Table 6 Patient-level 5-fold cross-validation results for all model configurations.
From: Machine learning approaches overcome imbalanced clinical data for intraoral free flap monitoring
Model configuration | Accuracy | Precision | Recall | F1 score |
|---|---|---|---|---|
Proposed (weighted focal loss) | 0.9379 ± 0.0173 | 0.9421 ± 0.0152 | 0.9379 ± 0.0173 | 0.9080 ± 0.0253 |
Focal loss without weight | 0.8956 ± 0.0229 | 0.8987 ± 0.0185 | 0.8956 ± 0.0229 | 0.8962 ± 0.0099 |
Weighted cross-entropy loss | 0.8634 ± 0.0748 | 0.9039 ± 0.0221 | 0.8634 ± 0.0748 | 0.8789 ± 0.0524 |
CE loss without weight | 0.9339 ± 0.0197 | 0.9187 ± 0.0380 | 0.9339 ± 0.0197 | 0.9060 ± 0.0263 |
