Stabilized Direct Learning for Efficient Estimation of Individualized Treatment Rules

• Activity Number: 291 - Contributed Poster Presentations: Biometrics Section
• Type: Contributed
• Date/Time: Tuesday, August 9, 2022 : 10:30 AM to 12:20 PM
• Sponsor: Biometrics Section
• Author(s): Kushal S Shah* and Haoda Fu and Michael Kosorok
• Companies: University of North Carolina at Chapel Hill and Eli Lilly and Company and University of North Carolina at Chapel Hill
• Keywords: D-Learning; Heteroscedasticity; Individualized treatment rule; Multi-arm treatments; Precision medicine; Statistical machine learning

Abstract

In recent years, precision medicine has seen many advancements. Significant focus has been placed on algorithms to estimate individualized treatment rules (ITR), which map from patient covariates to the space of available treatments with the goal of maximizing patient outcome. Direct Learning (D-Learning), a recent one-step method, estimates the ITR by directly modeling the treatment-covariate interaction. However, when the variance of the outcome is heterogeneous with respect to treatment and covariates, D-Learning does not leverage this structure. We propose Stabilized Direct Learning (SD-Learning), which utilizes heteroscedasticity in the error term through a residual reweighting which models the residual variance via flexible machine learning algorithms such as XGBoost and random forests. We also develop an internal cross-validation scheme which determines the best residual model amongst competing models. SD-Learning improves the efficiency of D-Learning estimates in binary and multi-arm treatment scenarios. The method is simple to implement and an easy way to improve existing algorithms within the D-Learning family.