Bayesian Dynamic Survival Model with Covariate-Varying Coefficients and Its Applications to Epidemiologic Research

• Activity Number: 36
• Type: Contributed
• Date/Time: Sunday, August 9, 2015 : 2:00 PM to 3:50 PM
• Sponsor: Biometrics Section
• Author(s): Jianghua He*
• Companies: University of Kansas Medical Center
• Keywords: Time-varying coefficient ; survival analysis ; 2-dimensional smoothing ; Body mass index ; mortality

Abstract

For epidemiologic Research, traditional analysis approaches such as the logistic regression model and the Cox proportional model ignore the potential changing relationship over time. Various time-varying coefficients survival models have been proposed in the literature to overcome such limitations. The Bayesian Dynamic Survival Model approaches the issue from the Bayesian perspective and is more flexible in modeling time-varying coefficients. Sometimes, the time-varying coefficients may also change with ontinuous covariates. A Bayesian dynamic survival model with covariate-varying coefficients is proposed to model flexible associations in both the time and the covariate dimensions. The implementation of such a model can be conducted easily based on a transformation of survival data. This Bayesian dynamic survival model is applied to examine the association of body weight and mortality in a Meta analyses based on multiple cohorts with different characteristics.