; preanalysis gate: Hair/fit differences are observational; any artifact filtering must be trained without test participants; data usability is a separate endpoint.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Hair/fit differences are observational; any artifact filtering must be trained without test participants; data usability is a separate endpoint.
Vary plausible reference misclassification, verification-selection and event-matching window under a prespecified sensitivity family.
Report subgroup usable-data coverage and calibration; avoid interpreting an imprecise null difference as equivalence.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Hair/fit differences are observational; any artifact filtering must be trained without test participants; data usability is a separate endpoint.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/105; question: Can consumer EEG indices distinguish neural engagement from eye, jaw and fit artifacts across diverse users?; aims: Compare consumer indices with simultaneous research EEG and peripheral artifact channels.
Randomize task engagement and benign artifact conditions within participants.
Validate frozen features across new users and repeated home sessions.; design verbatim planning only: Planning assumption: 90 adults with deliberately varied hair characteristics and head fit complete two laboratory visits; 40 undertake a short home-repeatability substudy. Devices are used for recording only.; outcomes verbatim: Primary: out-of-participant discrimination of randomized task conditions after artifact control. Secondary: usable-data proportion, repeatability, setup time and subgroup measurement error.; analysis verbatim inherited not authoritative if refined: Prevent trial-level leakage across participants; use paired device comparisons and prespecified fit contrasts. Distinguish randomized artifact effects from observational demographic differences and correct feature-selection multiplicity.; falsification verbatim: Indices primarily tracking artifacts, or loss of performance on new participants, would undermine cognitive-state interpretation.; source ids unverified leads: neurofeedback-adhd-meta-2025
neurofeedback-regulation-review-2026
wearables-living-review; prospectus precedence: No

N40 — Data minimization for useful and understandable EEG feedback

Rank: 193
Cluster: Neurotechnology/neurostimulation
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: PRED_NI; allocation or exposure unit: paired computational representations of the same EEG; separate explanation-format randomization; independent sampling unit: participant; repeated measurement unit: EEG session/model prediction; consent response for separate experiment; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: prediction, calibration or model transportability; prediction gain is not clinical efficacy, mechanistic necessity or causal treatment selection; specific constraint: Margin must be justified before confirmation; computational utility is not consent comprehension and retaining fewer features does not guarantee anonymity.
Primary estimand: minimized-versus-full representation predictive-utility difference within a pilot-justified noninferiority margin
Candidate primary outcome: prespecified held-out feedback utility score
Primary finalized: No
Mathematical model: type: paired noninferiority of predictive utility, with justified margin and independent held-out participant bootstrap/simulation; formulas: Gain = mean_test[L(Y,f_base(X)) - L(Y,f_new(X))], with both predictions frozen before test outcomes; bootstrap independent held-out units.
For a binary target assess E[Y | p_hat=p] versus p and calibration intercept/slope logit Pr(Y=1)=a_cal+b_cal logit(p_hat); for a continuous target use Y=a_cal+b_cal y_hat+error plus scale-appropriate calibration/loss.
For utility where larger is better, Delta_U=U_minimized-U_full; declare noninferiority only if the prespecified one-sided lower confidence bound exceeds -M_NI.; variables: Y: prespecified held-out feedback utility score; i: participant; t: EEG session/model prediction; consent response for separate experiment; the registered visit/time-window index, not another independent unit; A: paired computational representations of the same EEG; separate explanation-format randomization; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: minimized-versus-full representation predictive-utility difference within a pilot-justified noninferiority margin; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; L: one proper or scientifically justified primary loss/scoring rule fixed for the actual outcome type; positive Gain favors new model; f base,f new: locked baseline/new prediction functions using only information available at the registered prediction time; test: untouched person/family/lab/site/time set as specified per thesis; p hat,y hat: frozen probability or continuous prediction; a cal,b cal: calibration intercept and slope, estimated on evaluation data for assessment; using them to update the model requires another evaluation set; U minimized,U full: paired held-out predictive utility under the registered representations; M NI: pilot/externally justified utility-loss margin fixed before confirmation; no numerical margin supplied; family code: PRED_NI
Sample size design: family: paired noninferiority of predictive utility, with justified margin and independent held-out participant bootstrap/simulation; formula: Simulate or bootstrap the independent-unit test loss/calibration/decision criterion; plan cases/events, noncases and independent sites/families/programs separately. No universal events-per-variable or participant N is asserted.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Margin must be justified before confirmation; computational utility is not consent comprehension and retaining fewer features does not guarantee anonymity.
Train/tune/test separation at the actual biological/participant/family/time level; no future, postoutcome or target-derived predictors.
Validation support must cover intended deployment; calibration and missing/failing measurements are reported with discrimination/loss.
A validation subset selected by culture success, assay viability, consent or data completeness is a different target unless the full strategy is explicitly evaluated.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Meaningful utility loss or unchanged comprehension would undermine the proposed design; persistent sensitive-information leakage would challenge privacy benefit.; preanalysis gate: Margin must be justified before confirmation; computational utility is not consent comprehension and retaining fewer features does not guarantee anonymity.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Margin must be justified before confirmation; computational utility is not consent comprehension and retaining fewer features does not guarantee anonymity.
Repeat loss/calibration with independent-cluster resampling, prespecified subgroup/transport shifts and plausible missingness.
Compare fixed simple baselines, calibration and uncertainty coverage; report abstention/failure/turnaround rather than deleting hard cases.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Frozen prediction time/horizon, proper loss, baseline model, feature availability, independent split/holdout and target deployment population.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Margin must be justified before confirmation; computational utility is not consent comprehension and retaining fewer features does not guarantee anonymity.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/192; question: Can on-device EEG feature extraction preserve useful feedback while reducing unnecessary neural-data retention and improving informed consent?; aims: Quantify feedback performance under progressively minimized data representations.
Validate utility on held-out participants and repeat sessions.
Randomize consent explanations and measure comprehension and sharing preferences.; design verbatim planning only: Planning assumption: 120 adults complete two EEG sessions and a communication experiment. Computational comparisons use the same recordings; consent randomization changes explanation format, never undisclosed sharing.; outcomes verbatim: Primary: noninferiority of minimized-feature feedback against a prespecified utility margin established in the pilot. Secondary: consent comprehension, retention preferences, subgroup utility and setup burden.; analysis verbatim inherited not authoritative if refined: Separate computational performance from causal explanation effects; use participant-level holdouts and uncertainty intervals. Prespecify privacy proxies and correct representation comparisons; less retained data does not guarantee anonymity.; falsification verbatim: Meaningful utility loss or unchanged comprehension would undermine the proposed design; persistent sensitive-information leakage would challenge privacy benefit.; source ids unverified leads: bci-clinical-translation-2026
digital-phenotyping-relapse-review-2026
wearables-remote-monitoring-2026; prospectus precedence: No

P01 — Does HRV change the right training decisions?

Rank: 23
Cluster: Exercise/cardiorespiratory adaptation
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT; allocation or exposure unit: participant randomized coach access to HRV versus symptoms/history alone; independent sampling unit: participant with coach clustering; repeated measurement unit: CPET and training-decision visit; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Training load and symptoms after randomization may mediate adaptation; no biological efficacy attribution to the HRV metric alone.
Primary estimand: HRV-guidance offer effect on peak oxygen uptake under the common coaching strategy
Candidate primary outcome: laboratory peak oxygen uptake at the end of the training block
Primary finalized: No
Mathematical model: type: contrast precision/power by design-matrix or simulation planning; formulas: E(Y_it | A_i, X_i, Y_i0) = alpha_t + z(A_i)^T theta_t + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Delta = sum_t w_t c^T theta_t; estimate by baseline-adjusted marginal standardization, with unit/cluster-appropriate inference.; variables: Y: laboratory peak oxygen uptake at the end of the training block; i: participant with coach clustering; t: CPET and training-decision visit; the registered visit/time-window index, not another independent unit; A: participant randomized coach access to HRV versus symptoms/history alone; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: HRV-guidance offer effect on peak oxygen uptake under the common coaching strategy; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; z(A): prespecified indicators for the actual randomized arms; two arms reduce to one indicator; family code: RCT
Sample size design: family: contrast precision/power by design-matrix or simulation planning; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Training load and symptoms after randomization may mediate adaptation; no biological efficacy attribution to the HRV metric alone.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: No incremental adaptation despite reliable signal capture and frequent decision changes would reject the proposed information benefit.; preanalysis gate: Training load and symptoms after randomization may mediate adaptation; no biological efficacy attribution to the HRV metric alone.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Training load and symptoms after randomization may mediate adaptation; no biological efficacy attribution to the HRV metric alone.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Training load and symptoms after randomization may mediate adaptation; no biological efficacy attribution to the HRV metric alone.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/22; question: Does physiological guidance improve aerobic adaptation beyond an equally responsive coach using training history?; aims: Quantify reproducible HRV deviations before supervised exercise sessions.
Randomize access to HRV when coaches select session intensity.
Test whether altered decisions mediate measured aerobic adaptation.; design verbatim planning only: Population model: Previously inactive adults entering a supervised aerobic training program. Comparator: Identical coaches receive training history and symptoms, with or without standardized HRV. Prospective causal strategy: Parallel randomization for one training block; outcome assessors remain blinded and decisions are logged prospectively. Sample planning: Plan 120 participants; pilot decision frequency, attrition, and CPET variance determine final sample justification.; outcomes verbatim: Primary: laboratory peak oxygen uptake change. Secondary: ventilatory threshold, completed work, adverse events, and decision concordance.; analysis verbatim inherited not authoritative if refined: Baseline-adjusted intention-to-treat models account for medication. Prespecify one primary endpoint; exploratory mediation addresses time-varying symptoms and load without claiming randomized mediators.; falsification verbatim: No incremental adaptation despite reliable signal capture and frequent decision changes would reject the proposed information benefit.; source ids unverified leads: hrv-exercise-prescription-2025
wearables-living-review; prospectus precedence: No

P02 — Can wearable fitness scores detect adaptation equitably?

Rank: 32
Cluster: Exercise/cardiorespiratory adaptation
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: AGREE_CHANGE; allocation or exposure unit: device fitness estimates compared with repeated CPET before/after training; independent sampling unit: participant with independent site validation; repeated measurement unit: device, CPET replicate and visit; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: measurement agreement/reliability; condition contrasts are causal only where the condition itself was randomized; specific constraint: Correlated baseline/follow-up reference error inflates responder misclassification; participant and site holdouts, firmware and informative device failure matter.
Primary estimand: device-specific error in measured fitness change relative to CPET change
Candidate primary outcome: device fitness change minus reference CPET change
Primary finalized: No
Mathematical model: type: paired hierarchical agreement/bias/limits precision with independent participants/donors/labs and replicate allocation; formulas: D_ij = S_ij - R_ij = beta_0 + beta_context^T C_ij + u_i + b_lab + epsilon_ij.
Context-specific agreement limits use beta_0 + beta_context^T C +/- z_(1-alpha/2) sqrt(Var(D_new | target context)); distinguish a new person from a repeat reading.
If reference error is material, S_ij = T_ij + b_S + e_S,ij and R_ij = T_ij + b_R + e_R,ij; identifiability needs reference validation/replicates or stated constraints.
D_i = (S_i1-S_i0)-(R_i1-R_i0); Var(D_i) propagates all baseline/follow-up and reference cross-covariances.; variables: Y: device fitness change minus reference CPET change; i: participant with independent site validation; t: device, CPET replicate and visit; the registered visit/time-window index, not another independent unit; A: device fitness estimates compared with repeated CPET before/after training; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: device-specific error in measured fitness change relative to CPET change; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; S ij: sensor/test result in the stated endpoint unit; R ij: paired validated reference result in the same unit; D ij: paired test-minus-reference error, matching the per-thesis outcome; C ij: prespecified context, firmware/device and subgroup features; T ij: latent underlying quantity; not automatically identifiable; b S,b R: test and reference biases; e S,e R: test and reference measurement errors; b lab: laboratory effect when multiple laboratories actually sampled; D new: prediction error for the prespecified future measurement context; family code: AGREE_CHANGE
Sample size design: family: paired hierarchical agreement/bias/limits precision with independent participants/donors/labs and replicate allocation; formula: Plan confidence width for bias and agreement limits by hierarchical simulation; n readings is not n independent people. Acceptance limits in physical units must be independently justified.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Correlated baseline/follow-up reference error inflates responder misclassification; participant and site holdouts, firmware and informative device failure matter.
Reference calibration, matched time/unit and appropriately modeled reference error; correlation is not agreement.
Future-context prediction assumptions must match held-out persons/labs/conditions; missing/failing readings are outcomes of the measurement system.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Calibration that fails on held-out participants or worsens subgroup error would disconfirm a generalizable improvement.; preanalysis gate: Correlated baseline/follow-up reference error inflates responder misclassification; participant and site holdouts, firmware and informative device failure matter.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Correlated baseline/follow-up reference error inflates responder misclassification; participant and site holdouts, firmware and informative device failure matter.
Assess proportional/heteroscedastic error, reference uncertainty and subgroup-specific failure, not just a global correlation.
Repeat estimates with and without questionable quality strata under predeclared rules; report excluded/failing measurements explicitly.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Correlated baseline/follow-up reference error inflates responder misclassification; participant and site holdouts, firmware and informative device failure matter.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/31; question: Do consumer fitness estimates track true training changes across skin pigmentation, body size, and baseline fitness?; aims: Estimate repeatability of laboratory and wearable fitness measurements.
Compare longitudinal change agreement across prespecified demographic strata.
Externally validate a transparent calibration model after training.; design verbatim planning only: Population model: Adults spanning fitness, body size, skin pigmentation, and socioeconomic groups in standardized aerobic training. Comparator: Two consumer devices versus repeated cardiopulmonary exercise testing and a simple field test. Prospective causal strategy: Prospective validation with standardized training; hold out participants and a second recruitment site for calibration testing. Sample planning: Plan 180 participants, including 50 repeat-test volunteers; subgroup precision and device access determine the final sample.; outcomes verbatim: Primary: error in fitness change relative to CPET. Secondary: responder misclassification, missingness, usability, and cost per valid measurement.; analysis verbatim inherited not authoritative if refined: Use agreement intervals and hierarchical error models. Account for exposure and firmware; correct prespecified device comparisons and report subgroup uncertainty and informative missingness.; falsification verbatim: Calibration that fails on held-out participants or worsens subgroup error would disconfirm a generalizable improvement.; source ids unverified leads: cardiorespiratory-fitness-meta-2025
wearables-living-review
wearables-remote-monitoring-2026; prospectus precedence: No

P03 — Distributed exercise under real scheduling constraints

Rank: 37
Cluster: Exercise/cardiorespiratory adaptation
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT_NI; allocation or exposure unit: participant randomized distributed versus continuous exercise program; independent sampling unit: participant; repeated measurement unit: baseline/final CPET and scheduling report; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: No asserted margin or powered N; include assignment and supportive adherence analyses without equating adherence conditioning with randomization.
Primary estimand: distributed-program minus continuous-program mean fitness difference relative to a justified noninferiority margin
Candidate primary outcome: laboratory peak oxygen uptake change
Primary finalized: No
Mathematical model: type: active-comparator noninferiority precision/power with justified margin and missingness/adherence sensitivity; formulas: Delta=E[Y | assigned simple/distributed program]-E[Y | assigned sensor/continuous comparator], on the registered fitness-improvement scale where larger is better.
Declare noninferiority only if the prespecified one-sided lower confidence bound for Delta exceeds -M_NI; estimate assignment effect and a separately justified supportive adherence estimand.; variables: Y: laboratory peak oxygen uptake change; i: participant; t: baseline/final CPET and scheduling report; the registered visit/time-window index, not another independent unit; A: participant randomized distributed versus continuous exercise program; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: distributed-program minus continuous-program mean fitness difference relative to a justified noninferiority margin; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; M NI: independently justified maximum acceptable loss in the actual aerobic-fitness unit, fixed before outcome analysis; no numerical margin supplied; family code: RCT_NI
Sample size design: family: active-comparator noninferiority precision/power with justified margin and missingness/adherence sensitivity; formula: Use design-specific simulation or independent-unit resampling for the stated estimand and target confidence width/power; preserve all allocation, censoring, clustering and measurement-error structure.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: No asserted margin or powered N; include assignment and supportive adherence analyses without equating adherence conditioning with randomization.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Assay sensitivity, active comparator fidelity and a justified margin are required; missingness, crossover and selection can bias toward apparent noninferiority.
Per-protocol comparisons do not automatically preserve randomization; define adherence strategy and confounding assumptions.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Inferior adaptation despite similar completed work would reject physiological interchangeability; no adherence gain would reject the scheduling mechanism.; preanalysis gate: No asserted margin or powered N; include assignment and supportive adherence analyses without equating adherence conditioning with randomization.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: No asserted margin or powered N; include assignment and supportive adherence analyses without equating adherence conditioning with randomization.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
No asserted margin or powered N; include assignment and supportive adherence analyses without equating adherence conditioning with randomization.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/36; question: Can brief distributed activity produce comparable aerobic adaptation when total prescribed work and coaching are matched?; aims: Measure scheduling barriers and baseline exercise-test repeatability.
Randomize distributed versus continuous sessions with matched prescribed work.
Separate adherence-mediated effects from adaptation per completed work.; design verbatim planning only: Population model: Inactive shift workers and caregivers able to undertake screened, moderate exercise. Comparator: Distributed sessions versus a continuous-session program with identical exercise modes and staff contact. Prospective causal strategy: Parallel pragmatic trial across a defined training block; blinded CPET assessment follows a common pretest routine. Sample planning: Plan 140 participants; pilot adherence and variance inform a justified noninferiority margin and final sample.; outcomes verbatim: Primary: peak oxygen uptake change. Secondary: completed work, scheduling burden, blood pressure, enjoyment, and retention.; analysis verbatim inherited not authoritative if refined: Separate intention-to-treat and adherence estimands. Adjust for shifts, caregiving, fitness, and occupational activity; per-work comparisons remain selection-sensitive, with secondary endpoint multiplicity controlled.; falsification verbatim: Inferior adaptation despite similar completed work would reject physiological interchangeability; no adherence gain would reject the scheduling mechanism.; source ids unverified leads: cardiorespiratory-fitness-meta-2025
hrv-exercise-prescription-2025
behavior-stress-sleep-rct-2026; prospectus precedence: No

P04 — Training intensity boundaries versus generic heart-rate zones

Rank: 62
Cluster: Exercise/cardiorespiratory adaptation
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT_INT; allocation or exposure unit: participant randomized CPET-threshold versus generic heart-rate prescription; independent sampling unit: participant; repeated measurement unit: threshold/CPET and prescription assessment; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Baseline disagreement measured with error and prone to regression to mean; actual completed work is postassignment.
Primary estimand: prescription assignment effect and one baseline zone-disagreement interaction on threshold workload
Candidate primary outcome: ventilatory-threshold workload at the fixed follow-up
Primary finalized: No
Mathematical model: type: interaction-specific precision/power with modifier distribution and reliability; simulate the actual interaction; formulas: E(Y_it | A_i, B_i, X_i) = alpha_t + tau_t A_i + eta_t B_i + kappa_t A_i B_i + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Treatment contrast at baseline B=b is tau_t + kappa_t b; the interaction target is kappa_t, not a difference in within-subgroup p-values.; variables: Y: ventilatory-threshold workload at the fixed follow-up; i: participant; t: threshold/CPET and prescription assessment; the registered visit/time-window index, not another independent unit; A: participant randomized CPET-threshold versus generic heart-rate prescription; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: prescription assignment effect and one baseline zone-disagreement interaction on threshold workload; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; B: baseline disagreement between individual CPET threshold prescription and generic heart-rate zones; centering/coding and any measurement-error model fixed before evaluation; eta: baseline modifier association; kappa: assignment-by-baseline-modifier interaction coefficient on the stated scale; family code: RCT_INT
Sample size design: family: interaction-specific precision/power with modifier distribution and reliability; simulate the actual interaction; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure
modifier prevalence/distribution and variance
modifier reliability and overlap
interaction size or precision target; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Baseline disagreement measured with error and prone to regression to mean; actual completed work is postassignment.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The modifier must be measured before assignment/initiation; interaction is scale dependent and requires support within modifier strata.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: No advantage in the high-disagreement subgroup, with verified domain separation, would challenge the rationale for routine laboratory personalization.; preanalysis gate: Baseline disagreement measured with error and prone to regression to mean; actual completed work is postassignment.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Baseline disagreement measured with error and prone to regression to mean; actual completed work is postassignment.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Vary modifier measurement-error assumptions and retain a continuous modifier unless a cutoff was independently fixed.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Baseline disagreement measured with error and prone to regression to mean; actual completed work is postassignment.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/61; question: Does individual ventilatory-threshold prescription reduce undertraining and excessive exertion compared with generic heart-rate zones?; aims: Map disagreement between generic zones and measured thresholds.
Randomize threshold-based or generic intensity prescriptions with equal contact.
Test baseline disagreement as a prespecified treatment-effect modifier.; design verbatim planning only: Population model: Inactive middle-aged adults without exercise-limiting disease after clinical screening. Comparator: CPET-derived thresholds versus conventional age-based heart-rate guidance, using identical exercise modes. Prospective causal strategy: Assessor-blinded parallel training trial; independent staff verify session work and prescription fidelity. Sample planning: Provisionally recruit 130 adults; repeat CPET in a pilot subgroup estimates threshold error and the interaction sample requirement.; outcomes verbatim: Primary: ventilatory-threshold workload change. Secondary: peak oxygen uptake, excessive-exertion episodes, adherence, and implementation cost.; analysis verbatim inherited not authoritative if refined: Estimate the assignment-based total effect, adjusting for baseline threshold and baseline heart-rate medication. Register one baseline-disagreement interaction and one primary endpoint; distinguish effect modification from regression to the mean. Report completed training work separately, with any work-mediated analysis exploratory and explicit about postrandomization confounding.; falsification verbatim: No advantage in the high-disagreement subgroup, with verified domain separation, would challenge the rationale for routine laboratory personalization.; source ids unverified leads: cardiorespiratory-fitness-meta-2025
hrv-exercise-prescription-2025
metabolic-flexibility-prehabilitation-2025; prospectus precedence: No

P05 — Are aerobic nonresponders reproducible?

Rank: 132
Cluster: Exercise/cardiorespiratory adaptation
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RESPONSE_VAR; allocation or exposure unit: block-start timing randomized; same training prescription repeated within person; independent sampling unit: participant; repeated measurement unit: training block and repeated CPET test; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: variance/reliability and response-reproducibility estimation; noisy change labels do not establish stable individual causal effects; specific constraint: Random-slope variance may be unidentified with few repeats; irreversible adaptation changes later baseline, so consistency is not exact individual causal response.
Primary estimand: reproducibility/covariance of participant-specific training-associated changes after measurement/secular variation correction
Candidate primary outcome: oxygen-uptake change in each repeated block
Primary finalized: No
Mathematical model: type: variance-component/reliable-change interval precision by hierarchical simulation; independent donors/persons and replicate allocation; formulas: Y_ibr = mu + u_i + beta_b + v_ib + e_ibr; estimate donor/person, day/block, person-by-block and technical residual components using the actual crossed/nested replication.
For two repeated measurements, reliable-change threshold = z_(1-alpha*/2) sqrt[Var(Y_i2-Y_i1 | no persistent change)]; include covariance instead of assuming independent errors.
Observed block change D_ib=delta_b+R_ib+e_ib; Cov(D_i1,D_i2)=Cov(R_i1,R_i2)+Cov(e_i1,e_i2). Estimate error covariance from repeated CPET/wait-list data before calling the remaining covariance reproducible response.; variables: Y: oxygen-uptake change in each repeated block; i: participant; t: training block and repeated CPET test; the registered visit/time-window index, not another independent unit; A: block-start timing randomized; same training prescription repeated within person; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: reproducibility/covariance of participant-specific training-associated changes after measurement/secular variation correction; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; b: collection day or repeated training block; r: technical/retest replicate; u i: stable unit-level component; beta b: fixed day/block/period shift; v ib: unit-by-day/block deviation; e ibr: technical/retest error; alpha*: clock-family/reliability error rate after registered multiplicity rule; D ib: observed oxygen-uptake change in training block b; R ib: latent training-associated change after specified secular adjustment; not directly observed individual causal response; e ib: change-measurement error, including shared baseline/retest covariance; family code: RESPONSE_VAR
Sample size design: family: variance-component/reliable-change interval precision by hierarchical simulation; independent donors/persons and replicate allocation; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Random-slope variance may be unidentified with few repeats; irreversible adaptation changes later baseline, so consistency is not exact individual causal response.
Enough independent units and repeated occasions to separate variance components; fit/identifiability cannot be assumed from a planning sample count.
Reliable change excludes modeled short-term noise, not all biological confounding or regression to the mean.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Large stable between-person response variance after error correction, replicated across blocks, would reject the hypothesis of predominantly transient nonresponse.; preanalysis gate: Random-slope variance may be unidentified with few repeats; irreversible adaptation changes later baseline, so consistency is not exact individual causal response.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Random-slope variance may be unidentified with few repeats; irreversible adaptation changes later baseline, so consistency is not exact individual causal response.
Profile/bootstrap variance-component intervals, boundary estimates and covariance structures; report false-positive classifications under negative controls.
Vary missingness/retest reliability and avoid declaring responders from effect direction alone.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Random-slope variance may be unidentified with few repeats; irreversible adaptation changes later baseline, so consistency is not exact individual causal response.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/131; question: Does apparent low responsiveness persist when the same person repeats a standardized training block?; aims: Estimate laboratory test error before training begins.
Repeat a standardized training block after monitored detraining.
Partition stable individual response from transient and adherence variation.; design verbatim planning only: Population model: Previously inactive adults completing two supervised aerobic training blocks. Comparator: Each participant repeats the same prescription; a wait-list observational interval estimates secular variability. Prospective causal strategy: Prospective replicated-intervention design with randomized block-start timing and blinded repeated CPET measurements. Sample planning: Plan 90 completers and allow recruitment for substantial long-term attrition; pilot estimates determine whether random-slope variance is identifiable.; outcomes verbatim: Primary: reproducibility of individual oxygen-uptake change. Secondary: threshold response, work adherence, and responder-label stability.; analysis verbatim inherited not authoritative if refined: Hierarchical measurement-error models separate participant, block, and residual variance. Account for fitness and illness; avoid noisy responder dichotomies and prespecify few modifiers.; falsification verbatim: Large stable between-person response variance after error correction, replicated across blocks, would reject the hypothesis of predominantly transient nonresponse.; source ids unverified leads: n-of-1-methods-review-2025
digital-n-of-1-physiology-2026
cardiorespiratory-fitness-meta-2025; prospectus precedence: No

P06 — Concurrent training order and muscle oxygenation

Rank: 86
Cluster: Exercise/cardiorespiratory adaptation
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: KINETICS; allocation or exposure unit: participant randomized same-session versus separated concurrent training; independent sampling unit: participant; repeated measurement unit: submaximal oxygen-uptake time sample and training visit; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Kinetic curve identifiability and measurement timing must be validated; NIRS and oxygen uptake do not independently prove a peripheral mechanism.
Primary estimand: assignment effect on the prespecified submaximal oxygen-uptake kinetic parameter
Candidate primary outcome: time constant or registered summary of oxygen-uptake kinetics
Primary finalized: No
Mathematical model: type: two-stage nonlinear kinetic-parameter and randomized arm-contrast precision with curve-error propagation; formulas: VO2_i(t)=V0_i+Amp_i[1-exp(-(t-Td_i)/K_i)] for t>Td_i under a validated kinetic model; pre-delay behavior follows the registered baseline model.
Compare the prespecified K (or another externally fixed kinetic summary) between randomized training arms with baseline adjustment, propagating curve-estimation uncertainty.; variables: Y: time constant or registered summary of oxygen-uptake kinetics; i: participant; t: submaximal oxygen-uptake time sample and training visit; the registered visit/time-window index, not another independent unit; A: participant randomized same-session versus separated concurrent training; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: assignment effect on the prespecified submaximal oxygen-uptake kinetic parameter; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; VO2 i(t): measured oxygen uptake over the standardized submaximal task; V0 i: baseline uptake; Amp i: response amplitude; Td i: time delay; K i: kinetic time constant; identifiable only with suitable time sampling and model fit; family code: KINETICS
Sample size design: family: two-stage nonlinear kinetic-parameter and randomized arm-contrast precision with curve-error propagation; formula: Use design-specific simulation or independent-unit resampling for the stated estimand and target confidence width/power; preserve all allocation, censoring, clustering and measurement-error structure.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Kinetic curve identifiability and measurement timing must be validated; NIRS and oxygen uptake do not independently prove a peripheral mechanism.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Fix kinetic endpoint and curve model before treatment comparison; failed curves are missing measurements with possible treatment-related selection, not exclusions by convenience.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Precise equivalence in completed work, oxygen-uptake kinetics and muscle-oxygenation responses would challenge the proposed schedule-dependent advantage; these measurements alone cannot falsify a specific central-versus-peripheral pathway.; preanalysis gate: Kinetic curve identifiability and measurement timing must be validated; NIRS and oxygen uptake do not independently prove a peripheral mechanism.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Kinetic curve identifiability and measurement timing must be validated; NIRS and oxygen uptake do not independently prove a peripheral mechanism.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Kinetic curve identifiability and measurement timing must be validated; NIRS and oxygen uptake do not independently prove a peripheral mechanism.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/85; question: Does separating strength and endurance sessions change aerobic adaptation and muscle-oxygenation responses compared with same-session training?; aims: Quantify acute work loss under alternative session arrangements.
Compare longitudinal aerobic adaptation with prescribed work matched.
Relate muscle oxygenation changes to central exercise-test measures.; design verbatim planning only: Population model: Recreational adults beginning combined resistance and aerobic training. Comparator: Same-session strength then endurance versus sessions separated within the weekly schedule. Prospective causal strategy: Randomized parallel intervention with blinded CPET, monitored work, and standardized test timing after recovery. Sample planning: Plan 100 participants; acute pilot variance and schedule adherence determine the final trial size.; outcomes verbatim: Primary: submaximal oxygen-uptake kinetics. Secondary: peak oxygen uptake, near-infrared muscle oxygenation, strength, and training completion.; analysis verbatim inherited not authoritative if refined: Baseline-adjusted assignment-based models examine exploratory work mediation. Account for baseline habitual activity and nutrition and for NIRS adipose-thickness and measurement effects; control secondary physiological multiplicity. Treat oxygen-uptake kinetics and NIRS as complementary responses, not proof that changes arise from peripheral rather than central mechanisms.; falsification verbatim: Precise equivalence in completed work, oxygen-uptake kinetics and muscle-oxygenation responses would challenge the proposed schedule-dependent advantage; these measurements alone cannot falsify a specific central-versus-peripheral pathway.; source ids unverified leads: resistance-training-volume-meta-2025
cardiorespiratory-fitness-meta-2025
metabolic-flexibility-prehabilitation-2025; prospectus precedence: No

P07 — Does metabolic flexibility add predictive information?

Rank: 209
Cluster: Exercise/cardiorespiratory adaptation
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: PRED; allocation or exposure unit: baseline fuel-switching marker evaluated before standardized training; independent sampling unit: participant with independent cohort validation; repeated measurement unit: baseline calorimetry and later CPET; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: prediction, calibration or model transportability; prediction gain is not clinical efficacy, mechanistic necessity or causal treatment selection; specific constraint: Only pretraining information enters primary prediction; posttraining adherence is future leakage for baseline decision use.
Primary estimand: out-of-sample fitness-change prediction gain from baseline flexibility beyond fixed pretraining variables
Candidate primary outcome: peak oxygen-uptake change after training
Primary finalized: No
Mathematical model: type: external/nested predictive-validation precision with target prevalence/variance, independent-unit holdout coverage and model complexity; formulas: Gain = mean_test[L(Y,f_base(X)) - L(Y,f_new(X))], with both predictions frozen before test outcomes; bootstrap independent held-out units.
For a binary target assess E[Y | p_hat=p] versus p and calibration intercept/slope logit Pr(Y=1)=a_cal+b_cal logit(p_hat); for a continuous target use Y=a_cal+b_cal y_hat+error plus scale-appropriate calibration/loss.; variables: Y: peak oxygen-uptake change after training; i: participant with independent cohort validation; t: baseline calorimetry and later CPET; the registered visit/time-window index, not another independent unit; A: baseline fuel-switching marker evaluated before standardized training; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: out-of-sample fitness-change prediction gain from baseline flexibility beyond fixed pretraining variables; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; L: one proper or scientifically justified primary loss/scoring rule fixed for the actual outcome type; positive Gain favors new model; f base,f new: locked baseline/new prediction functions using only information available at the registered prediction time; test: untouched person/family/lab/site/time set as specified per thesis; p hat,y hat: frozen probability or continuous prediction; a cal,b cal: calibration intercept and slope, estimated on evaluation data for assessment; using them to update the model requires another evaluation set; family code: PRED
Sample size design: family: external/nested predictive-validation precision with target prevalence/variance, independent-unit holdout coverage and model complexity; formula: Simulate or bootstrap the independent-unit test loss/calibration/decision criterion; plan cases/events, noncases and independent sites/families/programs separately. No universal events-per-variable or participant N is asserted.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Only pretraining information enters primary prediction; posttraining adherence is future leakage for baseline decision use.
Train/tune/test separation at the actual biological/participant/family/time level; no future, postoutcome or target-derived predictors.
Validation support must cover intended deployment; calibration and missing/failing measurements are reported with discrimination/loss.
A validation subset selected by culture success, assay viability, consent or data completeness is a different target unless the full strategy is explicitly evaluated.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: A reproducible, externally validated improvement beyond simple measurements would falsify the hypothesis of negligible added value.; preanalysis gate: Only pretraining information enters primary prediction; posttraining adherence is future leakage for baseline decision use.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Only pretraining information enters primary prediction; posttraining adherence is future leakage for baseline decision use.
Repeat loss/calibration with independent-cluster resampling, prespecified subgroup/transport shifts and plausible missingness.
Compare fixed simple baselines, calibration and uncertainty coverage; report abstention/failure/turnaround rather than deleting hard cases.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Frozen prediction time/horizon, proper loss, baseline model, feature availability, independent split/holdout and target deployment population.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Only pretraining information enters primary prediction; posttraining adherence is future leakage for baseline decision use.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/208; question: Can laboratory fuel-switching measurements predict training adaptation beyond fitness, body composition, and habitual activity?; aims: Compare repeatability of exercise and meal-based fuel-switching measures.
Measure their incremental prediction during standardized aerobic training.
Validate a parsimonious model in a separate cohort.; design verbatim planning only: Population model: Adults across a range of adiposity and glucose regulation undergoing clinically screened exercise training. Comparator: Models with standard clinical and fitness variables versus models additionally including indirect-calorimetry fuel switching. Prospective causal strategy: Prospective observational biomarker study embedded within a standardized intervention; prediction is separated explicitly from causal mediation. Sample planning: Plan 160 development participants and an external cohort; simulation constrains model complexity and final sample.; outcomes verbatim: Primary: out-of-sample prediction error for peak oxygen-uptake change. Secondary: measurement reliability, calibration, and participant burden.; analysis verbatim inherited not authoritative if refined: For the primary pretraining prediction comparison, use only prespecified information available before training: baseline diet, medication, activity, body composition and fitness, with or without baseline fuel-switching measures. Analyze later adherence/exposure-adjusted associations separately as explanatory analyses. Use nested validation and shrinkage, limit predictors, and report external calibration across metabolic-status groups.; falsification verbatim: A reproducible, externally validated improvement beyond simple measurements would falsify the hypothesis of negligible added value.; source ids unverified leads: metabolic-flexibility-t2d-meta-2025
metabolic-flexibility-prehabilitation-2025
cardiorespiratory-fitness-meta-2025; prospectus precedence: No

P08 — Exerkine trajectories that survive mechanistic validation

Rank: 240
Cluster: Exercise/cardiorespiratory adaptation
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: ASSOC_LAB; allocation or exposure unit: human exercise/rest samples observed; endothelial assay exposures/controls randomized separately; independent sampling unit: human participant/plasma donor; independent cell batch/model; repeated measurement unit: blood time point and cell-assay replicate; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: association or prospective prediction; temporal order and model adjustment alone do not identify a mechanism or treatment effect; specific constraint: Hemoconcentration/work confound circulating signals; cell blocking controls support pathway dependence in model only, not whole-person therapy.
Primary estimand: replicated human association of prioritized exerkine change with vascular adaptation; separate controlled cell-response contrast
Candidate primary outcome: flow-mediated-dilation change paired with prioritized blood-signal change
Primary finalized: No
Mathematical model: type: longitudinal association/reliability or held-out prediction precision simulation, not causal-efficacy power; formulas: Y_it = alpha + u_i + beta_W (B_it-Bbar_i) + beta_B Bbar_i + gamma^T X_it + f_time(t) + b_study + epsilon_it; u_i is a random intercept with a declared conditional mean/covariance model, not unrestricted fixed person effects.
B_observed,it = B_true,it + e_it; estimate measurement variance from replicates/validation or report sensitivity rather than assume it known.; variables: Y: flow-mediated-dilation change paired with prioritized blood-signal change; i: human participant/plasma donor; independent cell batch/model; t: blood time point and cell-assay replicate; the registered visit/time-window index, not another independent unit; A: human exercise/rest samples observed; endothelial assay exposures/controls randomized separately; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: replicated human association of prioritized exerkine change with vascular adaptation; separate controlled cell-response contrast; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; B it: change in the prioritized exerkine signal, locked after discovery and before replication; Bbar i: person/donor mean over a prespecified observation schedule; beta W,beta B: within-unit and between-unit association coefficients, not causal effects; X it: explicitly time-ordered baseline/premeasurement covariates; posttreatment adjustment changes interpretation; f time(t): prespecified sampling/recovery/calendar trajectory; b study: study/site/batch effect where relevant; B true,e it: latent predictor and measurement error, requiring identification constraints; family code: ASSOC_LAB
Sample size design: family: longitudinal association/reliability or held-out prediction precision simulation, not causal-efficacy power; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Hemoconcentration/work confound circulating signals; cell blocking controls support pathway dependence in model only, not whole-person therapy.
Associations need a clear sampling/time frame; causal interpretation is not supplied by this model.
Measurement error, reverse causation, time-varying confounding and selected observation may remain despite adjustment.
Randomized cell-assay perturbations identify only the controlled in-vitro contrast; human blood-signal/function association is a separate estimate.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Failure to replicate associations or loss of plasma effects under nonspecific controls would reject the nominated mechanism.; preanalysis gate: Hemoconcentration/work confound circulating signals; cell blocking controls support pathway dependence in model only, not whole-person therapy.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Hemoconcentration/work confound circulating signals; cell blocking controls support pathway dependence in model only, not whole-person therapy.
Vary lag order/window and measurement-error reliability within a prespecified family; include negative-control windows when meaningful.
Separate within- and between-unit effects, missingness patterns and source-study replication; do not use cell/assay counts as independent donor counts.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Hemoconcentration/work confound circulating signals; cell blocking controls support pathway dependence in model only, not whole-person therapy.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/239; question: Which acute exercise-induced blood signals predict later vascular adaptation and reproduce functional effects in endothelial models?; aims: Replicate acute protein trajectories across standardized exercise visits.
Test prospective association with training-induced vascular adaptation.
Evaluate prioritized participant-plasma effects in endothelial cell models.; design verbatim planning only: Population model: Screened adults undergoing supervised aerobic training, plus established human endothelial cell lines. Comparator: Rest-day versus exercise-day samples; cell assays compare pre/post plasma with prespecified pathway-blocking controls. Prospective causal strategy: Repeated human sampling followed by blinded assays; cell perturbations test pathway dependence without inferring whole-person therapeutic efficacy. Sample planning: Plan 70 participants and 30 replication volunteers; assay variance constrains prioritized signals and final recruitment.; outcomes verbatim: Primary: replicated association with flow-mediated dilation change. Secondary: oxygen uptake, endothelial function assays, and signal stability.; analysis verbatim inherited not authoritative if refined: Control adiposity, hemoconcentration, medication, and achieved work. Apply discovery false-discovery control, lock replication targets, and model cell batch and repeated donor samples.; falsification verbatim: Failure to replicate associations or loss of plasma effects under nonspecific controls would reject the nominated mechanism.; source ids unverified leads: exercise-mimetics-review-2026
autophagy-exercise-human-meta-2023
cardiorespiratory-fitness-meta-2025; prospectus precedence: No

P09 — Cleaner exercise routes without losing adaptation

Rank: 133
Cluster: Exercise/cardiorespiratory adaptation
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT_COUNT; allocation or exposure unit: participant randomized supported lower-exposure route encouragement versus feasible usual routes; independent sampling unit: participant with neighborhood clustering; repeated measurement unit: scheduled training session and exposure-monitor interval; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Exposure is encouraged rather than assigned; IV estimates need exclusion/monotonicity; route gradient/weather and missed opportunities require prespecification.
Primary estimand: route-support offer effect on symptom-free completed training sessions
Candidate primary outcome: count of symptom-free completed training sessions in the fixed block
Primary finalized: No
Mathematical model: type: overdispersed count contrast planning by simulation, with follow-up/cluster/zero inflation assumptions; formulas: Y_it ~ NegativeBinomial(mu_it,phi); log(mu_it)=alpha_t+z(A_i)^T theta_t+beta Y_i0+gamma^T X_i+b_cluster(i)+u_i.
Delta = standardized E[Y | do(A=1)] - E[Y | do(A=0)] under randomization; report ratio exp(c^T theta) only when the log-mean model supports it.; variables: Y: count of symptom-free completed training sessions in the fixed block; i: participant with neighborhood clustering; t: scheduled training session and exposure-monitor interval; the registered visit/time-window index, not another independent unit; A: participant randomized supported lower-exposure route encouragement versus feasible usual routes; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: route-support offer effect on symptom-free completed training sessions; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; mu it: conditional mean of the actual per-thesis count; phi: overdispersion parameter, with Var(Y)=mu+mu^2/phi under this parameterization; z(A): prespecified indicators for the actual arms or randomized policy; family code: RCT_COUNT
Sample size design: family: overdispersed count contrast planning by simulation, with follow-up/cluster/zero inflation assumptions; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Exposure is encouraged rather than assigned; IV estimates need exclusion/monotonicity; route gradient/weather and missed opportunities require prespecification.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: No exposure separation would invalidate the intervention test; separated exposure without tolerance or adherence benefit would challenge the hypothesis.; preanalysis gate: Exposure is encouraged rather than assigned; IV estimates need exclusion/monotonicity; route gradient/weather and missed opportunities require prespecification.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Exposure is encouraged rather than assigned; IV estimates need exclusion/monotonicity; route gradient/weather and missed opportunities require prespecification.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Exposure is encouraged rather than assigned; IV estimates need exclusion/monotonicity; route gradient/weather and missed opportunities require prespecification.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/132; question: Does access to lower-pollution exercise routes improve aerobic training tolerance while preserving completed exercise work?; aims: Map personal exposure differences between feasible local routes.
Randomize supported route choice during a standardized training program.
Estimate whether exposure reduction changes tolerance and adherence.; design verbatim planning only: Population model: Urban adults beginning walking or cycling training, including residents of highly exposed neighborhoods. Comparator: Supported lower-exposure routes versus usual feasible routes, with equal coaching and exercise targets. Prospective causal strategy: Randomized parallel encouragement trial; personal exposure monitors verify separation without assigning deliberately hazardous exposure. Sample planning: Plan 150 participants across neighborhoods; pilot compliance and exposure variance determine final sample justification.; outcomes verbatim: Primary: symptom-free completed training sessions. Secondary: peak oxygen uptake, personal particulate exposure, blood pressure, and travel burden.; analysis verbatim inherited not authoritative if refined: Account for weather, gradient, deprivation, respiratory status, and work. Report intention-to-treat effects; instrumental-variable exposure estimates require explicit encouragement assumptions and sensitivity analyses.; falsification verbatim: No exposure separation would invalidate the intervention test; separated exposure without tolerance or adherence benefit would challenge the hypothesis.; source ids unverified leads: cardiorespiratory-fitness-meta-2025
wearables-remote-monitoring-2026
rwd-digital-platform-cohort-2026; prospectus precedence: No

P10 — Minimum-data personalization for accessible aerobic training

Rank: 11
Cluster: Exercise/cardiorespiratory adaptation
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT_NI; allocation or exposure unit: participant randomized transparent minimum-data versus sensor-intensive adaptation; independent sampling unit: participant with site clustering; repeated measurement unit: CPET and retention assessment; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: No unchosen margin; equal support and supplied access define comparator; differential dropout and adherence analyses can weaken apparent noninferiority.
Primary estimand: simple-program minus sensor-program fitness difference against a justified noninferiority margin
Candidate primary outcome: measured aerobic-fitness change
Primary finalized: No
Mathematical model: type: active-comparator noninferiority precision/power with justified margin and missingness/adherence sensitivity; formulas: Delta=E[Y | assigned simple/distributed program]-E[Y | assigned sensor/continuous comparator], on the registered fitness-improvement scale where larger is better.
Declare noninferiority only if the prespecified one-sided lower confidence bound for Delta exceeds -M_NI; estimate assignment effect and a separately justified supportive adherence estimand.; variables: Y: measured aerobic-fitness change; i: participant with site clustering; t: CPET and retention assessment; the registered visit/time-window index, not another independent unit; A: participant randomized transparent minimum-data versus sensor-intensive adaptation; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: simple-program minus sensor-program fitness difference against a justified noninferiority margin; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; M NI: independently justified maximum acceptable loss in the actual aerobic-fitness unit, fixed before outcome analysis; no numerical margin supplied; family code: RCT_NI
Sample size design: family: active-comparator noninferiority precision/power with justified margin and missingness/adherence sensitivity; formula: Use design-specific simulation or independent-unit resampling for the stated estimand and target confidence width/power; preserve all allocation, censoring, clustering and measurement-error structure.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: No unchosen margin; equal support and supplied access define comparator; differential dropout and adherence analyses can weaken apparent noninferiority.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Assay sensitivity, active comparator fidelity and a justified margin are required; missingness, crossover and selection can bias toward apparent noninferiority.
Per-protocol comparisons do not automatically preserve randomization; define adherence strategy and confounding assumptions.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: A clinically meaningful sensor-program advantage despite equal support would falsify the proposed sufficiency of simple feedback.; preanalysis gate: No unchosen margin; equal support and supplied access define comparator; differential dropout and adherence analyses can weaken apparent noninferiority.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: No unchosen margin; equal support and supplied access define comparator; differential dropout and adherence analyses can weaken apparent noninferiority.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
No unchosen margin; equal support and supplied access define comparator; differential dropout and adherence analyses can weaken apparent noninferiority.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/10; question: Can a simple, transparent exercise adaptation rule match a sensor-intensive program in underserved adults?; aims: Co-design low-literacy exercise feedback with community participants.
Compare simple-rule and wearable-intensive adaptive programs.
Measure data burden, engagement, and subgroup fitness outcomes.; design verbatim planning only: Population model: Inactive adults recruited through community services with devices and connectivity supplied when needed. Comparator: Transparent perceived-exertion/work rules versus HRV/activity-sensor adaptation, with identical human support. Prospective causal strategy: Pragmatic randomized parallel trial; both algorithms are frozen before recruitment and fitness assessors remain blinded. Sample planning: Provisionally enroll 160 adults; pilot retention and CPET variance establish a defensible noninferiority margin and final sample.; outcomes verbatim: Primary: measured aerobic-fitness change. Secondary: retention, participant burden, implementation cost, accessibility, and missing-data patterns.; analysis verbatim inherited not authoritative if refined: Adjust for baseline fitness, accommodations, and site. Prespecify intention-to-treat noninferiority and supportive adherence analyses; examine differential missingness without equating disengagement with biological nonresponse.; falsification verbatim: A clinically meaningful sensor-program advantage despite equal support would falsify the proposed sufficiency of simple feedback.; source ids unverified leads: hrv-exercise-prescription-2025
smartphone-personalized-exercise-rct-2025
wearables-remote-monitoring-2026; prospectus precedence: No

P11 — Collagen versus complete protein in Achilles rehabilitation

Rank: 87
Cluster: Muscle/connective tissue/injury
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT; allocation or exposure unit: participant randomized collagen adjunct versus total-protein/energy-matched complete protein during common rehabilitation; independent sampling unit: participant with clinic effects; repeated measurement unit: tendon-function visit; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Actual diet/rehabilitation adherence are process measures; total-protein/energy matching does not imply equal amino-acid profiles or permit a placebo/no-nutrition interpretation.
Primary estimand: collagen-assignment effect on tendon-specific function at the fixed primary time
Candidate primary outcome: validated tendon-specific functional score
Primary finalized: No
Mathematical model: type: contrast precision/power by design-matrix or simulation planning; formulas: E(Y_it | A_i, X_i, Y_i0) = alpha_t + z(A_i)^T theta_t + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Delta = sum_t w_t c^T theta_t; estimate by baseline-adjusted marginal standardization, with unit/cluster-appropriate inference.; variables: Y: validated tendon-specific functional score; i: participant with clinic effects; t: tendon-function visit; the registered visit/time-window index, not another independent unit; A: participant randomized collagen adjunct versus total-protein/energy-matched complete protein during common rehabilitation; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: collagen-assignment effect on tendon-specific function at the fixed primary time; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; z(A): prespecified indicators for the actual randomized arms; two arms reduce to one indicator; family code: RCT
Sample size design: family: contrast precision/power by design-matrix or simulation planning; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Actual diet/rehabilitation adherence are process measures; total-protein/energy matching does not imply equal amino-acid profiles or permit a placebo/no-nutrition interpretation.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: A replicated functional advantage over complete protein, exceeding measurement uncertainty, would reject the hypothesis of no tendon-specific benefit.; preanalysis gate: Actual diet/rehabilitation adherence are process measures; total-protein/energy matching does not imply equal amino-acid profiles or permit a placebo/no-nutrition interpretation.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Actual diet/rehabilitation adherence are process measures; total-protein/energy matching does not imply equal amino-acid profiles or permit a placebo/no-nutrition interpretation.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Actual diet/rehabilitation adherence are process measures; total-protein/energy matching does not imply equal amino-acid profiles or permit a placebo/no-nutrition interpretation.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/86; question: Does collagen add tendon-specific functional benefit beyond matched complete protein during evidence-based Achilles rehabilitation?; aims: Establish repeatability of tendon mechanics and patient function.
Compare blinded nutritional adjuncts during identical progressive rehabilitation.
Test whether structural changes correspond to meaningful symptom improvement.; design verbatim planning only: Population model: Adults with clinician-confirmed midportion Achilles tendinopathy participating in supervised rehabilitation. Comparator: Collagen adjunct versus matched complete protein; a dietitian maintains comparable total protein and energy. Prospective causal strategy: Double-blind randomized parallel trial with blinded ultrasound assessors and standardized rehabilitation progression. Sample planning: Plan 100 participants through two clinics; pilot recruitment and function-score variance determine final sample.; outcomes verbatim: Primary: validated tendon-specific functional score. Secondary: loading tolerance, tendon stiffness, activity participation, and adverse events.; analysis verbatim inherited not authoritative if refined: Estimate the intention-to-treat effect, adjusting for baseline symptom duration, function and diet. Prespecify one primary assessment; report rehabilitation adherence as a process outcome rather than a routine primary-model covariate. Mechanistic or adherence analyses remain exploratory, and missingness receives sensitivity analysis.; falsification verbatim: A replicated functional advantage over complete protein, exceeding measurement uncertainty, would reject the hypothesis of no tendon-specific benefit.; source ids unverified leads: collagen-tendon-rct-2025
collagen-tendon-middle-age-rct-2025
collagen-umbrella-review-2026; prospectus precedence: No

P12 — Should tendon images change rehabilitation loading?

Rank: 36
Cluster: Muscle/connective tissue/injury
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: CLUSTER; allocation or exposure unit: clinician randomized symptom/function-only versus additional ultrasound information, as specified by source aim two; independent sampling unit: clinician with patients nested or crossed by treating clinician; repeated measurement unit: patient tendon-function visit and prospective clinician decision log; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Clinician is the allocation unit; define patient attribution/crossing and contamination; imaging change is not independently randomized from clinical decisions.
Primary estimand: clinician imaging-information strategy effect on tendon-specific function
Candidate primary outcome: validated tendon-specific functional-improvement score
Primary finalized: No
Mathematical model: type: contrast precision/power by design-matrix or simulation planning; formulas: E(Y_it | A_i, X_i, Y_i0) = alpha_t + z(A_i)^T theta_t + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Delta = sum_t w_t c^T theta_t; estimate by baseline-adjusted marginal standardization, with unit/cluster-appropriate inference.; variables: Y: validated tendon-specific functional-improvement score; i: clinician with patients nested or crossed by treating clinician; t: patient tendon-function visit and prospective clinician decision log; the registered visit/time-window index, not another independent unit; A: clinician randomized symptom/function-only versus additional ultrasound information, as specified by source aim two; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: clinician imaging-information strategy effect on tendon-specific function; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; z(A): prespecified indicators for the actual randomized arms; two arms reduce to one indicator; family code: CLUSTER
Sample size design: family: contrast precision/power by design-matrix or simulation planning; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Clinician is the allocation unit; define patient attribution/crossing and contamination; imaging change is not independently randomized from clinical decisions.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Number of randomized clusters, cluster size imbalance and degrees of freedom govern inference; use small-sample correction/randomization inference when appropriate.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: No functional benefit despite frequent reliable image-driven changes would refute clinical utility; poorer confidence could reveal unintended harm.; preanalysis gate: Clinician is the allocation unit; define patient attribution/crossing and contamination; imaging change is not independently randomized from clinical decisions.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Clinician is the allocation unit; define patient attribution/crossing and contamination; imaging change is not independently randomized from clinical decisions.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Clinician is the allocation unit; define patient attribution/crossing and contamination; imaging change is not independently randomized from clinical decisions.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/35; question: Does adding ultrasound mechanics to symptom-guided rehabilitation improve persistent patellar tendinopathy outcomes?; aims: Quantify mechanical-measurement reliability during routine rehabilitation visits.
Randomize clinicians to symptom-only or additional imaging information.
Evaluate decision changes, function, and cost attributable to imaging.; design verbatim planning only: Population model: Recreational jumping-sport athletes with clinically diagnosed persistent patellar tendinopathy. Comparator: Standard symptom/function-guided loading versus the same rehabilitation with prespecified ultrasound-informed decisions. Prospective causal strategy: Pragmatic parallel trial with blinded endpoint assessment; clinicians record intended decisions before viewing images. Sample planning: Plan 110 athletes across sports clinics; feasibility data on recruitment and decision changes precede a formal sample calculation.; outcomes verbatim: Primary: tendon-specific functional improvement. Secondary: return to participation, loading progression, pain-related fear, imaging change, and costs.; analysis verbatim inherited not authoritative if refined: Adjust for sport exposure, duration, function, and clinician. Model clinic effects and contamination; distinguish exploratory decision mediation from nonrandom imaging-response associations.; falsification verbatim: No functional benefit despite frequent reliable image-driven changes would refute clinical utility; poorer confidence could reveal unintended harm.; source ids unverified leads: collagen-tendon-middle-age-rct-2025
collagen-tendon-rct-2025
belief-effects-health-review-2026; prospectus precedence: No

P13 — The marginal value of resistance-training volume in later life

Rank: 3
Cluster: Muscle/connective tissue/injury
Status: DRAFT — primary death estimand unresolved; not ready for registration
Design: model family: DEATH_PENDING; allocation or exposure unit: participant randomized low/middle/high supervised training volume; independent sampling unit: participant with center/trainer/session clustering; repeated measurement unit: twenty-four-week validated function assessment; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: primary estimand unresolved pending an explicit preregistration decision; ordinary survivor mean alone is not an overall treatment-policy effect; specific constraint: Death is not a zero SPPB or unobserved visit; define survival-incorporated or explicit survivor estimand before registration; middle arm informs shape without extra primary contrasts.
Primary estimand: highest-versus-lowest assignment effect on patient-important functional status after selecting the death strategy
Candidate primary outcome: twenty-four-week SPPB score when alive, with validated health-related inability scoring
Primary finalized: No
Mathematical model: type: unresolved: joint survival/function rank/composite simulation OR explicitly identified/bounded survivor-estimand precision; mean-score power cannot be finalized; formulas: No unconditional mean-score primary estimand exists until death handling is selected: Y^a is undefined when survival S^a=0.
Candidate 1: define ordered joint Z^a=(death status, living function/quality of life) with a patient-informed ordering; estimate the full F_a(z) and independent-draw superiority probability psi=Pr(Z_1>Z_0)+0.5Pr(Z_1=Z_0).
Candidate 2: SACE=E[Y^1-Y^0 | S^1=S^0=1]; this always-survivor causal estimand is not identified by comparing observed survivors without additional assumptions/bounds. Observed-survivor summaries must be labeled conditional.; variables: Y: twenty-four-week SPPB score when alive, with validated health-related inability scoring; i: participant with center/trainer/session clustering; t: twenty-four-week validated function assessment; the registered visit/time-window index, not another independent unit; A: participant randomized low/middle/high supervised training volume; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: highest-versus-lowest assignment effect on patient-important functional status after selecting the death strategy; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; a: assigned strategy, including clinician-directed changes per its treatment policy; S^a: potential survival indicator at the primary horizon under assignment a; Y^a: living-participant endpoint under a, undefined after death; Z^a: registered joint survival/functional or quality-of-life outcome if that option is chosen; Z 1,Z 0: independent draws from the arm-specific joint distributions; not paired individual benefit; F a: arm-specific joint-outcome distribution; psi: probabilistic superiority for independent draws, if selected; SACE: survivor average causal effect among the unobserved always-survivor principal stratum; family code: DEATH_PENDING
Sample size design: family: unresolved: joint survival/function rank/composite simulation OR explicitly identified/bounded survivor-estimand precision; mean-score power cannot be finalized; formula: Use design-specific simulation or independent-unit resampling for the stated estimand and target confidence width/power; preserve all allocation, censoring, clustering and measurement-error structure.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Death is not a zero SPPB or unobserved visit; define survival-incorporated or explicit survivor estimand before registration; middle arm informs shape without extra primary contrasts.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Do not impute a conventional living score after death or combine death with a missing living assessment.
If a rank/composite is selected, update question, effect measure, estimator and power together; if a survivor estimand is selected, specify its untestable assumptions/sensitivity.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Large additional functional gains without increased burden at higher volume would contradict the diminishing-return hypothesis.; preanalysis gate: Death is not a zero SPPB or unobserved visit; define survival-incorporated or explicit survivor estimand before registration; middle arm informs shape without extra primary contrasts.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Death is not a zero SPPB or unobserved visit; define survival-incorporated or explicit survivor estimand before registration; middle arm informs shape without extra primary contrasts.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose and justify the death/intercurrent-event estimand with patients/clinicians/statistician before registration; align score, validated inability rule, primary question and sample plan.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Death is not a zero SPPB or unobserved visit; define survival-incorporated or explicit survivor estimand before registration; middle arm informs shape without extra primary contrasts.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/2; question: In screened low-strength older adults, what is the twenty-four-week functional effect of the highest versus lowest of three feasible supervised resistance-training volume strategies?; aims: Establish reliable functional and strength measurements in novices.
Randomize progressive programs differing in supervised training volume.
Estimate marginal benefit relative to burden and adherence.; design verbatim planning only: Population model: Community-dwelling older adults with low strength but cleared for supervised resistance exercise. Comparator: Three progressive volume levels with matched exercise selection, technique support, and staff contact. Prospective causal strategy: Randomized parallel trial with blinded functional testing and prospectively logged actual work. Sample planning: Plan 150 participants across community centers; pilot attendance and functional-score variance determine final recruitment.; outcomes verbatim: Primary: twenty-four-week Short Physical Performance Battery result for the prespecified highest-versus-lowest assignment contrast. Secondary: strength, lean mass where feasible, gait, fatigue, falls, persistent pain, adverse events, attendance, completed work and participant time.; analysis verbatim inherited not authoritative if refined: For the provisional twenty-four-week functional endpoint, prioritize the highest-versus-lowest assignment contrast with baseline and center adjustment. Before registration, distinguish validated inability scores from unobserved assessments and death, choose the primary intercurrent-event strategy and align analysis/power. Do not impute a living functional score after death or adjust the main estimate for completed work/adherence; report process measures and exploratory mediation separately.; falsification verbatim: Large additional functional gains without increased burden at higher volume would contradict the diminishing-return hypothesis.; source ids unverified leads: resistance-training-volume-meta-2025
resistance-sarcopenia-meta-2026
protein-exercise-network-meta-2026; prospectus precedence: Yes; prospectus file: thesis-prospectuses.json; prospectus design and estimand: Use a three-arm parallel superiority trial with individual randomization stratified by center and baseline function. All arms receive progressive resistance work under the same clinical oversight; professionally specified volume levels differ in work allocation while exercise selection and technique support are matched. Equal staff contact includes neutral seated discussion rather than extra exercise in lower-volume arms. The candidate primary estimand is the highest-versus-lowest assignment effect on 24-week function regardless of missed sessions; the middle level informs shape with less precision. Apply the functional battery’s validated scoring rules for health-related inability and distinguish this from an assessment that was not obtained. Because function is undefined after death, finalize a survival-incorporated outcome or an explicitly defined survivor estimand before registration, and update the score-based question, analysis and power plan together. An observed-survivor score difference alone is not an overall causal treatment-policy effect.; prospectus analysis plan: For the candidate 24-week battery-score contrast, fit a baseline-adjusted model with center effects and appropriate trainer or session clustering. Finalize the estimator before enrollment to match validated inability scoring and the chosen primary death strategy; if a ranked or composite outcome is selected, use its corresponding effect measure and power calculation. Prespecify bounded-score and ceiling handling and examine an ordinal sensitivity model where appropriate. Estimate the single highest-versus-lowest primary contrast first; adjust secondary contrasts and endpoints according to a registered hierarchy. Frailty and sex interactions remain exploratory unless powered. Report assignment effects separately from adherence associations, which may be confounded by health, motivation and emerging pain.; precedence note: Expanded prospectus refinements govern this scaffold; verbatim earlier brief material is retained for traceability, not as a contradictory analysis instruction. F15 is explicitly two arms; F30 has no unjustified symptom noninferiority claim.

P14 — Do early lifting gains transfer to unpracticed strength tasks?

Rank: 109
Cluster: Muscle/connective tissue/injury
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT; allocation or exposure unit: participant randomized varied low-load practice versus fixed-lift rehearsal alongside common loaded training; independent sampling unit: participant; repeated measurement unit: unpracticed force task at endpoint/retention; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Equal familiarization and fixed held-out transfer task required; work/expectancy after randomization are outcomes or pathways, not primary covariates.
Primary estimand: varied-practice assignment effect on unpracticed isometric-force change
Candidate primary outcome: force on the prespecified unpracticed isometric task
Primary finalized: No
Mathematical model: type: contrast precision/power by design-matrix or simulation planning; formulas: E(Y_it | A_i, X_i, Y_i0) = alpha_t + z(A_i)^T theta_t + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Delta = sum_t w_t c^T theta_t; estimate by baseline-adjusted marginal standardization, with unit/cluster-appropriate inference.; variables: Y: force on the prespecified unpracticed isometric task; i: participant; t: unpracticed force task at endpoint/retention; the registered visit/time-window index, not another independent unit; A: participant randomized varied low-load practice versus fixed-lift rehearsal alongside common loaded training; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: varied-practice assignment effect on unpracticed isometric-force change; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; z(A): prespecified indicators for the actual randomized arms; two arms reduce to one indicator; family code: RCT
Sample size design: family: contrast precision/power by design-matrix or simulation planning; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Equal familiarization and fixed held-out transfer task required; work/expectancy after randomization are outcomes or pathways, not primary covariates.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Precise absence of unpracticed-task or retained benefit despite verified practice separation would challenge transfer; practiced-lift gains alone would not establish generalization.; preanalysis gate: Equal familiarization and fixed held-out transfer task required; work/expectancy after randomization are outcomes or pathways, not primary covariates.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Equal familiarization and fixed held-out transfer task required; work/expectancy after randomization are outcomes or pathways, not primary covariates.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Equal familiarization and fixed held-out transfer task required; work/expectancy after randomization are outcomes or pathways, not primary covariates.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/108; question: Does varied low-load movement practice improve transfer of early strength gains beyond fixed-lift rehearsal during otherwise identical resistance training?; aims: Establish repeatability and equal familiarization of transfer tests.
Randomize varied practice versus fixed-lift rehearsal alongside common loaded training.
Test retention after additional practice stops.; design verbatim planning only: Population model: Older adults new to resistance training. Comparator: Varied low-load movement practice versus fixed-lift rehearsal; both receive identical loaded training and contact. Prospective causal strategy: Randomized eight-week parallel trial with blinded assessors and equal test familiarization; four-week retention follows practice withdrawal while common training continues. Sample planning: Plan 120 participants; pilot transfer-test reliability and recruitment determine final sample.; outcomes verbatim: Primary: force change on an unpracticed isometric task. Secondary: practiced-lift strength, chair-rise performance, retention, completed work, and burden.; analysis verbatim inherited not authoritative if refined: Use baseline-adjusted intention-to-treat contrasts and one transfer endpoint. Separate practiced-task gains, report expectancy and completed work without postassignment adjustment, and control secondary multiplicity.; falsification verbatim: Precise absence of unpracticed-task or retained benefit despite verified practice separation would challenge transfer; practiced-lift gains alone would not establish generalization.; source ids unverified leads: protein-exercise-network-meta-2026
resistance-sarcopenia-meta-2026
precision-nutrition-rct-2024; prospectus precedence: No

P15 — When muscle mass misses functional improvement

Rank: 84
Cluster: Muscle/connective tissue/injury
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: PRED; allocation or exposure unit: training-associated imaging change evaluated against independently scored function; independent sampling unit: participant with external community-cohort validation; repeated measurement unit: operator/image replicate and function visit; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: prediction, calibration or model transportability; prediction gain is not clinical efficacy, mechanistic necessity or causal treatment selection; specific constraint: If predictors measured after training this is detection/concurrent validation, not a pretraining forecast; hydration/operator error and shrinkage require explicit modeling.
Primary estimand: held-out concurrent prediction/detection of functional improvement from muscle-quality measures beyond lean mass
Candidate primary outcome: prespecified functional-improvement endpoint
Primary finalized: No
Mathematical model: type: external/nested predictive-validation precision with target prevalence/variance, independent-unit holdout coverage and model complexity; formulas: Gain = mean_test[L(Y,f_base(X)) - L(Y,f_new(X))], with both predictions frozen before test outcomes; bootstrap independent held-out units.
For a binary target assess E[Y | p_hat=p] versus p and calibration intercept/slope logit Pr(Y=1)=a_cal+b_cal logit(p_hat); for a continuous target use Y=a_cal+b_cal y_hat+error plus scale-appropriate calibration/loss.; variables: Y: prespecified functional-improvement endpoint; i: participant with external community-cohort validation; t: operator/image replicate and function visit; the registered visit/time-window index, not another independent unit; A: training-associated imaging change evaluated against independently scored function; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: held-out concurrent prediction/detection of functional improvement from muscle-quality measures beyond lean mass; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; L: one proper or scientifically justified primary loss/scoring rule fixed for the actual outcome type; positive Gain favors new model; f base,f new: locked baseline/new prediction functions using only information available at the registered prediction time; test: untouched person/family/lab/site/time set as specified per thesis; p hat,y hat: frozen probability or continuous prediction; a cal,b cal: calibration intercept and slope, estimated on evaluation data for assessment; using them to update the model requires another evaluation set; family code: PRED
Sample size design: family: external/nested predictive-validation precision with target prevalence/variance, independent-unit holdout coverage and model complexity; formula: Simulate or bootstrap the independent-unit test loss/calibration/decision criterion; plan cases/events, noncases and independent sites/families/programs separately. No universal events-per-variable or participant N is asserted.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: If predictors measured after training this is detection/concurrent validation, not a pretraining forecast; hydration/operator error and shrinkage require explicit modeling.
Train/tune/test separation at the actual biological/participant/family/time level; no future, postoutcome or target-derived predictors.
Validation support must cover intended deployment; calibration and missing/failing measurements are reported with discrimination/loss.
A validation subset selected by culture success, assay viability, consent or data completeness is a different target unless the full strategy is explicitly evaluated.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: No added validation performance beyond a simple strength test would reject the case for additional ultrasound measures.; preanalysis gate: If predictors measured after training this is detection/concurrent validation, not a pretraining forecast; hydration/operator error and shrinkage require explicit modeling.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: If predictors measured after training this is detection/concurrent validation, not a pretraining forecast; hydration/operator error and shrinkage require explicit modeling.
Repeat loss/calibration with independent-cluster resampling, prespecified subgroup/transport shifts and plausible missingness.
Compare fixed simple baselines, calibration and uncertainty coverage; report abstention/failure/turnaround rather than deleting hard cases.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Frozen prediction time/horizon, proper loss, baseline model, feature availability, independent split/holdout and target deployment population.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
If predictors measured after training this is detection/concurrent validation, not a pretraining forecast; hydration/operator error and shrinkage require explicit modeling.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/83; question: Can low-cost muscle-quality measurements detect meaningful training benefits that lean-mass estimates overlook?; aims: Estimate repeatability across hydration and operator conditions.
Track muscle quality, mass, and function during training.
Validate a minimal clinically useful measurement combination externally.; design verbatim planning only: Population model: Older adults receiving a supervised resistance program, with a second community cohort for validation. Comparator: DXA lean-mass change versus ultrasound echo intensity, muscle thickness, and simple strength testing. Prospective causal strategy: Prospective repeated-measures validation embedded in training; imaging analysts remain blinded to function and visit order. Sample planning: Plan 140 development and 60 validation participants; reliability estimates constrain detectable change and model size.; outcomes verbatim: Primary: prediction of functional improvement. Secondary: measurement error, lean-mass/strength discordance, acquisition time, and cost.; analysis verbatim inherited not authoritative if refined: Control hydration, recent exercise, settings, adiposity, operator, and frailty. Participant-held-out validation, shrinkage, and one prespecified model comparison address multiplicity and overfitting.; falsification verbatim: No added validation performance beyond a simple strength test would reject the case for additional ultrasound measures.; source ids unverified leads: resistance-sarcopenia-meta-2026
protein-exercise-network-meta-2026
resistance-training-volume-meta-2025; prospectus precedence: No

P16 — Hamstring adaptation as an injury-risk mediator

Rank: 163
Cluster: Muscle/connective tissue/injury
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: CLUSTER_RATE; allocation or exposure unit: team randomized eccentric prevention versus matched general strength; independent sampling unit: team; repeated measurement unit: athlete exposure, recurrent injury and time-loss day; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Exposure and competing injuries may change under treatment; define marginal burden/exposure ratio rather than silently condition away prevented exposure; mediation likely sparse.
Primary estimand: assignment-strategy difference in time-loss hamstring burden per observed exposure, with calendar burden also reported
Candidate primary outcome: adjudicated hamstring time-loss days and at-risk exposure hours
Primary finalized: No
Mathematical model: type: cluster-randomized recurrent/time-loss burden simulation with exposure, team correlation and sparse-event/mediation limits; formulas: Burden_a=E[D_i^a]/E[H_i^a], where D is hamstring time-loss days and H observed at-risk exposure; estimate arm ratio of standardized totals with team-level uncertainty.
A recurrent/count model may use log E[D_ict]=log(H_ict)+alpha_t+tau A_c+gamma^T X_i+b_c+u_i, but report calendar-time burden and exposure separately because exposure can be treatment affected.; variables: Y: adjudicated hamstring time-loss days and at-risk exposure hours; i: team; t: athlete exposure, recurrent injury and time-loss day; the registered visit/time-window index, not another independent unit; A: team randomized eccentric prevention versus matched general strength; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: assignment-strategy difference in time-loss hamstring burden per observed exposure, with calendar burden also reported; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; D i: adjudicated hamstring time-loss days under the chosen recurrence/severity rule; H i: at-risk sports exposure hours under the same follow-up; fixed units and competing-injury rules required; A c: team assignment; Burden a: marginal strategy-specific burden/exposure ratio, not a pure biological injury propensity; family code: CLUSTER_RATE
Sample size design: family: cluster-randomized recurrent/time-loss burden simulation with exposure, team correlation and sparse-event/mediation limits; formula: Use design-specific simulation or independent-unit resampling for the stated estimand and target confidence width/power; preserve all allocation, censoring, clustering and measurement-error structure.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Exposure and competing injuries may change under treatment; define marginal burden/exposure ratio rather than silently condition away prevented exposure; mediation likely sparse.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Team randomization, correct recurrence/severity measurement and relevant competing injuries/withdrawal; rate and total calendar burden answer different questions.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Improved strength without reduced injury burden would invalidate strength as a sufficient protective surrogate in this setting.; preanalysis gate: Exposure and competing injuries may change under treatment; define marginal burden/exposure ratio rather than silently condition away prevented exposure; mediation likely sparse.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Exposure and competing injuries may change under treatment; define marginal burden/exposure ratio rather than silently condition away prevented exposure; mediation likely sparse.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Exposure and competing injuries may change under treatment; define marginal burden/exposure ratio rather than silently condition away prevented exposure; mediation likely sparse.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/162; question: Do changes in eccentric strength and muscle architecture explain injury reduction under a supervised prevention program?; aims: Validate field strength and ultrasound repeatability across teams.
Randomize teams to prevention training or matched general conditioning.
Link longitudinal adaptation to independently adjudicated injury burden.; design verbatim planning only: Population model: Adult recreational field-sport teams across several local clubs. Comparator: Supervised eccentric-focused prevention integrated into practice versus matched general strength conditioning. Prospective causal strategy: Cluster-randomized seasonal trial with blinded injury adjudication and recorded match/training exposure. Sample planning: Plan 20 teams and 300 athletes; uncertain injury-event rates may limit mediation to exploratory analysis.; outcomes verbatim: Primary: time-loss hamstring injury burden per exposure. Secondary: eccentric strength, muscle architecture, adherence, and program acceptability.; analysis verbatim inherited not authoritative if refined: Hierarchical recurrent-event models account for teams, prior injury, and exposure. Address competing injuries and contamination; exploratory mediation tests postrandomization confounding and sparse-event sensitivity.; falsification verbatim: Improved strength without reduced injury burden would invalidate strength as a sufficient protective surrogate in this setting.; source ids unverified leads: resistance-training-volume-meta-2025
collagen-tendon-rct-2025
smartphone-personalized-exercise-rct-2025; prospectus precedence: No

P17 — Truthful load feedback and pain-related avoidance

Rank: 24
Cluster: Muscle/connective tissue/injury
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT; allocation or exposure unit: participant randomized truthful load-feedback wording during common rehabilitation; independent sampling unit: participant with clinician effects; repeated measurement unit: fear/function assessment; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Underlying care/data held common; no fabricated risk; belief change and completed rehab are postassignment pathways.
Primary estimand: uncertainty-aware versus neutral numerical wording effect on pain-related fear
Candidate primary outcome: validated pain-related fear score
Primary finalized: No
Mathematical model: type: contrast precision/power by design-matrix or simulation planning; formulas: E(Y_it | A_i, X_i, Y_i0) = alpha_t + z(A_i)^T theta_t + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Delta = sum_t w_t c^T theta_t; estimate by baseline-adjusted marginal standardization, with unit/cluster-appropriate inference.; variables: Y: validated pain-related fear score; i: participant with clinician effects; t: fear/function assessment; the registered visit/time-window index, not another independent unit; A: participant randomized truthful load-feedback wording during common rehabilitation; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: uncertainty-aware versus neutral numerical wording effect on pain-related fear; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; z(A): prespecified indicators for the actual randomized arms; two arms reduce to one indicator; family code: RCT
Sample size design: family: contrast precision/power by design-matrix or simulation planning; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Underlying care/data held common; no fabricated risk; belief change and completed rehab are postassignment pathways.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: No fear reduction or increased delayed reporting of concerning symptoms would reject the proposed communication benefit.; preanalysis gate: Underlying care/data held common; no fabricated risk; belief change and completed rehab are postassignment pathways.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Underlying care/data held common; no fabricated risk; belief change and completed rehab are postassignment pathways.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Underlying care/data held common; no fabricated risk; belief change and completed rehab are postassignment pathways.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/23; question: Does the wording of accurate load feedback change pain-related fear and rehabilitation participation when the underlying activity data and prescribed rehabilitation are identical?; aims: Co-design accurate neutral and contextualized load explanations.
Randomize feedback wording during identical musculoskeletal rehabilitation.
Test whether fear change precedes participation and functional improvement.; design verbatim planning only: Population model: Adults with persistent nonspecific knee pain receiving clinician-directed strengthening. Comparator: Two truthful presentations of the same activity data: neutral numerical feedback or contextualized uncertainty-aware feedback. Prospective causal strategy: Parallel randomized communication trial with identical exercise care and blinded functional assessment; no fabricated risk scores. Sample planning: Plan 120 participants; an acceptability pilot estimates engagement and fear-score variance before final recruitment.; outcomes verbatim: Primary: validated pain-related fear score. Secondary: completed rehabilitation, knee function, symptom reporting, and inappropriate reassurance.; analysis verbatim inherited not authoritative if refined: Adjust for pain, beliefs, literacy, and clinician. Register one primary contrast; mediation remains exploratory and rare safety events cannot establish equivalence.; falsification verbatim: No fear reduction or increased delayed reporting of concerning symptoms would reject the proposed communication benefit.; source ids unverified leads: belief-effects-health-review-2026
nocebo-communication-review-2026
wearables-living-review; prospectus precedence: No

P18 — Smartphone movement feedback after knee reconstruction

Rank: 124
Cluster: Muscle/connective tissue/injury
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT; allocation or exposure unit: participant randomized movement feedback after separate algorithm-validation cohort; independent sampling unit: participant with therapist effects; repeated measurement unit: independent movement-quality assessment; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Lock algorithm before trial; do not mix validation participants into efficacy evaluation; reinjury is exploratory unless event-based design supports it.
Primary estimand: validated-feedback offer effect on independently assessed movement quality
Candidate primary outcome: prespecified independently assessed functional movement-quality score
Primary finalized: No
Mathematical model: type: contrast precision/power by design-matrix or simulation planning; formulas: E(Y_it | A_i, X_i, Y_i0) = alpha_t + z(A_i)^T theta_t + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Delta = sum_t w_t c^T theta_t; estimate by baseline-adjusted marginal standardization, with unit/cluster-appropriate inference.; variables: Y: prespecified independently assessed functional movement-quality score; i: participant with therapist effects; t: independent movement-quality assessment; the registered visit/time-window index, not another independent unit; A: participant randomized movement feedback after separate algorithm-validation cohort; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: validated-feedback offer effect on independently assessed movement quality; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; z(A): prespecified indicators for the actual randomized arms; two arms reduce to one indicator; family code: RCT
Sample size design: family: contrast precision/power by design-matrix or simulation planning; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Lock algorithm before trial; do not mix validation participants into efficacy evaluation; reinjury is exploratory unless event-based design supports it.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Feedback that improves app scores without improving independent tasks would refute meaningful transfer and suggest measurement gaming.; preanalysis gate: Lock algorithm before trial; do not mix validation participants into efficacy evaluation; reinjury is exploratory unless event-based design supports it.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Lock algorithm before trial; do not mix validation participants into efficacy evaluation; reinjury is exploratory unless event-based design supports it.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Lock algorithm before trial; do not mix validation participants into efficacy evaluation; reinjury is exploratory unless event-based design supports it.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/123; question: Does validated smartphone movement feedback improve home rehabilitation quality beyond therapist contact and exercise reminders?; aims: Validate smartphone measures against laboratory motion assessment.
Randomize feedback during clinician-approved home rehabilitation.
Test transfer from coached exercises to unprompted functional tasks.; design verbatim planning only: Population model: Adults in a clinician-defined rehabilitation phase after anterior cruciate ligament reconstruction. Comparator: Validated movement feedback versus identical app reminders and therapist contact without biomechanical feedback. Prospective causal strategy: Two-stage validation followed by an assessor-blinded randomized trial; clinicians retain authority over progression and sport clearance. Sample planning: Plan 50 validation and 100 trial participants; agreement limits and recruitment rates determine final scope.; outcomes verbatim: Primary: independently assessed functional movement quality. Secondary: strength symmetry, adherence, patient function, and false feedback frequency.; analysis verbatim inherited not authoritative if refined: Account for surgical technique, rehabilitation stage, sport, asymmetry, and therapist. Lock algorithms, separate validation participants, and correct secondary comparisons; reinjury remains exploratory.; falsification verbatim: Feedback that improves app scores without improving independent tasks would refute meaningful transfer and suggest measurement gaming.; source ids unverified leads: smartphone-personalized-exercise-rct-2025
wearables-living-review
wearable-fall-exercise-rct-2025; prospectus precedence: No

P19 — Immediate tendon analgesia versus durable rehabilitation

Rank: 67
Cluster: Muscle/connective tissue/injury
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: SELECTION_INT; allocation or exposure unit: acute loading method crossover before separate long-term method randomization; independent sampling unit: participant; repeated measurement unit: acute response by method and long-term tendon-function visit; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Response main effect is prognosis, not treatment selection; acute period/carryover and response error need propagation; no postrandom adherence adjustment.
Primary estimand: interaction between pre-randomization acute response contrast and randomized long-term method on tendon function
Candidate primary outcome: long-term tendon-function score and reliability-corrected pretrial analgesia contrast
Primary finalized: No
Mathematical model: type: effect-modification/replication calibration simulation with prediction uncertainty, repeated randomization and independent validation units; formulas: Y_i=alpha+tau A_i+eta P_i+kappa A_i P_i+beta Y_i0+gamma^T X_i+epsilon_i for the long-term randomized endpoint.
P_i is the frozen pretrial acute analgesia difference between methods, estimated with crossover period/carryover and measurement-error uncertainty; kappa tests selection, eta prognosis. Use population intercept alpha, adding random effects only for actual repeated/clustered outcomes.; variables: Y: long-term tendon-function score and reliability-corrected pretrial analgesia contrast; i: participant; t: acute response by method and long-term tendon-function visit; the registered visit/time-window index, not another independent unit; A: acute loading method crossover before separate long-term method randomization; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: interaction between pre-randomization acute response contrast and randomized long-term method on tendon function; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; P i: frozen pre-evaluation personal/phenotype effect prediction (or pretrial acute-response contrast for P19), with scale fixed before evaluation; eta: prognostic main effect of prediction; kappa: treatment-by-frozen-prediction coefficient, distinguishing predictive modification from prognosis; X it: registered predecision or baseline covariates; no future measurements; f time(t): predeclared practice/calendar trend; family code: SELECTION_INT
Sample size design: family: effect-modification/replication calibration simulation with prediction uncertainty, repeated randomization and independent validation units; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure
distribution/reliability of frozen predictions
true effect heterogeneity scenarios
within-person randomization schedule
independent validation sample/cluster coverage; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Response main effect is prognosis, not treatment selection; acute period/carryover and response error need propagation; no postrandom adherence adjustment.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Predictions are frozen before evaluation; noisy person-specific estimates are not observed counterfactual truth; training/evaluation dependence must be propagated.
A nonzero interaction does not by itself establish useful calibration or superiority over a population-only policy.
P_i is the pre-randomization acute response to method one minus response to method two; acute period/carryover and measurement error must be modeled separately.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Reliable acute responses that predict and modify randomized long-term treatment benefit would falsify the proposed disconnect.; preanalysis gate: Response main effect is prognosis, not treatment selection; acute period/carryover and response error need propagation; no postrandom adherence adjustment.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Response main effect is prognosis, not treatment selection; acute period/carryover and response error need propagation; no postrandom adherence adjustment.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Response main effect is prognosis, not treatment selection; acute period/carryover and response error need propagation; no postrandom adherence adjustment.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/66; question: Does an early reduction in tendon pain identify the loading method that produces better long-term function?; aims: Measure repeatability of pain responses to supervised loading.
Compare acute responses under counterbalanced loading conditions.
Test prediction of function during subsequent randomized rehabilitation.; design verbatim planning only: Population model: Adults with persistent clinically assessed patellar tendon pain. Comparator: Clinician-selected isometric versus dynamic loading within accepted rehabilitation options. Prospective causal strategy: An acute crossover phase precedes a separate parallel randomized rehabilitation phase; assessors remain blinded to early response. Sample planning: Plan 90 participants; pilot reliability and retention determine precision and washout scheduling.; outcomes verbatim: Primary: incremental prediction of tendon function from acute analgesia. Secondary: activity participation, pain trajectory, and tendon mechanics.; analysis verbatim inherited not authoritative if refined: Model acute crossover period/carryover and response reliability, then analyze the parallel randomized phase by assignment. Prespecify the contrast between acute responses to the two loading methods and its interaction with randomized long-term method as the treatment-selection test; distinguish this from a response main effect that predicts prognosis. Use only pre-randomization symptoms, expectations, sport exposure and analgesic history for primary adjustment; analyze later adherence separately. Limit model complexity, validate prediction and avoid selection from a single noisy response.; falsification verbatim: Reliable acute responses that predict and modify randomized long-term treatment benefit would falsify the proposed disconnect.; source ids unverified leads: collagen-tendon-rct-2025
collagen-tendon-middle-age-rct-2025
belief-effects-health-review-2026; prospectus precedence: No

P20 — The muscle–tendon adaptation timing gap in novices

Rank: 216
Cluster: Muscle/connective tissue/injury
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: LAGGED_ASSOC; allocation or exposure unit: standard-program cohort plus separate randomized feasibility pilot; independent sampling unit: participant; repeated measurement unit: lagged strength/tendon measurement and symptom interval; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: association or prospective prediction; temporal order and model adjustment alone do not identify a mechanism or treatment effect; specific constraint: Mismatch error and emerging pain can drive exposure changes; cohort temporal ordering is not causation; pilot tests fidelity/feasibility rather than injury prevention.
Primary estimand: association between predeclared lagged strength-tendon mismatch and later persistent symptoms
Candidate primary outcome: persistent tendon-symptom indicator preceded by measured mismatch
Primary finalized: No
Mathematical model: type: longitudinal association/reliability or held-out prediction precision simulation, not causal-efficacy power; formulas: logit Pr(Y_i,t+lag=1 | B_it,X_it)=alpha_i+beta_lag B_it+gamma^T X_it+f_time(t), with predictor measured before the symptom window.
B_observed,it = B_true,it + e_it; estimate measurement variance from replicates/validation or report sensitivity rather than assume it known.; variables: Y: persistent tendon-symptom indicator preceded by measured mismatch; i: participant; t: lagged strength/tendon measurement and symptom interval; the registered visit/time-window index, not another independent unit; A: standard-program cohort plus separate randomized feasibility pilot; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: association between predeclared lagged strength-tendon mismatch and later persistent symptoms; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; B it: reliability-corrected strength-versus-tendon mismatch defined before later symptom windows; Bbar i: person/donor mean over a prespecified observation schedule; beta W,beta B: within-unit and between-unit association coefficients, not causal effects; X it: explicitly time-ordered baseline/premeasurement covariates; posttreatment adjustment changes interpretation; f time(t): prespecified sampling/recovery/calendar trajectory; b study: study/site/batch effect where relevant; B true,e it: latent predictor and measurement error, requiring identification constraints; lag: independently chosen prospective lag; not tuned across many windows to obtain significance; family code: LAGGED_ASSOC
Sample size design: family: longitudinal association/reliability or held-out prediction precision simulation, not causal-efficacy power; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Mismatch error and emerging pain can drive exposure changes; cohort temporal ordering is not causation; pilot tests fidelity/feasibility rather than injury prevention.
Associations need a clear sampling/time frame; causal interpretation is not supplied by this model.
Measurement error, reverse causation, time-varying confounding and selected observation may remain despite adjustment.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: No reliable timing gap or no prospective symptom association would undermine the proposed connective-tissue readiness marker.; preanalysis gate: Mismatch error and emerging pain can drive exposure changes; cohort temporal ordering is not causation; pilot tests fidelity/feasibility rather than injury prevention.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Mismatch error and emerging pain can drive exposure changes; cohort temporal ordering is not causation; pilot tests fidelity/feasibility rather than injury prevention.
Vary lag order/window and measurement-error reliability within a prespecified family; include negative-control windows when meaningful.
Separate within- and between-unit effects, missingness patterns and source-study replication; do not use cell/assay counts as independent donor counts.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Mismatch error and emerging pain can drive exposure changes; cohort temporal ordering is not causation; pilot tests fidelity/feasibility rather than injury prevention.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/215; question: Do strength gains outpace tendon adaptation in new lifters, does this mismatch precede emerging symptoms, and is a function-based progression rule feasible compared with standard progression?; aims: Map strength and tendon-mechanics trajectories during novice training.
Test whether trajectory mismatch precedes clinically assessed symptoms.
Pilot a conservative function-based progression rule against standard progression.; design verbatim planning only: Population model: Adults beginning resistance training with no current lower-limb tendon disorder. Comparator: A longitudinal standard-program cohort followed by a small randomized progression-rule pilot. Prospective causal strategy: Repeated blinded measurements establish temporal ordering; the pilot tests implementability rather than definitive injury prevention. Sample planning: Plan 120 cohort participants and a 60-person pilot; symptom frequency determines whether further trials are justified.; outcomes verbatim: Primary: temporal association between mismatch and persistent symptoms. Secondary: strength, tendon stiffness, adherence, and progression-rule fidelity.; analysis verbatim inherited not authoritative if refined: Account for prior activity, body mass, occupation, and actual exposure. Use lagged measurement-error models; prespecify mismatch and treat injury events as exploratory.; falsification verbatim: No reliable timing gap or no prospective symptom association would undermine the proposed connective-tissue readiness marker.; source ids unverified leads: collagen-tendon-rct-2025
collagen-tendon-middle-age-rct-2025
resistance-training-volume-meta-2025; prospectus precedence: No

P21 — Passive heat as an adjunct to ordinary endurance training

Rank: 187
Cluster: Heat/cold/altitude/hypoxia
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT; allocation or exposure unit: participant randomized passive-heat adjunct versus matched thermoneutral rest with common endurance training; independent sampling unit: participant; repeated measurement unit: delayed CPET and postcessation visit; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Immediate plasma-volume/comfort changes cannot substitute for persistence; common recovery interval avoids transient-test artifact.
Primary estimand: heat-adjunct assignment effect on persistent peak oxygen uptake
Candidate primary outcome: laboratory peak oxygen uptake at the registered delayed endpoint
Primary finalized: No
Mathematical model: type: contrast precision/power by design-matrix or simulation planning; formulas: E(Y_it | A_i, X_i, Y_i0) = alpha_t + z(A_i)^T theta_t + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Delta = sum_t w_t c^T theta_t; estimate by baseline-adjusted marginal standardization, with unit/cluster-appropriate inference.; variables: Y: laboratory peak oxygen uptake at the registered delayed endpoint; i: participant; t: delayed CPET and postcessation visit; the registered visit/time-window index, not another independent unit; A: participant randomized passive-heat adjunct versus matched thermoneutral rest with common endurance training; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: heat-adjunct assignment effect on persistent peak oxygen uptake; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; z(A): prespecified indicators for the actual randomized arms; two arms reduce to one indicator; family code: RCT
Sample size design: family: contrast precision/power by design-matrix or simulation planning; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Immediate plasma-volume/comfort changes cannot substitute for persistence; common recovery interval avoids transient-test artifact.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Loss of all apparent benefit after fluid normalization would refute durable adaptation as the explanation.; preanalysis gate: Immediate plasma-volume/comfort changes cannot substitute for persistence; common recovery interval avoids transient-test artifact.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Immediate plasma-volume/comfort changes cannot substitute for persistence; common recovery interval avoids transient-test artifact.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Immediate plasma-volume/comfort changes cannot substitute for persistence; common recovery interval avoids transient-test artifact.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/186; question: Does supervised passive heat add durable aerobic benefit beyond identical endurance training and recovery time?; aims: Establish hydration-standardized testing and heat-session tolerability.
Randomize adjunct heat versus thermoneutral supervised rest.
Separate persistent adaptation from short-lived volume effects.; design verbatim planning only: Population model: Screened recreational adults completing the same aerobic training program. Comparator: Facility-supervised passive heat versus time- and attention-matched thermoneutral rest. Prospective causal strategy: Parallel randomized trial; blinded testing occurs after a standardized recovery interval and again after adjunct cessation. Sample planning: Plan 100 participants; a pilot estimates tolerability, attrition, and CPET variance before definitive sample calculation.; outcomes verbatim: Primary: persistent peak oxygen-uptake change. Secondary: plasma-volume estimates, exercise tolerance, vascular function, and adverse symptoms.; analysis verbatim inherited not authoritative if refined: Adjust for fitness, hydration, medication, ambient conditions, and training. Prespecify delayed primary testing, correct mechanistic endpoints, and test sensitivity to missed sessions.; falsification verbatim: Loss of all apparent benefit after fluid normalization would refute durable adaptation as the explanation.; source ids unverified leads: heat-thermotherapy-meta-2025
cardiorespiratory-fitness-meta-2025
exercise-electrolyte-rehydration-review-2025; prospectus precedence: No

P22 — Cold exposure timing and the adaptation–comfort tradeoff

Rank: 197
Cluster: Heat/cold/altitude/hypoxia
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT3; allocation or exposure unit: participant randomized immediate/delayed cold or thermoneutral immersion; independent sampling unit: participant; repeated measurement unit: strength/imaging endpoint; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Parallel allocation avoids irreversible training carryover; comfort/work are separate outcomes; other comparisons use registered multiplicity control.
Primary estimand: immediate-versus-delayed cold assignment effect on strength adaptation
Candidate primary outcome: prespecified strength adaptation endpoint
Primary finalized: No
Mathematical model: type: contrast precision/power by design-matrix or simulation planning; formulas: E(Y_it | A_i, X_i, Y_i0) = alpha_t + z(A_i)^T theta_t + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Delta = sum_t w_t c^T theta_t; estimate by baseline-adjusted marginal standardization, with unit/cluster-appropriate inference.; variables: Y: prespecified strength adaptation endpoint; i: participant; t: strength/imaging endpoint; the registered visit/time-window index, not another independent unit; A: participant randomized immediate/delayed cold or thermoneutral immersion; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: immediate-versus-delayed cold assignment effect on strength adaptation; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; z(A): prespecified indicators for the actual randomized arms; two arms reduce to one indicator; family code: RCT3
Sample size design: family: contrast precision/power by design-matrix or simulation planning; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Parallel allocation avoids irreversible training carryover; comfort/work are separate outcomes; other comparisons use registered multiplicity control.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Primary contrast is fixed before observing outcomes; the third arm does not create another uncorrected primary test.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Equivalent durable adaptation despite verified physiological timing differences would reject the proposed timing-dependent attenuation.; preanalysis gate: Parallel allocation avoids irreversible training carryover; comfort/work are separate outcomes; other comparisons use registered multiplicity control.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Parallel allocation avoids irreversible training carryover; comfort/work are separate outcomes; other comparisons use registered multiplicity control.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Parallel allocation avoids irreversible training carryover; comfort/work are separate outcomes; other comparisons use registered multiplicity control.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/196; question: Does the timing of supervised cold immersion change longer-term strength adaptation, and how do adaptation effects compare with changes in perceived comfort?; aims: Measure expectancy and acute physiological responses under standardized conditions.
Randomize exposure timing during identical resistance training.
Compare durable adaptation with perceived postexercise recovery.; design verbatim planning only: Population model: Healthy recreational adults entering supervised resistance training. Comparator: Immediate versus delayed cold exposure, plus thermoneutral immersion matched for contact and context. Prospective causal strategy: Parallel randomized trial with blinded strength and imaging assessments; exposure procedures remain within facility-approved safety practice. Sample planning: Plan 120 participants; pilot variation in adherence and strength change informs final sample and comparison priorities.; outcomes verbatim: Primary: strength adaptation. Secondary: muscle thickness, tendon mechanics, comfort, training completion, and adverse events.; analysis verbatim inherited not authoritative if refined: Estimate assignment-based total effects, adjusting for baseline strength, diet and preassignment expectancy. Prespecify the immediate-versus-delayed contrast and correct additional comparisons; report comfort and completed work as separate outcomes rather than routine postassignment covariates. Any expectancy/work mediation is exploratory; parallel allocation avoids training carryover.; falsification verbatim: Equivalent durable adaptation despite verified physiological timing differences would reject the proposed timing-dependent attenuation.; source ids unverified leads: cold-recovery-rct-2025
cold-metabolism-review-2026
resistance-training-volume-meta-2025; prospectus precedence: No

P23 — Does cold-induced expenditure survive appetite compensation?

Rank: 204
Cluster: Heat/cold/altitude/hypoxia
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: CROSS_ENERGY; allocation or exposure unit: participant randomized cold/thermoneutral repeated-period sequence; independent sampling unit: participant; repeated measurement unit: meal/calorimetry/activity day within period; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Component errors can be correlated; carryover/habituation and unmeasured intake/activity may dominate; body weight is secondary.
Primary estimand: cold-condition assignment effect on net estimated energy balance including compensation
Candidate primary outcome: energy intake minus measured expenditure under a fixed accounting horizon
Primary finalized: No
Mathematical model: type: paired/crossover contrast precision with period, carryover and within-unit covariance; simulation for actual design; formulas: g(E[Y_ipt | A_ip,X_i]) = alpha_i + tau A_ip + pi_p + rho A_i,p-1 + gamma^T X_i + f_time(t) + b_cluster(i,p).
Target the registered within-unit assignment contrast, standardized over period/history; direct and lag effects are distinct parameters.
Y_ip = I_ip - EE_ip; Var(Y_ip)=Var(I_ip)+Var(EE_ip)-2 Cov(I_ip,EE_ip).; variables: Y: energy intake minus measured expenditure under a fixed accounting horizon; i: participant; t: meal/calorimetry/activity day within period; the registered visit/time-window index, not another independent unit; A: participant randomized cold/thermoneutral repeated-period sequence; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: cold-condition assignment effect on net estimated energy balance including compensation; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; p: randomized crossover period/session/episode, not the outcome probability; pi p: period effect; include calendar/season/sequence terms supported by the design; rho: prespecified lagged-assignment/carryover parameter; cannot in general identify arbitrary persistent carryover in a short two-period trial; f time(t): prespecified within-period recovery/day trend; g: identity for mean outcomes, log for positive/count means, or logit for binary/proportion outcomes as specified below; I ip: estimated total energy intake over the fixed accounting horizon; EE ip: measured/estimated total energy expenditure over that same horizon; no double counting of activity expenditure; family code: CROSS_ENERGY
Sample size design: family: paired/crossover contrast precision with period, carryover and within-unit covariance; simulation for actual design; formula: For a simple two-condition paired mean only: Var(Delta_hat)=sigma_D^2/n_independent; actual cluster/period/lag models require design-specific simulation or c^T(D^T V^(-1)D)^(-1)c.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; sigma D^2: variance of independent-unit paired contrasts including measurement error; n independent: number of independent participants/teams/donors, not repeated samples; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Component errors can be correlated; carryover/habituation and unmeasured intake/activity may dominate; body weight is secondary.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Concealed randomized sequence, a scientifically justified recovery interval and no anticipation; period/carryover cannot be repaired solely by a nonsignificant test.
If carryover is plausible, report lag/period and first-period or otherwise valid design sensitivity; do not drop periods based only on a post hoc carryover p-value.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: A sustained expenditure surplus without intake or activity compensation would contradict the proposed offset mechanism.; preanalysis gate: Component errors can be correlated; carryover/habituation and unmeasured intake/activity may dominate; body weight is secondary.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Component errors can be correlated; carryover/habituation and unmeasured intake/activity may dominate; body weight is secondary.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Compare planned lag/carryover structures and a valid first-period sensitivity where its sample supports inference; assess informative absent periods.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Component errors can be correlated; carryover/habituation and unmeasured intake/activity may dominate; body weight is secondary.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/203; question: Does repeated controlled cold exposure change net energy balance after compensatory intake and movement are measured?; aims: Replicate acute expenditure and appetite responses across visits.
Track compensation during repeated randomized exposure periods.
Test persistence after exposure stops without prescribing weight loss.; design verbatim planning only: Population model: Screened physically active adults with stable weight and no eating-disorder risk. Comparator: Supervised cold versus thermoneutral exposure under otherwise comparable daily routines. Prospective causal strategy: Randomized repeated-period study with counterbalanced order, monitored washout, and standardized laboratory meal measurements. Sample planning: Plan 60 participants; pilot within-person variability and habituation determine period length and precision requirements.; outcomes verbatim: Primary: estimated net energy-balance difference. Secondary: measured intake, calorimetry, spontaneous activity, appetite, and tolerability.; analysis verbatim inherited not authoritative if refined: Model period/carryover, body composition, cold experience, and ambient exposure. Quantify intake uncertainty and prespecify compensation; body-weight change remains exploratory.; falsification verbatim: A sustained expenditure surplus without intake or activity compensation would contradict the proposed offset mechanism.; source ids unverified leads: cold-appetite-rct-2025
cold-metabolism-review-2026
wearables-living-review; prospectus precedence: No

P24 — Sweat electrolyte patches against whole-body balance

Rank: 88
Cluster: Heat/cold/altitude/hypoxia
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: AGREE; allocation or exposure unit: counterbalanced assessment environment; patch and whole-body references paired; independent sampling unit: participant; repeated measurement unit: patch site, visit, washdown and electrolyte assay; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: measurement agreement/reliability; condition contrasts are causal only where the condition itself was randomized; specific constraint: Local concentration times whole-body volume is a model requiring validation; contamination/recovery and repeat visits do not add independent participants.
Primary estimand: individual whole-body electrolyte-loss estimation error and out-of-condition agreement
Candidate primary outcome: patch-estimated minus reference whole-body loss in the fixed physical unit
Primary finalized: No
Mathematical model: type: paired hierarchical agreement/bias/limits precision with independent participants/donors/labs and replicate allocation; formulas: D_ij = S_ij - R_ij = beta_0 + beta_context^T C_ij + u_i + b_lab + epsilon_ij.
Context-specific agreement limits use beta_0 + beta_context^T C +/- z_(1-alpha/2) sqrt(Var(D_new | target context)); distinguish a new person from a repeat reading.
If reference error is material, S_ij = T_ij + b_S + e_S,ij and R_ij = T_ij + b_R + e_R,ij; identifiability needs reference validation/replicates or stated constraints.; variables: Y: patch-estimated minus reference whole-body loss in the fixed physical unit; i: participant; t: patch site, visit, washdown and electrolyte assay; the registered visit/time-window index, not another independent unit; A: counterbalanced assessment environment; patch and whole-body references paired; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: individual whole-body electrolyte-loss estimation error and out-of-condition agreement; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; S ij: sensor/test result in the stated endpoint unit; R ij: paired validated reference result in the same unit; D ij: paired test-minus-reference error, matching the per-thesis outcome; C ij: prespecified context, firmware/device and subgroup features; T ij: latent underlying quantity; not automatically identifiable; b S,b R: test and reference biases; e S,e R: test and reference measurement errors; b lab: laboratory effect when multiple laboratories actually sampled; D new: prediction error for the prespecified future measurement context; family code: AGREE
Sample size design: family: paired hierarchical agreement/bias/limits precision with independent participants/donors/labs and replicate allocation; formula: Plan confidence width for bias and agreement limits by hierarchical simulation; n readings is not n independent people. Acceptance limits in physical units must be independently justified.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Local concentration times whole-body volume is a model requiring validation; contamination/recovery and repeat visits do not add independent participants.
Reference calibration, matched time/unit and appropriately modeled reference error; correlation is not agreement.
Future-context prediction assumptions must match held-out persons/labs/conditions; missing/failing readings are outcomes of the measurement system.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Stable uncalibrated individual agreement across conditions would reject the proposed need for contextual recalibration.; preanalysis gate: Local concentration times whole-body volume is a model requiring validation; contamination/recovery and repeat visits do not add independent participants.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Local concentration times whole-body volume is a model requiring validation; contamination/recovery and repeat visits do not add independent participants.
Assess proportional/heteroscedastic error, reference uncertainty and subgroup-specific failure, not just a global correlation.
Repeat estimates with and without questionable quality strata under predeclared rules; report excluded/failing measurements explicitly.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Local concentration times whole-body volume is a model requiring validation; contamination/recovery and repeat visits do not add independent participants.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/87; question: Can a local sweat patch estimate individual whole-body electrolyte loss well enough to inform research hydration decisions?; aims: Validate patch measurements against laboratory chemistry under controlled sweating.
Compare local estimates with whole-body collection across conditions.
Test a locked calibration model in independent participants.; design verbatim planning only: Population model: Screened adults spanning body sizes and habitual activity levels during supervised exercise. Comparator: Commercial or research patches versus whole-body washdown, body-mass balance, and laboratory electrolyte assays. Prospective causal strategy: Prospective counterbalanced validation across facility-approved environments; no patch-guided drinking intervention occurs before validation. Sample planning: Plan 80 participants, with 30 repeat visits; assay and agreement precision determine final recruitment.; outcomes verbatim: Primary: individual loss-estimation error. Secondary: calibration drift, collection failures, site effects, usability, and subgroup error.; analysis verbatim inherited not authoritative if refined: Model rate, site, acclimation, body size, contamination, and repeated visits. Use agreement limits, independent testing, and prespecified acceptable error rather than correlation alone.; falsification verbatim: Stable uncalibrated individual agreement across conditions would reject the proposed need for contextual recalibration.; source ids unverified leads: sweat-biosensors-review-2025
sweat-multianalyte-microfluidic-2025
exercise-electrolyte-rehydration-review-2025; prospectus precedence: No

P25 — Body size and fitness in heat-acclimation variability

Rank: 218
Cluster: Heat/cold/altitude/hypoxia
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: OBS_TRAJECTORY; allocation or exposure unit: sex/body-size attributes observed within common supervised acclimation framework; independent sampling unit: participant; repeated measurement unit: standardized heat-strain assessment over acclimation; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: association or prospective prediction; temporal order and model adjustment alone do not identify a mechanism or treatment effect; specific constraint: Sex is not randomized; achieved strain may mediate adjusted differences; report total and assumption-dependent conditional associations separately.
Primary estimand: sex-disaggregated longitudinal adaptation association after prespecified baseline adjustment
Candidate primary outcome: change in standardized exercise heat-strain endpoint
Primary finalized: No
Mathematical model: type: longitudinal association/reliability or held-out prediction precision simulation, not causal-efficacy power; formulas: E(Y_it | S_i,X_i0)=alpha+u_i+f(t)+z(S_i)^T beta+z(S_i)^T g(t;theta)+gamma^T X_i0.
Estimate standardized sex-disaggregated baseline-to-final acclimation changes and their contrasts; t is the common assessment occasion. These are observational group associations, not causal effects of sex.; variables: Y: change in standardized exercise heat-strain endpoint; i: participant; t: standardized heat-strain assessment over acclimation; the registered visit/time-window index, not another independent unit; A: sex/body-size attributes observed within common supervised acclimation framework; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: sex-disaggregated longitudinal adaptation association after prespecified baseline adjustment; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; B it: prespecified molecular/physiological/state predictor paired with the per-thesis endpoint; exact definition/timing fixed before analysis; Bbar i: person/donor mean over a prespecified observation schedule; beta W,beta B: within-unit and between-unit association coefficients, not causal effects; X it: explicitly time-ordered baseline/premeasurement covariates; posttreatment adjustment changes interpretation; f time(t): prespecified sampling/recovery/calendar trajectory; b study: study/site/batch effect where relevant; B true,e it: latent predictor and measurement error, requiring identification constraints; S i: observed sex-related grouping/attributes under prespecified inclusive coding, not randomized treatment; z(S i): registered indicators/contrasts for supported groups; f(t): common acclimation-time trajectory; g(t;theta): prespecified group-by-acclimation-time deviations; X i0: baseline body size, fitness, medication and other pre-exposure variables; achieved exposure/strain adjustment is separately labeled conditional analysis; u i: participant random intercept with specified covariance; family code: OBS_TRAJECTORY
Sample size design: family: longitudinal association/reliability or held-out prediction precision simulation, not causal-efficacy power; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Sex is not randomized; achieved strain may mediate adjusted differences; report total and assumption-dependent conditional associations separately.
Associations need a clear sampling/time frame; causal interpretation is not supplied by this model.
Measurement error, reverse causation, time-varying confounding and selected observation may remain despite adjustment.
Sex is time invariant: do not try to estimate its within-person centered main effect. Group-by-time associations depend on supported groups and chosen conditioning set.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Large reproducible group differences unexplained by measured scaling factors would challenge the main explanatory hypothesis.; preanalysis gate: Sex is not randomized; achieved strain may mediate adjusted differences; report total and assumption-dependent conditional associations separately.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Sex is not randomized; achieved strain may mediate adjusted differences; report total and assumption-dependent conditional associations separately.
Vary lag order/window and measurement-error reliability within a prespecified family; include negative-control windows when meaningful.
Separate within- and between-unit effects, missingness patterns and source-study replication; do not use cell/assay counts as independent donor counts.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Sex is not randomized; achieved strain may mediate adjusted differences; report total and assumption-dependent conditional associations separately.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/217; question: Which apparent sex differences in heat adaptation remain after body size, fitness, and achieved heat strain are considered?; aims: Establish reliable strain and adaptation measurements across participants.
Compare responses under standardized relative-work and strain-matched conditions.
Validate transparent predictors without reproductive treatment or phase-based prescriptions.; design verbatim planning only: Population model: Screened adult athletes of different sexes and body sizes, recruited inclusively. Comparator: Counterbalanced familiarization conditions followed by the same supervised acclimation framework with clinically approved individual adjustment. Prospective causal strategy: Prospective repeated-measures physiology study; blinded analysis separates baseline differences from longitudinal adaptation. Sample planning: Plan 100 participants; pilot repeatability and predictor correlations limit the number of prespecified explanatory variables.; outcomes verbatim: Primary: change in standardized exercise heat strain. Secondary: sweating responses, plasma-volume estimates, tolerance, and prediction calibration.; analysis verbatim inherited not authoritative if refined: Model fitness, size, medication, training, and actual exposure. Report uncertain sex-disaggregated estimates without causal group claims; correct the prespecified interaction set.; falsification verbatim: Large reproducible group differences unexplained by measured scaling factors would challenge the main explanatory hypothesis.; source ids unverified leads: heat-thermotherapy-meta-2025
hyperhydration-heat-rct-2025
exercise-electrolyte-rehydration-review-2025; prospectus precedence: No

P26 — Relative-intensity pacing during supervised hypoxic exercise

Rank: 223
Cluster: Heat/cold/altitude/hypoxia
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: CROSS_PROP; allocation or exposure unit: participant randomized physiological-pacing/fixed-workload sequence; separate feasibility phase; independent sampling unit: participant; repeated measurement unit: fixed common-duration task time segment; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Strategy-specific prescription attainment is a different denominator; recovery/washout and safety stops are retained; no unsupervised-altitude claim.
Primary estimand: pacing-assignment effect on fraction of common task duration completed in intended physiological domain
Candidate primary outcome: time in intended domain divided by the same prespecified total task duration
Primary finalized: No
Mathematical model: type: paired/crossover contrast precision with period, carryover and within-unit covariance; simulation for actual design; formulas: g(E[Y_ipt | A_ip,X_i]) = alpha_i + tau A_ip + pi_p + rho A_i,p-1 + gamma^T X_i + f_time(t) + b_cluster(i,p).
Target the registered within-unit assignment contrast, standardized over period/history; direct and lag effects are distinct parameters.
Y_ip = D_ip/T_common; model bounded proportion with an appropriate mean model and cluster-robust inference, retaining safety-stopped remainder under a prespecified scoring rule.; variables: Y: time in intended domain divided by the same prespecified total task duration; i: participant; t: fixed common-duration task time segment; the registered visit/time-window index, not another independent unit; A: participant randomized physiological-pacing/fixed-workload sequence; separate feasibility phase; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: pacing-assignment effect on fraction of common task duration completed in intended physiological domain; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; p: randomized crossover period/session/episode, not the outcome probability; pi p: period effect; include calendar/season/sequence terms supported by the design; rho: prespecified lagged-assignment/carryover parameter; cannot in general identify arbitrary persistent carryover in a short two-period trial; f time(t): prespecified within-period recovery/day trend; g: identity for mean outcomes, log for positive/count means, or logit for binary/proportion outcomes as specified below; D ip: duration actually completed in the intended physiological domain; T common: the same prespecified task duration in both strategies; family code: CROSS_PROP
Sample size design: family: paired/crossover contrast precision with period, carryover and within-unit covariance; simulation for actual design; formula: For a simple two-condition paired mean only: Var(Delta_hat)=sigma_D^2/n_independent; actual cluster/period/lag models require design-specific simulation or c^T(D^T V^(-1)D)^(-1)c.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; sigma D^2: variance of independent-unit paired contrasts including measurement error; n independent: number of independent participants/teams/donors, not repeated samples; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Strategy-specific prescription attainment is a different denominator; recovery/washout and safety stops are retained; no unsupervised-altitude claim.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Concealed randomized sequence, a scientifically justified recovery interval and no anticipation; period/carryover cannot be repaired solely by a nonsignificant test.
If carryover is plausible, report lag/period and first-period or otherwise valid design sensitivity; do not drop periods based only on a post hoc carryover p-value.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: No improvement in common-task completion within the intended domain would refute the proposed pacing advantage; improved strategy-specific attainment alone would not establish preserved absolute work.; preanalysis gate: Strategy-specific prescription attainment is a different denominator; recovery/washout and safety stops are retained; no unsupervised-altitude claim.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Strategy-specific prescription attainment is a different denominator; recovery/washout and safety stops are retained; no unsupervised-altitude claim.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Compare planned lag/carryover structures and a valid first-period sensitivity where its sample supports inference; assess informative absent periods.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Strategy-specific prescription attainment is a different denominator; recovery/washout and safety stops are retained; no unsupervised-altitude claim.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/222; question: Does physiological pacing preserve training quality better than fixed external workload when exercising in controlled hypoxia?; aims: Map individual workload-to-strain changes under controlled hypoxia.
Compare fixed-work and relative-intensity pacing in counterbalanced sessions.
Pilot repeated-session adherence using the better-tolerated strategy.; design verbatim planning only: Population model: Medically screened recreational endurance athletes in an accredited hypoxia facility. Comparator: Fixed external workload versus individual physiological pacing during identical supervised environmental exposure. Prospective causal strategy: Randomized acute crossover with verified recovery, followed by a small parallel feasibility phase; neither phase tests unsupervised altitude use. Sample planning: Plan 50 crossover and 40 pilot participants; repeatability and intolerance frequency determine precision and later recruitment.; outcomes verbatim: Primary: proportion of a common prespecified task duration completed within the intended physiological domain. Secondary: absolute external work completed, symptoms, oxygenation, exertion, and session completion; strategy-specific prescription attainment is reported separately.; analysis verbatim inherited not authoritative if refined: Adjust for fitness, altitude experience, iron status, and order. Model period/carryover and one primary contrast; saturation shifts alone do not establish adaptation.; falsification verbatim: No improvement in common-task completion within the intended domain would refute the proposed pacing advantage; improved strategy-specific attainment alone would not establish preserved absolute work.; source ids unverified leads: intermittent-hypoxia-cognition-rct-2025
cardiorespiratory-fitness-meta-2025
metabolic-flexibility-prehabilitation-2025; prospectus precedence: No

P27 — Are normobaric and hypobaric exercise models interchangeable?

Rank: 238
Cluster: Heat/cold/altitude/hypoxia
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: CROSS_EQ; allocation or exposure unit: participant randomized verified normobaric/hypobaric/normoxic condition sequence; independent sampling unit: participant; repeated measurement unit: standardized ventilation measurement within period; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Match inspired partial pressure independently; achieved oxygenation is an outcome; nonsignificance does not establish interchangeability.
Primary estimand: normobaric-minus-hypobaric ventilation difference against justified equivalence bounds
Candidate primary outcome: ventilation at the fixed standardized relative workload
Primary finalized: No
Mathematical model: type: paired equivalence/TOST planning with justified asymmetric/symmetric bounds and carryover sensitivity; formulas: g(E[Y_ipt | A_ip,X_i]) = alpha_i + tau A_ip + pi_p + rho A_i,p-1 + gamma^T X_i + f_time(t) + b_cluster(i,p).
Target the registered within-unit assignment contrast, standardized over period/history; direct and lag effects are distinct parameters.
Equivalence requires CI_(1-2alpha)(Delta) entirely inside [-M_lower, M_upper], using two one-sided tests at alpha; a nonsignificant difference is insufficient.; variables: Y: ventilation at the fixed standardized relative workload; i: participant; t: standardized ventilation measurement within period; the registered visit/time-window index, not another independent unit; A: participant randomized verified normobaric/hypobaric/normoxic condition sequence; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: normobaric-minus-hypobaric ventilation difference against justified equivalence bounds; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; p: randomized crossover period/session/episode, not the outcome probability; pi p: period effect; include calendar/season/sequence terms supported by the design; rho: prespecified lagged-assignment/carryover parameter; cannot in general identify arbitrary persistent carryover in a short two-period trial; f time(t): prespecified within-period recovery/day trend; g: identity for mean outcomes, log for positive/count means, or logit for binary/proportion outcomes as specified below; M lower,M upper: independently justified practical/physiological equivalence bounds in the actual endpoint unit, not supplied here; alpha: one-sided test error rate fixed in registration; family code: CROSS_EQ
Sample size design: family: paired equivalence/TOST planning with justified asymmetric/symmetric bounds and carryover sensitivity; formula: For a simple two-condition paired mean only: Var(Delta_hat)=sigma_D^2/n_independent; actual cluster/period/lag models require design-specific simulation or c^T(D^T V^(-1)D)^(-1)c.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; sigma D^2: variance of independent-unit paired contrasts including measurement error; n independent: number of independent participants/teams/donors, not repeated samples; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Match inspired partial pressure independently; achieved oxygenation is an outcome; nonsignificance does not establish interchangeability.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Concealed randomized sequence, a scientifically justified recovery interval and no anticipation; period/carryover cannot be repaired solely by a nonsignificant test.
If carryover is plausible, report lag/period and first-period or otherwise valid design sensitivity; do not drop periods based only on a post hoc carryover p-value.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Narrow equivalence intervals across primary physiology and task response would reject the predicted model difference.; preanalysis gate: Match inspired partial pressure independently; achieved oxygenation is an outcome; nonsignificance does not establish interchangeability.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Match inspired partial pressure independently; achieved oxygenation is an outcome; nonsignificance does not establish interchangeability.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Compare planned lag/carryover structures and a valid first-period sensitivity where its sample supports inference; assess informative absent periods.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Externally justified equivalence bounds; credible comparator and uncertainty in paired differences.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Match inspired partial pressure independently; achieved oxygenation is an outcome; nonsignificance does not establish interchangeability.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/237; question: Do conditions matched for inspired oxygen partial pressure produce comparable exercise strain in normobaric and hypobaric research environments?; aims: Establish chamber-specific measurement reliability and participant blinding credibility.
Compare exercise physiology under counterbalanced matched-oxygen conditions.
Test whether remaining differences alter a standardized training-task response.; design verbatim planning only: Population model: Medically screened healthy adults studied in an established environmental chamber facility. Comparator: Normobaric and hypobaric hypoxia matched for inspired oxygen partial pressure, plus a normoxic reference under matched exercise tasks. Prospective causal strategy: Randomized crossover with independent gas/pressure verification, standardized recovery intervals, and blinded outcome analysis. Sample planning: Plan 40 completers; pilot within-person variation and chamber scheduling determine feasible precision and contrast prioritization.; outcomes verbatim: Primary: ventilation at standardized relative exercise workload. Secondary: oxygenation, cardiac response, perceived exertion, and completed work.; analysis verbatim inherited not authoritative if refined: Estimate total randomized condition effects using paired models for period/carryover and precondition altitude experience and hydration. Treat achieved oxygenation as a response, not a routine adjustment covariate; any oxygenation-mediated analysis is exploratory. Prespecify paired contrasts and reliability-based equivalence bounds; nonsignificance cannot establish interchangeability.; falsification verbatim: Narrow equivalence intervals across primary physiology and task response would reject the predicted model difference.; source ids unverified leads: intermittent-hypoxia-cognition-rct-2025
hbot-physical-performance-rct-2024
cardiorespiratory-fitness-meta-2025; prospectus precedence: No

P28 — Oxygenation traces versus meaningful altitude exercise tolerance

Rank: 234
Cluster: Heat/cold/altitude/hypoxia
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: PRED; allocation or exposure unit: future-session prediction in independently selected supervised altitude camp; independent sampling unit: participant and held-out camp; repeated measurement unit: pre-session prediction and later adjudicated modification; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: prediction, calibration or model transportability; prediction gain is not clinical efficacy, mechanistic necessity or causal treatment selection; specific constraint: Exclude concurrent adjudication symptoms from predictors; camp choice is observational; participants whose sessions stop supply safety-stop/missingness information.
Primary estimand: future-session intolerance prediction gain beyond saturation-only prediction at one frozen horizon
Candidate primary outcome: clinician-adjudicated intolerance requiring future-session modification
Primary finalized: No
Mathematical model: type: external/nested predictive-validation precision with target prevalence/variance, independent-unit holdout coverage and model complexity; formulas: Gain = mean_test[L(Y,f_base(X)) - L(Y,f_new(X))], with both predictions frozen before test outcomes; bootstrap independent held-out units.
For a binary target assess E[Y | p_hat=p] versus p and calibration intercept/slope logit Pr(Y=1)=a_cal+b_cal logit(p_hat); for a continuous target use Y=a_cal+b_cal y_hat+error plus scale-appropriate calibration/loss.; variables: Y: clinician-adjudicated intolerance requiring future-session modification; i: participant and held-out camp; t: pre-session prediction and later adjudicated modification; the registered visit/time-window index, not another independent unit; A: future-session prediction in independently selected supervised altitude camp; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: future-session intolerance prediction gain beyond saturation-only prediction at one frozen horizon; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; L: one proper or scientifically justified primary loss/scoring rule fixed for the actual outcome type; positive Gain favors new model; f base,f new: locked baseline/new prediction functions using only information available at the registered prediction time; test: untouched person/family/lab/site/time set as specified per thesis; p hat,y hat: frozen probability or continuous prediction; a cal,b cal: calibration intercept and slope, estimated on evaluation data for assessment; using them to update the model requires another evaluation set; family code: PRED
Sample size design: family: external/nested predictive-validation precision with target prevalence/variance, independent-unit holdout coverage and model complexity; formula: Simulate or bootstrap the independent-unit test loss/calibration/decision criterion; plan cases/events, noncases and independent sites/families/programs separately. No universal events-per-variable or participant N is asserted.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Exclude concurrent adjudication symptoms from predictors; camp choice is observational; participants whose sessions stop supply safety-stop/missingness information.
Train/tune/test separation at the actual biological/participant/family/time level; no future, postoutcome or target-derived predictors.
Validation support must cover intended deployment; calibration and missing/failing measurements are reported with discrimination/loss.
A validation subset selected by culture success, assay viability, consent or data completeness is a different target unless the full strategy is explicitly evaluated.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Reliable saturation alone matching enriched models across camps would falsify the proposed incremental physiological value.; preanalysis gate: Exclude concurrent adjudication symptoms from predictors; camp choice is observational; participants whose sessions stop supply safety-stop/missingness information.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Exclude concurrent adjudication symptoms from predictors; camp choice is observational; participants whose sessions stop supply safety-stop/missingness information.
Repeat loss/calibration with independent-cluster resampling, prespecified subgroup/transport shifts and plausible missingness.
Compare fixed simple baselines, calibration and uncertainty coverage; report abstention/failure/turnaround rather than deleting hard cases.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Frozen prediction time/horizon, proper loss, baseline model, feature availability, independent split/holdout and target deployment population.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Exclude concurrent adjudication symptoms from predictors; camp choice is observational; participants whose sessions stop supply safety-stop/missingness information.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/233; question: Do baseline ventilatory measures predict tolerable exercise at altitude better than a consumer oxygen-saturation trace?; aims: Validate consumer oxygenation against laboratory reference measurements.
Prospectively track medically supervised altitude-training participants.
Compare simple and physiology-enriched tolerance prediction models.; design verbatim planning only: Population model: Adults already enrolled in supervised altitude-training camps, with clinical clearance independent of the study. Comparator: Consumer saturation-only prediction versus models adding workload, symptoms, and laboratory ventilatory response. Prospective causal strategy: Prospective observational cohort; study measurements do not determine camp participation or override medical decisions. Sample planning: Plan 150 participants across camps; event frequency and held-out-site precision constrain predictors and final recruitment.; outcomes verbatim: Primary: clinician-adjudicated exercise intolerance requiring session modification. Secondary: measurement error, work completion, and prediction calibration.; analysis verbatim inherited not authoritative if refined: Prespecify a prediction time before each future session and use only information available then, including earlier symptom and saturation histories; exclude contemporaneous signs used to adjudicate that session's intolerance. Compare models for the same future-session horizon and clinical endpoint. Account for prior ascent/exposure, fitness, illness, medication and device conditions; validate by camp, examine informative missingness and separate prediction from causal adaptation.; falsification verbatim: Reliable saturation alone matching enriched models across camps would falsify the proposed incremental physiological value.; source ids unverified leads: intermittent-hypoxia-cognition-rct-2025
wearables-living-review
cardiorespiratory-fitness-meta-2025; prospectus precedence: No

P29 — Does heat distort cuffless blood-pressure feedback?

Rank: 43
Cluster: Heat/cold/altitude/hypoxia
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: CROSS_ERROR; allocation or exposure unit: participant randomized thermal-condition sequence with paired cuffless/reference readings; independent sampling unit: participant; repeated measurement unit: paired reading, device order and repeat visit; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Separate true reference-pressure change from error change; postcondition perfusion may mediate device error and is not primary total-effect adjustment.
Primary estimand: thermal-condition assignment effect on cuffless-minus-reference error
Candidate primary outcome: cuffless blood-pressure error in mmHg
Primary finalized: No
Mathematical model: type: paired/crossover contrast precision with period, carryover and within-unit covariance; simulation for actual design; formulas: g(E[Y_ipt | A_ip,X_i]) = alpha_i + tau A_ip + pi_p + rho A_i,p-1 + gamma^T X_i + f_time(t) + b_cluster(i,p).
Target the registered within-unit assignment contrast, standardized over period/history; direct and lag effects are distinct parameters.; variables: Y: cuffless blood-pressure error in mmHg; i: participant; t: paired reading, device order and repeat visit; the registered visit/time-window index, not another independent unit; A: participant randomized thermal-condition sequence with paired cuffless/reference readings; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: thermal-condition assignment effect on cuffless-minus-reference error; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; p: randomized crossover period/session/episode, not the outcome probability; pi p: period effect; include calendar/season/sequence terms supported by the design; rho: prespecified lagged-assignment/carryover parameter; cannot in general identify arbitrary persistent carryover in a short two-period trial; f time(t): prespecified within-period recovery/day trend; g: identity for mean outcomes, log for positive/count means, or logit for binary/proportion outcomes as specified below; family code: CROSS_ERROR
Sample size design: family: paired/crossover contrast precision with period, carryover and within-unit covariance; simulation for actual design; formula: For a simple two-condition paired mean only: Var(Delta_hat)=sigma_D^2/n_independent; actual cluster/period/lag models require design-specific simulation or c^T(D^T V^(-1)D)^(-1)c.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; sigma D^2: variance of independent-unit paired contrasts including measurement error; n independent: number of independent participants/teams/donors, not repeated samples; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Separate true reference-pressure change from error change; postcondition perfusion may mediate device error and is not primary total-effect adjustment.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Concealed randomized sequence, a scientifically justified recovery interval and no anticipation; period/carryover cannot be repaired solely by a nonsignificant test.
If carryover is plausible, report lag/period and first-period or otherwise valid design sensitivity; do not drop periods based only on a post hoc carryover p-value.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Stable narrow reference agreement across thermal states would reject the predicted temperature-related artifact.; preanalysis gate: Separate true reference-pressure change from error change; postcondition perfusion may mediate device error and is not primary total-effect adjustment.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Separate true reference-pressure change from error change; postcondition perfusion may mediate device error and is not primary total-effect adjustment.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Compare planned lag/carryover structures and a valid first-period sensitivity where its sample supports inference; assess informative absent periods.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Separate true reference-pressure change from error change; postcondition perfusion may mediate device error and is not primary total-effect adjustment.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/42; question: Are apparent blood-pressure improvements during heat sessions physiological or partly caused by temperature-sensitive measurement error?; aims: Establish reference and cuffless agreement before thermal exposure.
Track agreement during supervised heat and thermoneutral conditions.
Test whether baseline recalibration transfers across thermal states.; design verbatim planning only: Population model: Screened adults with varied baseline blood pressure and skin pigmentation. Comparator: Cuffless devices versus validated clinical reference measurements during counterbalanced facility-approved thermal conditions. Prospective causal strategy: Randomized crossover measurement study with blinded signal processing and prespecified device calibration procedures. Sample planning: Plan 70 participants with 25 repeat visits; agreement precision and subgroup recruitment determine final sample.; outcomes verbatim: Primary: change in cuffless measurement error across thermal states. Secondary: reference pressure, skin temperature, perfusion, and calibration drift.; analysis verbatim inherited not authoritative if refined: Model period, posture, hydration, pigmentation, perfusion, and device order. Repeated-measures agreement and prespecified contrasts distinguish true pressure change from algorithmic error.; falsification verbatim: Stable narrow reference agreement across thermal states would reject the predicted temperature-related artifact.; source ids unverified leads: cuffless-bp-aha-statement-2026
heat-thermotherapy-meta-2025
wearables-living-review; prospectus precedence: No

P30 — Transportable early warning of exertional heat intolerance

Rank: 213
Cluster: Heat/cold/altitude/hypoxia
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: PRED; allocation or exposure unit: blinded warning model evaluated before independent clinician safety decisions; independent sampling unit: participant and external site; repeated measurement unit: pre-horizon feature window and subsequent session modification; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: prediction, calibration or model transportability; prediction gain is not clinical efficacy, mechanistic necessity or causal treatment selection; specific constraint: Exclude decision-time safety-rule variables; safety stops censor later physiology; lead time and alert threshold fixed before test evaluation.
Primary estimand: calibration/false-alert burden of multisensor versus heart-rate warning for future modification
Candidate primary outcome: clinician-directed session-modification indicator after frozen warning horizon
Primary finalized: No
Mathematical model: type: external/nested predictive-validation precision with target prevalence/variance, independent-unit holdout coverage and model complexity; formulas: Gain = mean_test[L(Y,f_base(X)) - L(Y,f_new(X))], with both predictions frozen before test outcomes; bootstrap independent held-out units.
For a binary target assess E[Y | p_hat=p] versus p and calibration intercept/slope logit Pr(Y=1)=a_cal+b_cal logit(p_hat); for a continuous target use Y=a_cal+b_cal y_hat+error plus scale-appropriate calibration/loss.; variables: Y: clinician-directed session-modification indicator after frozen warning horizon; i: participant and external site; t: pre-horizon feature window and subsequent session modification; the registered visit/time-window index, not another independent unit; A: blinded warning model evaluated before independent clinician safety decisions; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: calibration/false-alert burden of multisensor versus heart-rate warning for future modification; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; L: one proper or scientifically justified primary loss/scoring rule fixed for the actual outcome type; positive Gain favors new model; f base,f new: locked baseline/new prediction functions using only information available at the registered prediction time; test: untouched person/family/lab/site/time set as specified per thesis; p hat,y hat: frozen probability or continuous prediction; a cal,b cal: calibration intercept and slope, estimated on evaluation data for assessment; using them to update the model requires another evaluation set; family code: PRED
Sample size design: family: external/nested predictive-validation precision with target prevalence/variance, independent-unit holdout coverage and model complexity; formula: Simulate or bootstrap the independent-unit test loss/calibration/decision criterion; plan cases/events, noncases and independent sites/families/programs separately. No universal events-per-variable or participant N is asserted.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Exclude decision-time safety-rule variables; safety stops censor later physiology; lead time and alert threshold fixed before test evaluation.
Train/tune/test separation at the actual biological/participant/family/time level; no future, postoutcome or target-derived predictors.
Validation support must cover intended deployment; calibration and missing/failing measurements are reported with discrimination/loss.
A validation subset selected by culture success, assay viability, consent or data completeness is a different target unless the full strategy is explicitly evaluated.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: No held-out improvement or unacceptable false-alert burden would reject practical utility despite apparent development-set accuracy.; preanalysis gate: Exclude decision-time safety-rule variables; safety stops censor later physiology; lead time and alert threshold fixed before test evaluation.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Exclude decision-time safety-rule variables; safety stops censor later physiology; lead time and alert threshold fixed before test evaluation.
Repeat loss/calibration with independent-cluster resampling, prespecified subgroup/transport shifts and plausible missingness.
Compare fixed simple baselines, calibration and uncertainty coverage; report abstention/failure/turnaround rather than deleting hard cases.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Frozen prediction time/horizon, proper loss, baseline model, feature availability, independent split/holdout and target deployment population.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Exclude decision-time safety-rule variables; safety stops censor later physiology; lead time and alert threshold fixed before test evaluation.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/212; question: Can a transparent multisensor model identify impending session intolerance better than heart rate alone across different settings?; aims: Create synchronized physiological and symptom records during approved exercise.
Develop a parsimonious model against a heart-rate baseline.
Validate predictions at a separate supervised field site.; design verbatim planning only: Population model: Screened adults in existing occupational or athletic heat-supervision programs. Comparator: Heart-rate-only warning versus a locked model adding skin temperature, work rate, and symptoms. Prospective causal strategy: Prospective blinded prediction study; clinicians apply existing safety rules without seeing experimental warnings. Sample planning: Plan 180 participants across sites; observed modification events determine model complexity and further recruitment.; outcomes verbatim: Primary: calibrated prediction of clinician-directed session modification. Secondary: false-alert burden, subgroup sensitivity, missingness, and device usability.; analysis verbatim inherited not authoritative if refined: Prespecify a clinically useful warning horizon and freeze all predictor streams before that horizon; exclude symptoms or safety-rule values first recorded at the modification decision. Account for fitness, acclimation, medication, clothing and environment using information available at prediction. Validate by site/participant, address safety-stop censoring and report lead time and false-alert burden without retrospective optimization of numerous thresholds.; falsification verbatim: No held-out improvement or unacceptable false-alert burden would reject practical utility despite apparent development-set accuracy.; source ids unverified leads: wearables-remote-monitoring-2026
sweat-biosensors-review-2025
hyperhydration-heat-rct-2025; prospectus precedence: No

P31 — Do readiness scores improve next-session performance?

Rank: 53
Cluster: Recovery/fatigue/athlete energy availability
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: MRT; allocation or exposure unit: coach-information decision moment randomized readiness score shown/hidden; independent sampling unit: athlete with coach clustering; decision moment randomized within athlete; repeated measurement unit: next-session standardized sprint assessment; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Prior disclosure affects training/availability; freeze eligible decision times, account for coach spillover and decision carryover.
Primary estimand: proximal score-disclosure excursion effect on next-session repeated-sprint performance
Candidate primary outcome: prespecified standardized next-session repeated-sprint endpoint
Primary finalized: No
Mathematical model: type: micro-randomized-trial power/precision simulation for availability, randomization probabilities, proximal correlation and lagged effects; formulas: E[Y_i,t+Delta(1)-Y_i,t+Delta(0) | I_it=1,S_it] = f(S_it)^T beta, a proximal excursion effect under the registered future-treatment policy.
Estimate beta using centered A_it-p_it with known p_it=P(A_it=1 | H_it,I_it=1); a basic unmoderated moment is sum I_it(A_it-p_it)Y_i,t+Delta / sum I_it p_it(1-p_it), with prespecified weighting/augmentation as needed.; variables: Y: prespecified standardized next-session repeated-sprint endpoint; i: athlete with coach clustering; decision moment randomized within athlete; t: next-session standardized sprint assessment; the registered visit/time-window index, not another independent unit; A: coach-information decision moment randomized readiness score shown/hidden; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: proximal score-disclosure excursion effect on next-session repeated-sprint performance; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; I it: availability/eligibility indicator decided before current randomization; H it: observed predecision history including prior prompts/breaks; p it: known, positive randomized prompt/show probabilities among available moments; S it: prespecified predecision moderator vector; Delta: fixed proximal outcome delay/window, not a mean difference in this formula; f(S)^T beta: registered moderation model; beta is estimated with participant/coach dependence; family code: MRT
Sample size design: family: micro-randomized-trial power/precision simulation for availability, randomization probabilities, proximal correlation and lagged effects; formula: Simulate the actual availability/history/randomization process and estimate Var(beta_hat) with participant/coach dependence; moment count alone is not the independent sample size.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Prior disclosure affects training/availability; freeze eligible decision times, account for coach spillover and decision carryover.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Sequential randomization at eligible moments, positivity of prompt/no prompt, consistency and prespecified availability; proximal effect does not automatically identify a long-term strategy effect.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: Consistent performance improvement with fewer inappropriate cancellations would falsify the hypothesis of negligible added decision value.; preanalysis gate: Prior disclosure affects training/availability; freeze eligible decision times, account for coach spillover and decision carryover.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Prior disclosure affects training/availability; freeze eligible decision times, account for coach spillover and decision carryover.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Vary availability models, lagged effects, proximal windows and informative postprompt missingness without searching a favorable window.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Prior disclosure affects training/availability; freeze eligible decision times, account for coach spillover and decision carryover.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/52; question: Does showing a coach a consumer readiness score improve near-term recovery decisions beyond symptoms and recent training load?; aims: Validate score repeatability against standardized performance testing.
Randomize score visibility at eligible coaching decisions.
Estimate performance effects and unnecessary training cancellations.; design verbatim planning only: Population model: Recreational team-sport athletes under an existing supervised training schedule. Comparator: Coaches receive symptoms and recent load, with or without the consumer readiness score. Prospective causal strategy: Micro-randomized information trial over a season; standardized next-session performance is assessed independently, and clinical safety rules remain unchanged. Sample planning: Plan 100 athletes; pilot decision frequency and within-athlete dependence determine final precision.; outcomes verbatim: Primary: standardized next-session repeated-sprint performance. Secondary: modified sessions, perceived recovery, missed training, and coach confidence.; analysis verbatim inherited not authoritative if refined: Model athlete/coach clustering, prior workload, injury, and time trends. Prespecify proximal effects, account for decision carryover, and avoid multiple discoveries from proprietary components.; falsification verbatim: Consistent performance improvement with fewer inappropriate cancellations would falsify the hypothesis of negligible added decision value.; source ids unverified leads: wearables-living-review
digital-n-of-1-physiology-2026
wearable-stress-measurement-review-2026; prospectus precedence: No

P32 — Scheduled versus symptom-triggered deloading

Rank: 105
Cluster: Recovery/fatigue/athlete energy availability
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT; allocation or exposure unit: participant randomized symptom-triggered versus fixed-calendar deload strategy; independent sampling unit: participant; repeated measurement unit: recovered-strength and fatigue assessment; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Strategies differ through completed work and timing; do not adjust those pathways away; fatigue dropout sensitivity essential.
Primary estimand: deload-strategy assignment effect on recovered strength after the block
Candidate primary outcome: strength performance after common standardized recovery
Primary finalized: No
Mathematical model: type: contrast precision/power by design-matrix or simulation planning; formulas: E(Y_it | A_i, X_i, Y_i0) = alpha_t + z(A_i)^T theta_t + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Delta = sum_t w_t c^T theta_t; estimate by baseline-adjusted marginal standardization, with unit/cluster-appropriate inference.; variables: Y: strength performance after common standardized recovery; i: participant; t: recovered-strength and fatigue assessment; the registered visit/time-window index, not another independent unit; A: participant randomized symptom-triggered versus fixed-calendar deload strategy; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: deload-strategy assignment effect on recovered strength after the block; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; z(A): prespecified indicators for the actual randomized arms; two arms reduce to one indicator; family code: RCT
Sample size design: family: contrast precision/power by design-matrix or simulation planning; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Strategies differ through completed work and timing; do not adjust those pathways away; fatigue dropout sensitivity essential.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: More persistent fatigue or worse recovered performance under trigger-based timing would reject the proposed efficiency advantage.; preanalysis gate: Strategies differ through completed work and timing; do not adjust those pathways away; fatigue dropout sensitivity essential.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Strategies differ through completed work and timing; do not adjust those pathways away; fatigue dropout sensitivity essential.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Strategies differ through completed work and timing; do not adjust those pathways away; fatigue dropout sensitivity essential.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/104; question: Does a transparent fatigue-triggered deload preserve performance better than a fixed calendar during recreational strength training?; aims: Establish reliable fatigue and standardized performance measurements.
Randomize calendar-based versus transparent trigger-based deload timing.
Compare recovered performance and total training opportunity.; design verbatim planning only: Population model: Experienced recreational lifters following supervised, noncompetitive training blocks. Comparator: Fixed-calendar deloads versus clinician-approved triggers using symptoms and submaximal performance. Prospective causal strategy: Parallel randomized trial with the same permitted deload options, coaching contact, and blinded outcome testing after standardized recovery. Sample planning: Plan 110 participants; pilot trigger frequency, retention, and performance variance guide a formal sample calculation.; outcomes verbatim: Primary: recovered strength performance after the training block. Secondary: persistent fatigue, completed work, injury symptoms, and satisfaction.; analysis verbatim inherited not authoritative if refined: Adjust for strength, habitual volume, nutrition, and occupational work. Register one primary contrast; distinguish strategy from completed work and examine fatigue-related dropout sensitivity.; falsification verbatim: More persistent fatigue or worse recovered performance under trigger-based timing would reject the proposed efficiency advantage.; source ids unverified leads: resistance-training-volume-meta-2025
n-of-1-methods-review-2025
hrv-exercise-prescription-2025; prospectus precedence: No

P33 — Whole-food refueling versus commercial recovery products

Rank: 10
Cluster: Recovery/fatigue/athlete energy availability
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: CROSS_EQ; allocation or exposure unit: participant randomized whole-food/commercial refueling sequence; independent sampling unit: participant; repeated measurement unit: next-session performance in each recovery period; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Match relevant nutrient/fluid provision and verify intake; washout and preceding work standardized; pragmatic acceptability phase is a separate target.
Primary estimand: whole-food-minus-commercial performance difference against justified equivalence bounds
Candidate primary outcome: standardized next-session exercise-performance endpoint
Primary finalized: No
Mathematical model: type: paired equivalence/TOST planning with justified asymmetric/symmetric bounds and carryover sensitivity; formulas: g(E[Y_ipt | A_ip,X_i]) = alpha_i + tau A_ip + pi_p + rho A_i,p-1 + gamma^T X_i + f_time(t) + b_cluster(i,p).
Target the registered within-unit assignment contrast, standardized over period/history; direct and lag effects are distinct parameters.
Equivalence requires CI_(1-2alpha)(Delta) entirely inside [-M_lower, M_upper], using two one-sided tests at alpha; a nonsignificant difference is insufficient.; variables: Y: standardized next-session exercise-performance endpoint; i: participant; t: next-session performance in each recovery period; the registered visit/time-window index, not another independent unit; A: participant randomized whole-food/commercial refueling sequence; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: whole-food-minus-commercial performance difference against justified equivalence bounds; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; p: randomized crossover period/session/episode, not the outcome probability; pi p: period effect; include calendar/season/sequence terms supported by the design; rho: prespecified lagged-assignment/carryover parameter; cannot in general identify arbitrary persistent carryover in a short two-period trial; f time(t): prespecified within-period recovery/day trend; g: identity for mean outcomes, log for positive/count means, or logit for binary/proportion outcomes as specified below; M lower,M upper: independently justified practical/physiological equivalence bounds in the actual endpoint unit, not supplied here; alpha: one-sided test error rate fixed in registration; family code: CROSS_EQ
Sample size design: family: paired equivalence/TOST planning with justified asymmetric/symmetric bounds and carryover sensitivity; formula: For a simple two-condition paired mean only: Var(Delta_hat)=sigma_D^2/n_independent; actual cluster/period/lag models require design-specific simulation or c^T(D^T V^(-1)D)^(-1)c.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; sigma D^2: variance of independent-unit paired contrasts including measurement error; n independent: number of independent participants/teams/donors, not repeated samples; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Match relevant nutrient/fluid provision and verify intake; washout and preceding work standardized; pragmatic acceptability phase is a separate target.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Concealed randomized sequence, a scientifically justified recovery interval and no anticipation; period/carryover cannot be repaired solely by a nonsignificant test.
If carryover is plausible, report lag/period and first-period or otherwise valid design sensitivity; do not drop periods based only on a post hoc carryover p-value.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: A reproducible performance deficit despite nutrient matching would challenge equivalence and prompt examination of food-form mechanisms.; preanalysis gate: Match relevant nutrient/fluid provision and verify intake; washout and preceding work standardized; pragmatic acceptability phase is a separate target.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Match relevant nutrient/fluid provision and verify intake; washout and preceding work standardized; pragmatic acceptability phase is a separate target.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Compare planned lag/carryover structures and a valid first-period sensitivity where its sample supports inference; assess informative absent periods.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Externally justified equivalence bounds; credible comparator and uncertainty in paired differences.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Match relevant nutrient/fluid provision and verify intake; washout and preceding work standardized; pragmatic acceptability phase is a separate target.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/9; question: Can affordable whole-food refueling match a composition-matched commercial product for recovery between demanding exercise sessions?; aims: Co-design acceptable nutrient-matched refueling options with athletes.
Compare recovery during counterbalanced standardized exercise visits.
Evaluate feasibility and cost during ordinary team training.; design verbatim planning only: Population model: Recreational endurance and team-sport athletes without conditions requiring individualized clinical nutrition. Comparator: Dietitian-designed whole foods versus commercial refueling, matched for relevant nutrients and fluid. Prospective causal strategy: Randomized crossover laboratory comparison with standardized previsit intake and recovery, followed by a pragmatic acceptability phase. Sample planning: Plan 60 crossover participants; within-person performance variability informs the equivalence margin and final sample requirement.; outcomes verbatim: Primary: next-session standardized exercise performance. Secondary: gastrointestinal symptoms, perceived recovery, optional glycogen measurement, cost, and food access.; analysis verbatim inherited not authoritative if refined: Model period/carryover, diet, product use, expectancy, and prior work. Prespecify equivalence bounds, standardize preceding nutrition, and separate optional mechanistic analyses from functional recovery.; falsification verbatim: A reproducible performance deficit despite nutrient matching would challenge equivalence and prompt examination of food-form mechanisms.; source ids unverified leads: exercise-electrolyte-rehydration-review-2025
precision-nutrition-rct-2024
female-athlete-nutrition-review-2025; prospectus precedence: No

P34 — Fueling support and bone-stress vulnerability in women runners

Rank: 28
Cluster: Recovery/fatigue/athlete energy availability
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT_COUNT; allocation or exposure unit: runner randomized dietitian/food-planning support versus standard education; independent sampling unit: participant with team/club clustering if shared; repeated measurement unit: training session and clinical referral; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: No deficit induced; injury incidence and turnover markers remain separate exploratory endpoints; observed training exposure can change under support.
Primary estimand: fueling-support offer effect on symptom-free completed training sessions
Candidate primary outcome: count of symptom-free completed sessions over the fixed season
Primary finalized: No
Mathematical model: type: overdispersed count contrast planning by simulation, with follow-up/cluster/zero inflation assumptions; formulas: Y_it ~ NegativeBinomial(mu_it,phi); log(mu_it)=alpha_t+z(A_i)^T theta_t+beta Y_i0+gamma^T X_i+b_cluster(i)+u_i.
Delta = standardized E[Y | do(A=1)] - E[Y | do(A=0)] under randomization; report ratio exp(c^T theta) only when the log-mean model supports it.; variables: Y: count of symptom-free completed sessions over the fixed season; i: participant with team/club clustering if shared; t: training session and clinical referral; the registered visit/time-window index, not another independent unit; A: runner randomized dietitian/food-planning support versus standard education; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: fueling-support offer effect on symptom-free completed training sessions; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; mu it: conditional mean of the actual per-thesis count; phi: overdispersion parameter, with Var(Y)=mu+mu^2/phi under this parameterization; z(A): prespecified indicators for the actual arms or randomized policy; family code: RCT_COUNT
Sample size design: family: overdispersed count contrast planning by simulation, with follow-up/cluster/zero inflation assumptions; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: No deficit induced; injury incidence and turnover markers remain separate exploratory endpoints; observed training exposure can change under support.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: No improvement despite verified uptake and increased dietary adequacy would challenge the proposed shortfall mechanism.; preanalysis gate: No deficit induced; injury incidence and turnover markers remain separate exploratory endpoints; observed training exposure can change under support.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: No deficit induced; injury incidence and turnover markers remain separate exploratory endpoints; observed training exposure can change under support.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
No deficit induced; injury incidence and turnover markers remain separate exploratory endpoints; observed training exposure can change under support.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/27; question: Does dietitian-supported fueling improve training tolerance and bone-stress vulnerability markers in women distance runners?; aims: Validate repeated intake and training-expenditure measurement feasibility.
Randomize proactive fueling support versus standard athlete education.
Track training continuity and clinically adjudicated bone-stress symptoms.; design verbatim planning only: Population model: Adult women recreational or collegiate distance runners, with clinical referral for suspected energy deficiency. Comparator: Proactive dietitian consultations and practical food planning versus standard education; both groups retain full clinical care. Prospective causal strategy: Parallel randomized support trial across a training season; no deficit is induced and outcome adjudicators remain blinded. Sample planning: Plan 140 runners; pilot retention and marker variance determine final sample; injury incidence remains exploratory.; outcomes verbatim: Primary: symptom-free completed training. Secondary: dietary adequacy, fatigue, bone-turnover markers, and clinician-confirmed bone-stress injuries.; analysis verbatim inherited not authoritative if refined: Adjust for injury history, training, body size, food access, and clinical history. Correct secondary biomarkers, model exposure, and quantify intake/expenditure error.; falsification verbatim: No improvement despite verified uptake and increased dietary adequacy would challenge the proposed shortfall mechanism.; source ids unverified leads: female-athlete-nutrition-review-2025
female-menstrual-health-activity-meta-2025
endocrine-vitamin-d-guideline-2024; prospectus precedence: No

P35 — Detecting underfueling in male endurance athletes

Rank: 214
Cluster: Recovery/fatigue/athlete energy availability
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: DIAG; allocation or exposure unit: prospective blinded index assessment against independent clinical pathway; independent sampling unit: participant with club-held-out validation; repeated measurement unit: multimodal measurement and adjudicated case; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: diagnostic/benchmark agreement or classification validity; reader/interface/navigation effects are causal only in their separately randomized component; specific constraint: Reference is imperfect; avoid incorporating index score into adjudication; no single biomarker gold standard or induced restriction.
Primary estimand: sensitivity, calibration and referral burden for actionable underfueling versus fixed single-marker alternatives
Candidate primary outcome: independently adjudicated actionable-underfueling status
Primary finalized: No
Mathematical model: type: diagnostic/benchmark sensitivity, false-alert and calibration precision with independent cases/noncases and cluster/reference uncertainty; formulas: Sensitivity = TP/(TP+FN); false-alert burden = FP/(registered nonevent time/opportunities); report participant/family-clustered uncertainty and failed/indeterminate cases.
For probability scores, report calibration and a fixed proper score in participant/source-disjoint evaluation, with uncertain reference labels modeled or bounded.; variables: Y: independently adjudicated actionable-underfueling status; i: participant with club-held-out validation; t: multimodal measurement and adjudicated case; the registered visit/time-window index, not another independent unit; A: prospective blinded index assessment against independent clinical pathway; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: sensitivity, calibration and referral burden for actionable underfueling versus fixed single-marker alternatives; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; TP,FN,FP: true-positive, false-negative and false-positive counts relative to the independently defined reference and matching window, not an assumed perfect gold standard; reference: per-thesis independent reference, including adjudication uncertainty; nonevent time/opportunities: fixed at-risk denominator; event specificity and false alerts per hour are distinct targets; family code: DIAG
Sample size design: family: diagnostic/benchmark sensitivity, false-alert and calibration precision with independent cases/noncases and cluster/reference uncertainty; formula: An independent-case rough starting point is n_case ~= z_(1-alpha*/2)^2 Se(1-Se)/h_Se^2; replace with interval-based/clustered simulation for the actual design. Plan noncases/observation time and reference failures separately.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; Se: plausible sensitivity from prior evidence or pilot, not an asserted study result; h Se: desired sensitivity interval half-width; n case: independent positive cases; repeated events within a person require correlation adjustment; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Reference is imperfect; avoid incorporating index score into adjudication; no single biomarker gold standard or induced restriction.
Reference independence, fixed thresholds/window and an explicit denominator; incorporation/verification bias can invalidate apparent accuracy.
Participant/family-level test splits and paired comparisons; report abstention, assay failure and indeterminate truth rather than force every case into a binary label.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: No added accuracy beyond a simple symptom/intake history would reject the proposed value of multimodal measurement.; preanalysis gate: Reference is imperfect; avoid incorporating index score into adjudication; no single biomarker gold standard or induced restriction.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Reference is imperfect; avoid incorporating index score into adjudication; no single biomarker gold standard or induced restriction.
Vary plausible reference misclassification, verification-selection and event-matching window under a prespecified sensitivity family.
Report subgroup usable-data coverage and calibration; avoid interpreting an imprecise null difference as equivalence.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Reference is imperfect; avoid incorporating index score into adjudication; no single biomarker gold standard or induced restriction.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/213; question: Can a repeated multimodal assessment identify clinically actionable underfueling better than a single biomarker or wearable score?; aims: Estimate reliability of candidate measurements during ordinary training.
Compare blinded multimodal assessments with independent clinical adjudication.
Validate a minimal screening set in a separate club cohort.; design verbatim planning only: Population model: Adult male endurance athletes across performance levels, without inducing dietary restriction. Comparator: Single-marker and consumer-score approaches versus repeated multimodal research assessments. Prospective causal strategy: Prospective diagnostic-accuracy cohort; independent clinicians use a prespecified assessment pathway and receive safety-critical findings immediately. Sample planning: Plan 180 athletes; actionable-case prevalence and external-validation precision determine recruitment and predictor limits.; outcomes verbatim: Primary: accuracy for independently adjudicated actionable underfueling. Secondary: fatigue, training interruption, referral burden, and measurement acceptability.; analysis verbatim inherited not authoritative if refined: Account for workload, illness, body size, medication, and food access. Separate index tests from adjudication, validate by club, and quantify imperfect-reference uncertainty.; falsification verbatim: No added accuracy beyond a simple symptom/intake history would reject the proposed value of multimodal measurement.; source ids unverified leads: female-athlete-nutrition-review-2025
wearable-stress-measurement-review-2026
metabolic-flexibility-prehabilitation-2025; prospectus precedence: No

P36 — Training the practical tolerance of exercise fueling

Rank: 74
Cluster: Recovery/fatigue/athlete energy availability
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: RCT; allocation or exposure unit: participant randomized progressive food-fueling practice versus equal-contact education; independent sampling unit: participant; repeated measurement unit: uncoached standardized challenge and symptom interval; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Tolerated intake is postassignment and may mediate symptom/performance; challenge must have common externally fixed scoring and safety-stop rules.
Primary estimand: practice-offer effect on gastrointestinal symptom burden during the uncoached challenge
Candidate primary outcome: prespecified gastrointestinal symptom-burden summary during common challenge
Primary finalized: No
Mathematical model: type: contrast precision/power by design-matrix or simulation planning; formulas: E(Y_it | A_i, X_i, Y_i0) = alpha_t + z(A_i)^T theta_t + beta_t Y_i0 + gamma_t^T X_i + b_cluster(i) + u_i.
Delta = sum_t w_t c^T theta_t; estimate by baseline-adjusted marginal standardization, with unit/cluster-appropriate inference.; variables: Y: prespecified gastrointestinal symptom-burden summary during common challenge; i: participant; t: uncoached standardized challenge and symptom interval; the registered visit/time-window index, not another independent unit; A: participant randomized progressive food-fueling practice versus equal-contact education; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: practice-offer effect on gastrointestinal symptom burden during the uncoached challenge; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; z(A): prespecified indicators for the actual randomized arms; two arms reduce to one indicator; family code: RCT
Sample size design: family: contrast precision/power by design-matrix or simulation planning; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Tolerated intake is postassignment and may mediate symptom/performance; challenge must have common externally fixed scoring and safety-stop rules.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: No transferred symptom or intake improvement despite practice adherence would refute the proposed adaptation mechanism.; preanalysis gate: Tolerated intake is postassignment and may mediate symptom/performance; challenge must have common externally fixed scoring and safety-stop rules.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Tolerated intake is postassignment and may mediate symptom/performance; challenge must have common externally fixed scoring and safety-stop rules.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Tolerated intake is postassignment and may mediate symptom/performance; challenge must have common externally fixed scoring and safety-stop rules.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/73; question: Does supervised progressive practice with ordinary exercise foods reduce gastrointestinal-limited fatigue compared with one-time nutrition education?; aims: Characterize reproducible gastrointestinal symptoms during standardized sessions.
Randomize supported fueling practice versus matched-contact education.
Test whether improved tolerance transfers to uncoached exercise.; design verbatim planning only: Population model: Recreational endurance athletes reporting recurrent, clinically assessed exercise-related gastrointestinal discomfort. Comparator: Dietitian-supervised practice using ordinary foods and beverages versus equal-contact nutrition education without structured practice. Prospective causal strategy: Parallel randomized intervention with standardized exercise challenges and blinded performance assessment; all participants retain symptom evaluation. Sample planning: Plan 100 athletes; pilot symptom variance, adherence, and performance determine final sample justification.; outcomes verbatim: Primary: gastrointestinal symptom burden during the uncoached challenge. Secondary: tolerated intake, completed work, fatigue, and clinical referrals.; analysis verbatim inherited not authoritative if refined: Adjust for diet, gastrointestinal history, heat, anxiety, and intensity. Prespecify one primary challenge, correct secondary endpoints, and acknowledge nonrandom mediation by intake.; falsification verbatim: No transferred symptom or intake improvement despite practice adherence would refute the proposed adaptation mechanism.; source ids unverified leads: exercise-electrolyte-rehydration-review-2025
female-athlete-nutrition-review-2025
precision-nutrition-rct-2024; prospectus precedence: No

P37 — Team food access as a causal recovery intervention

Rank: 34
Cluster: Recovery/fatigue/athlete energy availability
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: STEPPED; allocation or exposure unit: team rollout timing randomized for site food access; independent sampling unit: team; repeated measurement unit: athlete-session and calendar rollout period; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Randomized sequence still needs calendar/season effects and adequate teams per step; spillover and changing team composition limit identification.
Primary estimand: food-access rollout assignment effect on fatigue-related missed/modified sessions
Candidate primary outcome: count of fatigue-related missed or modified sessions per fixed calendar period
Primary finalized: No
Mathematical model: type: stepped-wedge cluster count simulation by randomized sequence, period, cluster size and delayed intervention effect; formulas: Y_ict ~ NegativeBinomial(mu_ict,phi); log(mu_ict)=alpha_c+lambda_t+tau A_ct+gamma^T X_i+u_i, with team rollout sequence randomized.
A_ct is the assigned availability at calendar period t; prespecify delayed/ramp-up effect and count denominator. Use sequence-aware small-cluster inference and sensitivity to heterogeneous period effects.; variables: Y: count of fatigue-related missed or modified sessions per fixed calendar period; i: team; t: calendar rollout period; A: team rollout timing randomized for site food access; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: randomized team; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; A ct: assigned intervention availability according to randomized rollout timing, regardless of individual uptake; lambda t: calendar/season period effect; alpha c: team baseline effect; phi: negative-binomial overdispersion; mu ict: expected fatigue-related missed/modified-session count; family code: STEPPED
Sample size design: family: stepped-wedge cluster count simulation by randomized sequence, period, cluster size and delayed intervention effect; formula: Use design-specific simulation or independent-unit resampling for the stated estimand and target confidence width/power; preserve all allocation, censoring, clustering and measurement-error structure.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Randomized sequence still needs calendar/season effects and adequate teams per step; spillover and changing team composition limit identification.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Enough independent teams and overlap across rollout periods; calendar effects must be separated from treatment and spillover/composition tracked.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: No improvement among athletes whose measured food access increases would challenge the proposed structural-fueling mechanism.; preanalysis gate: Randomized sequence still needs calendar/season effects and adequate teams per step; spillover and changing team composition limit identification.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Randomized sequence still needs calendar/season effects and adequate teams per step; spillover and changing team composition limit identification.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Randomized sequence still needs calendar/season effects and adequate teams per step; spillover and changing team composition limit identification.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/33; question: Does improving the training-site food environment reduce fatigue and missed sessions beyond individual nutrition education?; aims: Map barriers to obtaining meals around scheduled training.
Randomize team-level access to practical posttraining food provision.
Estimate effects on fatigue, participation, and access inequalities.; design verbatim planning only: Population model: Adult university and community athletes across mixed sports and socioeconomic backgrounds. Comparator: Accessible food provision plus standard education versus the same education and existing services. Prospective causal strategy: Cluster-randomized staggered rollout across teams; rollout timing is randomized and all teams eventually receive the service. Sample planning: Plan 16 teams and 240 athletes; pilot intracluster correlation and uptake determine precision and rollout periods.; outcomes verbatim: Primary: fatigue-related missed or modified sessions. Secondary: food security, dietary adequacy, performance, attendance, and implementation cost.; analysis verbatim inherited not authoritative if refined: Model teams, season, academic/work demands, exposure, and food access. Prespecify access interaction, adjust rollout time, and quantify spillover and secondary endpoint multiplicity.; falsification verbatim: No improvement among athletes whose measured food access increases would challenge the proposed structural-fueling mechanism.; source ids unverified leads: female-athlete-nutrition-review-2025
precision-nutrition-rct-2024
behavior-stress-sleep-rct-2026; prospectus precedence: No

P38 — Compression recovery beyond device expectation

Rank: 205
Cluster: Recovery/fatigue/athlete energy availability
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: CROSS; allocation or exposure unit: participant randomized approved/active-feeling comparator compression sequence; independent sampling unit: participant; repeated measurement unit: standardized next-session performance visit; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Blinding beliefs are postassignment; active comparator may have physiological activity; expectation mediation is assumption-dependent.
Primary estimand: compression assignment effect on next-session objective performance
Candidate primary outcome: standardized next-session performance endpoint
Primary finalized: No
Mathematical model: type: paired/crossover contrast precision with period, carryover and within-unit covariance; simulation for actual design; formulas: g(E[Y_ipt | A_ip,X_i]) = alpha_i + tau A_ip + pi_p + rho A_i,p-1 + gamma^T X_i + f_time(t) + b_cluster(i,p).
Target the registered within-unit assignment contrast, standardized over period/history; direct and lag effects are distinct parameters.; variables: Y: standardized next-session performance endpoint; i: participant; t: standardized next-session performance visit; the registered visit/time-window index, not another independent unit; A: participant randomized approved/active-feeling comparator compression sequence; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: compression assignment effect on next-session objective performance; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; p: randomized crossover period/session/episode, not the outcome probability; pi p: period effect; include calendar/season/sequence terms supported by the design; rho: prespecified lagged-assignment/carryover parameter; cannot in general identify arbitrary persistent carryover in a short two-period trial; f time(t): prespecified within-period recovery/day trend; g: identity for mean outcomes, log for positive/count means, or logit for binary/proportion outcomes as specified below; family code: CROSS
Sample size design: family: paired/crossover contrast precision with period, carryover and within-unit covariance; simulation for actual design; formula: For a simple two-condition paired mean only: Var(Delta_hat)=sigma_D^2/n_independent; actual cluster/period/lag models require design-specific simulation or c^T(D^T V^(-1)D)^(-1)c.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; sigma D^2: variance of independent-unit paired contrasts including measurement error; n independent: number of independent participants/teams/donors, not repeated samples; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Blinding beliefs are postassignment; active comparator may have physiological activity; expectation mediation is assumption-dependent.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Concealed randomized sequence, a scientifically justified recovery interval and no anticipation; period/carryover cannot be repaired solely by a nonsignificant test.
If carryover is plausible, report lag/period and first-period or otherwise valid design sensitivity; do not drop periods based only on a post hoc carryover p-value.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: A reproducible performance advantage accompanied by physiological change despite comparable expectations would reject the primarily perceptual explanation.; preanalysis gate: Blinding beliefs are postassignment; active comparator may have physiological activity; expectation mediation is assumption-dependent.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Blinding beliefs are postassignment; active comparator may have physiological activity; expectation mediation is assumption-dependent.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Compare planned lag/carryover structures and a valid first-period sensitivity where its sample supports inference; assess informative absent periods.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Blinding beliefs are postassignment; active comparator may have physiological activity; expectation mediation is assumption-dependent.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/204; question: Does lower-limb compression improve objective between-session recovery after expectations and sensory context are measured?; aims: Validate a credible low-compression comparator and expectation measures.
Compare recovery under randomized counterbalanced device conditions.
Explore whether limb-volume change accompanies subsequent performance, while distinguishing this proxy from direct venous-function measurement.; design verbatim planning only: Population model: Screened recreational athletes after a standardized supervised exercise task. Comparator: Clinically approved compression versus an active-feeling low-compression comparator matched for device contact and duration. Prospective causal strategy: Assessor-blinded crossover with standardized recovery periods, predeclared washout, and postvisit blinding assessment. Sample planning: Plan 60 completers; pilot within-person performance variability and comparator credibility determine final precision.; outcomes verbatim: Primary: next-session standardized performance. Secondary: perceived recovery, limb-volume change, discomfort, expectancy, and adverse symptoms.; analysis verbatim inherited not authoritative if refined: Model period/carryover, work, swelling, and device use. Register one primary contrast; expectation mediation remains exploratory because beliefs are not independently randomized.; falsification verbatim: A reproducible performance advantage accompanied by physiological change despite comparable expectations would reject the primarily perceptual explanation.; source ids unverified leads: belief-effects-health-review-2026
cold-recovery-rct-2025
n-of-1-methods-review-2025; prospectus precedence: No

P39 — Active recovery during dense competition schedules

Rank: 157
Cluster: Recovery/fatigue/athlete energy availability
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: CLUSTER_CROSS; allocation or exposure unit: team randomized active-recovery/rest weekend sequence; independent sampling unit: team, with athletes nested; repeated measurement unit: repeated sprint day within separated simulation weekend; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Recovery verified before next weekend; game exposure is postassignment if strategy changes participation; shared team treatment requires team-level uncertainty.
Primary estimand: team-strategy assignment effect on the registered repeated-sprint trajectory summary
Candidate primary outcome: repeated-sprint performance across days under fixed time weights
Primary finalized: No
Mathematical model: type: paired/crossover contrast precision with period, carryover and within-unit covariance; simulation for actual design; formulas: g(E[Y_ipt | A_ip,X_i]) = alpha_i + tau A_ip + pi_p + rho A_i,p-1 + gamma^T X_i + f_time(t) + b_cluster(i,p).
Target the registered within-unit assignment contrast, standardized over period/history; direct and lag effects are distinct parameters.; variables: Y: repeated-sprint performance across days under fixed time weights; i: team, with athletes nested; t: repeated sprint day within separated simulation weekend; the registered visit/time-window index, not another independent unit; A: team randomized active-recovery/rest weekend sequence; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: team-strategy assignment effect on the registered repeated-sprint trajectory summary; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; p: randomized crossover period/session/episode, not the outcome probability; pi p: period effect; include calendar/season/sequence terms supported by the design; rho: prespecified lagged-assignment/carryover parameter; cannot in general identify arbitrary persistent carryover in a short two-period trial; f time(t): prespecified within-period recovery/day trend; g: identity for mean outcomes, log for positive/count means, or logit for binary/proportion outcomes as specified below; family code: CLUSTER_CROSS
Sample size design: family: paired/crossover contrast precision with period, carryover and within-unit covariance; simulation for actual design; formula: For a simple two-condition paired mean only: Var(Delta_hat)=sigma_D^2/n_independent; actual cluster/period/lag models require design-specific simulation or c^T(D^T V^(-1)D)^(-1)c.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; sigma D^2: variance of independent-unit paired contrasts including measurement error; n independent: number of independent participants/teams/donors, not repeated samples; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Recovery verified before next weekend; game exposure is postassignment if strategy changes participation; shared team treatment requires team-level uncertainty.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Concealed randomized sequence, a scientifically justified recovery interval and no anticipation; period/carryover cannot be repaired solely by a nonsignificant test.
If carryover is plausible, report lag/period and first-period or otherwise valid design sensitivity; do not drop periods based only on a post hoc carryover p-value.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: A sustained performance advantage across days with equal game exposure would reject the predicted absence of objective benefit.; preanalysis gate: Recovery verified before next weekend; game exposure is postassignment if strategy changes participation; shared team treatment requires team-level uncertainty.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Recovery verified before next weekend; game exposure is postassignment if strategy changes participation; shared team treatment requires team-level uncertainty.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Compare planned lag/carryover structures and a valid first-period sensitivity where its sample supports inference; assess informative absent periods.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Recovery verified before next weekend; game exposure is postassignment if strategy changes participation; shared team treatment requires team-level uncertainty.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/156; question: Does low-load active recovery preserve repeated-match performance better than quiet rest when time and staff attention are matched?; aims: Establish reliable performance and symptom measures under ordinary training.
Randomize recovery strategies across matched tournament simulations.
Test cumulative rather than single-day recovery effects.; design verbatim planning only: Population model: Adult recreational team-sport athletes participating in supervised, noninjurious tournament simulations. Comparator: Clinician-approved low-load active recovery versus time-matched quiet rest with equivalent staff contact. Prospective causal strategy: Team-cluster crossover across separated simulation weekends, with randomized order and verified return to baseline before the next period. Sample planning: Plan 12 teams and 140 athletes; pilot team correlation, carryover, and weekend availability determine final scope.; outcomes verbatim: Primary: repeated-sprint performance trajectory across days. Secondary: jump performance, soreness, perceived fatigue, participation, and adverse symptoms.; analysis verbatim inherited not authoritative if refined: Model athlete/team clustering, period/order, game exposure, diet, and prior load. Prespecify cumulative interaction, verify washout, and distinguish perceived from objective recovery.; falsification verbatim: A sustained performance advantage across days with equal game exposure would reject the predicted absence of objective benefit.; source ids unverified leads: cold-recovery-rct-2025
digital-n-of-1-physiology-2026
hrv-exercise-prescription-2025; prospectus precedence: No

P40 — Athlete fatigue summaries that improve clinical assessment

Rank: 130
Cluster: Recovery/fatigue/athlete energy availability
Status: DRAFT — preregistration scaffold; parameters, assumptions, source/access verification unresolved
Design: model family: CLUSTER_BINARY; allocation or exposure unit: clinician randomized full longitudinal fatigue summary versus usual assessment; independent sampling unit: clinician; repeated measurement unit: patient consultation and independent adjudication; experimental unit rule: The independently allocated participant/day/cluster/culture/trajectory is the experimental unit for its assignment. The biological/source unit and independent transport units may differ; the two fields above are intentional. Technical repeats, cells, variants, time points and repeated tasks do not add independent people/donors/lots/trials.; interpretation: randomized assignment contrast, provided the intended randomization is implemented and the outcome/intercurrent-event strategy is finalized; specific constraint: Adjudication must avoid incorporation bias; summary component associations cannot isolate wearable contribution without separate component randomization.
Primary estimand: summary-offer assignment effect on independently judged appropriate management
Candidate primary outcome: appropriate-management decision indicator under a fixed rubric
Primary finalized: No
Mathematical model: type: binary risk-difference precision/power, simulated with actual cluster/trajectory structure; formulas: logit Pr(Y_i=1 | A_i,X_i) = alpha + z(A_i)^T theta + gamma^T X_i + b_cluster(i).
RD = E_X[p_hat(1,X) - p_hat(0,X)]; report absolute risk difference as well as any odds ratio.; variables: Y: appropriate-management decision indicator under a fixed rubric; i: clinician; t: patient consultation and independent adjudication; the registered visit/time-window index, not another independent unit; A: clinician randomized full longitudinal fatigue summary versus usual assessment; code 1 for the first named active strategy and 0 for its specified comparator, or use frozen arm indicators for multiarm designs; X: prespecified baseline or predecision prognostic variables only; variable set and timing require a DAG/data dictionary; c: fixed primary contrast vector representing: summary-offer assignment effect on independently judged appropriate management; b c,u i: cluster and unit effects, included only for clustering actually present; c as a subscript denotes cluster, whereas contrast vector c is bold in mathematical use; epsilon: residual error with within-unit covariance estimated or prespecified, never assumed independent across all repeated observations; alpha,beta,gamma,theta,tau: intercept, baseline/covariate, arm and interaction parameters estimated from data; no numerical values assumed; w t: prespecified nonnegative time-summary weights summing to one; a single primary endpoint has weight one; logit(p): log[p/(1-p)]; p hat(a,X): fitted outcome probability under arm a standardized to the prespecified population; RD: marginal risk difference; z(A): prespecified indicators for the actual arms or randomized policy; family code: CLUSTER_BINARY
Sample size design: family: binary risk-difference precision/power, simulated with actual cluster/trajectory structure; formula: For Gaussian identity-link linear contrasts only: Var(Delta_hat) = c^T (D^T V^(-1) D)^(-1) c. For nonlinear marginal risks/ratios/counts, use estimator-specific information/sandwich covariance with the delta method or simulate the actual estimator. Choose CI width/power for justified h/delta.; variables: D: planned analysis design matrix, including actual arm/period/cluster terms; V: planned outcome covariance from pilot/data assumptions; alpha*: registered error rate after the primary multiplicity rule; h: desired confidence-interval half-width on the stated estimand scale; delta: externally justified target difference, not estimated from the same confirmatory outcomes; z q: standard-normal q quantile for an approximation; small-cluster/exact methods may replace it; required inputs: independent units and allocation ratios
residual variance/correlation
effect or precision target
missingness
multiplicity
model/cluster structure; powered n: Not established; warning: No powered sample size is asserted. The formula is a planning family; actual model, access and pilot assumptions must be fixed and simulated where necessary.
Assumptions and nonidentification: Adjudication must avoid incorporation bias; summary component associations cannot isolate wearable contribution without separate component randomization.
Consistency of the specified offer/strategy, correct time zero and no unmodeled interference across independently allocated units.
Use allocation-aware uncertainty and a prespecified estimator targeting the chosen outcome scale; model fit is not proof of identification.
Clinician/team allocation requires cluster-level uncertainty and small-cluster sensitivity, not a patient-level nominal sample count.
The supplied sources have not been freshly medically verified. Novelty, source accuracy, access and ethics/regulatory feasibility remain pending; this scaffold cannot supply those facts.
Hypothesis refutation/confirmation requires a registered decision margin and sufficient precision/fidelity; a nonsignificant result alone does not establish equality or absence.
Verification gate: registered scientific challenge: No management improvement, or more low-value testing without actionable findings, would refute clinical utility.; preanalysis gate: Adjudication must avoid incorporation bias; summary component associations cannot isolate wearable contribution without separate component randomization.; decision rule pending: Translate this challenge into one registered effect/precision, equivalence, calibration, detection, agreement or joint-success rule appropriate to the stated model. Bounds/thresholds need scientific justification; no favorable number is invented here.; failure of verification: If units/provenance, assay/reference validity, allocation/eligibility, temporal separation or independent replication gates fail, downgrade to the narrower descriptive/measurement target; do not announce the original causal/clinical claim as tested.; does not establish: Clinical efficacy, safety, biological rejuvenation, mechanistic necessity, causality, external validity or novelty beyond the explicitly identified target.
Sensitivity checks: Adjudication must avoid incorporation bias; summary component associations cannot isolate wearable contribution without separate component randomization.
Change the residual/link model without changing the registered estimand; report absolute contrasts with uncertainty.
Assess missing living-participant outcomes with arm-specific pattern-mixture/tipping-point analyses rather than a single MAR claim.
Apply a fixed primary contrast/outcome hierarchy and multiplicity rule; label exploratory pathways/subgroups distinctly.
Audit provenance, exclusions, timing, physical units, outcome transformations and independent-unit resampling; do not allow technical repeats to inflate sample size.
Requires data or decision: Validated endpoint instrument/scoring, physical units, primary horizon and clinically/scientifically relevant precision or difference.
Allocation probabilities/sequence, actual cluster structure, variance/covariance and plausible missingness from appropriate pilot or accessible data.
Choose the statistical decision margin, confidence/error rate and primary multiplicity rule before seeing confirmatory outcomes; no numerical effect size or powered N is supplied.
Confirm participant/donor/lot/trial access, facility/supervisor capacity, lawful data rights and a current primary-source/prior-art review; inherited planning counts are not access or power evidence.
Define censoring, death, health-related inability, rescue, switching and missing-living-outcome handling where relevant; do not combine biologically undefined outcomes with missing measurements.
Adjudication must avoid incorporation bias; summary component associations cannot isolate wearable contribution without separate component randomization.
Source brief: brief file: biohacking-thesis-dossier.json; brief json pointer: /ideas/129; question: Does a structured longitudinal fatigue summary help clinicians identify actionable causes without increasing low-value testing?; aims: Characterize trajectories preceding presentations with persistent exertional fatigue.
Randomize clinician access to structured longitudinal summaries.
Measure actionable findings, assessment burden, and safe return planning.; design verbatim planning only: Population model: Adult athletes seeking sports-medicine assessment for persistent fatigue, with urgent symptoms triaged immediately. Comparator: Standard clinical assessment versus the same care plus a structured history, workload, symptom, and wearable summary. Prospective causal strategy: Pragmatic clinician-level randomized trial; independent adjudicators classify actionable findings and all groups receive usual diagnostic care. Sample planning: Plan 180 consultations across three clinics; pilot case mix, clustering, and adjudication agreement determine final recruitment.; outcomes verbatim: Primary: independently adjudicated appropriate management decisions. Secondary: testing burden, referral timing, symptom course, and patient understanding.; analysis verbatim inherited not authoritative if refined: Estimate the randomized full-summary effect, accounting for clinician assignment/clustering and baseline severity, sport, load and food access. Prespecify appropriateness criteria and address incorporation/reference bias. Associations with individual summary components are exploratory and cannot isolate the causal contribution of wearable data without a separately randomized component comparison.; falsification verbatim: No management improvement, or more low-value testing without actionable findings, would refute clinical utility.; source ids unverified leads: wearables-post-acute-2026
wearable-hrv-inflammation-2026
wearable-stress-measurement-review-2026; prospectus precedence: No