# clone project   
git clone git@github.com:haochunchang/explain-ECG-diagnosis.git
cd ECG-diagnosis

# install project
pip install -e .
pip install -r requirements.txt

# run main script
python main.py --help

The data is from The PTB Diagnostic ECG Database.

The database contains 549 records from 290 subjects. Each subject is represented by 1 ~ 5 records

Each record includes 15 simultaneously measured signals:

• The conventional 12 leads (i, ii, iii, avr, avl, avf, v1, v2, v3, v4, v5, v6)
• The 3 Frank lead ECGs (vx, vy, vz).

Each signal is digitized at 1000 samples per second.

• Preprocessing
  • Splits records into a fixed chunk size, treating each chunk as individual sample.
• Model
  • Use 1D convolutional filters and fully-connected layers to learn signal features.
• Evaluation
  • Using Accuracy, F1 score and confusion matrix on testing dataset.
• Interpretation
  • Modified Gradient-weighted Class Activation Mapping (Grad-CAM).
  • Modified Local Interpretable Model-agnostic Explanations (LIME).