- Fix max_len_nonzro_seq_in_batch
- Change LR: Exponential decay scheduler (AS)
- More simple model (MI)
- Random crop, Normalize individually, Add smooth input, Border Loss (AS)
- Fscore params: average{‘micro’, ‘macro’, ‘samples’,’weighted’, ‘binary’} ?
dynamic_ecg.py- building a labeled ECG-line over time. Command line arguments: id - patient id
- Deep learning and the electrocardiogram: review of the current state-of-the-art, 2021
- Robust detection of atrial fibrillation from short-term ECG using convolutional neural networks, 2020
- Deep Learning Approach for Highly Specific Atrial Fibrillation and Flutter Detection based on RR Intervals, 2019
- Deep Ensemble Detection of Congestive Heart Failure using Shortterm RR Intervals, 2018
- A Deep Learning Approach for Real-Time Detection of Atrial Fibrillation, 2018
| Quantity | Mean | Std |
|---|---|---|
| rr | 641.282 | 121.321 |
| anomaly_rr | 656.356 | 28.0538 |
| observation_time | 162468 | 235341 |
| min_max_obs_time | 19792 | 1.86858e+06 |
| observation_ticks | 264.135 | 411.681 |
| min_max_obs_ticks | 32 | 3661 |
| anomaly_ticks | 11.9082 | 3.96988 |
| min_max_anomaly_ticks | 6 | 41 |
| anomaly_time | 7776.85 | 2790.6 |
| min_max_anomaly_time | 3220 | 31712 |
| intra_ticks | 40.6251 | 64.7626 |
| min_max_intra_ticks | 1 | 687 |
| intra_time | 25935.7 | 39943.9 |
| min_max_intra_time | 544 | 436520 |