Bayesian Evidence Synthesis and Simulations for Design and Decision-Making

• Activity Number: 677
• Type: Topic Contributed
• Date/Time: Thursday, August 13, 2015 : 10:30 AM to 12:20 PM
• Sponsor: Section on Bayesian Statistical Science
• Author(s): Forrest Williamson*
• Companies: Eli Lilly and Company
• Keywords: Bayesian ; clinical trial simulation ; predictive probability

Abstract

Efficient drug development requires designs and decisions to be informed by the totality emerging evidence around the drug in question. Evidence to support key hypotheses is derived from many sources (e.g. external literature, internal trial results, and other sources). Bayesian evidence synthesis leverages these data sources enabling the design of more robust trials and more informed decision making.   This talk focuses on designing and assessing the risk of clinical trials using results of Bayesian evidence synthesis. Bayesian predictive simulation methods are described to calculate the probability that a trial will succeed, along with other metrics (e.g. probability of selecting a dose or population) relevant to assessing how well the trial is expected to meet its goals. Methods for predicting the operating characteristics of the trial (e.g. false positive/true positive) are described and evaluated in a Bayesian framework. Application of the methodology is done in two examples, one for an exploratory trial and one for a confirmatory trial. These examples illustrate how Bayesian methods can improve our designs and decision making during the drug development cycle.