Primary Submission Category: Machine Learning and Causal Inference
Automated, Efficient, and Model-Free Covariate Adjustment under Stratified Randomization
Authors: Raphael Kim, Michele Santacatterina, Ramin Zabih, Ivan Diaz,
Presenting Author: Raphael Kim*
Covariate adjustment and stratified randomization have shown to improve precision and power in clinical trials. Recently, methods have been
proposed to provide valid asymptotics for covariate adjustment using M-estimation under stratified randomization. However, leveraging efficiency gains with these methods require pre-specification of a small set of the covariates that are most predictive of the outcome, which is difficult in practice since most trials measure dozens of baseline covariates that are predictive of the outcome. In this work, we build on existing literature to propose an automatic, efficient, and model-free covariate adjustment method which permits data-driven covariate selection. Through extensive simulations and analysis, we showcase the simplicity and improved precision of our method when the covariate set is not known a priori.