A0653
Title: Meta-learners to analyze treatment heterogeneity in survival data
Authors: Ying Ding - University of Pittsburgh (United States) [presenting]
Abstract: An important aspect of precision medicine focuses on characterizing diverse responses to treatment due to unique patient characteristics, also known as heterogeneous treatment effects (HTE), and identifying beneficial subgroups with enhanced treatment effects. Estimating HTE with right-censored data in observational studies remains challenging. A meta-learner-based procedure is proposed with pseudo-outcomes for analyzing HTE in survival data, which includes a pseudo-outcome-based meta-learner framework for estimating HTE, a variable importance metric for identifying predictive variables to HTE, and a data-adaptive procedure to select subgroups with enhanced treatment effects. The proposed procedure is applied to analyze subgroup treatment heterogeneity of a written asthma action plan (WAAP) on time-to-ED (Emergency Department) return due to asthma exacerbation, using a large EHR dataset with visit records expanded from pre- to post-COVID-19 pandemic. Vulnerable subgroups of patients are identified as having poorer asthma outcomes but enhanced benefits from WAAP and characterized patient profiles. The research offers valuable insights for healthcare policymakers and providers in advocating influenza vaccination and strategic WAAP distribution to particularly vulnerable groups during a disruptive public health event, ultimately enhancing pediatric asthma control.
