Table 2 Predictive performance of each model, measured by test accuracy (%) and AUROC, stratified by race/ethnicity
| Â | Test Accuracy (%) | AUROC | ||||||
|---|---|---|---|---|---|---|---|---|
All | White | Black | Hispanic | All | White | Black | Hispanic | |
k-NN | 96.0 | 97.0 | 95.0 | 96.0 | 0.612 | 0.691 | 0.661 | 0.621 |
Logistic regression | 96.0 | 97.0 | 95.0 | 97.0 | 0.730 | 0.738 | 0.683 | 0.791 |
Decision tree | 96.1 | 97.0 | 95.0 | 97.0 | 0.647 | 0.721 | 0.661 | 0.709 |
Gradient-boosted decision tree | 96.1 | 96.6 | 95.0 | 96.8 | 0.739 | 0.759 | 0.701 | 0.838 |
Random forest | 96.1 | 96.6 | 95.1 | 96.8 | 0.740 | 0.748 | 0.644 | 0.744 |