Scripps Research Develops ECG-CLIP AI for Heart Disease Detection

Scripps Research scientists published a study in Lancet Digital Health on September 1, 2026, introducing ECG-CLIP, an AI foundation model that detects heart disease using 91% less labeled data.

The model was trained on over 1.7 million ECGs from 540,000 people and requires only about a dozen confirmed examples of a disease to achieve high diagnostic performance.

ECG-CLIP outperformed existing models in detecting acute myocardial infarction and cardiac amyloidosis, while also predicting atrial fibrillation and 30-day survival outcomes after surgery.