Accurate likelihood inference on the area under the ROC curve for small sample.

The accuracy of a diagnostic test with continuous-scale results is of high importance in clinical medicine. Receiver operating characteristics (ROC)curves, and in particular the area under the curve (AUC), are widely used to examine the effectiveness of diagnostic markers. Classical likelihood-based inference about the AUC has been widely studied under various parametric assumptions, but it is well-known that it can be inaccurate when the sample size is small, in particular in the presence of unknown parameters. The aim of this paper is to propose and discuss modern higher-order likelihood based procedures to obtain accurate point estimators and confidence intervals for the AUC. The accuracy of the proposed methodology is illustrated by simulation studies. Moreover, two real data examples are used to illustrate the application of the proposed methods.