🫀 PULSE-HF: Predicting Worsening Left Ventricular Function in Heart Failure Patients from ECGs

PULSE-HF is a deep learning model that forecasts whether a patient's left ventricular ejection fraction (LVEF) will fall below 40% within one year, using a standard 12-lead ECG and prior LVEF measurements. It is designed for patients with a history of heart failure.

This repository hosts the model weights. Code for preprocessing, training, and evaluation is on GitHub: mit-ccrg/PULSE-HF.

📄 Paper: Bergamaschi*, Yau*, Chandak* et al. Forecasting left ventricular systolic dysfunction in heart failure with artificial intelligence. eClinicalMedicine, 2026. Read it here (*equal contribution)

---

Citation

@article{bergamaschi2026pulsehf,
  title={Forecasting left ventricular systolic dysfunction in heart failure with artificial intelligence},
  author={Bergamaschi, Teya and Yau, Tiffany and Chandak, Payal and Kyereme-Tuah, Abena and Hung, Judy and Gaggin, Hanna and Kohane, Isaac S. and Stultz, Collin M.},
  journal={eClinicalMedicine},
  year={2026},
  volume={92},
  doi={10.1016/j.eclinm.2026.103783},
}

Contact

GitHub issues.