Utilização de técnicas bayesianas em modelos de regressão de Poisson para dados de contagem longitudinais e dados de contagem com medidas repetidas apresentando excesso de zeros

Abstract

In medical and biological researches we often .nd count data. For longitudinal count data, usual Poisson regression models, assuming independence among observations, are not applicable because of the correlation of these measures. This work presents hierarchical Bayesian models considering random e¤ects to analyze longitudinal count data. A Normal and a Gamma distribution are considered to these e¤ects besides the mixture of Normal distributions. We also present zero in.ated Poisson (ZIP) regression models for repeated measures. Markov Chain Monte Carlo (MCMC) is used to estimate the parameters. Keywords: Longitudinal Count Data; Poisson Regression Model; Zero In.ated Model; Hierarchical Model; Bayesian Analysis; MCMC Methods.