Interpretable ECG heart age and serial electrical aging in 399,786 electrocardiograms.

Background

Interpretable ECG Heart Age models estimate the Heart Age Gap, the difference between ECG-predicted and chronological age. Whether a single measurement of accumulated burden predicts the future pace of electrical aging is unknown.

Objectives

To derive and internally validate an interpretable ECG Heart Age from routine 12‑lead measurements, quantify within-person aging using a measurement-based Electrical Aging Score (mEAS), and test whether baseline Heart Age Gap predicts subsequent mEAS progression.

Methods

We analyzed 399,786 standard 12‑lead ECGs from 128,605 patients recorded from April 2015 to June 2025. ECG Heart Age was calculated from six routine ECG measurements and sex, compared with a sex-specific expected value from strict-normal ECGs, and internally validated in a 20% held-out sample. Annualized mEAS slopes in 7885 patients with at least three ECGs spanning more than 5 years were modeled using piecewise mixed-effects regression.

Results

In held-out strict-normal ECGs (n = 15,701 from 10,439 patients), ECG Heart Age matched chronological age (median gap -0.19 years), whereas abnormal ECGs appeared older (median gap +6.58 years; area under the curve 0.79). All aging markers increased over time (all p < 0.001), and mEAS progression accelerated 3.3-fold after age 65. Baseline Heart Age Gap did not predict subsequent mEAS slope (r = -0.08). In a 3-year landmark analysis, higher early mEAS slope was associated with later left bundle branch block and paced rhythm but not atrial fibrillation or flutter.

Conclusions

ECG Heart Age reflects accumulated electrical burden, whereas serial mEAS reflects ongoing pace. Baseline burden did not predict subsequent pace, so a single ECG does not convey both; whether the two index distinct biological processes remains untested. External validation against clinical outcomes is required before clinical use.