Novel tests for evaluating two ROC curves under paired samples.

Disease prevention is important and can be accomplished by developing diagnostic tests. The receiver operating characteristic (ROC) curve and the area under the ROC curve (AUC) are used to assess the accuracy of diagnostic tests. The assessment for the superiority between evaluating two diagnostic tests is needed when comparing two diagnostic tests. Existing tests are constructed by comparing two AUCs under the paired samples. Nevertheless, it is problematic when two ROC curves are crossing. This article proposes a test that takes into account the possible correlation between pairs. Simulations are conducted to evaluate the feasibility of the test.