Table 4.
Prediction performance of different machine learning models in training and validation cohort.
| Classifiers | Training cohort | Validation cohort | ||||||
|---|---|---|---|---|---|---|---|---|
| AUC (95% CI) | Accuracy | Sensitivity | Specificity | AUC (95% CI) | Accuracy | Sensitivity | Specificity | |
| SVM | 0.791 (0.781–0.802) | 82.60% | 75.30% | 90.30% | 0.776 (0.766–0.786) | 73.80% | 71.30% | 76.60% |
| KNN | 0.765 (0.754–0.775) | 77.20% | 70.50% | 84.20% | 0.716 (0.706–0.726) | 67.70% | 65.10% | 70.50% |
| RF | 0.711 (0.699–0.723) | 77.60% | 69.00% | 86.90% | 0.685 (0.673–0.697) | 68.20% | 67.40% | 69.20% |
| DT | 0.729 (0.718–0.740) | 75.10% | 67.70% | 83.20% | 0.647 (0.634–0.661) | 64.70% | 61.60% | 68.10% |
| LR | 0.783 (0.773–0.794) | 81.50% | 74.70% | 88.70% | 0.761 (0.752–0.771) | 72.60% | 69.60% | 72.50% |
SVM, support vector machine; KNM, k-nearest neighbor); RF, random forest; DT, decision tree; LR, logistic regression; AUC, the area under the curve.