Table 2 Comparison between evaluation indicators of different models
From: Machine learning-based risk prediction model for arteriovenous fistula stenosis
Model | F1 Score (95% CI) | Accuracy (95% CI) | Precision (95% CI) | Recall (95% CI) | AUC (95% CI) | Specificity (95% CI) |
|---|---|---|---|---|---|---|
RF | 0.789 (0.739–0.832) | 0.721 (0.667–0.773) | 0.745 (0.681–0.800) | 0.839 (0.781–0.890) | 0.826 (0.774–0.870) | 0.527 (0.436–0.623) |
XGBoost | 0.811 (0.763–0.852) | 0.773 (0.725–0.818) | 0.840 (0.782–0.892) | 0.785 (0.723–0.845) | 0.829 (0.785–0.880) | 0.754 (0.670–0.830) |
KNN | 0.727 (0.677–0.776) | 0.634 (0.581–0.687) | 0.678 (0.617–0.740) | 0.786 (0.724–0.844) | 0.638 (0.574–0.700) | 0.381 (0.287–0.477) |
DT | 0.771 (0.718–0.822) | 0.764 (0.715–0.811) | 0.968 (0.934–0.993) | 0.641 (0.573–0.714) | 0.802 (0.765–0.843) | 0.965 (0.929–0.992) |
SVM | 0.792 (0.748–0.833) | 0.711 (0.660–0.760) | 0.718 (0.659–0.779) | 0.884 (0.837–0.929) | 0.803 (0.756–0.854) | 0.428 (0.333–0.515) |
LR | 0.775 (0.729–0.818) | 0.704 (0.650–0.756) | 0.734 (0.673–0.794) | 0.823 (0.764–0.876) | 0.787 (0.733–0.835) | 0.509 (0.414–0.606) |
ANN | 0.793 (0.743–0.838) | 0.747 (0.698–0.794) | 0.807 (0.749–0.866) | 0.781 (0.717–0.839) | 0.778 (0.725–0.832) | 0.692 (0.615–0.776) |