Bayesian modeling of Dupuytren disease using copula Gaussian graphical models

Dupuytren disease is a fibroproliferative disorder with unknown etiology that often progresses and eventually can cause permanent contractures of the affected fingers. Most of the researches on severity of the disease and the phenotype of this disease are observational studies without concrete statistical analyses. There is a lack of multivariate analysis for the disease taking into account potential risk factors. In this paper, we provide a novel Bayesian framework to discover potential risk factors and which fingers are jointly affected. Copula Gaussian graphical modeling is one potential way to discover the underlying conditional independence of variables in mixed data. Our Bayesian approach is based on copula Gaussian graphical models. We embed a graph selection procedure inside a semiparametric Gaussian copula. We carry out the posterior inference by using an efficient sampling scheme which is a trans-dimensional MCMC approach based on birth-death process. We implemented the method as a general purpose in the R package BDgraph.