Activity Number:
|
484
- Applied Bayesian Methodology
|
Type:
|
Contributed
|
Date/Time:
|
Wednesday, August 10, 2022 : 2:00 PM to 3:50 PM
|
Sponsor:
|
International Society for Bayesian Analysis (ISBA)
|
Abstract #323471
|
|
Title:
|
Multilevel Bayesian Joint Modeling for a Longitudinal Marker and Time-to-Recurrent Event to Characterize Heterogeneity in Multi-Center Studies
|
Author(s):
|
Grace Chen Zhou* and Seongho Song and Pedro M. Afonso and Eleni-Rosalina Andrinopoulou and Rhonda Szczesniak
|
Companies:
|
University of Cincinnati and University of Cincinnati and Erasmus Medical Center and Erasmus Medical Center and Cincinnati Children's Hospital Medical Center
|
Keywords:
|
multi-center;
recurrent event;
joint model;
Hamiltonian Monte Carlo;
medical monitoring;
CF Registry data
|
Abstract:
|
Pulmonary exacerbation can occur repeatedly in the same patient living with lung disease. It is of great clinical interest to monitor and predict the probability of next occurrence using available information of a longitudinal marker. Given the fact that lung function has been measured extensively from multi-center cohorts, we propose a joint longitudinal and time-to-recurrent model with center-specific effects under a Bayesian perspective, with the aim to deepen epidemiological understanding of a long-term lung disease progression. Our study explores time-dependent trajectory value and slope as association structures that combine information across a longitudinal marker and recurrent hazards, and account for two types of risk intervals, which are known as calendar time and gap time. The adaptability of our formulation is assessed through a simulation study and a motivating application for cystic fibrosis registry data is provided. All posterior samplings are carried out by Hamiltonian Monte Carlo via a lightweight R interface to Stan.
|
Authors who are presenting talks have a * after their name.