Table 2.
Performance comparisons of six machine learning methods.
| Model | Accuracy | Sensitivity | Specificity | Precision | F1 Score | ROC-AUC | PR-AUC |
|---|---|---|---|---|---|---|---|
| Artificial neural network | 85.2% | 67.5% | 91.7% | 75.7% | 71.4% | 84.0% | 76.0% |
| Decision tree | 87.7% | 66.2% | 93.6% | 79.7% | 72.4% | 84.0% | 79.0% |
| Logistic regression | 83.1% | 64.5% | 98.3% | 93.4% | 76.3% | 86.0% | 84.0% |
| Random forest | 86.8% | 67.5% | 91.7% | 75.7% | 71.4% | 84.0% | 76.0% |
| Support vector machine | 86.8% | 64.2% | 98.8% | 95.5% | 76.8% | 88.0% | 70.0% |
| XGBoost | 85.8% | 62.7% | 98.6% | 94.3% | 75.3% | 85.0% | 82.0% |
Seven indicators (accuracy, sensitivity, specificity, precision, F1 score, ROC-AUC, and PR-AUC) for machine learning models were used to evaluate the results of the six models (artificial neural network, decision tree, logistic regression, random forest, support vector machine, and XGBoost). ROC-AUC: Receiver operating characteristic curve area under the curve. PR-AUC: Precision–recall curve area under the curve.