Table 1 Configuration of arrhythmia detection model. n indicates the batch size
From: Continual learning framework for a multicenter study with an application to electrocardiogram
| Â | Kernel | Size |
|---|---|---|
ECG Processing Layer | Â | Â |
 Input |  | (n, 12, 5000) |
 Block1 (stride = 2) | (11, 12, 8 × 12 = 96) | (n, 96, 1250) |
 Block2 (stride = 2) | (7, 96, 96) | (n, 96, 313) |
 Block3 (stride = 2) | (5, 96, 96) | (n, 96, 80) |
 Block4 (stride = 1) | (5, 96, 96)×2 | (n, 96, 80) |
 Block5 (stride = 1) | (5, 96, 96)×2 | (n, 96, 80) |
 Block6 (stride = 1) | (5, 96, 16 × 12 = 192) | (n, 192, 40) |
 Block7 (stride = 1) | (5, 192, 192) | (n, 192, 40) |
 Block8 (stride = 1) | (5, 192, 192) | (n, 192, 40) |
 Block9 (stride = 1) | (5, 192, 32 × 12 = 384) | (n, 384, 20) |
 Block10 (stride = 1) | (5, 384, 384) | (n, 384, 20) |
 Block11 (stride = 1) | (5, 384, 384) | (n, 384, 20) |
Patient Information Processing Layer | Â | |
 Input |  | (n, 2) |
 Dense1 | (2, 32) | (n, 32) |
 Dense2 | (32, 64) | (n, 64) |
 Dense3 | (64, 64) | (n, 64) |
Arrhythmia Detection Layer | Â | |
 Input |  | (n, 384 × 20 + 64 = 7744) |
 Dense1 | (7744, 512) | (n, 512) |
 Dense2 | (512, 256) | (n, 256) |
 Dense3 | (256, 2) | (n, 2) |