Table 4. The comparison of the five-fold cross-validation results of the logistic regression and the neural networks models.
AUC, area under the curve
| Training | Testing | ||||||||
| AUC | Sensitivity | Specificity | Accuracy | AUC | Sensitivity | Specificity | Accuracy | ||
| Logistic regression | Fold 1 | 0.879 (0.861 to 0.897) | 48.1% (42.9 to 53.3) | 94.3% (92.8 to 95.5) | 83.9% (82.1 to 85.7) | 0.840 (0.798 to 0.883) | 39.1% (29.1 to 49.9) | 94.7% (91.6 to 96.9) | 82.2% (78.2 to 85.8) |
| Fold 2 | 0.880 (0.862 to 0.898) | 50% (44.8 to 55.2) | 94.3% (92.8 to 95.5) | 84.4 (82.5 to 86.1) | 0.839 (0.797 to 0.881) | 40.2% (30.1 to 51) | 92.8% (89.4 to 95.4) | 81% (76.9 to 84.7) | |
| Fold 3 | 0.873 (0.855 to 0.892) | 48.4% (43.1 to 53.6) | 95.1% (93.8 to 96.3) | 84.7% (82.9 to 86.4) | 0.868 (0.831 to 0.905) | 41.3% (31.3 to 52.1) | 93.7% (90.5 to 96.1) | 82% (77.9 to 85.6) | |
| Fold 4 | 0.873 (0.854 to 0.892) | 46.9% (41.7 to 52.1) | 94.7% (93.3 to 95.8) | 83.9% (82.1 to 85.7) | 0.876 (0.840 to 0.912) | 46.2% (35.6 to 56.9) | 95.9% (93.1 to 97.8) | 84.8% (81 to 88.2) | |
| Fold 5 | 0.858 (0.839 to 0.878) | 42.5% (37.4 to 47.7) | 94.5% (93.1 to 95.7) | 82.9% (81 to 84.7) | 0.918 (0.889 to 0.948) | 63.7% (53 to 73.7) | 93.4% (90.1 to 95.9) | 86.8% (83.1 to 89.9) | |
| Mean | 0.873 (0.862 to 0.884) | 47.2% (43.7 to 50.7) | 94.6% (94.2 to 95) | 84% (83.1 to 84.8) | 0.868 (0.828 to 0.908) | 46.1% (33.4 to 58.8) | 94.1% (92.6 to 95.6) | 83.4% (80.4 to 86.3) | |
| Neural networks | Fold 1 | 0.877 (0.859 to 0.896) | 52.2% (46.9 to 57.4) | 92.9% (91.3 to 94.2) | 83.8% (81.9 to 85.5) | 0.845 (0.803 to 0.887) | 44.6% (34.2 to 55.3) | 93.1% (89.7 to 95.6) | 82.2% (78.2 to 85.8) |
| Fold 2 | 0.879 (0.861 to 0.897) | 47.8% (42.6 to 53.1) | 95.7% (94.4 to 96.7) | 85% (83.2 to 86.7) | 0.845 (0.805 to 0.886) | 40.2% (30.1 to 51) | 95.3 (92.4 to 97.3) | 83% (79 to 86.5) | |
| Fold 3 | 0.861 (0.841 to 0.881) | 47.5% (42.3 to 52.8) | 93.3 (91.7 to 94.6) | 83.1% (81.1 to 84.8) | 0.866 (0.828 to 0.904) | 43.5% (33.2 to 54.2) | 92.5% (89 to 95.1) | 81.5 (77.4 to 85.1) | |
| Fold 4 | 0.870 (0.851 to 0.889) | 50.7% (45.4 to 55.9) | 92.1% (90.5 to 93.5) | 82.8% (80.9 to 84.6) | 0.883 (0.848 to 0.918) | 52.8% (42 to 63.3) | 93.7% (90.5 to 96.1) | 84.6% (80.7 to 88) | |
| Fold 5 | 0.859 (0.840 to 0.879) | 45.8% (40.6 to 51) | 93.3% (91.7 to 94.6) | 82.6% (80.7 to 84.4) | 0.904 (0.873 to 0.934) | 50.6% (39.9 to 61.2) | 93.7% (90.5 to 96.1) | 84.1% (80.2 to 87.5) | |
| Mean | 0.869 (0.858 to 0.880) | 48.8%(45.6 to 52) | 93.5% (91.8 to 95.1) | 83.5% (82.3 to 84.7) | 0.868 (0.837 to 0.900) | 46.3% (39.9 to 52.8) | 93.7% (92.4 to 95) | 83.1% (81.5to 84.7) | |