
Title: A Bayesian Sequential Design for Clinical Trials
Abstract: This talk presents a comprehensive overview of a Bayesian sequential design for phase III clinical trials, highlighting its advantages in ethical efficiency, statistical power, and resource allocation. It covers three key Bayesian frameworks: BSDAR for continuous outcomes, BSD4TEO for time-to-event outcomes, and BSDEF for exponential family outcomes. Each design incorporates adaptive randomization and alpha spending functions to control the overall Type I error rate while allowing for early stopping due to efficacy or futility. The BSDAR method is illustrated through a diabetic clinical trial, demonstrating reduced sample size and improved allocation efficiency. BSD4TEO is applied to both simulated data and a real-world CGD dataset, showcasing its performance in survival analysis using Bayes factors for interim decision-making. BSDEF is evaluated through simulation studies for count outcomes, comparing its power with that of frequentist group sequential designs. Overall, the Bayesian sequential design offers greater efficiency by allowing early trial termination, while the use of alpha spending functions helps maintain control over the type I error rate.