Comparative study of statistical methods for clustered ROC data: nonparametric methods and multiple outputation methods

Abstract

In clustered receiver operating characteristic (ROC) data each patient has several normal and abnormal observations. Within the same cluster, observations are naturally correlated. Several nonparametric methods have been proposed in the literature to handle clustered data structure, but their performances on simulated and real datasets have not been compared. Recently, a multiple outputation method has been considered for clustered data in areas other than diagnostic accuracy to account for within-cluster correlation. The multiple outputation method offers a resampling-based alternative for one sample clustered data with or without covariates, or for hypothesis testing in two sample clustered data. The method does not require a specific within-cluster correlation structure and yields a valid estimator while accounting for the within-cluster correlations. This paper contributes to the literature by introducing the multiple outputation method to the ROC setting, and empirically comparing the performance of these clustered ROC curve methods. The performance of these methods is also evaluated through two real examples.

Keywords:

• ROC curve
• Wilcoxon statistics
• diagnostic tests
• empirical estimators
• multiple outputation

Acknowledgments

The authors wish to thank Dr. Rosner for providing the Sorbinil Retinopathy Trial eye rating dataset and Dr. Zhou for providing the Detection of Glaucomatous Deterioration eye rating dataset. This work is supported in part by the Intramural Research Program of the National Institutes of Health and the US Social Security Administration. The opinions expressed in this article are the author's own and do not reflect the view of the National Institutes of Health, the Department of Health and Human Services, US Social Security Administration, or the United States government.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work is supported in part by the Intramural Research Program of the National Institutes of Health and the US Social Security Administration.

Notes on contributors

Zhuang Miao

Dr. Zhuang Miao received his Ph.D. in Statistics from George Mason University in 2014. His areas of research have focused on diagnostic medicine, nonclinical statistics and clinical trials. He is currently a mathematical statistician working in FDA.

Liansheng Larry Tang

Dr. Liansheng Larry Tang is a statistician specializing in statistical methodology and collaborative research. His current methodological research areas include statistical methods in forensics, diagnostic medicine, group sequential designs and substance abuse research and criminology. He received his Ph.D. in Statistics from Southern Methodist University in 2005. He did postdoctoral training in the Department of Biostatistics at University of Washington. His research is currently supported by National Institute of Justice and National Institutes of Health.

Ao Yuan

Dr. Ao Yuan got his PhD in Statistics from University of British Columbia. His research interests include parametric-semiparametric-nonparametric inference, Bayesian inference, biostatistics and statistical genetics.