Zuse Institute Berlin (Weimann et al.)
Model weights not public. Contact creators for more information.
Self-supervised ECG representation-learning method that adapts Joint-Embedding Predictive Architecture (JEPA) -- originally developed for images -- to 1D electrocardiogram signals. A Vision Transformer encoder (ViT-XS/S/B) is pretrained to predict masked temporal segments of the ECG directly in latent feature space, using a masking strategy tailored to time-series, on more than 1 million ECGs pooled from MIMIC-IV-ECG, CODE-15%, PTB-XL, Chapman-Shaoxing, CPSC2018/Extra, Georgia, PTB, and St-Petersburg-INCART. After fine-tuning on PTB-XL, the ViT-S/JEPA model reaches 0.945 AUC on the all-statements diagnostic task, exceeding prior self-supervised ECG baselines including CPC and ST-MEM. Developed at the Zuse Institute Berlin.
Architecture
Vision Transformer
Vision Transformer encoder pretrained with Joint-Embedding Predictive Architecture (JEPA) using a temporal-masking strategy adapted for 1D ECG signals; fine-tuned for downstream classification
Framework
PyTorch
Added to catalog
2026-08-10
Publicly available 15% subset of the CODE (Clinical Outcomes in Digital Electrocardiology) dataset from the Telehealth Network of Minas Gerais; full CODE dataset (>2 million exams) is request-only.
12-lead ECG multi-label arrhythmia-classification dataset released for the 2018 China Physiological Signal Challenge.
45,152 12-lead, 10-second ECGs from Chapman University / Shaoxing People's Hospital / Ningbo First Hospital with arrhythmia diagnoses; ~56% male / 44% female.
Georgia 12-lead ECG Challenge (G12EC) database, part of the PhysioNet/CinC 2020-2021 Challenge training data.
800,035 12-lead ECG-report pairs from 161,352 subjects at Beth Israel Deaconess Medical Center.
268 subjects with 12-lead ECGs plus patient information; commonly used as a small external-validation cohort for ECG-based diagnosis models.
52% male / 48% female; age range 0-95 (median ~62). 21,837 10-second 12-lead ECG records.
75 annotated 30-minute 12-lead ambulatory ECG recordings from 32 subjects, collected at the St. Petersburg Institute of Cardiological Technics; one of three corpora (with CODE and Chapman-Shaoxing-Ningbo, ~8M ECGs total) used to pretrain xECG.
Multi-label ECG diagnostic statement classification (PTB-XL 'all statements' task), evaluated via linear probing and fine-tuning