The Youden Index helps turn probability estimates into binary predictions when both true-positive and true-negative performance matter. It is the difference between the true positive rate and false positive rate, equivalently sensitivity plus specificity minus one. section 5.1.3
For a classifier detecting CYP3A4 inhibitors, evaluate possible decision thresholds along the ROC curve and choose the threshold that maximizes the Youden Index. This jointly favors sensitivity and specificity rather than optimizing only precision or recall. The logistic-regression example produces an optimal binarization threshold of 0.505. section 5.1.3 section 5.5
CYP3A4 inhibitor classification
A logistic regressor outputs inhibitor probabilities. Comparing thresholds with the ROC curve identifies 0.505 as the operating point selected by Youden’s Index, converting probabilities into predicted inhibitor and non-inhibitor classes. section 5.1.3
