Abstract:
|
A major challenge in developing precision medicines is the identification and confirmation of patient subgroups where an investigational regimen has a positive benefit-risk balance. In biopharmaceutical development, exploring these patient subgroups of potential interest are usually achieved by constructing decision rules (signature) using single or multiple biomarkers in a data driven fashion, accompanied by rigorous statistical performance evaluation to account for potential overfitting issues inherent in subgroup searching. In this presentation, we will discuss general considerations in exploratory subgroup analysis, investigate statistical methods for subgroup identification, and propose statistical principles for the subgroup performance assessment. An example of subgroup identification for an immunology disease treatment leading to regulatory label inclusion will also be presented.
|