Statistics Ph.D. Dissertation Defense - Rebecca Knowlton

This 2025 Dissertation Defense will be held on Thursday, April 3 from 2:00 p.m. to 4:00 p.m. with Rebecca Knowlton. This event will be hybrid. If you are able to attend in person, it will be held in WEL 5.204. If you need the Zoom link, please email stat.admin@austin.utexas.edu.

Title: Heterogeneous Surrogate Markers in Clinical Trials and Real-World Settings

Advisor: Dr. Layla Parast

Abstract: Surrogate markers are often used in clinical trials to replace primary outcomes for earlier, less costly, or less burdensome evaluations of treatment effectiveness. However, traditional methods for validating surrogate markers often assume homogeneity in surrogate strength across all individuals, which may not hold in practice. To address this limitation, this dissertation develops novel frameworks for assessing and leveraging heterogeneity in surrogate strength. First, we introduce a framework to assess how the proportion of treatment effect explained by a surrogate varies as a function of baseline covariates, using parametric and semiparametric estimation procedures, and we include a formal test for heterogeneity and subgroup identification procedures. Then, we present Efficient Testing using Surrogate Information (ETSI), a nonparametric method for testing treatment effects when the surrogate is valid only for certain subgroups, with tools to optimize future study design. Finally, we extend the evaluation of surrogate strength to non-randomized data, using meta-learners to handle confounding and identify individuals for whom the surrogate is a valid replacement. Throughout, we demonstrate the performance of the proposed methodology via simulation studies and applications to real health datasets. These contributions provide a comprehensive set of tools for evaluating heterogeneous surrogacy, broadening the scope of surrogate marker validation.