Compare the Discriminatory Ability of Biomarkers Subject to LOD Using Semiparametric Transformation Model

• Activity Number: 363
• Type: Contributed
• Date/Time: Tuesday, August 11, 2015 : 10:30 AM to 12:20 PM
• Sponsor: Biometrics Section
• Author(s): Lixuan Yin* and Guoqing Diao and Aiyi Liu
• Companies: George Mason University and George Mason University and NICHD/NIH
• Keywords: ROC curve ; semiparametric transformation model ; limit of detection

Abstract

Receiver operating characteristic (ROC) curve is a commonly used tool for evaluating the performance of diagnostic biomarkers, whose measures are often subject to a limit of detection (LOD). Standard approaches assume joint normality when comparing the discriminatory ability of correlated biomarkers by testing the equality of the areas under the ROC curves (AUC). Violation of the distribution assumption may lead to biased estimators of AUCs and incorrect type I error rates and powers. We consider a semiparametric transformation model allowing for heteroscedasticity, under which a biomarker is assumed to be normally distributed after an unknown transformation. We develop efficient likelihood-based procedures for estimating the unknown parameters and performing hypothesis testing. The proposed estimators are shown to be consistent, asymptotically normal, and asymptotically efficient. The proposed test accurately controls the type I error rates and is more powerful than its parametric counterpart. Extensive simulation studies and an application to a real data set are provided.