In medical research, continuous markers are widely employed in diagnostic tests to distinguish diseased and non-diseased subjects. The accuracy of such diagnostic tests is commonly assessed using the receiver operating characteristic (ROC) curve. To
summarize an ROC curve and determine its optimal cut-point, the Youden index is popularly used. In literature, estimation of the Youden index has been widely studied via various statistical modeling strategies on the conditional density. This paper proposes a new model-free estimation method, which directly estimates the covariate-adjusted cut-point without estimating the conditional density. Consequently, covariate-adjusted Youden index can be estimated based on the estimated cutpoint. The proposed method formulates the estimation problem in a large margin classification framework, which allows flexible modeling of the covariate-adjusted Youden index through kernel machines. The advantage of the proposed method is demonstrated in a variety of simulated experiments as well as a real application to Pima Indians diabetes study.
In clinical trials, minimum clinically important difference (MCID) has attracted increasing interest as an important supportive clinical and statistical inference tool. Many estimation methods have been developed based on various intuitions, while li
ttle theoretical justification has been established. This paper proposes a new estimation framework of MCID using both diagnostic measurements and patient-reported outcomes (PROs). It first provides a precise definition of population-based MCID so that estimating such a MCID can be formulated as a large margin classification problem. The framework is then extended to personalized MCID to allow individualized thresholding value for patients whose clinical profiles may affect their PRO responses. More importantly, we show that the proposed estimation framework is asymptotically consistent, and a finite-sample upper bound is established for its prediction accuracy compared against the ideal MCID. The advantage of our proposed method is also demonstrated in a variety of simulated experiments as well as applications to two benchmark datasets and two phase-3 clinical trials.
