Electronic Theses and Dissertations

Yilun Sun

Date

2023

Document Type

Dissertation

Degree Name

Doctor of Philosophy

Department

Mathematical Sciences

Committee Chair

EBENEZER GEORGE

Committee Member

Motomi Mori

Committee Member

Li Tang

Committee Member

Majid Noroozi

Abstract

Precision medicine has emerged as a promising approach to advancing global healthcare. However, the traditional clinical trial paradigm requires novel adaptations to adjust to challenges posed by advances in precision medicine. To address this issue, master protocol designs, such as basket trials, have been created to improve flexibility and efficiency in clinical trials. These novel designs often involve multiple disease subgroups and frequently suffer from a limited number of participants. To overcome this challenge, it is necessary to borrow information across patient subgroups to estimate response rates within subgroups in clinical trials. Bayesian methods are among the most commonly recommended approaches for facilitating borrowing strategies. However, existing Bayesian methods often ignore variation in subgroup sizes and the heterogeneity between subgroups caused by multiple factors. In this dissertation, we propose a Bayesian procedure for analyzing concurrent response rates data. Our approach clusters subgroups based on outcomes, conducts flexible information borrowing based on subgroup similarities, accounts for multiple factors, and filters out the contribution of subgroups with small sizes.

Comments

Data is provided by the student.

Library Comment

Dissertation or thesis originally submitted to ProQuest

Notes

Open Access

Download

Share

COinS