Table 4 Performance comparison of CNN-LSTM-SE model on ECG datasets divided by different durations
From: Heart failure classification using deep learning to extract spatiotemporal features from ECG
Split seconds | Acc(%) | PPV(%) | Sen(%) | Spe(%) |
|---|---|---|---|---|
2 s | 97.472 | 97.471 | 97.426 | 99.043 |
3 s | 97.748 | 97.4675 | 97.936 | 99.1595 |
4 s | 98.114 | 97.891 | 98.254 | 99.2805 |
5 s | 98.332 | 98.356 | 98.4665 | 99.3655 |
6 s | 98.65 | 98.491 | 98.7575 | 99.4845 |
8 s | 98.942 | 98.967 | 98.9565 | 99.5945 |
9 s | 98.812 | 98.7465 | 99.034 | 99.5525 |
10s | 98.706 | 98.917 | 98.704 | 99.4965 |
12 s | 99.09 | 98.9855 | 99.033 | 99.649 |
15 s | 98.938 | 98.937 | 99.11 | 99.597 |
18 s | 98.778 | 98.433 | 98.682 | 99.5405 |
20s | 98.37 | 97.969 | 97.7495 | 99.4055 |