CFE 2019: Start Registration
View Submission - CMStatistics
Title: Robust inference for the covariate-specific ROC curve and its associated summary indices Authors:  Vanda Lourenco - Faculty of Sciences and Technology - New University of Lisbon (Portugal) [presenting]
Miguel de Carvalho - CEAUL (Centro de Estatistica e Aplicacoes), Universidade de Lisboa (Portugal)
Vanda Inacio - University of Edinburgh (United Kingdom)
Abstract: Accurate diagnosis of disease is of critical importance in health care and medical research. The receiver operating characteristic (ROC) curve is the most popular tool for evaluating the discriminatory ability of continuous biomarkers. In practice, the performance of a test/biomarker can depend on covariates (e.g., age and/or gender). In order to take covariate information into account, the covariate-specific ROC curve has been proposed as a way of evaluating how the accuracy of the biomarker changes as a function of such covariates. We develop a robust and flexible model for conducting inference about the covariate-specific ROC curve and its associated covariate-specific summary indices, that safeguards against atypical biomarker observations while accommodating for nonlinear covariates effects. Specifically, we postulate a location-scale regression model for the test outcomes in each group, combining additive B-splines regression and M-estimation for the mean function with the residuals being estimated via a weighted empirical distribution function. Simulation results show that our approach successfully recovers the true covariate-specific ROC curve and corresponding summary indices on a variety of data contamination scenarios. The adequacy of the method is further illustrated using data on age-specific accuracy of glucose as a biomarker of diabetes.