Contrast-Based Optimal Treatment Rule Estimation with Composite Survival Data

• Activity Number: 38 - Recent Advances in Adaptive Treatment Strategy Estimation
• Type: Invited
• Date/Time: Sunday, August 7, 2022 : 4:00 PM to 5:50 PM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Ai Ni* and Xiaohan Guo
• Companies: The Ohio State University and The Ohio State University
• Keywords: composite survival outcome; contrast function; Individualized treatment rule; outcome weighted learning; personalized medicine; win ratio

Abstract

An optimal treatment rule (OTR) assigns an optimal treatment to subjects based on their personal characteristics. An archetypal OTR estimation approach is the outcome weighted learning (OWL), which is based on a weighted classification framework. OWL has been developed for time to a single event of interest. In biomedical research, time to a composite of multiple event types is often of interest. In this study, we propose a contrast-weighted learning (CWL) method to estimate OTR with composite survival outcomes. CWL naturally incorporates priorities of multiple event types into OTR estimation. When the contrast is win indicator, the weight in the CWL coincides with the Proportion in Favor of Treatment in the win ratio literature. We establish theoretical properties of the estimated OTR including Fisher consistency. We conduct simulations to evaluate the finite sample performance of CWL in comparison to the time-to-first-event analysis. We apply the proposed CWL method to an EMR dataset to estimate an OTR to assign either direct oral anticoagulants or warfarin to atrial fibrillation patients to maximize their time to a composite of cardiovascular events.