TARGET-AI ECG-image ViT

ecg-clip-beit-base-384

Yale School of Medicine (CarDS Lab)

12-lead ECG image

Filter catalog by Modality:
ECG

Structural heart disease (composite)

Filter catalog by Disease / Trait:
Structural Heart & Cardiomyopathy

Embedding

Filter catalog by Task Type:
Representation Learning

Vision Transformer

Filter catalog by Architecture:
Transformer

PyTorch

Filter catalog by Framework:
PyTorch

CC BY-NC 4.0

Filter catalog by License:
Non-commercial / Research-only

file_copy

View model files code View code

BEiT-base Vision Transformer that embeds images of standard 12-lead ECG printouts into a representation space, enabling zero-shot screening for structural heart disease by comparing a new ECG against reference case/control embedding centroids rather than requiring task-specific training. Trained on private Yale New Haven Health System ECG images and validated against the public EchoNext dataset. Aimed at scanned or legacy ECG images still common in EHR systems. Developed by Yale's CarDS Lab.

memory Specifications

gavel License

CC BY-NC 4.0

check_small Open source close_small No commercial use 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

TARGET-AI: a foundational approach for the targeted deployment of artificial intelligence electrocardiography in the electronic health record open_in_new

2025 · original paper

DOI: 10.1056/AIoa2500588

database Training & evaluation data

UK Biobank External Validation Cohort (TARGET-AI)

validation

public 33,518 subjects · United Kingdom

Yale New Haven Health ECG-Echo Paired Cohort (TARGET-AI)

train

public 159,322 subjects · USA

754,533 ECG-echo pairs from 159,322 individuals

Yale Temporally-Distinct Evaluation Cohort (TARGET-AI)

test

public 5,198 subjects · USA

science Capabilities & performance