A new joint model of longitudinal and time-to-event data
Conference
64th ISI World Statistics Congress
Format: IPS Abstract
Keywords: "competing risks"
Session: IPS 137 - Recent advances in modeling time-to-event data and longitudinal data
Monday 17 July 2 p.m. - 3:40 p.m. (Canada/Eastern)
Abstract
Recent discoveries have emphasized the importance of within-subject (WS) visit-to-visit variability of longitudinal biomarkers as significant risk factors for health outcomes. This talk introduces a novel joint model that incorporates a longitudinal biomarker with heterogeneous WS variability and a competing risks time-to-event outcome. The proposed model provides a valuable framework for testing heterogeneity in WS variability, exploring the association between WS variability and survival outcomes, and enabling dynamic prediction of survival by considering both the individual mean and WS variability of the biomarker. We present an expectation-maximization algorithm for semiparametric maximum likelihood estimation, along with a profile-likelihood method for standard error estimation and inference. Moreover, we have developed efficient computational algorithms specifically tailored for analyzing biobank-scale data with tens of thousands of subjects. Through simulation results, we demonstrate the advantages of our method over traditional joint models. To illustrate the practical implications of our approach, we apply it to the Multi-Ethnic Study of Atherosclerosis (MESA) data, yielding intriguing findings.