# Deep representation learning of electrocardiogram reveals biological insights in cardiac phenotypes and cardiovascular diseases

## TL;DR

This study uses deep learning on ECG data to uncover new biological insights and genetic links to heart health and disease.

## Contribution

A novel deep learning approach generates interpretable latent factors from ECGs, revealing new genetic associations and cardiovascular insights.

## Key findings

- Latent factors from ECG data correlate with traditional and emerging cardiac biomarkers.
- 170 genetic loci were identified, including 29 new ones linked to ECG phenotypes.
- Latent factors show strong associations with conduction, rhythm, and structural heart disorders.

## Abstract

Conventional approaches to analyzing electrocardiograms (ECG) in discrete parameters (such as the PR interval) ignored the high dimensionality of data omitted subtle but relevant information. We applied a variational auto-encoder to learn the underlying distributions of the ECG of 41,927 UK Biobank participants, generating 32-dimensional representation (latent factors). The latent factors showed correlations to conventional ECG parameters and strong associations to cardiac phenotypes estimated from magnetic resonance imaging. We found definitive associations of the latent factors to conduction, rhythm, and structural disorders (all p < 4.51 × 10−308) and additionally value in mortality prediction. Genome wide association study (GWAS) of the latent factors, revealed 170 genetic loci with 29 not previously associated with electrocardiographic phenotypes. Further characterization of the genetic signals suggested involvement in cardiac development, contractility, and electrophysiology. Our results supported that the deep representation learning of 12-lead ECG could provide clinically meaningful and interpretable insights into cardiovascular biology and health.

•AI-derived latent factors of the ECG linked to traditional and emerging biomarkers•Latent factors showed strong clinical associations with cardiovascular outcomes•GWAS identified 164 genetic loci with strong genetic correlations to outcomes•Enrichment analyses supported heart-specificity of current genetic discoveries

AI-derived latent factors of the ECG linked to traditional and emerging biomarkers

Latent factors showed strong clinical associations with cardiovascular outcomes

GWAS identified 164 genetic loci with strong genetic correlations to outcomes

Enrichment analyses supported heart-specificity of current genetic discoveries

Cardiovascular medicine; Artificial intelligence