“Model Selection and Combination for Estimating Heterogeneous Treatment Effects“
Abstract: Researchers often believe that a treatment’s effect on a response may be heterogeneous with respect to certain baseline covariates. This is an important premise of personalized medicine. Within a given set of regression models or procedures, those that best estimate the regression function may not be best for estimating the conditional effect of a treatment given covariates; therefore, there is a need for methods of model selection and combination targeted to treatment effect estimation. We propose a modification of cross-validation that targets estimation of the conditional treatment effect. This modified cross-validation can be used to select one of the candidate models or to properly weight the models for combination. This talk will outline a statistical framework for conditional treatment effect estimation, describe our methods and their theoretical properties, illustrate the methods using data from a clinical trial, and discuss future research directions.
For inquiries contact Chengjie Xiong.