Receiver operating characteristic (ROC) curve for the Bayes linear model

Copyright information:Taken from "A multivariate Bayesian model for assessing morbidity after coronary artery surgery"Critical Care 2006;10(3):R94-R94.Published online 17 Jul 2006PMCID:PMC1550964. The ROC curve (bold continuous line) and its 95% confidence interval (fine continuous lines) were derived from the training set by a maximum-likelihood estimation procedure by assuming binormal distribution of the data. The corresponding area under the ROC curve was 0.80 (with a 95% confidence interval ranging from 0.75 to 0.83). The cross indicates the point where SE and SP are equal. The dashed line represents the empirical ROC curve obtained from testing data. FPF, false-positive fraction; SE, sensitivity; SP, specificity; TPF, true-positive fraction.