Table 3 The machine learning performance of different omics
Modality | ACC | Precise | Recall | F1 score | AUC | PR-AUC | Optimal operating point |
|---|---|---|---|---|---|---|---|
EEG | 78.05% | 85.71% | 63.16% | 0.7273 | 0.8182 | 0.7328 | (0.2273, 0.7368) |
Gut microbiota | 75.61% | 66.67% | 89.47% | 0.7826 | 0.7895 | 0.6577 | (0. 4091, 0.9474) |
Metabolomic | 82.93% | 87.50% | 73.68% | 0.8000 | 0.8780 | 0.8490 | (0.0455, 0.7368) |
EEG + metabolomic | 87.81% | 88.89% | 84.21% | 0.8649 | 0.8947 | 0.8364 | (0.0909, 0.8421) |
EEG + gut microbiota | 78.05% | 72.73% | 84.21% | 0.7805 | 0.8014 | 0.7252 | (0.1346, 0.6842) |
Metabolomic + gut microbiota | 85.37% | 93.33% | 73.68% | 0.8235 | 0.8852 | 0.8388 | (0.0455, 0.7368) |
EEG + metabolomic + gut microbiota | 92.69% | 94.44% | 89.47% | 0.9198 | 0.9641 | 0.9136 | (0.0455, 0.9474) |