Table 5.
Performance of different methods in SZ/HC classification with 10-fold cross-validation.
| Method | ACC(%) | SEN(%) | SPE(%) | f1(%) | AUC(%) |
|---|---|---|---|---|---|
| Nearest Neighbors | 64.6(0.6) | 69.1(0.8) | 61.6(0.6) | 61.0(0.8) | 64.8(0.6) |
| AdaBoost | 72.3(1.1) | 72.4(1.1) | 72.1(1.1) | 72.8(1.0) | 72.2(1.1) |
| Naive Bayes | 69.6(0.5) | 67.5(0.4) | 72.7(0.6) | 72.1(0.5) | 69.5(0.5) |
| Gaussian Process | 64.3(0.8) | 68.5(1.0) | 61.4(0.7) | 60.9(0.9) | 64.4(0.8) |
| Linear SVM | 77.1(0.9) | 77.4(1.0) | 76.8(0.9) | 77.4(0.9) | 77.1(0.9) |
| Deep Neural Net | 80.3(0.5) | 80.8(0.9) | 79.8(0.9) | 80.5(0.5) | 80.3(0.5) |
| GAN | 82.1(0.7) | 78.1(1.5) | 86.2(1.1) | 81.6(0.8) | 82.3(0.7) |
Table 5 The performance of different methods in SZ/HC classification with 10-folds cross-validation. SVM, support vector machine; No FM, No feature matching; GAN, Generative Adversarial Networks; ACC, Accuracy; SEN, sensitivity; SPE, specificity; F1, F-score; AUC, area under curve; SZ: schizophrenia; HC: healthy control.