cuongvng / melep-ecg

MELEP: A Novel Predictive Measure of Transferability in Multi-Label ECG Diagnosis

MELEP: A Novel Predictive Measure of Transferability in Multi-Label ECG Diagnosis

This repository contains code and resources related to the implementation of the paper MELEP: A Novel Predictive Measure of Transferability in Multi-Label ECG Diagnosis.

Installation

git clone https://github.com/cuongvng/melep-ecg.git
cd melep-ecg
pip install -r requirements.txt

Instructions

Dataset Acquisition

The paper used publicly available datasets: PTB-XL, CPSC2018, Georgia, and Chapman-Shaoxing-Ningbo (CSN). Raw datasets can be found here. Processed PTB-XL and CSN datasets for experiments can be found at the corresponding folder.

Pretraining Models

We provided pre-trained models used for our experiments at this link. Please follow instructions in this repo if you want to pre-train those models from scratch.

Running experiments

After installing required packages, download datasets and models, run the corresponding scripts to reproduce our experiments. For example:

python ptbxl_resnet_transfer.py

Citation & Acknowledgements

If you find this work helpful, please cite:

@article{nguyen2023melep,
  title={MELEP: A Novel Predictive Measure of Transferability in Multi-Label ECG Analysis},
  author={Nguyen, Cuong V and Duong, Hieu Minh and Do, Cuong D},
  journal={arXiv preprint arXiv:2311.04224},
  year={2023}
}

License

This project is licensed under the CC-BY-4.0 license.