Table 1 Binary iEEG classification results of LSTM compared to previously used MLP14.
From: Bispectrum and Recurrent Neural Networks: Improved Classification of Interictal and Preictal States
Subject ID | # Sz. | This Work LSTM | Bou Assi et al.14 MLP | ||||
---|---|---|---|---|---|---|---|
Test SS (%) | Test SP (%) | Accuracy (%) | Accuracy (%) | ||||
Train | Test | Train a | Test a | ||||
1 | 17 | 86.67 | 98.33 | 95.83 | 92.50 | 84.23 | 76.71 |
2 | 17 | 56.29 | 87.05 | 75.00 | 71.67 | 71.71 | 67.23 |
3 | 11 | 91.67 | 97.92 | 95.69 | 94.71 | 90.89 | 90.40 |
Total | 45 | 78.21 | 94.43 | 88.84 | 86.29 | 82.28 | 78.11 |