Table 3.
Area under the receiving characteristic curve (AUROC) with 95% confidence intervals, to compare and select the best model
| Models | Primary outcome (highest care level within 24 hours) |
Secondary outcome (care level at the 24th hour) |
||
|---|---|---|---|---|
| Full feature | Simple feature | Full feature | Simple feature | |
| Gradient Boosting* | 0.88 (0.87–0.89) | 0.82 (0.80–0.83) | 0.86 (0.85–0.87) | 0.81 (0.79–0.82) |
| Random Forest | 0.86 (0.85–0.87) | 0.78 (0.77–0.80) | 0.84 (0.82–0.85) | 0.78 (0.76–0.79) |
| Logistic Regression (elastic net) | 0.84 (0.82–0.85) | 0.79 (0.77–0. 80) | 0.82 (0.80–0.83) | 0.78 (0.77–0.80) |
| Feed-Forward Neural Networks | 0.85 (0.83–0.86) | 0.77 (0.76–0.79) | 0.82 (0.81–0.84) | 0.78 (0.76–0.79) |
Notes: The baseline of AUROC is fixed at 0.5, which is equal to random guessing.
*Best models with highest AUROC .