Zetyra: A Validated Suite of Statistical Calculators for Efficient Clinical Trial Design

Modern clinical trial design methods—such as covariate adjustment, group sequential monitoring, Bayesian predictive probability, adaptive sample size re-estimation, response-adaptive randomization, and master protocol designs—can substantially improve trial efficiency. However, adoption remains limited due to fragmented software ecosystems, opaque validation practices, and high licensing costs. Zetyra is an integrated web-based platform providing 18 validated statistical calculators for clinical trial design across four tiers (Free, Evidence Pro, Bayesian Toolkit add-on, Enterprise). Frequentist methods CUPED variance reduction for baseline-adjusted power and sample size calculations Group sequential design with O'Brien–Fleming, Pocock, and alpha-spending boundaries Survival power analysis using Schoenfeld and Freedman formulations Blinded sample size re-estimation (Kieser–Friede) Unblinded sample size re-estimation using promising-zone methodology (Mehta–Pocock) Single-arm sample size re-estimation for binary endpoints Response-adaptive randomization (DBCD, Thompson Sampling, Neyman, Pocock–Simon) Master protocol designs: basket (BHM/EXNEX), umbrella, and platform (MAMS) Bayesian methods Predictive probability of success (PPoS) for interim go/no-go decisions Prior elicitation tools Bayesian borrowing from external data (power priors, commensurate priors) Single-arm and two-arm Bayesian trial design Bayesian sequential monitoring Validation framework The platform was validated through 896 automated tests across 42 validation scripts, benchmarked against established statistical software (gsDesign, rpact), analytical solutions, Zhou & Ji (2024), and replications of published clinical trials (PACIFIC, MONALEESA-7, HPTN 083, I-SPY 2, STAMPEDE, REMAP-CAP). The maximum observed numerical deviation was 0.034 in z-score scale for group sequential boundary calculations. All validation code and test cases are publicly available under an MIT license. Illustrative applications Example case studies demonstrate potential design efficiencies, including: ~30% sample size reduction through covariate adjustment (ρ ≈ 0.55) ~24% cost reduction via early stopping under group sequential monitoring Improved Phase II decision-making using Bayesian predictive probability More patients allocated to better-performing arms via response-adaptive randomization Shared infrastructure savings through master protocol designs (basket, umbrella, platform) Changes in v2.3 (May 2026) relative to v2.0 (March 2026): Expanded from 13 to 18 calculators: added single-arm sample size re-estimation, response-adaptive randomization, and master protocol designs (basket, umbrella, platform) Validation suite expanded from 499 tests across 25 scripts to 896 tests across 42 scripts Consolidated main white paper and Bayesian Toolkit white paper into a single 132-page document Updated conditional power SSR citation (Qian, 2026a) to reflect peer review status at Statistics in Biopharmaceutical Research Added FDA January 2026 draft Bayesian guidance context throughout This release includes: A 132-page technical white paper describing the statistical methods and validation framework A 4-page executive brief summarizing the platform and use cases