A Simulation Study for a Novel Joint Model with Integrated Fractional Brownian Motion

• Activity Number: 87 - Survival and Longitudinal/Clustered Data Analysis
• Type: Contributed
• Date/Time: Monday, August 9, 2021 : 10:00 AM to 11:50 AM
• Sponsor: Biometrics Section
• Author(s): Anushka Kumarihamy Palipana* and Seongho Song and Nishant Gupta and Rhonda Szczesniak
• Companies: University of Cincinnati and University of Cincinnati and University of Cincinnati and Cincinnati Children's Hospital Medical Center
• Keywords: Joint modeling; Cox proportional hazards model

Abstract

Joint models can be used to study a longitudinal biomarker process, the effect of covariates on the biomarker process, the effect of covariates, and the biomarker process on the terminal/recurrent event process. Many studies have used the random-effects model to study the biomarker progress over time. However, the high variability arising from biological processes cannot be effectively explained by using a simple random-effects model. Our novel approach uses a linear mixed-effects model with an integrated fractional Brownian motion (IFBM) process to capture within-subject variation, and allowing the model to follow a multivariate Normal distribution. We use a Cox proportional hazards model for the survival sub-model. The two sub-models are joined through a shared parameter, which depicts the association of the biomarker process with the event process. Bayesian methods are used to fit the model. We perform a simulation study of our joint model with scaled integrated Brownian motion and a joint model without a stochastic process.