|
Abstract:
|
Advances in science and technology are providing unprecedented opportunities for new approaches to disease diagnosis and treatment. However, this has had little impact on the attrition and cost of drug development; and in some sense, the barriers to successful development have never been higher. This highlights the need to capitalize on more efficient statistical strategies and clinical trial designs that leverage all available data. One statistical strategy is Bayesian methods which use the concept of prior information - a rational approach given that subsequent studies are designed on the basis of data and treatment response distribution from previous trial - to improve precision of treatment response estimates in subsequent trials. Its use could translate to potentially smaller sample size, increased probability of selecting the right dose/patient population, and/or the ability to terminate trials early for efficacy or futility. In this talk, we will re-examine two trial examples, one that succeeded and one that failed to illustrate decisions that the Bayesian methodology could have improved. Simulation results and interpretations from the 2 case examples will be discussed.
|
