|
Activity Number:
|
570
- Joint Modeling of Longitudinal and Survival Data
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Wednesday, July 31, 2019 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Biometrics Section
|
|
Abstract #307037
|
Presentation
|
|
Title:
|
Joint Modeling of Multivariate Longitudinal Outcomes and Multiple Time-To-Events in Presence of Informative Censoring
|
|
Author(s):
|
Md Akhtar Hossain* and Alexander C McLain and Hrishikesh Chakraborty
|
|
Companies:
|
University of South Carolina and University of South Carolina and Duke Clinical Research Institute, Duke University
|
|
Keywords:
|
Longitudinal data;
Multivariate joint model;
Informative censoring;
Terminal event;
HIV/AIDS
|
|
Abstract:
|
Joint modeling of longitudinal and time-to-event data has been studied extensively in the past decade. We present a joint model for multivariate longitudinal outcomes and multiple time-to-events in presence of informative censoring by a terminal event. We modeled the longitudinal outcomes using a multivariate partially linear mixed effects model with skew-normal errors. The time-to-events are modeled using Cox proportional hazard models with Gaussian frailties to account for their associations. For estimating the model parameters, we proposed a Bayesian approach to jointly model the longitudinal and time-to-event processes linked through shared random effects. We also present a Bayesian framework for dynamic predictions of future longitudinal outcome trajectories and time-to-events risks. The proposed model and dynamic prediction methods are evaluated using statistical simulations. An application of the proposed model and dynamic prediction methods is demonstrated using the South Carolina HIV/AIDS surveillance data. We jointly modeled the longitudinally observed viral loads, CD4 counts, and HIV/AIDS co-morbidities with death as the terminal and informative censoring event.
|
Authors who are presenting talks have a * after their name.
