Table 2. F1 scores of different algorithms to predict postoperative complications among patient undergoing thoracic surgery with one-lung ventilation.
| Classification algorithm | Dataset | F1 score without SMOTE | F1 score with SMOTE |
|---|---|---|---|
| Random forests | Combined | 0.6604 | 0.6783 |
| Intraoperative | 0.6642 | 0.6537 | |
| Preoperative | 0.6643 | 0.5807 | |
| Support vector machine | Combined | 0.7338 | 0.7647 |
| Intraoperative | 0.6627 | 0.8298 | |
| Preoperative | 0.6627 | 0.5596 | |
| Logistic regression | Combined | 0.6903 | 0.7163 |
| Intraoperative | 0.6627 | 0.6028 | |
| Preoperative | 0.6627 | 0.6175 |
SMOTE, synthetic minority over-sampling technique.