Skip to main content
. 2026 Aug 12;17:1907116. doi: 10.3389/fimmu.2026.1907116

Figure 3.

Illustration with two data visualizations: Panel A, a ROC curve comparing logistic ENI-Score model performance for training, internal, and external cohorts, showing AUC values of zero point eight seven three, zero point eight one six, and zero point eight three zero; Panel B, a bar chart of AUC by cohort for various machine learning methods, with training, internal, external, and five-fold cross-validation groups consistently demonstrating high AUC values across all models.

Discrimination of ten machine-learning models: (A) logistic ENI-Score ROC curves across the training, internal, and external cohorts; (B) training, internal, external, and cross-validated AUC by model.