Table 6 Comparison of the accuracy of the subject-specific ablation results for challenging-to-distinguish subjects using baseline algorithms on CHB-MIT dataset.
From: Synchronization-based graph spatio-temporal attention network for seizure prediction
Feature selection | Model | Chb2 | Chb5 | Chb6 | Chb9 | Chb10 | Chb14 | Chb16 | Chb18 | Chb21 |
|---|---|---|---|---|---|---|---|---|---|---|
Spatio | GCN | 90.6 | 73.63 | 65.82 | 81.08 | 67.06 | 53.92 | 65.44 | 72.27 | 75.15 |
Spatio | GAT | 93.07 | 92.83 | 95.1 | 92.96 | 97.09 | 96.15 | 96.75 | 95.56 | 92.82 |
Temporal | Transformer | 92.61 | 93.42 | 93.77 | 93.11 | 94.49 | 92.77 | 94.37 | 95.66 | 93.75 |
Spatio-temporal | GCN + Transformer | 93.85 | 78.45 | 70.95 | 88.16 | 73.92 | 71.38 | 85.56 | 87.27 | 79.46 |
Spatio-temporal | GAT + LSTM | 94.27 | 90.93 | 93.5 | 90.93 | 94.34 | 96.31 | 91.5 | 94.84 | 90.59 |
Spatio-temporal | SGSTAN | 97.63 | 96.93 | 95.8 | 97.69 | 98.99 | 98.85 | 98.44 | 98.09 | 95.93 |
