Zhi Yuh Ou Yang

Title: Joint modelling of longitudinal  and survival data
Date: Monday, July 15th, 2024
Time: 9:30am
Location: LIB 2020 & Zoom
Supervised by: Dr. Rachel Altman

Abstract: Joint models for time-to-event and longitudinal data characterize the association between features of a longitudinal process and the time to an event. For this project, we conduct a comprehensive simulation study that examines the performance of estimators of the association parameter in the joint model under different data-generating mechanisms. We consider four estimators: a two-stage estimator that assumes a Weibull baseline hazard function, a two-stage estimator that makes minimal assumptions on the baseline hazard function, a maximum likelihood estimator that assumes a Weibull baseline hazard function, and a maximum likelihood estimator that makes minimal assumptions on the baseline hazard function. We evaluate the convergence rates of the estimation methods, the biases of the estimators of the association parameter, and the coverage probabilities of confidence intervals for the association parameter. We then apply our findings to the analysis of the association between the time till resumption of ovulation following childbirth and adiponectin (an energy marker) in a small sample of Indigenous Guatemalan women.