Abstract:
|
Subjective Prior Elicitation in drug development is growing in popularity in both simulation work and in Bayesian models. This approach allows us to use expert information on quantities of interest when data is not present or gaps exist in the current models. In this talk, we review practical issues that have arisen when we have elicited priors, including the following: (1) what type of problems can subjective priors be used in and what should their role be; (2) who should be interviewed and how many experts are needed. Different modes of performing the elicitations and the role of technology; and, (3) how to assess the informativeness of a prior and its impact on any further modeling. We conclude with the pros and cons of using elicited priors.
|