Logistic regression models were used for these analyses. AUC: area under the receiver operating characteristic (ROC) curve. True positive rate (= sensitivity) indicates the percentage of patients correctly classified in the EF group (EF ≤ 40% or EF > 40%). False positive rate (= 1-specificity) indicates the percentage of patients misclassified in the EF group. Optimal classification models are shown; all other combination of genes provided lower classification performances.