Graduation Year

2025

Document Type

Thesis

Degree

M.S.C.S.

Degree Name

MS in Computer Science (M.S.C.S.)

Degree Granting Department

Computer Science and Engineering

Major Professor

John Templeton, Ph.D.

Committee Member

Shaun Canavan, Ph.D.

Committee Member

Sengbae Kim, Ph.D.

Committee Member

Adrian Kohut, Ph.D.

Keywords

Graph Attention Networks, View Correlation Discovery Network, Similarity Networks, Data Preprocessing, Multi-Modal Learning

Abstract

The prediction of uterine cancer recurrence is very important for assisting women in reducing the cancer risks and also for the growing field of personalized medicine. The primary aim of this thesis is to investigate the integration of various omics data alongside clinical and therapeutic information to predict survival in uterine cancer. The combination is very important for understanding the risk factors, including clinical aspects, genetics, and the treatment schedule, in order to prescribe the appropriate way to reduce the risk of recurrence, make clinical interactions easier, and enhance personalized patient care. This study utilizes the publicly accessible TCGA dataset, which includes multiple types of omics data, including mRNA, DNA Methylation, and Copy Number Alterations, alongside clinical data such as patient demographics, treatment details, and surgical information. This study involves employing a modern Graph Attention Networks model for generating embeddings, which were subsequently input into an ensemble classifier comprising MLP, SVM, Random Forest, and XGBoost classifiers. The performance of various classifiers and the ensemble model is analyzed to demonstrate the efficiency of each model using metrics like accuracy, AUC score, F-1 score, precision, recall, sensitivity and specificity. This strategy yielded a significant improvement in the accuracy of predicting uterine cancer recurrence.

Download

Share

COinS