jwc-rad / ecg-ai

ecg-ai

This repository provides PyTorch implementation of deep learning model for estimating age from electrocardiogram (ECG). This was the winning solution for the MAIC ECG AI 2023 challenge.

Installation

Create a virtual environment and activate it.

conda create -n ecgai python=3.9
conda activate ecgai

Install required packages

git clone https://github.com/jwc-rad/ecg-ai.git
cd ecg-ai
pip install -r requirements.txt

cd ..
git clone https://github.com/jwc-rad/MISLight.git
cd MISLight
pip install -e .

cd ..ecg-ai

Dataset

The current configurations assume the input shape as 12x5000, a 10-second 12-lead ECG with the sampling frequence of 500Hz. There are some public ECG datasets with the same data format as follows:

• A large scale 12-lead electrocardiogram database for arrhythmia study
• PTB-XL

Prepare data

Train

python train.py experiment=maic_v1 paths.data_root_dir=${PATH_TO_DATA_DIRECTORY}

• For more options, get started from config/experiment/maic_v1.yaml