Advanced biomarker analysis for early Alzheimer's detection: a 3-class classification approach

Abstract: The receiver operating characteristic (ROC) curve is an important tool for the discrimination of two populations. However, in many settings, the diagnostic decision is not limited to a binary choice. ROC surfaces are considered as a natural generalization of ROC curves in three-class diagnostic problems and the Volume Under the ROC Surface (VUS) was proposed as an index for the assessment of the diagnostic accuracy of the marker under consideration. In this paper, we propose an overlap measure (OVL) in the case of three-class diagnostic problems. Specifically, parametric and non-parametric approaches for the estimation of OVL are introduced. We evaluate this measure through simulations and compare it with the well-known measure given by VUS. Furthermore, our proposal is applied to the clinical diagnosis of early stage Alzheimer's disease.