Machine learning models delivered the strongest performance across nearly all evaluation metrics. CHAID and CART provided the highest and most stable sensitivity, accuracy and discriminatory power, ...