Comparing empirical ROC curves using a Java application: CERCUS

Abstract

Receiver Operating Characteristic (ROC) analysis is a methodology that has gained much popularity in our days, especially in Medicine, since through the ROC curves, it provides a useful tool to evaluate and specify problems in the performance of a diagnostic indicator. The area under empirical ROC curve (AUC) it’s an indicator that can be used to compare two or more ROC curves. This work arose from the necessity of the existence of software that allows the calculation of the necessary measures to compare systems based on ROC curves. Several software, commercial and non-commercial, are available to perform the calculation of the measures associated to the ROC analysis. However, they present some flaws, especially when there is a need to compare independent samples with different dimensions, or also to compare two ROC curves that intersect. In this paper is presented a new application called CERCUS (Comparison of Empirical ROC Curves). This was developed using a programming language (Java) and stands out for the possibility of comparing two or more ROC curves that cross each other. The main objective of CERCUS is the calculation of several ROC estimates using different methods and make the ROC curves comparison, even if there is an intersection, either for independent or paired samples. It also allows the graph representation of the ROC curve in a unitary plan as well the graph of the area between curves in comparison. This paper presents the program’s versatility in data entry, test menus and visualization of graphs and results.