Title: Comparing ROC curves conditioned to a multidimensional covariate
Authors: Aris Fanjul Hevia - Universidad de Santiago de Compostela (Spain) [presenting]
Wenceslao Gonzalez-Manteiga - University of Santiago de Compostela (Spain)
Juan-Carlos Pardo-Fernandez - Universidade de Vigo (Spain)
Ingrid Van Keilegom - KU Leuven (Belgium)
Abstract: The comparison of Receiver Operating Characteristic (ROC) curves is a commonly accepted way of comparing the discriminatory capability of different diagnostic variables of a certain disease. There are several methodologies discussed in the literature for making that comparison, although most of them do not consider the possible effect that the presence of covariates can have in the performance of the test. A new test is proposed for comparing ROC curves conditioned to the value of a multidimensional covariate. Projections will be used for transforming the problem into a one dimensional approach easier to handle. Some simulations of the new methodology and an illustration on a real dataset application are presented.