Use of Bayesian Methods in Making Pharmaceutical Internal Decisions

• Activity Number: 556
• Type: Topic Contributed
• Date/Time: Wednesday, August 1, 2012 : 2:00 PM to 3:50 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Mani Lakshminarayanan*+
• Companies: Merck
• Keywords: Prior information ; Predictive probability ; Historical data ; Adaptive design ; Posterior distribution

Abstract

UResearch and development of pharmaceutical products evolve several stages of design and analysis of trials (involving both animals and humans) that need to be subjected to intermediate and final decisions in the development continuum. In this day and age of getting safe and efficacious pharmaceutical products to patients in dire need, it is imminent that pharmaceutical industries are faced with making timely decisions to keep their product portfolio in par with their competitors. Most of these portfolio specific decisions may require calculating probabilities of specific quantities such as [true treatment effect > pre-specified value| data] or estimating probability of success at Phase-III (with errors or bounds) given the current data (say Phase-II). Use of historical data may also prove to be critical in decision making. Bayesian methods present a natural framework for capturing and updating information that can be used to reduce uncertainty in decision problems. In this presentation, Bayesian methods will be explored in support of making go/no go decisions at various decision points in the drug development process. Examples will be used as illustrations.