Figure 5.
ROC curve analyses for short-term andlong-termfacial nerve function damage, (A) ROC curve analysis yielded a cut-off delta FNMEP value of 30 V (sensitivity, 91.3%; specificity, 98.6%) for predicting short-term facial nerve function damage. The AUC is 0.976 (95% confidence interval: 0.936–1.000; Youden index: 0.899). (B) ROC curve analysis yielded a cut-off value of 75 V (sensitivity, 100%; specificity, 98.8%) for predicting long-term facial nerve function damage. The AUC is 0.992 (95% confidence interval: 0.975–1.000; Youden index: 0.988). AUC = area under the curve, FNMEP = facial nerve motor-evoked potential, ROC = receiver operating characteristic.