Skip to main content
. 2026 Sep 11;14:1927235. doi: 10.3389/fcell.2026.1927235

FIGURE 3.

Eight-panel composite image compares multiple machine learning models for prediction performance. Panels A and B show ROC curves for training and test cohorts, with area under curve values for each model. Panels C and D display calibration curves for training and test cohorts, indicating observed event percentages by bin midpoint. Panels E and F present DCA curves, depicting standardized net benefit versus high-risk threshold. Panels G and H contain heatmaps of DeLong test p-values for model pair comparisons in training and test cohorts, with color legend indicating significance levels. Each panel is clearly labeled with a corresponding letter.

Efficacy of microboime-based models (A,B) The ROC curves of different machine learning models in the training and test cohort. (C,D) The calibration curves of different machine learning models in the training and test cohort. (E,F) The DCA curves of different machine learning models in the training and test cohort. (G,H) The heap map of AUCs comparison of different models by DeLong test in the training and test cohort.