Table 3 Comparison of the performance of EEG and audio data fusion models at different depths of GCN layers
From: An adaptive multi-graph neural network with multimodal feature fusion learning for MDD detection
Ablation condition | GCN LayerCount | ACC(\(\%\)) | PRE(\(\%\)) | REC(\(\%\)) | F1 Score(\(\%\)) |
---|---|---|---|---|---|
EEG | 4 | 86.11 | 85.85 | 84.26 | 85.05 |
3 | 90.06 | 90.20 | 88.46 | 89.32 | |
2 | 84.72 | 87.50 | 84.26 | 85.85 | |
1 | 81.94 | 83.18 | 82.41 | 82.79 | |
0 | 73.15 | 76.04 | 67.59 | 71.57 | |
Audio | 4 | 62.50 | 68.32 | 63.89 | 66.03 |
3 | 90.48 | 92.36 | 90.48 | 91.41 | |
2 | 83.80 | 88.00 | 84.15 | 84.62 | |
1 | 81.94 | 83.18 | 82.41 | 82.79 | |
0 | 73.61 | 72.64 | 71.30 | 71.96 |