Table 2 Performance of risk models
Prediction | # Cases (prevalence) | Model | AUROC (95% CI) | AUPRC (95% CI) | F1 Value (95% CI) |
|---|---|---|---|---|---|
In-patient mortality | 41 (0.36%) | XGBoost | 0.866 (0.777, 0.943) | 0.031 (0.015, 0.072) | 0.057 (0.000, 0.167) |
Logistic regression | 0.914 (0.811, 0.956) | 0.044 (0.019, 0.114) | 0.044 (0.028, 0.063) | ||
30-day readmission | 941 (8.20%) | XGBoost | 0.610 (0.587, 0.635) | 0.116 (0.104, 0.132) | 0.122 (0.078, 0.156) |
Logistic regression | 0.622 (0.599, 0.645) | 0.130 (0.113, 0.149) | 0.189 (0.171, 0.206) | ||
Length-of-stay | N/A | XGBoost | 0.841 (0.833, 0.847) | 0.741 (0.729, 0.753) | 0.666 (0.654, 0.678) |
Logistic regression | 0.822 (0.815, 0.829) | 0.719 (0.707, 0.730) | 0.646 (0.634, 0.658) |