Bayesian update and the use of predictive probability in adaptive clinical trials

Abstract

Clinical trial is a prescribed learning process for identifying safe and effective treatments. Bayesian methods naturally take the “learn as we go” approach and are uniquely suitable for such learning. In my talk, I will first illustrate the concept of Bayesian update using simple models such as the beta-binomial model and the normal-inverse Gaussian model. Then, I will introduce the concept of using predictive probability to monitor clinical trial. Predictive probability can be applied in single-arm Phase IIA trials [1] and randomized Phase IIB trials.[2] Lastly, I will illustrate its application in randomized multi-arm screening trials under the platform design. [3] The proposed platform-based approach consolidates inter-study control arms enabling investigators to assign more new patients to novel therapies. The process accommodates mid-trial modifications to the study arms that allow both dropping poorly performing agents as well as incorporating new agents. Adaptive clinical trial designs increase the study efficiency, allow more flexible trial conduct, and treat more patients with more effective treatments in the trial with desirable frequentist properties.