Fig. 1From: A novel model to label delirium in an intensive care unit from clinician actionsComparison of ROC curves for 5 binary classifiers for presence or absence of delirium in the critical care setting. By AUC, logistic regression (unadjusted odds) outperformed supervised binary classification by naïve Bayes, support vector machines (SVM), Classification and Regression Trees (CART), and random forestsBack to article page