Model checking for ROC regression analysis.

摘要: The receiver operating characteristic (ROC) curve is a prominent tool for characterizing the accuracy of continuous diagnostic test. To account factors that might influence test accuracy, various ROC regression methods have been proposed. However, as in any analysis, when assumed models do not fit data well, these may render invalid and misleading results. date, practical model-checking techniques suitable validating existing are yet available. In this article, we develop cumulative residual-based procedures to graphically numerically assess goodness some commonly used models, show how specific components can be examined within framework. We derive asymptotic null distributions residual processes discuss resampling approximate practice. illustrate our with dataset from cystic fibrosis registry.