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

Background: Conventional electrocardiogram (ECG) analysis faces a persistent dichotomy: expert-driven features ensure interpretability but lack sensitivity to latent patterns, while deep learning offers high accuracy but functions as a black box with high data dependency. We introduce ECGomics, a systematic paradigm and open-source platform for the multidimensional deconstruction of cardiac signals into digital biomarker. Methods: Inspired by the taxonomic rigor of genomics, ECGomics deconstructs cardiac activity across four dimensions: Structural, Intensity, Functional, and Comparative. This taxonomy synergizes expert-defined morphological rules with data-driven latent representations, effectively bridging the gap between handcrafted features and deep learning embeddings. Results: We operationalized this framework into a scalable ecosystem consisting of a web-based research platform and a mobile-integrated solution (https://github.com/PKUDigitalHealth/ECGomics). The web platform facilitates high-throughput analysis via precision parameter configuration, high-fidelity data ingestion, and 12-lead visualization, allowing for the systematic extraction of biomarkers across the four ECGomics dimensions. Complementarily, the mobile interface, integrated with portable sensors and a cloud-based engine, enables real-time signal acquisition and near-instantaneous delivery of structured diagnostic reports. This dual-interface architecture successfully transitions ECGomics from theoretical discovery to decentralized, real-world health management, ensuring professional-grade monitoring in diverse clinical and home-based settings. Conclusion: ECGomics harmonizes diagnostic precision, interpretability, and data efficiency. By providing a deployable software ecosystem, this paradigm establishes a robust foundation for digital biomarker discovery and personalized cardiovascular medicine.

💡 Research Summary

The paper introduces “ECGomics,” an open‑source, multi‑dimensional framework that transforms raw electrocardiogram (ECG) recordings into a rich set of digital biomarkers by integrating expert‑driven rule‑based features with deep‑learning‑derived embeddings. Recognizing the long‑standing trade‑off between interpretability (hand‑crafted morphological features) and predictive power (black‑box deep neural networks), the authors propose a taxonomy inspired by genomics that decomposes cardiac electrical activity into four inter‑related dimensions: Structural, Intensity, Functional, and Comparative.

Structural ECGomics captures classic morphological descriptors (e.g., P‑wave duration, QRS amplitude) using the ENCASE pipeline, which implements clinically validated rule sets. Intensity ECGomics quantifies the signal’s energy distribution and non‑linear dynamics through spectral analysis and R‑R interval skewness measures. Functional ECGomics assesses autonomic and physiological performance via heart‑rate variability, QRS fragmentation, and related metrics extracted by the FeatureDB framework. Comparative ECGomics leverages pre‑trained Net1D‑based models (CardioLearn? and ECGFounder) to generate high‑dimensional latent representations that are benchmarked against large population datasets, enabling the estimation of “ECG age,” disease risk scores, and systemic health indices.

Technically, the workflow consists of three stages. First, raw 12‑lead ECGs are processed through a hybrid pipeline that simultaneously extracts expert‑defined features and deep embeddings. Second, the resulting feature vectors are mapped onto a broad biomedical landscape, including demographics, laboratory biomarkers, imaging phenotypes, and multi‑omics data (genomics, transcriptomics, proteomics, metabolomics). This stage uncovers electro‑genetic signatures and systemic associations, positioning the heart as a holistic sensor of metabolic, renal, and endocrine dysfunction. Third, predictive models—primarily gradient‑boosted trees (XGBoost)—integrate the multi‑dimensional feature set to stratify risk for acute cardiovascular events, chronic kidney disease, anemia, and other non‑cardiac conditions with high accuracy (AUROC >0.88 in reported experiments).

To promote accessibility, the authors deliver both a web‑based research portal and a mobile application. The web interface supports high‑throughput batch uploads, precise parameter configuration, and 12‑lead visualization, while the mobile app connects to portable ECG sensors and a cloud engine for real‑time acquisition and near‑instantaneous diagnostic reports. All code and documentation are publicly available on GitHub (https://github.com/PKUDigitalHealth/ECGomics), allowing researchers worldwide to reproduce and extend the pipeline without requiring specialized hardware or deep‑learning expertise.

Empirical validation demonstrates that the Structural, Intensity, and Functional dimensions correlate strongly with established clinical markers, whereas the Comparative dimension yields novel biomarkers such as “cardiac age” and predicts laboratory values (e.g., NT‑proBNP) and imaging metrics (e.g., left ventricular ejection fraction). Integration of these features improves disease detection beyond conventional ECG interpretation, and correlation analyses reveal specific genetic variants linked to ECG waveform alterations, underscoring the platform’s potential for electro‑genomic research.

In summary, ECGomics bridges the gap between interpretability and predictive performance by providing a standardized, scalable, and open ecosystem for ECG‑based digital biomarker discovery. Its four‑dimensional taxonomy, combined with robust analytical pipelines and user‑friendly interfaces, positions it as a foundational tool for precision cardiology, systemic health monitoring, and future multi‑omics integration.