Improved Estimation and Optimal Linear Combination on Youden Index

Diagnostic cut-off point of biomarker measurements is needed for classifying a random subject to be either diseased or healthy. However, such cut-off point is usually unknown and needs to be estimated by some optimization criteria, among which, Youden index has been widely adopted in practice. Youden index, defined as max (sensitivity + specificity -1), directly measures the largest total diagnostic accuracy a biomarker can achieve. Therefore, it is desirable to estimate the optimal cut-off point associated with Youden index. The first topic of this talk proposed to apply ranked set sampling approach to improve the estimation accuracy of Youden index and its associated cut-off point. In practice, usually multiple biomarkers are measured on the same subject for disease diagnosis. In case a primary biomarker is found, we may apply regression techniques to adjust the primary biomarker values given other biomarkers considered as covariates, or we may rank the primary measurements based on other covariate values thus utilizing the ranked set sampling approach. However, if all biomarkers are of similar importance, combining these biomarkers into a single score would be the best option to achieve satisfying diagnostic accuracy. In the second part of the talk, the idea of using Youden index as an objective function for searching the optimal linear combination is discussed. The resulting combined score is the optimal linear combination corresponding to the largest estimate of Youden index among all possible linear combinations