Table 3 P-values for pairwise comparisons among the top three best-performing models (RF, XGBoost, SVC).

From: Predictive factors of hypoglycemia in type 2 diabetes: a prospective study using machine learning

Metrics

Models

RF

XGBoost

SVC

AUPRC

RF

–

0.1

 < 0.001

XGBoost

0.1

–

 < 0.001

F1

RF

–

0.6

0.8

XGBoost

0.6

–

0.6

AUC

RF

–

0.8

0.7

XGBoost

0.8

–

0.5

Accuracy

RF

–

0.3

0.4

XGBoost

0.3

–

0.8

Brier score

RF

–

0.2

0.006

XGBoost

0.2

–

0.006

  1. Pairwise comparison of AUC scores was performed using DeLong’s test, while bootstrapping methods were applied to compare additional metrics. RF random forest, XGBoost eXtreme gradient boosting, SVC support vector classifier, KNN k-nearest neighbors, LR logistic regression.
Back to article page