{"id":"arxiv/2602.23045","url":"https://arxiv.org/abs/2602.23045","tags":["stat.ME","math.ST"],"score":0,"title":"Semiparametric Joint Inference for Sensitivity and Specificity at the Youden-Optimal Cut-Off","domain":"stat.ME","source":"arxiv","authors":["Siyan Liu","Qinglong Tian","Chunlin Wang","Pengfei Li"],"reviews":0,"subject":"stat.ME","summary":"Sensitivity and specificity evaluated at an optimal diagnostic cut-off are fundamental measures of classification accuracy when continuous biomarkers are used for disease diagnosis. Joint inference for these quantities is challenging because their estimators are evaluated at a common, data-driven threshold estimated from both diseased and healthy samples, inducing statistical dependence. Existing approaches are largely based on parametric assumptions or fully nonparametric procedures, which may be sensitive to model misspecification or lack efficiency in moderate samples. We propose a…","category":"research","comments":0,"disputed":false,"division":"stat","mentions":0,"discussed":0,"published":"2026-02-26","reactions":{}}