Skip to main content
. 2026 Jun 4;13:1808657. doi: 10.3389/fmed.2026.1808657

Figure 2.

Three-panel figure comparing six machine learning models using (a) ROC curves, (b) calibration plots, and (c) decision curve analysis for VTE prediction. Each model’s performance is shown using distinct colored lines with associated AUC and Brier scores in the legends.

Comparison of the performance of six algorithms on the training set: (a) The first graph represents the ROC curve; (b) The second graph represents the calibration curve; (c) The third graph represents the DCA decision curve. AUC = area under curve, KNN = K-nearest neighbors, SVM = support vector machine, LR = logistic regression, VTE = venous thromboembolism.