Non Parametric ROC Summary Statistics

Abstract

Receiver operating characteristic (ROC) curves are useful statistical tools for medical diagnostic testing. It has been proved its capability to assess diagnostic marker’s ability to distinguish between healthy and diseased subjects and to compare different diagnostic markers. In this paper we introduce non parametric ROC summary statistics to assess a ROC curve across the entire range of F P F s ∈ (0, 1) as well as over a restricted range of F P F s and compare them with some existing ones through a simulation study and through some real data examples. We also show their capability to compare two diagnostic markers.