HeartLang

Peking University (PKUDigitalHealth)

12-lead ECG

Filter catalog by Modality:
ECG

General Purpose / Multi-task

Filter catalog by Disease / Trait:
General / Foundation

Multi-label classification

Filter catalog by Task Type:
Classification

Transformer

Filter catalog by Architecture:
Transformer

PyTorch

Filter catalog by Framework:
PyTorch

MIT

Filter catalog by License:
Permissive

file_copy

View model files code View code

Treats ECGs as a language: a QRS-Tokenizer converts raw waveforms into discrete heartbeat 'words' from a learned 8,192-entry vocabulary, and a spatio-temporal transformer (ST-ECGFormer) is pretrained via masked-sentence modeling over these tokens. Evaluated for robust, competitive performance across six public ECG datasets and published at ICLR 2025. Developed by Peking University's digital health group, pretrained on MIMIC-IV-ECG.

memory Specifications

gavel License

MIT

check_small Open source check_small Commercial use OK check_small Modifications allowed Attribution required No share-alike requirement

License for model weights only. Associated code may be licensed seperately, check code source for specific terms.

description Publication

Reading Your Heart: Learning ECG Words and Sentences via Pre-training ECG Language Model open_in_new

2025 · original paper

database Training & evaluation data

MIMIC-IV-ECG open_in_new

train

public 161,352 subjects · USA

800,035 12-lead ECG-report pairs from 161,352 subjects at Beth Israel Deaconess Medical Center.

science Capabilities & performance