Artificial intelligence electrocardiogram-predicted biological age gap and mortality: Capturing dynamic risk with...

Area of Science:

• Cardiology
• Gerontology
• Artificial Intelligence

Background:

• Artificial intelligence (AI) can predict biological age from electrocardiograms (ECGs), offering prognostic value for mortality.
• Serial ECG measurements are accessible and inexpensive, potentially improving individual risk stratification.

Purpose of the Study:

• To determine if repeated AI-derived biological age measurements reveal divergence from chronological aging.
• To assess if serial biological age improves all-cause mortality hazard estimates.

Main Methods:

• Retrospective cohort study of 46,960 cardiology patients with at least two ECGs.
• AI estimated biological age from each ECG; biological age gap (biological vs. chronological age) was calculated.
• Survival analysis used Cox proportional-hazards models, comparing single vs. multiple ECG models with log-likelihood ratio tests and C-indices.

Main Results:

• The mean biological aging rate was 0.7 ± 4.1 years/year.
• An increasing biological age gap correlated with higher mortality risk; negative gaps showed a protective effect.
• The multiple-ECG model demonstrated superior predictive accuracy (C-index 0.763 vs. 0.747) over the single-ECG model, with accuracy increasing up to 10 ECGs.

Conclusions:

• Biological aging often diverges from chronological aging in patients.
• While a single ECG's AI-derived biological age predicts mortality, serial measurements significantly enhance predictive accuracy.
• Serial biological age estimation using ECGs can refine risk assessment and guide personalized medical care.