Outlive
LongevityResearchHub

AI Reads Your Heart's Age to Predict Heart Disease Risk

AI-Derived ECG Age Gap from ECGFounder as a Digital Biomarker for Predicting Cardiovascular Risk

TL;DR

Researchers used artificial intelligence to analyze ECGs and calculate a person's cardiac "biological age"—how old their heart appears versus their calendar age. They found that people whose hearts looked older than they actually were had significantly higher risk of heart attacks and strokes, suggesting this AI-derived metric could help identify high-risk patients before they get sick.

Why This Matters

A simple heart test could help doctors spot who's at highest risk for heart attacks or strokes years before problems start.

Credibility Assessment Preliminary — 37/100
Study Design
Rigor of the research methodology
8/20
Sample Size
Whether the study was sufficiently powered
15/20
Peer Review
Review status and journal reputation
3/20
Replication
Has this finding been independently reproduced?
5/20
Transparency
Funding disclosure and data availability
6/20
Overall
Sum of all five dimensions
37/100

What this means

This AI tool shows promise for identifying heart disease risk before symptoms appear, but it's preliminary research that needs peer review and validation before doctors would use it clinically. The exciting concept—reading your heart's 'biological age'—could eventually help guide prevention, but we don't yet know if the AI is detecting real aging or just statistical patterns.

Red Flags: Major: Preprint without peer review. Moderate: Modest model performance (r=0.55 age prediction); unclear mechanisms underlying associations; no discussion of clinical utility or cost-effectiveness; limited transparency on hyperparameter selection and potential for overfitting; zero citations (very recent).

Cardiovascular disease kills more people globally than any other cause, yet we still struggle to identify who will have a heart attack or stroke before it happens. Traditional risk factors like cholesterol and blood pressure are helpful but imperfect. This study explores whether artificial intelligence can extract hidden aging signals from a routine ECG—a simple, non-invasive electrical recording of the heart—to predict future cardiovascular problems.

The researchers trained a deep learning model called ECGFounder on nearly 27,000 healthy people's ECGs to establish what a "normal" heart's electrical signature looks like at each age. The model learned to predict a person's age from their ECG pattern. Importantly, they then tested whether discrepancies between predicted age and actual age—the "age gap"—could predict future cardiac events. They analyzed 67,824 ECGs from 63,512 UK Biobank participants and validated findings in an independent Chinese hospital cohort.

The results are striking: for every additional year the AI predicted someone's heart to be older than their actual age, cardiovascular risk jumped 13%. People whose hearts looked more than 6 years older than expected had 4.5 times higher risk of major cardiac events. Conversely, those whose hearts appeared younger than expected had protective effects. These associations held even after accounting for traditional risk factors like age, smoking, and diabetes.

However, several limitations deserve emphasis. First, this is a preprint—not yet peer-reviewed—so findings require independent verification. Second, the model explains only modest variance in age (r=0.55), meaning individual predictions have substantial uncertainty. Third, we don't know the biological mechanisms: does this reflect genuine accelerated aging, or just ECG patterns associated with hidden disease? The study is essentially correlational—demonstrating association, not causation. Finally, clinical utility remains unclear: would this information change treatment decisions beyond existing risk scores?

For longevity research, this represents an intriguing direction: using AI to extract aging signals from existing medical data could enable cheap, scalable screening. However, the field must move carefully from association to actionable biology. Does accelerated cardiac age reflect cellular senescence? Mitochondrial dysfunction? Or just ECG remodeling from undetected hypertension? Understanding mechanism would transform this from a statistical predictor into a true biological biomarker and potential intervention target.

This work exemplifies modern longevity science: mining big datasets and AI models to identify novel aging biomarkers. But biomarkers are only useful if they guide intervention. The next phase requires prospective validation, mechanistic studies, and ultimately randomized trials showing that targeting the underlying biology improves outcomes.

View Original Source

0 Comments