Table 4.
Performance of the leave-one-site-out method in MDD/HC classification.
| Method | ACC(%) | SEN(%) | SPE(%) | f1(%) | AUC(%) |
|---|---|---|---|---|---|
| Nearest Neighbors | 55.5(2.0) | 57.3(2.5) | 53.9(4.6) | 55.5(4.4) | 55.6(1.2) |
| AdaBoost | 47.9(1.4) | 49.3(5.2) | 47.1(1.5) | 42.1(1.6) | 48.3(2.0) |
| Naive Bayes | 55.1(6.8) | 59.5(9.5) | 52.8(6.9) | 48.2(12.7) | 55.6(6.3) |
| Gaussian Process | 56.9(5.0) | 59.2(4.6) | 55.0(6.8) | 55.8(8.9) | 57.1(2.6) |
| Linear SVM | 55.9(6.2) | 59.0(2.7) | 53.7(9.0) | 52.4(15.8) | 56.1(4.8) |
| Deep Neural Net | 55.0(3.4) | 60.8(5.0) | 52.4(3.1) | 44.7(2.9) | 55.6(1.8) |
| GAN | 64.3(2.9) | 61.5(3.1) | 70.8(6.5) | 70.5(1.5) | 63.8(3.4) |
Table 4 The performance of the leave-one-site-out method in MDD/HC classification. SVM, support vector machine; GAN, Generative Adversarial Networks; ACC, Accuracy; SEN, sensitivity; SPE, specificity; F1, F-score; AUC, area under curve; MDD: major depressive disorder; HC: healthy control.