Date of Award

2022

Degree Type

Thesis

Degree Name

Master of Science in Statistics

Department

Computer Science and Statistics

First Advisor

Yichi Zhang

Abstract

A treatment regimen is a decision rule that can be used to guide clinicians in connecting patients' information to the feasible treatment strategy. Estimating optimal treatment regimens has gained increasing attention in the statistics field. New models based on machine learning approaches, such as outcome weighted learning (OWL) and entropy learning (EL), were suggested to overcome the computational challenge of the doubly robust estimation due to nonconvexity and non-differentiability. However, substituting the 0-1 loss by a smooth loss leads to the lack of estimation consistency within the class under consideration. To solve this problem, concordance-assisted learning (CAL) was then proposed to estimate optimal treatment regime through pairwise comparison. However, the concordance function is discontinuous, and the optimization is computationally expensive. In this study, we propose a new method which estimates parameters using a weighted logistic regression based on the pairwise comparison between any two individuals. We obtain the standard errors of the estimated coefficients using bootstrap approach. Simulation results under different scenarios and application to the ACTG175 data both demonstrate that the proposed pairwise comparison learning (PCL) method can estimate optimal treatment regimens successfully and outperforms existing approaches.

Creative Commons License

Creative Commons Attribution 4.0 License
This work is licensed under a Creative Commons Attribution 4.0 License.

Share

COinS
 
 

To view the content in your browser, please download Adobe Reader or, alternately,
you may Download the file to your hard drive.

NOTE: The latest versions of Adobe Reader do not support viewing PDF files within Firefox on Mac OS and if you are using a modern (Intel) Mac, there is no official plugin for viewing PDF files within the browser window.