A novel framework that transforms ECG interpretation by combining supervised pre-training with multimodal learning to diagnose cardiac conditions more efficiently.

• 127 cardiac conditions accurately identified with less labeled data
• Achieves state-of-the-art performance across multiple downstream tasks
• Multimodal approach integrates different ECG formats (single-lead, 12-lead, etc.)
• Reduces dependency on extensive task-specific fine-tuning

This research addresses cardiovascular diseases—a leading cause of global mortality—by making ECG diagnostics more accessible, accurate, and efficient for healthcare providers.

SuPreME: A Supervised Pre-training Framework for Multimodal ECG Representation Learning