ECG classification

Electrocardiography (ECG) classification based on the PTB-XL dataset.

Goal

The goal of this repo is to explore the PTB-XL dataset and try to classify ECGs.

Data

The data used in this work is from PhysioNet and it is available here.

You can manually download it or run:

wget -r -N -c -np https://physionet.org/files/ptb-xl/1.0.2/

Setup

Poetry is used to manage the requirements. Visit Poetry for installation details.

Tip: if you are using VS Code run poetry config virtualenvs.in-project true to generate the .env in the root

With Poetry installed, one just needs to run poetry install from the root of the repo.

EDA

The first goal is to explore the data. Our data exploration consists on the following steps:

• load the data
• visualize target data
• preprocess the data
  • apply filters
  • estimate the baseline wander
• estimate the QRS complex
• evaluate the data split suggested by Physionet
• data augmentation techniques

The EDA is done in notebooks/eda.ipynb. Once we run the notebook, we have a clear understanding on how is the data and how is it split.

Model

Work in progress.