Flexible Statistical Methods for Early Clinical Trials
View Presentation View Presentation
*Yaning Wang, FDA 

Keywords: estimation, trial design, modelling, exposure-response

The focus of early clinical trials is to learn rather than confirm. Therefore, the statistical methods for these trials should be driven by estimation instead of hypothesis testing. During the review process, various efficient methods have been applied by FDA and recommended to the sponsors for analyzing early trials and planning for later trials. These methods include straightforward responder analysis, Bayesian survival modelling, nonlinear mixed effect pharmacokinetic/pharmacodynamic modelling, mechanistic-based disease modelling, etc. The ultimate goal of these analyses is to learn about the exposure-response to aid in designing pivotal trials.