A BAYESIAN-APPROACH TO A GENERAL REGRESSION-MODEL FOR ROC CURVES

A fully Bayesian approach to a general nonlinear ordinal regression mo del for ROC-curve analysis is presented. Samples from the marginal pos terior distributions of the model parameters are obtained by a Markov- chain Monte Carlo (MCMC) technique-Gibbs sampling. These samples facil itate the calculation of point estimates and credible regions as well as inferences for the associated areas under the ROC curves. The analy sis of an example using freely available software shows that the use o f noninformative vague prior distributions for all model parameters yi elds posterior summary statistics very similar to the conventional max imum-likelihood estimates. Clinically important advantages of this Bay esian approach are: the possible inclusion of prior knowledge and beli efs into the ROC analysis (via the prior distributions), the possible calculation of the posterior predictive distribution of a future patie nt outcome, and the potential to address questions such as: ''What is the probability that a certain diagnostic test is better in one settin g than in another?''.