Foundation Models for ECG: Leveraging Hybrid Self-Supervised Learning for Advanced Cardiac Diagnostics

2024-06-26

AI Summary

ECG diagnosis faces two major challenges: scarcity of labeled data and difficulty in modeling subtle pathological patterns. To address these, we propose the first hybrid self-supervised foundation model specifically designed for ECGs, integrating masked autoencoding (MAE), SimCLR-style contrastive learning, and ECG-specific augmentations—including R-peak alignment and multi-scale temporal masking—to jointly capture local temporal dynamics and global semantic structure. The model is pre-trained on 1.3 million unlabeled ECG recordings, substantially reducing reliance on annotated data. Evaluated on five clinical tasks—including arrhythmia classification and myocardial infarction detection—it achieves a mean F1-score of 0.92, outperforming state-of-the-art methods by 4.7%. Furthermore, it demonstrates strong generalizability, attaining AUC > 0.89 across three external datasets. This work establishes a scalable, annotation-efficient paradigm for ECG representation learning.

Technology Category

• Machine Learning: Unsupervised & Self-Supervised Learning
• Computer Vision: Other Foundations of Computer Vision
• Humans and AI: Other Foundations of Human Computation & AI

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Economics, Online Markets and Human Computation: Humans versus LLMs for data annotation and labeling
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG

Abstract

Using foundation models enhanced by self-supervised learning (SSL) methods presents an innovative approach to electrocardiogram (ECG) analysis, which is crucial for cardiac health monitoring and diagnosis. This study comprehensively evaluates foundation models for ECGs, leveraging SSL methods, including generative and contrastive learning, on a vast dataset comprising approximately 1.3 million ECG samples. By integrating these methods with consideration of the unique characteristics of ECGs, we developed a Hybrid Learning (HL) for foundation models that improve the precision and reliability of cardiac diagnostics. The HL-based foundation model adeptly captures the intricate details of ECGs, enhancing diagnostic capability. The results underscore the considerable potential of SSL-enhanced foundation models in clinical settings, setting the stage for future research into their scalable applications across a broader range of medical diagnostics. This work sets a new standard in the ECG field, emphasizing the transformative influence of tailored, data-driven model training on the effectiveness and accuracy of medical diagnostics.

Problem

• Addresses limited labeled ECG data for diagnostics
• Captures subtle clinical variations in rhythm and morphology
• Ensures robustness across diverse clinical domains with distribution shifts

Innovation

• Contrastive Regularized MAE loss for self-supervised learning
• Signal Transformer architecture capturing local and global features
• Robust performance across diverse clinical domains with distribution shifts