Modelos de sobrevivência bivariados baseados na cópula PVF

Abstract

An alternative developed to study associations among multivariate survival times is the use of models based on copula functions. In this work, we use the survival model derived from the PVF copula, based on the Power Variance Function distribution, to model the dependence of bivariate data in the presence of covariates and censored observations. For inferential purposes, we perform a Bayesian approach using Monte Carlo Markov Chain (MCMC) methods. Some discussions about model selection criteria are presented. In order to detect influential observations, we used the Bayesian method of deletion influence analysis of cases based on divergence psi. Finally, we show the applicability of the proposed models to simulated and real datasets.