Statistics Ph.D. Dissertation Defense - Ciara Nugent

This 2024 Dissertation Defense will be held on Friday, July 26, from 8:30 a.m. to 10:30 a.m. with Ciara Nugent. This event will be virtual. If you need the Zoom link, please email stat.admin@austin.utexas.edu.
 

Title: A Decision Theoretic Approach to Combining Inference Across Data Sources with Applications to Subgroup Analysis in Clinical Trials

Advisor: Peter Mueller

Abstract: Many research questions can not be answered by a single scientific study or source of data.  To address this challenge, researchers develop principled ways to combine information from multiple studies to reach a conclusion. A very common instance of this problem is the combination of results across multiple clinical studies, commonly known as meta-analysis, usually with the understanding that results from individual studies are only available as published summary statistics. In this thesis I consider principled Bayesian decision theoretic approaches to the general problem of combining inference across multiple studies. After a discussion of the general problem I focus on the particular example that arises when the desired inference is subgroup analysis, that is, inference about how results for subpopulations of interest differ from the results of the larger study population.  In the last chapter I consider an extension to a typical meta-analysis problem. Methodologically, the proposed solutions start with a decision theoretic approach to combining inference from two studies through the use of a utility function. Building on this foundation I then consider a more complex utility function for Bayesian population finding. Finally, expanding upon the same utility function I then consider more than two studies to propose a similar solution for a general meta-analysis problem.