[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7EBWGAY8Y6429YSWZXESKXB":3,"skill-same:01M7EBWGAY8Y6429YSWZXESKXB":62},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":16,"categories":17,"keywords":18,"publishedAt":19,"linkOnly":20,"markdownContent":21,"markdownBody":22,"frontMatter":23,"refName":31,"commitSha":32,"skillDirectory":33,"forksCount":14,"repositoryHtmlUrl":34,"repositoryHomepageUrl":34,"ownerHtmlUrl":35,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":36,"install":37,"resources":49},"01M7EBWGAY8Y6429YSWZXESKXB","bio-clinical-biostatistics-bayesian-trials",null,"Designs Bayesian clinical trials including Phase I dose-finding (BOIN, CRM, EWOC, mTPI-2), meta-analytic-predictive (MAP) priors with robust mixtures for external data borrowing, EXNEX for basket trials, hierarchical models for safety AE (Berry-Berry), Bayesian platform trials (I-SPY 2, GBM AGILE, REMAP-CAP), and posterior probability stopping rules. Covers FDA Bayesian Devices Guidance (2010), FDA Bayesian Methodology in Drugs Draft (January 2026), BOIN Fit-for-Purpose qualification (December 2021), and Project Optimus dose-optimisation. Use when designing dose-finding studies, platform trials, or sensitivity analyses with informative priors.","Mr-Nobody-Anonymous\u002FAll-skills","Mr-Nobody-Anonymous","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F184253119?v=4","01M7AEXFJHKTYQW5R0ZCQQTWFV","anonymous","01M7C00M3GQR4QKP9GZA42GFAK",0,"MIT","https:\u002F\u002Fgithub.com\u002FMr-Nobody-Anonymous\u002FAll-skills\u002Fblob\u002Fmain\u002Fawesome_skills\u002Fbiology\u002Fbio-clinical-biostatistics-bayesian-trials\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: bio-clinical-biostatistics-bayesian-trials\ndescription: Designs Bayesian clinical trials including Phase I dose-finding (BOIN, CRM, EWOC, mTPI-2), meta-analytic-predictive (MAP) priors with robust mixtures for external data borrowing, EXNEX for basket trials, hierarchical models for safety AE (Berry-Berry), Bayesian platform trials (I-SPY 2, GBM AGILE, REMAP-CAP), and posterior probability stopping rules. Covers FDA Bayesian Devices Guidance (2010), FDA Bayesian Methodology in Drugs Draft (January 2026), BOIN Fit-for-Purpose qualification (December 2021), and Project Optimus dose-optimisation. Use when designing dose-finding studies, platform trials, or sensitivity analyses with informative priors.\ntool_type: r\nprimary_tool: RBesT\ngoal_approach_exempt: true\nsource: \"https:\u002F\u002Fgithub.com\u002FGPTomics\u002FbioSkills\"\nsource_repository: \"GPTomics\u002FbioSkills\"\nsource_path: \"clinical-biostatistics\u002Fbayesian-trials\u002FSKILL.md\"\nlicense: \"MIT\"\nimported_at: \"2026-09-20\"\n---\n\n## Version Compatibility\n\nReference examples tested with: R `RBesT` 1.7+ (Roche), `OncoBayes2` 0.8+ (Novartis), `BOIN` 2.7+, `dfcrm` 0.2-2+, `escalation` 0.1+, `trialr` 0.1.6+, `bayesDP`, `psborrow2` (FDA-supported), `rstan` \u002F `cmdstanr`, `brms`. Legacy: `JAGS`, `WinBUGS`.\n\nBefore using code patterns, verify installed versions match. If versions differ:\n- R: `packageVersion('\u003Cpkg>')` then `?function_name`\n- Confirmatory regulatory work: validate against pinned package versions in submission\n\nIf code throws an error, introspect the installed package and adapt the example to match the actual API rather than retrying.\n\n# Bayesian Clinical Trials\n\n**\"Design a Bayesian clinical trial\"** -> Specify a prior, likelihood, and decision rule with frequentist operating characteristics demonstrated via simulation; for dose-finding use FDA-endorsed BOIN; for borrowing use robust MAP priors; for adaptive platforms use posterior probability of efficacy stopping with simulation-calibrated thresholds.\n\n## Regulatory Status -- The 2024-2026 Bayesian Pivot\n\n**FDA 2010 CDRH Bayesian Devices Guidance** (Feb 5 2010): the only Bayesian-specific FDA guidance until January 2026. Why devices were ahead: CDRH's PMA pathway permits one pivotal trial and accepts borrowing from prior\u002FOUS data more readily than CDER. Example: Edwards SAPIEN (PARTNER B, PMA P100041, Nov 2011) was approved on a single randomized pivotal trial (TAVR vs standard therapy in inoperable patients); the later SAPIEN 3 intermediate-risk PMA used a propensity-score comparison of a single-arm cohort against PARTNER IIA surgical controls -- illustrating CDRH's acceptance of non-randomized\u002Fborrowed comparisons.\n\n**FDA January 2026 CDER Bayesian Methodology Draft** (FDA-2025-D-3217; comment period closed March 13 2026): first-ever drug-side Bayesian guidance. Explicit that Bayesian primary inference in pivotals is acceptable provided:\n- Prospective specification\n- Simulation-based operating characteristics (including frequentist Type-I error under null scenarios — agency still wants calibration)\n- Justified priors\n- Code\u002Fdata sufficient for FDA replication\n\n**Project Optimus (FDA OCE, launched 2021; final dose-optimisation guidance Aug 2024):** rewrites Phase I\u002FII oncology by requiring randomised dose comparison before registration. Has made multi-arm randomised dose-finding (BOIN-12, gBOIN-ET) much more important than classic MTD-finding.\n\n**FDA BOIN Fit-for-Purpose qualification (December 2021):** first formal FDA endorsement of a specific dose-finding design under the Drug Development Tools program.\n\n**ICH E20 (Step 2b\u002F3 draft June 2025; NOT final)** treats Bayesian as a legitimate analytic framework but requires demonstration of acceptable frequentist operating characteristics (Type-I, power) over a pre-specified parameter space.\n\n## Algorithmic Taxonomy\n\n| Method | Use case | Software | Strength | Fails when |\n|--------|----------|----------|----------|------------|\n| BOIN | Phase I MTD | R `BOIN` (Yuan) | **FDA Fit-for-Purpose 2021**; pre-tabulated decisions; no bedside Bayesian software | Statistically less efficient than CRM under correct skeleton |\n| mTPI-2 \u002F Keyboard | Phase I MTD | R `escalation`; R `Keyboard` | Default replacement for mTPI; fixes Ockham bias | Tabulated; transparency |\n| CRM | Phase I MTD | R `dfcrm`, `trialr` | Most efficient under correct skeleton | Skeleton mis-specification biases MTD |\n| EWOC | Phase I MTD | R `ewoc`, `dfcrm` | Explicit overdose-control constraint (P(dose>MTD) \u003C= 0.25) | More conservative than CRM in small trials |\n| BOIN-12 \u002F gBOIN-ET | Phase 1b dose-optimisation (Project Optimus) | R `BOIN` extensions | Multi-arm randomised dose comparison | Requires explicit efficacy + toxicity scoring |\n| MAP prior | Borrowing from historical control arms | R `RBesT::gMAP` | Industry-standard borrowing | Sample-size of MAP prior must be calibrated (Schmidli 2014) |\n| Robust MAP | Borrowing with prior-data conflict protection | R `RBesT::robustify` | Adds vague component (weight 0.1-0.3) to detach if conflict | Mixture weight choice affects borrowing |\n| EXNEX | Basket trial across rare-disease strata | R `bhmbasket`; OncoBayes2 | Avoids HM catastrophic borrowing; mixture 0.5\u002F0.5 default (Neuenschwander 2016) | Default weights may over-borrow |\n| Dixon-Simon shrinkage | Subgroup analysis | Custom Stan\u002Fbrms | Honest about no qualitative interaction prior | Prior on tau drives results |\n| Berry-Berry 3-level hierarchical | AE multiplicity (AE within PT within SOC) | R `c212`; JMP Clinical | Tames safety multiplicity | Spike-and-slab tuning matters |\n| Posterior probability stopping | Adaptive sequential | Custom; FACTS commercial | Bayesian likelihood-principle compatible | Threshold calibration via simulation |\n| Predictive probability of success | End-of-Phase-2 go\u002Fno-go | Custom Stan | Decision-theoretic; integrates over posterior | Requires Phase 3 design specified |\n| Spiegelhalter skeptical\u002Fenthusiastic prior | Sensitivity for regulatory pivotals | Custom | Frames regulator-vs-sponsor evidence | Prior elicitation effort |\n| Power prior | Pediatric extrapolation borrowing from adults | R `bayesDP`, `psborrow2` | Partial borrowing with discount gamma | gamma choice (Jan 2026 FDA draft: 0.3-0.6) |\n\n**Postdoc reading list:**\n\n- FDA 2010 *Guidance for Industry: Use of Bayesian Statistics in Medical Device Clinical Trials* (Feb 5 2010)\n- FDA 2026 Draft *Use of Bayesian Methodology in Clinical Trials* (FDA-2025-D-3217, Jan 2026)\n- Berry SM, Carlin BP, Lee JJ, Müller P 2010 *Bayesian Adaptive Methods for Clinical Trials* (CRC)\n- Schmidli H, Gsteiger S, Roychoudhury S, O'Hagan A, Spiegelhalter D, Neuenschwander B 2014 *Biometrics* 70:1023 (MAP + robust MAP)\n- Weber S, Li Y, Seaman J, Kakizume T, Schmidli H 2021 *J Stat Softw* 100:19 (RBesT)\n- Neuenschwander B, Wandel S, Roychoudhury S, Bailey S 2016 *Pharm Stat* 15:123 (EXNEX)\n- Liu S, Yuan Y 2015 *J R Stat Soc C* 64:507 (BOIN)\n- O'Quigley J, Pepe M, Fisher L 1990 *Biometrics* 46:33 (CRM)\n- Babb J, Rogatko A, Zacks S 1998 *Stat Med* 17:1103 (EWOC)\n- Ji Y, Liu P, Li Y, Bekele BN 2010 *Clin Trials* 7:653 (mTPI)\n- Guo W, Wang SJ, Yang S, Lynn H, Ji Y 2017 *Contemp Clin Trials* 58:23 (mTPI-2 \u002F Keyboard)\n- Berry SM, Broglio KR, Groshen S, Berry DA 2013 *Clin Trials* 10:720 (basket trial hierarchical)\n- Berry SM, Berry DA 2004 *Biometrics* 60:418 (three-level AE hierarchical)\n- Spiegelhalter DJ, Freedman LS, Parmar MKB 1994 *JRSS-A* 157:357 (skeptical\u002Fenthusiastic prior framework)\n- Rugo HS et al 2016 *NEJM* 375:23 (I-SPY 2 veliparib-carboplatin)\n- Angus DC et al 2020 *JAMA* (REMAP-CAP COVID rationale)\n\n## Decision Tree by Scenario\n\n| Scenario | Recommended approach | Why |\n|----------|---------------------|-----|\n| Phase 1 oncology, single-agent MTD | BOIN with target DLT 30%; cohort size 3 | FDA Fit-for-Purpose 2021; tabulated escalation |\n| Phase 1 oncology, combination (2 agents) | BLRM with EXNEX in OncoBayes2 | Multi-dimensional dose; industry standard at Novartis\u002FRoche |\n| Phase 1b\u002F2 dose-optimisation (Project Optimus) | BOIN-12 or gBOIN-ET; randomised 2-dose comparison | Aug 2024 FDA dose-optimisation guidance |\n| Phase 3 with historical control arms available | Robust MAP via RBesT; gMAP() + robustify() | Industry standard borrowing with prior-data conflict protection |\n| Basket trial across rare-disease strata | EXNEX (0.5 EX \u002F 0.5 NEX mixture) via OncoBayes2 | Avoids HM catastrophic borrowing |\n| Pediatric extrapolation from adult data | Power prior with discount gamma 0.3-0.6 | working convention; the FDA Bayesian Jan 2026 draft does not prescribe a specific gamma range -- check the draft for the current language before quoting |\n| Phase 3 trial with single arm + RWE comparator | Propensity-score-integrated power prior via psborrow2 | FDA-supported package for external controls |\n| Adaptive trial wanting posterior-probability stopping | Custom Stan model + simulation-calibrated threshold | Bayesian likelihood-principle compatible; no penalty for repeated looks |\n| End-of-Phase-2 go\u002Fno-go | Predictive probability of success in Phase 3 | Integrates posterior over Phase 3 design |\n| Hypothesis-generating safety AE analysis (>100 PTs) | Berry-Berry 3-level hierarchical (AE within PT within SOC) | Tames multiplicity; spike-and-slab on log OR |\n| Subgroup analysis post-signal | Bayesian shrinkage (Dixon-Simon, RBesT) | Hemmings-Koch 2019: shrinkage for replication planning, NOT signal generation |\n| Regulatory pivotal sensitivity | Spiegelhalter skeptical-prior framework | Frames \"evidence for regulators\" vs \"evidence for sponsor\" |\n\n## Phase I Dose-Finding -- BOIN, CRM, mTPI-2\n\n### BOIN (FDA-preferred operational)\n\n```r\nlibrary(BOIN)\n\n# Generate escalation table for protocol\nboundary_table \u003C- get.boundary(\n    target = 0.30,           # target DLT rate\n    ncohort = 10,            # 10 cohorts -> max 30 patients with size 3\n    cohortsize = 3,\n    n.earlystop = 12,        # stop early at lowest dose if 12 patients show futility\n    p.saf = 0.6 * 0.30,      # \"safe\" escalation boundary\n    p.tox = 1.4 * 0.30       # \"toxic\" de-escalation boundary\n)\nprint(boundary_table)\n# Pre-printed at investigator desk; no bedside Bayesian software\n\n# Operating characteristics simulation\noc_boin \u003C- get.oc(\n    target = 0.30,\n    p.true = c(0.05, 0.10, 0.20, 0.30, 0.40, 0.55),  # true DLT per dose\n    ncohort = 10,\n    cohortsize = 3,\n    ntrial = 1000\n)\nprint(oc_boin)\n# Reports: MTD selection accuracy, overdose risk, average sample size\n```\n\n**BOIN's transparency-over-modelling philosophy:** unlike CRM, BOIN does NOT use information from intermediate dose levels in a model-based way. The Jin-Yuan vs Neuenschwander\u002FMozgunov debate (Stat Med, Pharm Stat, since ~2018): BLRM\u002FCRM are statistically more efficient under correct model; BOIN is operationally simpler and more transparent.\n\n### CRM with calibrated skeleton\n\n```r\nlibrary(dfcrm)\n\nprior_skeleton \u003C- getprior(halfwidth = 0.05, target = 0.30, nu = 3, nlevel = 6)\n# Lee-Cheung 2009 indifference-interval calibration\n\ncrm_sim \u003C- crmsim(\n    PI = c(0.05, 0.10, 0.20, 0.30, 0.40, 0.55),\n    prior = prior_skeleton,\n    target = 0.30,\n    n = 30,\n    x0 = 1,                  # starting dose\n    nsim = 1000,\n    method = 'bayes',\n    model = 'logistic'\n)\nprint(crm_sim)\n```\n\n**Skeleton mis-specification is the canonical CRM failure mode.** Lee-Cheung 2009 indifference-interval method gives a systematic calibration approach.\n\n### EWOC (overdose control)\n\n```r\n# Babb-Rogatko-Zacks 1998: explicit P(dose > MTD) \u003C= alpha (default 0.25)\n# Implementation in dfcrm::ewoc; or `ewoc` package\n```\n\n## MAP Priors and RBesT\n\n**Schmidli et al 2014 *Biometrics* 70:1023:** Meta-Analytic-Predictive prior. Fit random-effects meta-analysis of historical control arms; derive predictive distribution for new control arm; use as informative prior. Effective sample size from history typically 20-80% of new control arm.\n\n```r\nlibrary(RBesT)\n\n# Historical control data (4 prior studies)\nhistorical_data \u003C- data.frame(\n    study = c('s1', 's2', 's3', 's4'),\n    n = c(40, 35, 50, 45),\n    r = c(8, 6, 12, 9)         # responders\n)\n\n# Fit MAP via gMAP (Stan-based random-effects meta-analysis)\nmap_prior \u003C- gMAP(\n    cbind(r, n - r) ~ 1 | study,\n    data = historical_data,\n    family = binomial,\n    tau.dist = 'HalfNormal',\n    tau.prior = 0.5,           # between-study SD prior\n    beta.prior = cbind(0, 2)    # weakly informative on logit response\n)\nprint(map_prior)\n\n# Approximate posterior with mixture for downstream computation\nmap_mix \u003C- automixfit(map_prior, Nc = 2)\nprint(map_mix)\n\n# Effective sample size\ness(map_mix)\n\n# Robust MAP: add vague mixture component (weight 0.1-0.3) to guard against prior-data conflict\nrobust_map \u003C- robustify(map_mix, weight = 0.2, mean = 0.5, n = 1)\nprint(robust_map)\ness(robust_map)\n```\n\n**Robust MAP rationale:** if the new data disagree with historical (prior-data conflict), the mixture down-weights the informative component automatically. The mixture weight on the informative component is a tuning choice and should be varied in a pre-specified sensitivity analysis.\n\n## EXNEX for Basket Trials\n\n**Neuenschwander, Wandel, Roychoudhury, Bailey 2016 *Pharm Stat* 15:123:** Mixture of exchangeable (shared mean+variance) + non-exchangeable (per-basket independent), typically weighted 0.5\u002F0.5. Avoids HM catastrophic borrowing when one basket truly different.\n\n```r\nlibrary(OncoBayes2)  # Novartis-developed; canonical EXNEX implementation\n\n# Or simplified via bhmbasket\nlibrary(bhmbasket)\n\n# Conceptual: each basket has its own posterior, with shrinkage governed by exchangeability mixture\n# Default weights 0.5 EX \u002F 0.5 NEX\n# Sensitivity over weights (0.1, 0.3, 0.5, 0.7, 0.9) is essential\n```\n\n## Bayesian Platform Trials\n\n### I-SPY 2 (Rugo et al 2016 *NEJM* 375:23)\n\nNeoadjuvant breast cancer; 10 biomarker-defined subtypes × multiple arms; Bayesian RAR; **graduation criterion = posterior predictive probability of success in 300-patient Phase 3 ≥ 0.85.** Berry Consultants designed engine.\n\n```r\n# Conceptual implementation requires custom Stan or FACTS (Berry Consultants commercial)\n\n# Pseudocode:\n# 1. Fit hierarchical model to platform data: response ~ arm + biomarker_subtype + arm:subtype\n# 2. Posterior draws of treatment effect by subtype\n# 3. For each draw, simulate Phase 3 trial: n=300, treatment vs control, observed effect\n# 4. Compute proportion of draws meeting Phase 3 success criterion\n# 5. If proportion >= 0.85, arm graduates\n```\n\n### REMAP-CAP (Angus 2020 *JAMA*)\n\nSevere pneumonia, repurposed for COVID-19; Bayesian factorial multi-domain design. Generated corticosteroid signal independently of RECOVERY.\n\n### Drop-the-loser vs promising-the-winner\n\n- **Adaptive arm-dropping (futility):** posterior P(beating control) drops below threshold -> close. Mathematically straightforward.\n- **\"Promising-the-winner\":** selection bias. Bias-adjusted estimators (Robertson 2023; conditional MLE) standard in I-SPY 2 reports.\n\n## Hierarchical Models for Safety Multiplicity (Berry-Berry 2004)\n\n**Berry SM, Berry DA 2004 *Biometrics* 60:418:** three-level hierarchical model for AE multiplicity (AE within MedDRA PT within SOC); spike-and-slab on the log OR. Tames the FDA-feared multiplicity in safety summaries.\n\n```r\nlibrary(c212)  # Berry-Berry implementation\n\n# Conceptual: each AE has log OR drawn from spike-and-slab prior\n# Spike at 0 (no effect); slab as N(mu_SOC, sigma_SOC)\n# SOC-level parameters from N(mu_overall, sigma_overall)\n# Borrowing within SOC; shrinkage toward 0 if no evidence\n\n# JMP Clinical also implements this for industry use\n```\n\n## Power Priors for Borrowing\n\n```r\nlibrary(bayesDP)\nlibrary(psborrow2)  # FDA-supported package\n\n# Power prior: combines current data L(theta | D_current) with historical L(theta | D_hist)^gamma\n# gamma in [0, 1]; gamma = 0 = no borrowing; gamma = 1 = full pooling\n\n# Typical pediatric extrapolation: gamma = 0.3 to 0.6 per FDA Bayesian Jan 2026 draft\n```\n\n## External Control Arms and Real-World Evidence (RWE)\n\n**The 2024-2026 regulatory shift:** FDA has materially expanded acceptance of external\u002Fhistorical\u002Fsynthetic control arms in rare disease, paediatric, and accelerated-approval settings. Key documents: FDA 2018 RWE Framework (and 2024 enhancements), FDA 2023 Considerations for Use of RWE\u002FRWD for Regulatory Decisions, EMA Reflection Paper on Use of RWE in Regulatory Decision-Making (effective 2024). Bayesian methods are the natural fit because historical data become prior information rather than concurrent control.\n\n### Methodology taxonomy\n\n| Method | Borrowing mechanism | Discount control | When to use |\n|--------|---------------------|------------------|-------------|\n| Power prior (Ibrahim-Chen 2000) | Likelihood of historical data raised to power gamma | gamma in [0, 1] fixed or modelled | When historical data is single source; gamma ~ Beta in adaptive power prior |\n| Robust MAP (Schmidli 2014) | Meta-analytic-predictive prior + vague mixture | Mixture weight (typ 0.1-0.3) | Multiple historical control arms; standard for borrowing |\n| Commensurate prior (Hobbs 2011) | Conditional model on agreement parameter | Tau estimated from data | When agreement between historical and current is data-determined |\n| Propensity-integrated power prior | Power prior weighted by PS overlap | gamma * (PS-trimmed overlap) | RWE comparator with covariate imbalance |\n| Doubly robust ATT via causal inference | IPW + outcome regression | n\u002Fa | RWE comparator; identifies marginal ATT |\n\n### psborrow2 — the FDA-supported RWE framework\n\nThe `psborrow2` package (Genentech \u002F Bayer \u002F FDA-Janssen collaboration; CRAN 2024+) is the canonical R implementation for propensity-score-integrated Bayesian Dynamic Borrowing. **The skeleton below illustrates the workflow conceptually; verify exact function names and arguments against the current `psborrow2` vignette before use** (the package API has evolved through 2024-2026).\n\n```r\nlibrary(psborrow2)\n\n# Define external and internal data\next_data \u003C- data.frame(usubjid = ..., trt = 0, outcome = ..., covariates = ...)\nint_data \u003C- data.frame(usubjid = ..., trt = 0 | 1, outcome = ..., covariates = ...)\n\n# Create borrowing design\nborrowing_design \u003C- borrowing_full(\n    method_name = \"BDB\",  # Bayesian Dynamic Borrowing\n    ext_flag_col = \"ext\",\n    tau_prior = prior_gamma(0.001, 0.001)  # weakly informative on borrowing\n)\n\n# Outcome model (Cox for TTE; logistic for binary)\noutcome_model \u003C- outcome_surv_exponential(\n    time_var = \"time\",\n    cens_var = \"cens\",\n    baseline_prior = prior_normal(0, 100),\n    trt_prior = prior_normal(0, 100)\n)\n\n# Run Bayesian analysis with covariate adjustment + borrowing\nresult \u003C- create_analysis_obj(\n    data_matrix = borrow_obj,\n    outcome = outcome_model,\n    borrowing = borrowing_design,\n    covariates = c(\"age\", \"ecog\", \"baseline_severity\")\n)\nmcmc_result \u003C- mcmc_sample(result, n_chains = 4, n_iter = 4000)\n```\n\n### Operational rules (FDA 2024-2025 RWE practice)\n\n1. **Pre-specify the RWE source** and document acquisition (registry, EHR, claims, RWD vendor)\n2. **Demonstrate comparability** via propensity-score overlap (standardised mean differences \u003C0.25 for key prognostic factors)\n3. **Apply discount priors** — full pooling (gamma=1) is regulatory-rejected; typical discount gamma 0.3-0.6\n4. **Sensitivity over borrowing strength** — report results at multiple gamma or mixture weights\n5. **Tipping-point analysis on prior-data agreement** — at what discount does the conclusion flip?\n6. **E-value or bound for unmeasured confounding** (VanderWeele-Ding 2017) — required for FDA submissions; reports the minimum strength of unmeasured confounding that could overturn the result\n\n### When RWE is NOT acceptable\n\n- Trial sponsor and RWE source have meaningful incentive misalignment (e.g., RWE from non-disinterested source)\n- RWE captured before standard-of-care evolved (constancy violation, similar to NI biocreep)\n- Outcome definitions differ between RWE and current trial (variable harmonisation impossible)\n- Censoring patterns in RWE differ structurally from trial (administrative vs disease-driven)\n- Highly variable baseline characteristics impossible to balance via propensity weighting\n\n### Recent decisive cases (2024-2026)\n\n- **Zynteglo (FDA 2022, ongoing post-market):** beta-thalassemia gene therapy; single-arm trial vs natural history RWE comparator\n- **Skysona (FDA 2022):** cerebral adrenoleukodystrophy; RWE natural-history comparator\n- **Multiple ultra-rare disease accelerated approvals 2024-2025:** RWE\u002Fexternal control increasingly accepted in \u003C100-patient trials\n\n## Spiegelhalter Skeptical\u002FEnthusiastic Priors\n\n**Spiegelhalter, Freedman, Parmar 1994 *JRSS-A* 157:357:** the trip-wire \u002F skeptical-prior framework. Pre-specify a skeptical prior centred at the null and an enthusiastic prior centred at the alternative; stopping requires the skeptic to be convinced (posterior under skeptical prior exceeds threshold).\n\n**Frames \"evidence for regulators\" vs \"evidence for sponsor\" in Bayesian language**; still cited in modern Bayesian-trial protocols.\n\n```r\n# Skeptical prior: N(0, sd_sk) — centred at null\n# Enthusiastic prior: N(delta_alt, sd_en) — centred at clinically meaningful effect\n# Decision: stop for efficacy if P(theta > 0 | skeptical posterior) > 0.975\n#           stop for futility if P(theta \u003C delta_alt | enthusiastic posterior) > 0.80\n```\n\n## Per-Method Failure Modes\n\n### CRM with mis-calibrated skeleton\n\n- **Trigger:** Default or arbitrary skeleton without indifference-interval calibration.\n- **Mechanism:** Skeleton dictates target dose; mis-calibration biases MTD.\n- **Symptom:** MTD selection differs systematically from clinical expectation.\n- **Fix:** Calibrate via Lee-Cheung 2009; or switch to BOIN.\n\n### MAP prior with prior-data conflict\n\n- **Trigger:** Historical control rate differs substantially from observed current control.\n- **Mechanism:** Informative MAP prior pulls toward historical; current data poorly fit.\n- **Symptom:** Posterior dominated by prior; current data evidence under-weighted.\n- **Fix:** Robust MAP with mixture weight 0.2-0.3; verify prior-data conflict via posterior predictive checks.\n\n### EXNEX with default 0.5\u002F0.5 weights\n\n- **Trigger:** Default mixture weights without sensitivity.\n- **Mechanism:** 50% EX weight allows substantial borrowing even when basket differs.\n- **Symptom:** Detected differential basket \"softened\" by borrowing.\n- **Fix:** Sensitivity analysis over weights (0.1, 0.3, 0.5, 0.7, 0.9); report range.\n\n### Posterior probability stopping without simulation-calibrated threshold\n\n- **Trigger:** Stopping rule P(theta > 0 | data) > 0.975 applied without Type-I simulation.\n- **Mechanism:** Bayesian rule may not control frequentist Type-I in regulatory sense.\n- **Symptom:** FDA review flags lack of Type-I demonstration.\n- **Fix:** Simulate under null; calibrate threshold so frequentist Type-I = nominal.\n\n### I-SPY 2 graduation criterion without bias correction\n\n- **Trigger:** Graduated arm's effect estimate reported uncorrected.\n- **Mechanism:** Selection on PP > 0.85 inflates estimate.\n- **Symptom:** Phase 3 confirmation finds smaller effect than platform suggested.\n- **Fix:** Bias-correction via conditional MLE or hierarchical Bayesian; cite Robertson 2023.\n\n### Bayesian shrinkage for signal discovery (Dane vs Hemmings)\n\n- **Trigger:** Hierarchical model fit during signal discovery rather than replication planning.\n- **Mechanism:** Shrinkage pre-emptively damps heterogeneity being searched for.\n- **Symptom:** Signal detected by causal forest gets shrunken to null in shrinkage analysis.\n- **Fix:** Hemmings-Koch 2019 position — shrinkage for replication planning, not signal generation; cite Dane et al 2019 EFSPI white paper + critique.\n\n### Power prior with gamma = 1 (full pooling)\n\n- **Trigger:** Full pooling of historical and current data.\n- **Mechanism:** Ignores between-study heterogeneity; biases estimate.\n- **Symptom:** Overconfident posterior; cross-validation reveals poor fit.\n- **Fix:** Working-convention discount gamma 0.3-0.6 (the FDA Bayesian Jan 2026 draft does not prescribe a specific range); sensitivity over gamma.\n\n### WinBUGS reproducibility\n\n- **Trigger:** Submission contains WinBUGS code without containerised environment.\n- **Mechanism:** Older Windows-only software; reproducibility fragile.\n- **Symptom:** Reviewer cannot replicate analysis.\n- **Fix:** Migrate to Stan (`rstan`\u002F`cmdstanr`); Docker\u002Frenv-pinned environment; include seeds + posterior diagnostics (R-hat \u003C1.01, ESS >1000 per chain).\n\n## Quantitative Thresholds\n\n| Threshold | Source | Rationale |\n|-----------|--------|-----------|\n| FDA BOIN Fit-for-Purpose qualification (Dec 2021) | FDA Drug Development Tools program | First formal FDA dose-finding endorsement |\n| Target DLT rate 30% (Phase 1 oncology) | Standard convention | Modal target across oncology Phase 1 |\n| MAP prior effective sample size 20-80% of new control | Schmidli 2014 | Borrowing strength typical range |\n| Robust MAP mixture weight 0.1-0.3 | Schmidli 2014 | Guards against prior-data conflict |\n| EXNEX default 0.5 EX \u002F 0.5 NEX | Neuenschwander 2016 | Standard starting weight; sensitivity required |\n| I-SPY 2 graduation PP >= 0.85 | I-SPY 2 operational reports (Rugo\u002FPark 2016) | Bayesian platform standard |\n| Power prior gamma 0.3-0.6 for pediatric extrapolation | working convention; the FDA Bayesian Jan 2026 draft does not prescribe a specific range | Partial borrowing default |\n| Stan R-hat \u003C1.01, ESS >1000 per chain | Vehtari 2021 *Bayesian Analysis* | Posterior convergence criteria |\n| EWOC overdose constraint P(dose > MTD) \u003C= 0.25 | Babb-Rogatko-Zacks 1998 | Safety floor |\n\n## Common Errors\n\n| Error \u002F symptom | Cause | Solution |\n|-----------------|-------|----------|\n| CRM with arbitrary skeleton | No calibration | Lee-Cheung 2009 indifference-interval; or BOIN |\n| MAP without prior-data conflict check | Posterior dominated by prior | Robust MAP; PP-check; sensitivity over mixture weight |\n| EXNEX with single weight scheme | No sensitivity | Weights 0.1, 0.3, 0.5, 0.7, 0.9; report range |\n| Posterior probability stopping without Type-I sim | Regulatory rejection | Simulate under null; calibrate threshold |\n| I-SPY 2 graduated arm reported uncorrected | Selection bias | Conditional MLE; cite Robertson 2023 |\n| Bayesian shrinkage for signal discovery | Hemmings-Koch critique | Shrinkage for replication only |\n| Power prior gamma = 1 | Full pooling | Discount 0.3-0.6 per FDA 2026 draft |\n| WinBUGS without containerisation | Reproducibility | Stan + Docker\u002Frenv-pinned |\n| BOIN vs CRM comparison without simulation OCs | Apples-to-oranges | Compare OCs over same true DLT rates |\n| FDA cited for Bayesian drugs guidance pre-2026 | Confusion | FDA 2010 is DEVICES; FDA 2026 (draft) is drugs |\n\n## Anticipated Reviewer Pushback\n\n| Pushback | Response |\n|----------|----------|\n| \"Type-I error control?\" | Simulation under null demonstrates frequentist Type-I = nominal at threshold chosen; documented in SAP appendix |\n| \"Prior justification?\" | MAP from historical control arms via gMAP; robust mixture weight 0.2 for prior-data conflict; sensitivity over prior provided |\n| \"Why BOIN over CRM?\" | BOIN Fit-for-Purpose qualified Dec 2021; pre-tabulated escalation; no bedside Bayesian software; OCs comparable to CRM in simulation |\n| \"EXNEX weight sensitivity?\" | Reported over weights 0.1, 0.3, 0.5, 0.7, 0.9; results stable; primary at 0.5\u002F0.5 per Neuenschwander 2016 |\n| \"Power prior gamma?\" | Discount 0.5 per FDA Bayesian Jan 2026 draft; sensitivity over 0.3-0.7 provided |\n| \"Posterior probability threshold?\" | Calibrated via simulation to frequentist Type-I 0.025 one-sided; cite Berry 2010 |\n| \"Stan reproducibility?\" | Docker container + renv-pinned R + Stan version; seeds provided; R-hat \u003C1.01, ESS >2000 per parameter |\n| \"Bias correction on platform graduation?\" | Conditional MLE applied to estimate Phase 3 effect; cite Robertson 2023 |\n| \"Why not frequentist instead?\" | Bayesian framework permits borrowing (rare disease, pediatric); working convention; the FDA Bayesian Jan 2026 draft does not prescribe a specific gamma range -- check the draft for the current language before quoting primary inference with simulation calibration |\n\n## References\n\n- Babb J, Rogatko A, Zacks S. 1998. Cancer Phase I clinical trials: efficient dose escalation with overdose control. *Stat Med* 17:1103-1120.\n- Berry SM, Berry DA. 2004. Accounting for multiplicities in assessing drug safety: a three-level hierarchical mixture model. *Biometrics* 60:418-426.\n- Berry SM, Broglio KR, Groshen S, Berry DA. 2013. Bayesian hierarchical modeling of patient subpopulations: efficient designs of Phase II oncology clinical trials. *Clin Trials* 10:720-734.\n- Berry SM, Carlin BP, Lee JJ, Müller P. 2010. *Bayesian Adaptive Methods for Clinical Trials*. CRC.\n- Dane A, Spencer A, Rosenkranz G, Lipkovich I, Parke T. 2019. Subgroup analysis and interpretation for phase 3 confirmatory trials: EFSPI\u002FPSI white paper. *Pharm Stat* 18:126-139.\n- FDA. 2010. Guidance for Industry: Use of Bayesian Statistics in Medical Device Clinical Trials.\n- FDA. 2021. BOIN Drug Development Tool Fit-for-Purpose Qualification.\n- FDA. 2026. Use of Bayesian Methodology in Clinical Trials. Draft Guidance (FDA-2025-D-3217).\n- Guo W, Wang SJ, Yang S, Lynn H, Ji Y. 2017. A Bayesian interval dose-finding design addressing Ockham's razor: mTPI-2. *Contemp Clin Trials* 58:23-33.\n- Hemmings R, Koch A. 2019. Commentary on Dane et al. *Pharm Stat* 18:140-144.\n- Ji Y, Liu P, Li Y, Bekele BN. 2010. A modified toxicity probability interval method for dose-finding trials. *Clin Trials* 7:653-663.\n- Liu S, Yuan Y. 2015. Bayesian optimal interval designs for phase I clinical trials. *JRSS-C* 64:507-523.\n- Neuenschwander B, Wandel S, Roychoudhury S, Bailey S. 2016. Robust exchangeability designs for early phase clinical trials with multiple strata. *Pharm Stat* 15:123-134.\n- O'Quigley J, Pepe M, Fisher L. 1990. Continual reassessment method: a practical design for phase 1 clinical trials in cancer. *Biometrics* 46:33-48.\n- Rugo HS et al. 2016. Adaptive randomization of veliparib-carboplatin treatment in breast cancer. *NEJM* 375:23-34.\n- Robertson DS, Lee KM, López-Kolkovska BC, Villar SS. 2023. Response-adaptive randomization in clinical trials: from myths to practical considerations. *Stat Sci* 38:185-208.\n- Schmidli H, Gsteiger S, Roychoudhury S, O'Hagan A, Spiegelhalter D, Neuenschwander B. 2014. Robust meta-analytic-predictive priors in clinical trials with historical control information. *Biometrics* 70:1023-1032.\n- Spiegelhalter DJ, Freedman LS, Parmar MKB. 1994. Bayesian approaches to randomized trials. *JRSS-A* 157:357-387.\n- Vehtari A et al. 2021. Rank-normalization, folding, and localization: an improved R-hat for assessing convergence. *Bayesian Analysis*.\n- Weber S, Li Y, Seaman J, Kakizume T, Schmidli H. 2021. Applying meta-analytic-predictive priors with the R Bayesian evidence synthesis tools. *J Stat Softw* 100:19.\n\n## Related Skills\n\n- clinical-biostatistics\u002Fadaptive-designs - Group-sequential, SSR, platform trials\n- clinical-biostatistics\u002Fsubgroup-analysis - Bayesian shrinkage for HTE (Dixon-Simon, Berry)\n- clinical-biostatistics\u002Fpower-and-sample-size - Bayesian SS via predictive probability of success\n- clinical-biostatistics\u002Fmultiplicity-graphical - Berry-Berry AE hierarchical\n- clinical-biostatistics\u002Ftrial-reporting - Bayesian inference reporting per CONSORT 2025\n- clinical-biostatistics\u002Fmissing-data-sensitivity - Bayesian rbmi imputation\n- machine-learning\u002Fbiomarker-discovery - Bayesian HTE for biomarker subgroups\n- experimental-design\u002Fsample-size - General methods\n","## Version Compatibility\n\nReference examples tested with: R `RBesT` 1.7+ (Roche), `OncoBayes2` 0.8+ (Novartis), `BOIN` 2.7+, `dfcrm` 0.2-2+, `escalation` 0.1+, `trialr` 0.1.6+, `bayesDP`, `psborrow2` (FDA-supported), `rstan` \u002F `cmdstanr`, `brms`. Legacy: `JAGS`, `WinBUGS`.\n\nBefore using code patterns, verify installed versions match. If versions differ:\n- R: `packageVersion('\u003Cpkg>')` then `?function_name`\n- Confirmatory regulatory work: validate against pinned package versions in submission\n\nIf code throws an error, introspect the installed package and adapt the example to match the actual API rather than retrying.\n\n# Bayesian Clinical Trials\n\n**\"Design a Bayesian clinical trial\"** -> Specify a prior, likelihood, and decision rule with frequentist operating characteristics demonstrated via simulation; for dose-finding use FDA-endorsed BOIN; for borrowing use robust MAP priors; for adaptive platforms use posterior probability of efficacy stopping with simulation-calibrated thresholds.\n\n## Regulatory Status -- The 2024-2026 Bayesian Pivot\n\n**FDA 2010 CDRH Bayesian Devices Guidance** (Feb 5 2010): the only Bayesian-specific FDA guidance until January 2026. Why devices were ahead: CDRH's PMA pathway permits one pivotal trial and accepts borrowing from prior\u002FOUS data more readily than CDER. Example: Edwards SAPIEN (PARTNER B, PMA P100041, Nov 2011) was approved on a single randomized pivotal trial (TAVR vs standard therapy in inoperable patients); the later SAPIEN 3 intermediate-risk PMA used a propensity-score comparison of a single-arm cohort against PARTNER IIA surgical controls -- illustrating CDRH's acceptance of non-randomized\u002Fborrowed comparisons.\n\n**FDA January 2026 CDER Bayesian Methodology Draft** (FDA-2025-D-3217; comment period closed March 13 2026): first-ever drug-side Bayesian guidance. Explicit that Bayesian primary inference in pivotals is acceptable provided:\n- Prospective specification\n- Simulation-based operating characteristics (including frequentist Type-I error under null scenarios — agency still wants calibration)\n- Justified priors\n- Code\u002Fdata sufficient for FDA replication\n\n**Project Optimus (FDA OCE, launched 2021; final dose-optimisation guidance Aug 2024):** rewrites Phase I\u002FII oncology by requiring randomised dose comparison before registration. Has made multi-arm randomised dose-finding (BOIN-12, gBOIN-ET) much more important than classic MTD-finding.\n\n**FDA BOIN Fit-for-Purpose qualification (December 2021):** first formal FDA endorsement of a specific dose-finding design under the Drug Development Tools program.\n\n**ICH E20 (Step 2b\u002F3 draft June 2025; NOT final)** treats Bayesian as a legitimate analytic framework but requires demonstration of acceptable frequentist operating characteristics (Type-I, power) over a pre-specified parameter space.\n\n## Algorithmic Taxonomy\n\n| Method | Use case | Software | Strength | Fails when |\n|--------|----------|----------|----------|------------|\n| BOIN | Phase I MTD | R `BOIN` (Yuan) | **FDA Fit-for-Purpose 2021**; pre-tabulated decisions; no bedside Bayesian software | Statistically less efficient than CRM under correct skeleton |\n| mTPI-2 \u002F Keyboard | Phase I MTD | R `escalation`; R `Keyboard` | Default replacement for mTPI; fixes Ockham bias | Tabulated; transparency |\n| CRM | Phase I MTD | R `dfcrm`, `trialr` | Most efficient under correct skeleton | Skeleton mis-specification biases MTD |\n| EWOC | Phase I MTD | R `ewoc`, `dfcrm` | Explicit overdose-control constraint (P(dose>MTD) \u003C= 0.25) | More conservative than CRM in small trials |\n| BOIN-12 \u002F gBOIN-ET | Phase 1b dose-optimisation (Project Optimus) | R `BOIN` extensions | Multi-arm randomised dose comparison | Requires explicit efficacy + toxicity scoring |\n| MAP prior | Borrowing from historical control arms | R `RBesT::gMAP` | Industry-standard borrowing | Sample-size of MAP prior must be calibrated (Schmidli 2014) |\n| Robust MAP | Borrowing with prior-data conflict protection | R `RBesT::robustify` | Adds vague component (weight 0.1-0.3) to detach if conflict | Mixture weight choice affects borrowing |\n| EXNEX | Basket trial across rare-disease strata | R `bhmbasket`; OncoBayes2 | Avoids HM catastrophic borrowing; mixture 0.5\u002F0.5 default (Neuenschwander 2016) | Default weights may over-borrow |\n| Dixon-Simon shrinkage | Subgroup analysis | Custom Stan\u002Fbrms | Honest about no qualitative interaction prior | Prior on tau drives results |\n| Berry-Berry 3-level hierarchical | AE multiplicity (AE within PT within SOC) | R `c212`; JMP Clinical | Tames safety multiplicity | Spike-and-slab tuning matters |\n| Posterior probability stopping | Adaptive sequential | Custom; FACTS commercial | Bayesian likelihood-principle compatible | Threshold calibration via simulation |\n| Predictive probability of success | End-of-Phase-2 go\u002Fno-go | Custom Stan | Decision-theoretic; integrates over posterior | Requires Phase 3 design specified |\n| Spiegelhalter skeptical\u002Fenthusiastic prior | Sensitivity for regulatory pivotals | Custom | Frames regulator-vs-sponsor evidence | Prior elicitation effort |\n| Power prior | Pediatric extrapolation borrowing from adults | R `bayesDP`, `psborrow2` | Partial borrowing with discount gamma | gamma choice (Jan 2026 FDA draft: 0.3-0.6) |\n\n**Postdoc reading list:**\n\n- FDA 2010 *Guidance for Industry: Use of Bayesian Statistics in Medical Device Clinical Trials* (Feb 5 2010)\n- FDA 2026 Draft *Use of Bayesian Methodology in Clinical Trials* (FDA-2025-D-3217, Jan 2026)\n- Berry SM, Carlin BP, Lee JJ, Müller P 2010 *Bayesian Adaptive Methods for Clinical Trials* (CRC)\n- Schmidli H, Gsteiger S, Roychoudhury S, O'Hagan A, Spiegelhalter D, Neuenschwander B 2014 *Biometrics* 70:1023 (MAP + robust MAP)\n- Weber S, Li Y, Seaman J, Kakizume T, Schmidli H 2021 *J Stat Softw* 100:19 (RBesT)\n- Neuenschwander B, Wandel S, Roychoudhury S, Bailey S 2016 *Pharm Stat* 15:123 (EXNEX)\n- Liu S, Yuan Y 2015 *J R Stat Soc C* 64:507 (BOIN)\n- O'Quigley J, Pepe M, Fisher L 1990 *Biometrics* 46:33 (CRM)\n- Babb J, Rogatko A, Zacks S 1998 *Stat Med* 17:1103 (EWOC)\n- Ji Y, Liu P, Li Y, Bekele BN 2010 *Clin Trials* 7:653 (mTPI)\n- Guo W, Wang SJ, Yang S, Lynn H, Ji Y 2017 *Contemp Clin Trials* 58:23 (mTPI-2 \u002F Keyboard)\n- Berry SM, Broglio KR, Groshen S, Berry DA 2013 *Clin Trials* 10:720 (basket trial hierarchical)\n- Berry SM, Berry DA 2004 *Biometrics* 60:418 (three-level AE hierarchical)\n- Spiegelhalter DJ, Freedman LS, Parmar MKB 1994 *JRSS-A* 157:357 (skeptical\u002Fenthusiastic prior framework)\n- Rugo HS et al 2016 *NEJM* 375:23 (I-SPY 2 veliparib-carboplatin)\n- Angus DC et al 2020 *JAMA* (REMAP-CAP COVID rationale)\n\n## Decision Tree by Scenario\n\n| Scenario | Recommended approach | Why |\n|----------|---------------------|-----|\n| Phase 1 oncology, single-agent MTD | BOIN with target DLT 30%; cohort size 3 | FDA Fit-for-Purpose 2021; tabulated escalation |\n| Phase 1 oncology, combination (2 agents) | BLRM with EXNEX in OncoBayes2 | Multi-dimensional dose; industry standard at Novartis\u002FRoche |\n| Phase 1b\u002F2 dose-optimisation (Project Optimus) | BOIN-12 or gBOIN-ET; randomised 2-dose comparison | Aug 2024 FDA dose-optimisation guidance |\n| Phase 3 with historical control arms available | Robust MAP via RBesT; gMAP() + robustify() | Industry standard borrowing with prior-data conflict protection |\n| Basket trial across rare-disease strata | EXNEX (0.5 EX \u002F 0.5 NEX mixture) via OncoBayes2 | Avoids HM catastrophic borrowing |\n| Pediatric extrapolation from adult data | Power prior with discount gamma 0.3-0.6 | working convention; the FDA Bayesian Jan 2026 draft does not prescribe a specific gamma range -- check the draft for the current language before quoting |\n| Phase 3 trial with single arm + RWE comparator | Propensity-score-integrated power prior via psborrow2 | FDA-supported package for external controls |\n| Adaptive trial wanting posterior-probability stopping | Custom Stan model + simulation-calibrated threshold | Bayesian likelihood-principle compatible; no penalty for repeated looks |\n| End-of-Phase-2 go\u002Fno-go | Predictive probability of success in Phase 3 | Integrates posterior over Phase 3 design |\n| Hypothesis-generating safety AE analysis (>100 PTs) | Berry-Berry 3-level hierarchical (AE within PT within SOC) | Tames multiplicity; spike-and-slab on log OR |\n| Subgroup analysis post-signal | Bayesian shrinkage (Dixon-Simon, RBesT) | Hemmings-Koch 2019: shrinkage for replication planning, NOT signal generation |\n| Regulatory pivotal sensitivity | Spiegelhalter skeptical-prior framework | Frames \"evidence for regulators\" vs \"evidence for sponsor\" |\n\n## Phase I Dose-Finding -- BOIN, CRM, mTPI-2\n\n### BOIN (FDA-preferred operational)\n\n```r\nlibrary(BOIN)\n\n# Generate escalation table for protocol\nboundary_table \u003C- get.boundary(\n    target = 0.30,           # target DLT rate\n    ncohort = 10,            # 10 cohorts -> max 30 patients with size 3\n    cohortsize = 3,\n    n.earlystop = 12,        # stop early at lowest dose if 12 patients show futility\n    p.saf = 0.6 * 0.30,      # \"safe\" escalation boundary\n    p.tox = 1.4 * 0.30       # \"toxic\" de-escalation boundary\n)\nprint(boundary_table)\n# Pre-printed at investigator desk; no bedside Bayesian software\n\n# Operating characteristics simulation\noc_boin \u003C- get.oc(\n    target = 0.30,\n    p.true = c(0.05, 0.10, 0.20, 0.30, 0.40, 0.55),  # true DLT per dose\n    ncohort = 10,\n    cohortsize = 3,\n    ntrial = 1000\n)\nprint(oc_boin)\n# Reports: MTD selection accuracy, overdose risk, average sample size\n```\n\n**BOIN's transparency-over-modelling philosophy:** unlike CRM, BOIN does NOT use information from intermediate dose levels in a model-based way. The Jin-Yuan vs Neuenschwander\u002FMozgunov debate (Stat Med, Pharm Stat, since ~2018): BLRM\u002FCRM are statistically more efficient under correct model; BOIN is operationally simpler and more transparent.\n\n### CRM with calibrated skeleton\n\n```r\nlibrary(dfcrm)\n\nprior_skeleton \u003C- getprior(halfwidth = 0.05, target = 0.30, nu = 3, nlevel = 6)\n# Lee-Cheung 2009 indifference-interval calibration\n\ncrm_sim \u003C- crmsim(\n    PI = c(0.05, 0.10, 0.20, 0.30, 0.40, 0.55),\n    prior = prior_skeleton,\n    target = 0.30,\n    n = 30,\n    x0 = 1,                  # starting dose\n    nsim = 1000,\n    method = 'bayes',\n    model = 'logistic'\n)\nprint(crm_sim)\n```\n\n**Skeleton mis-specification is the canonical CRM failure mode.** Lee-Cheung 2009 indifference-interval method gives a systematic calibration approach.\n\n### EWOC (overdose control)\n\n```r\n# Babb-Rogatko-Zacks 1998: explicit P(dose > MTD) \u003C= alpha (default 0.25)\n# Implementation in dfcrm::ewoc; or `ewoc` package\n```\n\n## MAP Priors and RBesT\n\n**Schmidli et al 2014 *Biometrics* 70:1023:** Meta-Analytic-Predictive prior. Fit random-effects meta-analysis of historical control arms; derive predictive distribution for new control arm; use as informative prior. Effective sample size from history typically 20-80% of new control arm.\n\n```r\nlibrary(RBesT)\n\n# Historical control data (4 prior studies)\nhistorical_data \u003C- data.frame(\n    study = c('s1', 's2', 's3', 's4'),\n    n = c(40, 35, 50, 45),\n    r = c(8, 6, 12, 9)         # responders\n)\n\n# Fit MAP via gMAP (Stan-based random-effects meta-analysis)\nmap_prior \u003C- gMAP(\n    cbind(r, n - r) ~ 1 | study,\n    data = historical_data,\n    family = binomial,\n    tau.dist = 'HalfNormal',\n    tau.prior = 0.5,           # between-study SD prior\n    beta.prior = cbind(0, 2)    # weakly informative on logit response\n)\nprint(map_prior)\n\n# Approximate posterior with mixture for downstream computation\nmap_mix \u003C- automixfit(map_prior, Nc = 2)\nprint(map_mix)\n\n# Effective sample size\ness(map_mix)\n\n# Robust MAP: add vague mixture component (weight 0.1-0.3) to guard against prior-data conflict\nrobust_map \u003C- robustify(map_mix, weight = 0.2, mean = 0.5, n = 1)\nprint(robust_map)\ness(robust_map)\n```\n\n**Robust MAP rationale:** if the new data disagree with historical (prior-data conflict), the mixture down-weights the informative component automatically. The mixture weight on the informative component is a tuning choice and should be varied in a pre-specified sensitivity analysis.\n\n## EXNEX for Basket Trials\n\n**Neuenschwander, Wandel, Roychoudhury, Bailey 2016 *Pharm Stat* 15:123:** Mixture of exchangeable (shared mean+variance) + non-exchangeable (per-basket independent), typically weighted 0.5\u002F0.5. Avoids HM catastrophic borrowing when one basket truly different.\n\n```r\nlibrary(OncoBayes2)  # Novartis-developed; canonical EXNEX implementation\n\n# Or simplified via bhmbasket\nlibrary(bhmbasket)\n\n# Conceptual: each basket has its own posterior, with shrinkage governed by exchangeability mixture\n# Default weights 0.5 EX \u002F 0.5 NEX\n# Sensitivity over weights (0.1, 0.3, 0.5, 0.7, 0.9) is essential\n```\n\n## Bayesian Platform Trials\n\n### I-SPY 2 (Rugo et al 2016 *NEJM* 375:23)\n\nNeoadjuvant breast cancer; 10 biomarker-defined subtypes × multiple arms; Bayesian RAR; **graduation criterion = posterior predictive probability of success in 300-patient Phase 3 ≥ 0.85.** Berry Consultants designed engine.\n\n```r\n# Conceptual implementation requires custom Stan or FACTS (Berry Consultants commercial)\n\n# Pseudocode:\n# 1. Fit hierarchical model to platform data: response ~ arm + biomarker_subtype + arm:subtype\n# 2. Posterior draws of treatment effect by subtype\n# 3. For each draw, simulate Phase 3 trial: n=300, treatment vs control, observed effect\n# 4. Compute proportion of draws meeting Phase 3 success criterion\n# 5. If proportion >= 0.85, arm graduates\n```\n\n### REMAP-CAP (Angus 2020 *JAMA*)\n\nSevere pneumonia, repurposed for COVID-19; Bayesian factorial multi-domain design. Generated corticosteroid signal independently of RECOVERY.\n\n### Drop-the-loser vs promising-the-winner\n\n- **Adaptive arm-dropping (futility):** posterior P(beating control) drops below threshold -> close. Mathematically straightforward.\n- **\"Promising-the-winner\":** selection bias. Bias-adjusted estimators (Robertson 2023; conditional MLE) standard in I-SPY 2 reports.\n\n## Hierarchical Models for Safety Multiplicity (Berry-Berry 2004)\n\n**Berry SM, Berry DA 2004 *Biometrics* 60:418:** three-level hierarchical model for AE multiplicity (AE within MedDRA PT within SOC); spike-and-slab on the log OR. Tames the FDA-feared multiplicity in safety summaries.\n\n```r\nlibrary(c212)  # Berry-Berry implementation\n\n# Conceptual: each AE has log OR drawn from spike-and-slab prior\n# Spike at 0 (no effect); slab as N(mu_SOC, sigma_SOC)\n# SOC-level parameters from N(mu_overall, sigma_overall)\n# Borrowing within SOC; shrinkage toward 0 if no evidence\n\n# JMP Clinical also implements this for industry use\n```\n\n## Power Priors for Borrowing\n\n```r\nlibrary(bayesDP)\nlibrary(psborrow2)  # FDA-supported package\n\n# Power prior: combines current data L(theta | D_current) with historical L(theta | D_hist)^gamma\n# gamma in [0, 1]; gamma = 0 = no borrowing; gamma = 1 = full pooling\n\n# Typical pediatric extrapolation: gamma = 0.3 to 0.6 per FDA Bayesian Jan 2026 draft\n```\n\n## External Control Arms and Real-World Evidence (RWE)\n\n**The 2024-2026 regulatory shift:** FDA has materially expanded acceptance of external\u002Fhistorical\u002Fsynthetic control arms in rare disease, paediatric, and accelerated-approval settings. Key documents: FDA 2018 RWE Framework (and 2024 enhancements), FDA 2023 Considerations for Use of RWE\u002FRWD for Regulatory Decisions, EMA Reflection Paper on Use of RWE in Regulatory Decision-Making (effective 2024). Bayesian methods are the natural fit because historical data become prior information rather than concurrent control.\n\n### Methodology taxonomy\n\n| Method | Borrowing mechanism | Discount control | When to use |\n|--------|---------------------|------------------|-------------|\n| Power prior (Ibrahim-Chen 2000) | Likelihood of historical data raised to power gamma | gamma in [0, 1] fixed or modelled | When historical data is single source; gamma ~ Beta in adaptive power prior |\n| Robust MAP (Schmidli 2014) | Meta-analytic-predictive prior + vague mixture | Mixture weight (typ 0.1-0.3) | Multiple historical control arms; standard for borrowing |\n| Commensurate prior (Hobbs 2011) | Conditional model on agreement parameter | Tau estimated from data | When agreement between historical and current is data-determined |\n| Propensity-integrated power prior | Power prior weighted by PS overlap | gamma * (PS-trimmed overlap) | RWE comparator with covariate imbalance |\n| Doubly robust ATT via causal inference | IPW + outcome regression | n\u002Fa | RWE comparator; identifies marginal ATT |\n\n### psborrow2 — the FDA-supported RWE framework\n\nThe `psborrow2` package (Genentech \u002F Bayer \u002F FDA-Janssen collaboration; CRAN 2024+) is the canonical R implementation for propensity-score-integrated Bayesian Dynamic Borrowing. **The skeleton below illustrates the workflow conceptually; verify exact function names and arguments against the current `psborrow2` vignette before use** (the package API has evolved through 2024-2026).\n\n```r\nlibrary(psborrow2)\n\n# Define external and internal data\next_data \u003C- data.frame(usubjid = ..., trt = 0, outcome = ..., covariates = ...)\nint_data \u003C- data.frame(usubjid = ..., trt = 0 | 1, outcome = ..., covariates = ...)\n\n# Create borrowing design\nborrowing_design \u003C- borrowing_full(\n    method_name = \"BDB\",  # Bayesian Dynamic Borrowing\n    ext_flag_col = \"ext\",\n    tau_prior = prior_gamma(0.001, 0.001)  # weakly informative on borrowing\n)\n\n# Outcome model (Cox for TTE; logistic for binary)\noutcome_model \u003C- outcome_surv_exponential(\n    time_var = \"time\",\n    cens_var = \"cens\",\n    baseline_prior = prior_normal(0, 100),\n    trt_prior = prior_normal(0, 100)\n)\n\n# Run Bayesian analysis with covariate adjustment + borrowing\nresult \u003C- create_analysis_obj(\n    data_matrix = borrow_obj,\n    outcome = outcome_model,\n    borrowing = borrowing_design,\n    covariates = c(\"age\", \"ecog\", \"baseline_severity\")\n)\nmcmc_result \u003C- mcmc_sample(result, n_chains = 4, n_iter = 4000)\n```\n\n### Operational rules (FDA 2024-2025 RWE practice)\n\n1. **Pre-specify the RWE source** and document acquisition (registry, EHR, claims, RWD vendor)\n2. **Demonstrate comparability** via propensity-score overlap (standardised mean differences \u003C0.25 for key prognostic factors)\n3. **Apply discount priors** — full pooling (gamma=1) is regulatory-rejected; typical discount gamma 0.3-0.6\n4. **Sensitivity over borrowing strength** — report results at multiple gamma or mixture weights\n5. **Tipping-point analysis on prior-data agreement** — at what discount does the conclusion flip?\n6. **E-value or bound for unmeasured confounding** (VanderWeele-Ding 2017) — required for FDA submissions; reports the minimum strength of unmeasured confounding that could overturn the result\n\n### When RWE is NOT acceptable\n\n- Trial sponsor and RWE source have meaningful incentive misalignment (e.g., RWE from non-disinterested source)\n- RWE captured before standard-of-care evolved (constancy violation, similar to NI biocreep)\n- Outcome definitions differ between RWE and current trial (variable harmonisation impossible)\n- Censoring patterns in RWE differ structurally from trial (administrative vs disease-driven)\n- Highly variable baseline characteristics impossible to balance via propensity weighting\n\n### Recent decisive cases (2024-2026)\n\n- **Zynteglo (FDA 2022, ongoing post-market):** beta-thalassemia gene therapy; single-arm trial vs natural history RWE comparator\n- **Skysona (FDA 2022):** cerebral adrenoleukodystrophy; RWE natural-history comparator\n- **Multiple ultra-rare disease accelerated approvals 2024-2025:** RWE\u002Fexternal control increasingly accepted in \u003C100-patient trials\n\n## Spiegelhalter Skeptical\u002FEnthusiastic Priors\n\n**Spiegelhalter, Freedman, Parmar 1994 *JRSS-A* 157:357:** the trip-wire \u002F skeptical-prior framework. Pre-specify a skeptical prior centred at the null and an enthusiastic prior centred at the alternative; stopping requires the skeptic to be convinced (posterior under skeptical prior exceeds threshold).\n\n**Frames \"evidence for regulators\" vs \"evidence for sponsor\" in Bayesian language**; still cited in modern Bayesian-trial protocols.\n\n```r\n# Skeptical prior: N(0, sd_sk) — centred at null\n# Enthusiastic prior: N(delta_alt, sd_en) — centred at clinically meaningful effect\n# Decision: stop for efficacy if P(theta > 0 | skeptical posterior) > 0.975\n#           stop for futility if P(theta \u003C delta_alt | enthusiastic posterior) > 0.80\n```\n\n## Per-Method Failure Modes\n\n### CRM with mis-calibrated skeleton\n\n- **Trigger:** Default or arbitrary skeleton without indifference-interval calibration.\n- **Mechanism:** Skeleton dictates target dose; mis-calibration biases MTD.\n- **Symptom:** MTD selection differs systematically from clinical expectation.\n- **Fix:** Calibrate via Lee-Cheung 2009; or switch to BOIN.\n\n### MAP prior with prior-data conflict\n\n- **Trigger:** Historical control rate differs substantially from observed current control.\n- **Mechanism:** Informative MAP prior pulls toward historical; current data poorly fit.\n- **Symptom:** Posterior dominated by prior; current data evidence under-weighted.\n- **Fix:** Robust MAP with mixture weight 0.2-0.3; verify prior-data conflict via posterior predictive checks.\n\n### EXNEX with default 0.5\u002F0.5 weights\n\n- **Trigger:** Default mixture weights without sensitivity.\n- **Mechanism:** 50% EX weight allows substantial borrowing even when basket differs.\n- **Symptom:** Detected differential basket \"softened\" by borrowing.\n- **Fix:** Sensitivity analysis over weights (0.1, 0.3, 0.5, 0.7, 0.9); report range.\n\n### Posterior probability stopping without simulation-calibrated threshold\n\n- **Trigger:** Stopping rule P(theta > 0 | data) > 0.975 applied without Type-I simulation.\n- **Mechanism:** Bayesian rule may not control frequentist Type-I in regulatory sense.\n- **Symptom:** FDA review flags lack of Type-I demonstration.\n- **Fix:** Simulate under null; calibrate threshold so frequentist Type-I = nominal.\n\n### I-SPY 2 graduation criterion without bias correction\n\n- **Trigger:** Graduated arm's effect estimate reported uncorrected.\n- **Mechanism:** Selection on PP > 0.85 inflates estimate.\n- **Symptom:** Phase 3 confirmation finds smaller effect than platform suggested.\n- **Fix:** Bias-correction via conditional MLE or hierarchical Bayesian; cite Robertson 2023.\n\n### Bayesian shrinkage for signal discovery (Dane vs Hemmings)\n\n- **Trigger:** Hierarchical model fit during signal discovery rather than replication planning.\n- **Mechanism:** Shrinkage pre-emptively damps heterogeneity being searched for.\n- **Symptom:** Signal detected by causal forest gets shrunken to null in shrinkage analysis.\n- **Fix:** Hemmings-Koch 2019 position — shrinkage for replication planning, not signal generation; cite Dane et al 2019 EFSPI white paper + critique.\n\n### Power prior with gamma = 1 (full pooling)\n\n- **Trigger:** Full pooling of historical and current data.\n- **Mechanism:** Ignores between-study heterogeneity; biases estimate.\n- **Symptom:** Overconfident posterior; cross-validation reveals poor fit.\n- **Fix:** Working-convention discount gamma 0.3-0.6 (the FDA Bayesian Jan 2026 draft does not prescribe a specific range); sensitivity over gamma.\n\n### WinBUGS reproducibility\n\n- **Trigger:** Submission contains WinBUGS code without containerised environment.\n- **Mechanism:** Older Windows-only software; reproducibility fragile.\n- **Symptom:** Reviewer cannot replicate analysis.\n- **Fix:** Migrate to Stan (`rstan`\u002F`cmdstanr`); Docker\u002Frenv-pinned environment; include seeds + posterior diagnostics (R-hat \u003C1.01, ESS >1000 per chain).\n\n## Quantitative Thresholds\n\n| Threshold | Source | Rationale |\n|-----------|--------|-----------|\n| FDA BOIN Fit-for-Purpose qualification (Dec 2021) | FDA Drug Development Tools program | First formal FDA dose-finding endorsement |\n| Target DLT rate 30% (Phase 1 oncology) | Standard convention | Modal target across oncology Phase 1 |\n| MAP prior effective sample size 20-80% of new control | Schmidli 2014 | Borrowing strength typical range |\n| Robust MAP mixture weight 0.1-0.3 | Schmidli 2014 | Guards against prior-data conflict |\n| EXNEX default 0.5 EX \u002F 0.5 NEX | Neuenschwander 2016 | Standard starting weight; sensitivity required |\n| I-SPY 2 graduation PP >= 0.85 | I-SPY 2 operational reports (Rugo\u002FPark 2016) | Bayesian platform standard |\n| Power prior gamma 0.3-0.6 for pediatric extrapolation | working convention; the FDA Bayesian Jan 2026 draft does not prescribe a specific range | Partial borrowing default |\n| Stan R-hat \u003C1.01, ESS >1000 per chain | Vehtari 2021 *Bayesian Analysis* | Posterior convergence criteria |\n| EWOC overdose constraint P(dose > MTD) \u003C= 0.25 | Babb-Rogatko-Zacks 1998 | Safety floor |\n\n## Common Errors\n\n| Error \u002F symptom | Cause | Solution |\n|-----------------|-------|----------|\n| CRM with arbitrary skeleton | No calibration | Lee-Cheung 2009 indifference-interval; or BOIN |\n| MAP without prior-data conflict check | Posterior dominated by prior | Robust MAP; PP-check; sensitivity over mixture weight |\n| EXNEX with single weight scheme | No sensitivity | Weights 0.1, 0.3, 0.5, 0.7, 0.9; report range |\n| Posterior probability stopping without Type-I sim | Regulatory rejection | Simulate under null; calibrate threshold |\n| I-SPY 2 graduated arm reported uncorrected | Selection bias | Conditional MLE; cite Robertson 2023 |\n| Bayesian shrinkage for signal discovery | Hemmings-Koch critique | Shrinkage for replication only |\n| Power prior gamma = 1 | Full pooling | Discount 0.3-0.6 per FDA 2026 draft |\n| WinBUGS without containerisation | Reproducibility | Stan + Docker\u002Frenv-pinned |\n| BOIN vs CRM comparison without simulation OCs | Apples-to-oranges | Compare OCs over same true DLT rates |\n| FDA cited for Bayesian drugs guidance pre-2026 | Confusion | FDA 2010 is DEVICES; FDA 2026 (draft) is drugs |\n\n## Anticipated Reviewer Pushback\n\n| Pushback | Response |\n|----------|----------|\n| \"Type-I error control?\" | Simulation under null demonstrates frequentist Type-I = nominal at threshold chosen; documented in SAP appendix |\n| \"Prior justification?\" | MAP from historical control arms via gMAP; robust mixture weight 0.2 for prior-data conflict; sensitivity over prior provided |\n| \"Why BOIN over CRM?\" | BOIN Fit-for-Purpose qualified Dec 2021; pre-tabulated escalation; no bedside Bayesian software; OCs comparable to CRM in simulation |\n| \"EXNEX weight sensitivity?\" | Reported over weights 0.1, 0.3, 0.5, 0.7, 0.9; results stable; primary at 0.5\u002F0.5 per Neuenschwander 2016 |\n| \"Power prior gamma?\" | Discount 0.5 per FDA Bayesian Jan 2026 draft; sensitivity over 0.3-0.7 provided |\n| \"Posterior probability threshold?\" | Calibrated via simulation to frequentist Type-I 0.025 one-sided; cite Berry 2010 |\n| \"Stan reproducibility?\" | Docker container + renv-pinned R + Stan version; seeds provided; R-hat \u003C1.01, ESS >2000 per parameter |\n| \"Bias correction on platform graduation?\" | Conditional MLE applied to estimate Phase 3 effect; cite Robertson 2023 |\n| \"Why not frequentist instead?\" | Bayesian framework permits borrowing (rare disease, pediatric); working convention; the FDA Bayesian Jan 2026 draft does not prescribe a specific gamma range -- check the draft for the current language before quoting primary inference with simulation calibration |\n\n## References\n\n- Babb J, Rogatko A, Zacks S. 1998. Cancer Phase I clinical trials: efficient dose escalation with overdose control. *Stat Med* 17:1103-1120.\n- Berry SM, Berry DA. 2004. Accounting for multiplicities in assessing drug safety: a three-level hierarchical mixture model. *Biometrics* 60:418-426.\n- Berry SM, Broglio KR, Groshen S, Berry DA. 2013. Bayesian hierarchical modeling of patient subpopulations: efficient designs of Phase II oncology clinical trials. *Clin Trials* 10:720-734.\n- Berry SM, Carlin BP, Lee JJ, Müller P. 2010. *Bayesian Adaptive Methods for Clinical Trials*. CRC.\n- Dane A, Spencer A, Rosenkranz G, Lipkovich I, Parke T. 2019. Subgroup analysis and interpretation for phase 3 confirmatory trials: EFSPI\u002FPSI white paper. *Pharm Stat* 18:126-139.\n- FDA. 2010. Guidance for Industry: Use of Bayesian Statistics in Medical Device Clinical Trials.\n- FDA. 2021. BOIN Drug Development Tool Fit-for-Purpose Qualification.\n- FDA. 2026. Use of Bayesian Methodology in Clinical Trials. Draft Guidance (FDA-2025-D-3217).\n- Guo W, Wang SJ, Yang S, Lynn H, Ji Y. 2017. A Bayesian interval dose-finding design addressing Ockham's razor: mTPI-2. *Contemp Clin Trials* 58:23-33.\n- Hemmings R, Koch A. 2019. Commentary on Dane et al. *Pharm Stat* 18:140-144.\n- Ji Y, Liu P, Li Y, Bekele BN. 2010. A modified toxicity probability interval method for dose-finding trials. *Clin Trials* 7:653-663.\n- Liu S, Yuan Y. 2015. Bayesian optimal interval designs for phase I clinical trials. *JRSS-C* 64:507-523.\n- Neuenschwander B, Wandel S, Roychoudhury S, Bailey S. 2016. Robust exchangeability designs for early phase clinical trials with multiple strata. *Pharm Stat* 15:123-134.\n- O'Quigley J, Pepe M, Fisher L. 1990. Continual reassessment method: a practical design for phase 1 clinical trials in cancer. *Biometrics* 46:33-48.\n- Rugo HS et al. 2016. Adaptive randomization of veliparib-carboplatin treatment in breast cancer. *NEJM* 375:23-34.\n- Robertson DS, Lee KM, López-Kolkovska BC, Villar SS. 2023. Response-adaptive randomization in clinical trials: from myths to practical considerations. *Stat Sci* 38:185-208.\n- Schmidli H, Gsteiger S, Roychoudhury S, O'Hagan A, Spiegelhalter D, Neuenschwander B. 2014. Robust meta-analytic-predictive priors in clinical trials with historical control information. *Biometrics* 70:1023-1032.\n- Spiegelhalter DJ, Freedman LS, Parmar MKB. 1994. Bayesian approaches to randomized trials. *JRSS-A* 157:357-387.\n- Vehtari A et al. 2021. Rank-normalization, folding, and localization: an improved R-hat for assessing convergence. *Bayesian Analysis*.\n- Weber S, Li Y, Seaman J, Kakizume T, Schmidli H. 2021. Applying meta-analytic-predictive priors with the R Bayesian evidence synthesis tools. *J Stat Softw* 100:19.\n\n## Related Skills\n\n- clinical-biostatistics\u002Fadaptive-designs - Group-sequential, SSR, platform trials\n- clinical-biostatistics\u002Fsubgroup-analysis - Bayesian shrinkage for HTE (Dixon-Simon, Berry)\n- clinical-biostatistics\u002Fpower-and-sample-size - Bayesian SS via predictive probability of success\n- clinical-biostatistics\u002Fmultiplicity-graphical - Berry-Berry AE hierarchical\n- clinical-biostatistics\u002Ftrial-reporting - Bayesian inference reporting per CONSORT 2025\n- clinical-biostatistics\u002Fmissing-data-sensitivity - Bayesian rbmi imputation\n- machine-learning\u002Fbiomarker-discovery - Bayesian HTE for biomarker subgroups\n- experimental-design\u002Fsample-size - General methods\n",{"name":5,"source":24,"license":15,"tool_type":25,"description":7,"imported_at":26,"source_path":27,"primary_tool":28,"source_repository":29,"goal_approach_exempt":30},"https:\u002F\u002Fgithub.com\u002FGPTomics\u002FbioSkills","r","2026-09-20","clinical-biostatistics\u002Fbayesian-trials\u002FSKILL.md","RBesT","GPTomics\u002FbioSkills",true,"main","2d1750717747097b602acf30bd09a7cbd4b4c6ea","awesome_skills\u002Fbiology\u002Fbio-clinical-biostatistics-bayesian-trials","https:\u002F\u002Fgithub.com\u002FMr-Nobody-Anonymous\u002FAll-skills","https:\u002F\u002Fgithub.com\u002FMr-Nobody-Anonymous","2026-11-09 11:45:36",{"source":8,"skillName":5,"commands":38,"prompt":48,"zipAvailable":30},[39,42,45],{"tool":40,"command":41},"npx","npx skills add Mr-Nobody-Anonymous\u002FAll-skills --skill bio-clinical-biostatistics-bayesian-trials",{"tool":43,"command":44},"bunx","bunx skills add Mr-Nobody-Anonymous\u002FAll-skills --skill bio-clinical-biostatistics-bayesian-trials",{"tool":46,"command":47},"pnpm","pnpm dlx skills add Mr-Nobody-Anonymous\u002FAll-skills --skill bio-clinical-biostatistics-bayesian-trials","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FMr-Nobody-Anonymous\u002FAll-skills\u002Fblob\u002Fmain\u002Fawesome_skills\u002Fbiology\u002Fbio-clinical-biostatistics-bayesian-trials\u002FSKILL.md",[50,56],{"path":51,"resourceType":52,"resourceTypeDesc":53,"sizeBytes":54,"sourceHtmlUrl":55},"examples\u002Fbayesian_trials.R","OTHER","其他",8787,"https:\u002F\u002Fgithub.com\u002FMr-Nobody-Anonymous\u002FAll-skills\u002Fblob\u002Fmain\u002Fawesome_skills\u002Fbiology\u002Fbio-clinical-biostatistics-bayesian-trials\u002Fexamples\u002Fbayesian_trials.R",{"path":57,"resourceType":58,"resourceTypeDesc":59,"sizeBytes":60,"sourceHtmlUrl":61},"usage-guide.md","REFERENCE","参考文档",6352,"https:\u002F\u002Fgithub.com\u002FMr-Nobody-Anonymous\u002FAll-skills\u002Fblob\u002Fmain\u002Fawesome_skills\u002Fbiology\u002Fbio-clinical-biostatistics-bayesian-trials\u002Fusage-guide.md",[63,71,78,85,92,99],{"skillNo":64,"name":65,"nameZh":6,"description":66,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":67,"categories":68,"keywords":69,"publishedAt":70},"01M7FFFZPC2025JQQHPHDX154J","dei-assessment","Assess diversity, equity, and inclusion with metrics. TRIGGERS - Use when user needs help with dei-assessment related tasks.","https:\u002F\u002Fgithub.com\u002FMr-Nobody-Anonymous\u002FAll-skills\u002Fblob\u002Fmain\u002Fawesome_skills\u002Fhr-people\u002Fdei-assessment\u002FSKILL.md",[],[],"2026-11-16 10:05:34",{"skillNo":72,"name":73,"nameZh":6,"description":74,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":75,"categories":76,"keywords":77,"publishedAt":70},"01M7FFFWN8H4WA7A6Y398QKB6J","dei-action-plan","Create DEI action plans with goals and initiatives. TRIGGERS - Use when user needs help with dei-action-plan related tasks.","https:\u002F\u002Fgithub.com\u002FMr-Nobody-Anonymous\u002FAll-skills\u002Fblob\u002Fmain\u002Fawesome_skills\u002Fhr-people\u002Fdei-action-plan\u002FSKILL.md",[],[],{"skillNo":79,"name":80,"nameZh":6,"description":81,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":82,"categories":83,"keywords":84,"publishedAt":70},"01M7FFFSHTM0QZRVZ2NRJTJ0PY","culture-fit-assessment","Design culture add assessment processes. TRIGGERS - Use when user needs help with culture-fit-assessment related tasks.","https:\u002F\u002Fgithub.com\u002FMr-Nobody-Anonymous\u002FAll-skills\u002Fblob\u002Fmain\u002Fawesome_skills\u002Fhr-people\u002Fculture-fit-assessment\u002FSKILL.md",[],[],{"skillNo":86,"name":87,"nameZh":6,"description":88,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":89,"categories":90,"keywords":91,"publishedAt":70},"01M7FFFMA12RKEQA38KHJSXXPS","cowork-to-code-bridge","Use an already-installed, independently verified cowork-to-code bridge to run narrowly approved actions on the user's own macOS, Linux, or WSL2 machine through a local file queue.","https:\u002F\u002Fgithub.com\u002FMr-Nobody-Anonymous\u002FAll-skills\u002Fblob\u002Fmain\u002Fawesome_skills\u002Fhr-people\u002Fcowork-to-code-bridge\u002FSKILL.md",[],[],{"skillNo":93,"name":94,"nameZh":6,"description":95,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":96,"categories":97,"keywords":98,"publishedAt":70},"01M7FFFH1F6M8YRGW5V487GH1K","counter-offer-response","Draft counter-offer responses for retention. TRIGGERS - Use when user needs help with counter-offer-response related tasks.","https:\u002F\u002Fgithub.com\u002FMr-Nobody-Anonymous\u002FAll-skills\u002Fblob\u002Fmain\u002Fawesome_skills\u002Fhr-people\u002Fcounter-offer-response\u002FSKILL.md",[],[],{"skillNo":100,"name":101,"nameZh":6,"description":102,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":103,"categories":104,"keywords":105,"publishedAt":70},"01M7FFFDGEHV8Y7J9N40HZXW52","continuous-feedback-system","Design continuous feedback systems. TRIGGERS - Use when user needs help with continuous-feedback-system related tasks.","https:\u002F\u002Fgithub.com\u002FMr-Nobody-Anonymous\u002FAll-skills\u002Fblob\u002Fmain\u002Fawesome_skills\u002Fhr-people\u002Fcontinuous-feedback-system\u002FSKILL.md",[],[]]