Each week, we will be publishing layman’s abstracts of new articles from our prestigious portfolio of journals in statistics. The aim is to highlight the latest research to a broader audience in an accessible format.
Direct estimation of the area under the receiver operating characteristic curve with verification biased data. Statistics in Medicine. 2020; 4789-4820. https://doi.org/10.1002/sim.8753, .
In medical diagnostics, it’s expected that the true disease status of subjects can be identified based on screening test results. However, in many situations, not all subjects with given screening test results ultimately have their true disease status verified through a very accurate gold standard test. The missing of the verification of the true disease status might be based on results of diagnostic tests and other characteristics of subjects. The area under the ROC (AUC) curve is a widely used summary index of the accuracy of a diagnostic test. The AUC analysis based on partially verified subjects are usually biased. Authors in this paper consider the verification biased data under the assumption that the true disease status, if missing, is missing at random (MAR, which means the probability of a subject being verified does not depend on the subject’s disease status), and propose various new closed-form AUC estimators. They further compare the performance of the proposed approaches with existing methods by numerical study. Moreover, their methods are applied to analyze a data set from a Neonatal Hearing Screening study. The proposed AUC estimators have closed-form expressions and can be easily computed and directly applied in practice under the common MAR assumption for the verification biased data. Authors also provide the R code of the proposed methods for real applications. This is an interesting and meaningful work, which can help physicians evaluate the accuracy of a diagnostic test/system by directly estimating the AUC.