Bayesian separate and joint modeling for controlled clinical trial data using BUGS.

Venkatesan, P.; Ponnuraja, C.

Applied Bayesian Statistical Analysis. Proceedings of NSABSA, 2008;132-138.

Abstract: Many clinical trials and other medical studies generate both longitudinal (repeated measurements) and survival (time to event) data. The existing methods are inappropriate when the longitudinal variable is correlated. Earlier articles proposed a joint model for longitudinal and survival data, obtaining maximum likelihood estimates via the EM algorithm based on Bayesian approach implementing via Markov Chain Monte Carlo (MCMC) methods. The longitudinal and survival responses are assumed independent given a linking latent bivariate Gaussian process and available covariates. We use the approach to jointly model the longitudinal and survival data from a clinical trial comparing treatments and also its interactions. The joint Bayesian approach appears to offer significantly improved and enhanced estimation of survival times and other parameters of interest like gender, age and weight. In spite of the complexity the model, we find it to be relatively straight forward to implement and understand using the WinBUGS software.

Keywords: Bayesian Approach; Joint model; Markov chain Monte Carlo (MCMC); WinBUGS

Back to List of publications / Home