Firstly download the MIT BIH Arrhythmia dataset from https://www.physionet.org/content/mitdb/1.0.0/, named as "mit-bih-arrhythmia-database-1.0.0" Necessary packages can be installed using "settings and ECG picturing.ipynb" Use health.py to transform and save the data Then execute train.sh to train the model Performance metrics can be calculated by confusion matrix.ipynb Parameter can be adjusted in util.py AWN model can be modified in model/FFNN.py
Yuejin Li, Zhengdao Li, Kai Hwang. "Attentive Wavelet Neural Network for Time-Sensitive Applications: Illustrated by Arrhythmia Disease Detection".