ECGomics: An Open Platform for AI-ECG Digital Biomarker Discovery

Area of Science:

• Cardiology
• Biomedical Engineering
• Artificial Intelligence in Medicine

Background:

• Conventional electrocardiography (ECG) analysis struggles to balance expert interpretability with deep learning's predictive power.
• A need exists for a systematic framework integrating these approaches for comprehensive ECG analysis.

Purpose of the Study:

• To introduce ECGomics, a structured paradigm for ECG analysis.
• To bridge traditional feature engineering with deep learning using a unified taxonomy.

Main Methods:

• Deconstructing cardiac electrical signals into Structural, Intensity, Functional, and Comparative dimensions.
• Integrating expert-defined metrics with AI-derived latent embeddings to create digital biomarkers.
• Developing a scalable ecosystem including web, mobile, and API solutions.

Main Results:

• ECGomics demonstrated robust predictive performance across diverse clinical scenarios (e.g., atrial fibrillation, coronary stenosis).
• The framework maintained interpretability and required relatively low data.
• Validated flexibility and effectiveness in cardiovascular assessment.

Conclusions:

• ECGomics establishes an omics-level ECG representation system.
• Advances scalable precision cardiovascular assessment and data-driven health management.
• Provides a deployable digital biomarker ecosystem.