Degree
Master of Science
Program
Biostatistics
Supervisor
Yun-Hee Choi
Abstract
The purpose of this study was to compare shared frailty model, joint frailty model and joint nested frailty model in terms of model fitting and prediction accuracy, as applied to Lynch Syndrome family data. The specific question we wanted to address was how the intervals between screening visits affect the risk of developing colorectal cancer among Lynch Syndrome family members. We also addressed questions on how the screening process has an effect on mortality and risks of developing different stages of colorectal cancer. Results from the models show that joint nested frailty model is preferable. This model improves the prediction accuracy by jointly modeling recurrent screenings and terminal event at the same time accounts for both individual and familial correlation.
Recommended Citation
Yu, Bing, "A comparison of three prediction modelling approaches for clustered survival data with application to Lynch Syndrome Family" (2017). Electronic Thesis and Dissertation Repository. 5145.
https://ir.lib.uwo.ca/etd/5145