Joint Inference of ROC Measures for Diagnostic Biomarker/Test Evaluation
Presented by Jingjing Yin, PhD., Associate Professor of Biostatistics, Jiann-Ping Hsu College of Public Health, Georgia Southern University
Wednesday, March 8, 2023
4:00 PM ET
Virtual
Webex Meeting Link
It is common to use the summary ROC measures such as the area under the ROC curve (AUC) for diagnostic test accuracy evaluations and comparisons. We propose to use the AUC and the Youden index jointly for making inferences about diagnostic tests, as the two indices describe different aspects of the ROC curve. This can be done by first estimating the joint confidence region of the AUC and the Youden index. For deciding if a marker is achieving the targeted values, or comparing two biomarkers in a paired design, in terms of both the AUC and the Youden index, we can perform joint testing for such order-restricted hypotheses for which the traditional likelihood ratio test or its variant cannot apply. We propose and compare three testing procedures: 1) the intersection-union test; 2) the conditional test; and 3) the joint test. The performance of the proposed inference methods was evaluated and compared through simulations. The simulation results demonstrate that the proposed joint confidence region maintains the desired confidence level, and all three tests maintain the type I error under the null. Furthermore, among the three proposed testing methods, the conditional test is the preferred approach with markedly larger power consistently than the other two competing methods. In conclusion, estimating and testing jointly on the AUC and the Youden index gives more reliable results for biomarker evaluations than using a single index.
Speaker Bio:
Dr. Jingjing Yin has a background in statistical methods in medical diagnostics, parametric and nonparametric inference, order-restricted inference, and sampling design. Meanwhile, she has diverse collaborative research in the field of medical science, Epidemiology, Environmental health, and health communication and management. Beyond traditional biomedical statistical research, she has also applied her expertise to the emerging field of big data analytics, with a focus on text-mining social media data. She has published more than 80 peer-reviewed articles in various statistical or medical journals including Statistics in Medicine, Statistical Methods in Medical Research, Biometrical Journal, Pharmaceutical Statistics, and Journal of Biopharmaceutical Statistics.
Dr. Jingjing Yin is an Associate Professor of Biostatistics at Jiann-Ping Hsu College of Public Health, Georgia Southern University. She earned her Ph.D. (2014) and M.S. (2011) in Biostatistics from the University at Buffalo, and her B.A. (2009) from West China University of Medical Science in Chengdu, China.