Table 5 Performance comparison with traditional machine learning models.
From: Enhancing stroke risk prediction through class balancing and data augmentation with CBDA-ResNet50
Model | Accuracy (%) | Specificity (%) | Sensitivity (%) | Balanced Accuracy (%) | NPV | MCC |
|---|---|---|---|---|---|---|
Logistic Regression | 74.4 | 72.0 | 73.6 | 72.8 | 0.7317 | 0.4561 |
SVM | 72.6 | 73.7 | 68.3 | 71.0 | 0.6987 | 0.4193 |
Random Forest | 75.6 | 70.9 | 68.1 | 69.5 | 0.6883 | 0.3868 |
Decision Tree | 58.5 | 55.8 | 54.6 | 55.2 | 0.5500 | 0.1013 |
K-Nearest Neighbors | 72.4 | 69.7 | 73.9 | 71.8 | 0.7270 | 0.4351 |
Gradient Boosting | 71.1 | 63.5 | 67.9 | 69.1 | 0.6630 | 0.3123 |
CBDA-ResNet50 (Ours) | 97.8 | 98.5 | 97.9 | 98.2 | 0.9788 | 0.9627 |
