Bayesian variable selection using data driven reversible jump: an application to schizophrenia data

Abstract

Symptom based diagnosis are known to be limited specially concerning complex disorders such as schizophrenia. Modern attempts in providing predictive risk for such disease, to assist existing diagnosis tools, integrate genetic and brain information in what is known as imaging genetics. In this monography, our goal is both inferential and predictive. Regarding the inference, given the functional Magnetic Resonance Image and the Single Nucleotide Polymorphisms information of people diagnosed with schizophrenia and healthy people, we use a Bayesian probit model to select discriminating variables, while to estimate the predictive risk, the most promising models are combined using a Bayesian model averaging scheme. For these purposes, we propose an adaptive reversible jump markov chain monte carlo, named data driven reversible jump, for selecting the variables, estimating their effects and the predictive risk for future subjects.