Table 2.
5-fold XGBoost cross-validation results (training data) and test dataset.
Evaluation | Age group | Accuracy | F1 | AUROC | Sensitivity | Specificity | Precision | Negative to positive support ratio |
---|---|---|---|---|---|---|---|---|
Training (5-fold CV results) | Pediatric (0-14) | 0.969 ±0.001 | 0.024±0.047 | 0.833±0.022 | 0.014±0.029 | 0.998±0.002 | 0.067±0.133 | 33 |
Young(15-44) | 0.861±0.003 | 0.468±0.017 | 0.872±0.008 | 0.577±0.028 | 0.894±0.003 | 0.394±0.012 | 8.5 | |
Mid (45-64) | 0.742±0.008 | 0.585±0.009 | 0.844±0.009 | 0.813±0.021 | 0.722±0.014 | 0.457±0.009 | 3.3 | |
Older ≥ 65 | 0.699±0.012 | 0.565±0.006 | 0.815±0.011 | 0.838±0.023 | 0.657±0.021 | 0.426±0.010 | 3.5 | |
Test dataset | Pediatric (0-14) | 0.956 | 0.000 | 0.718 | 0.000 | 0.998 | 0.000 | 20 |
Young (15-44) | 0.791 | 0.482 | 0.796 | 0.609 | 0.825 | 0.399 | 5.3 | |
Mid (45-64) | 0.671 | 0.564 | 0.798 | 0.822 | 0.619 | 0.429 | 2.9 | |
Older ≥ 65 | 0.666 | 0.582 | 0.790 | 0.827 | 0.602 | 0.449 | 2.6 | |
Overall (all groups combined) | 0.736 | 0.545 | 0.814 | 0.752 | 0.731 | 0.427 | 3.8 |