All Times EDT
Keywords: assurance, conditional power, predictive power, promising zone
Bayesian predictive probabilities have become a ubiquitous tool for design and monitoring of clinical trials. The typical procedure is to average predictive probabilities over the prior or posterior distributions. In this poster, we highlight the limitations of relying solely on averaging, and propose the reporting of intervals or quantiles for the predictive probabilities. These intervals formalize the intuition that uncertainty decreases with more information. To demonstrate the practicality and generality of the proposed approach, we apply this method to a range of applications: - Phase 1 dose escalation in oncology - Early stopping for futility - Adaptive sample size re-estimation / Probability of Success - Dose response model selection and model averaging.