Estimation and Inference on the Volume Under the ROC Surface in a Clinical Region of Interest

• Activity Number: 246 - Improved Disease Classification Through Extensions of ROC Curve Estimation and Biomarker Characterization
• Type: Contributed
• Date/Time: Tuesday, August 9, 2022 : 8:30 AM to 10:20 AM
• Sponsor: Section on Medical Devices and Diagnostics
• Author(s): Kate Young* and Leonidas Bantis
• Companies: University of Kansas Medical Center and University of Kansas Medical Center
• Keywords: ROC Surface; 3-Class Classification; Kernel Density Estimators; Cox model; Box-Cox; Volume Under the Surface

Abstract

Summary measures of biomarker accuracy that employ the receiver operating characteristic (ROC) surface have been proposed for biomarkers that classify patients into one of three groups: healthy, early-stage, or advanced-stage disease. The well-known volume under the ROC surface (VUS) summarizes the overall discriminatory ability of a biomarker in such configurations. However, the VUS includes thresholds associated with clinically irrelevant true classification rates (TCRs). For example, due to the lethal nature of pancreatic cancer, thresholds associated with a low TCR for identifying patients with pancreatic cancer may be undesirable and not appropriate for use in a clinical setting. In this project, we study the properties of a more focused criterion, the partial volume under the ROC surface (pVUS), that summarizes the diagnostic accuracy of a marker in the three-class setting for regions restricted to only those of clinical interest. We propose methods for estimation and inference on the pVUS under parametric and non-parametric frameworks and apply these methods to the evaluation of potential biomarkers for the diagnosis of pancreatic cancer.