Regression Tree Methods for Precision Medicine

Abstract

In the fight against hard-to-treat diseases, such as cancer, it is often difficult to discover new treatments that benefit all subjects. For regulatory agency approval of new drugs, it is important to identify subgroups of subjects for whom the treatment has an enhanced effect. Regression trees are natural for this task because they partition the data space using patient biomarker and other characteristics. Two methods based on the GUIDE algorithm are proposed. Both are practically free of selection bias. They are applicable to randomized trials with two or more treatment arms, censored, multiple and longitudinal responses, and predictor variables with missing values. A bootstrap technique is used to construct confidence intervals of the treatment effects for post-selection inference.