ECG-Age
A research tool from the Bioinformatics & Omics Data Science Platform

Computational cardiology research interface

ECG-Age

Foundation-model estimation of biological cardiac age from ECG images, designed for exploratory research into age acceleration, morphology, and model-derived cardiac phenotypes.

ECGFounder PTB-XL calibrated Image and PDF input Research use only

ECG-derived age deviation predicts cardiovascular diseases across lead configurations and cohorts . Deniz Aydogdu, Farieda Gaber, Arash Sorooshmehr, Altuna Akalin. medRxiv, 2026.

Start analysis Method overview

Method overview

Biological cardiac age, inferred from ECG morphology.

We developed an interactive ECG analysis tool that estimates ECG-derived biological age from uploaded ECG graphics and reports task-specific prediction probabilities. By allowing users to paste or upload an ECG image directly, the tool provides rapid access to AI-based age estimation without requiring raw signal files.

The output includes both the predicted ECG age and associated probabilities across diagnostic categories, enabling users to explore patterns of cardiac aging and disease-related ECG morphology. This tool is designed to support research into ECG age acceleration, model interpretability, and the potential clinical utility of foundation-model-derived ECG biomarkers.

Powered by ECGFounder — a foundation model trained on over 10 million recordings (Li et al., NEJM AI 2025), with calibration on PTB-XL.

ECG-Age pipeline: PhysioNet datasets (PTB-XL, MIMIC-IV) feed an ECGFounder feature-extraction stage producing a 1,024-dim embedding and 150 task probabilities, an ElasticNet age model with bias correction and disease association, and an analysis stage covering age acceleration, age deviation, mortality risk and conduction bias — tested across 12-lead clinical, 6-lead Kardia, and 1-lead wearable ECGs.
The ECG-Age pipeline for ECG-derived age deviation analysis.

ECG specimen intake

Submit an ECG for ECG-Age inference

Upload a clear ECG image or PDF to digitize the tracing and generate a research report.

For research use only — not a medical device and not for diagnostic decisions.

  1. 01 Upload ECG image or PDF
  2. 02 Digitize and inspect lead layout
  3. 03 Extract ECGFounder representation
  4. 04 Report age and associated patterns
We do not save your ECG.
No ECG handy? Try a sample:
Add patient details — optional, improves age-gap interpretation
Upload an ECG file to continue.