ECG biometric signals hold identity under exercise and time changes
Learning Cardiac Features: ECG Biometrics Across Time and~Exercise
Artificial Intelligence
Summary
Your heartbeat signals (ECGs) have unique features that can identify you, like a fingerprint. This paper shows these features remain reliable even if you are exercising or tested at different times. The researchers trained a computer model on a large dataset to recognize these patterns despite changes caused by exercise or time between sessions. Their results showed low error rates, proving ECGs have a stable identity signature over various conditions.
What this means in practice
- •For healthcare device manufacturers: Develop wearable heart monitors that authenticate users accurately during daily activities including exercise, improving data security.$Commercial implications: Enables secure personal identification in consumer and medical heart-monitoring devices by proving resilience against physical and time variations.
- •For data security teams: Integrate ECG biometrics as a method to secure sensitive cardiac data by confirming user identity even under physiological changes.
Authors
Luca Thiebaud, Paul Chauchat, Mustapha Ouladsine, Stéphane Delliaux
Abstract
Electrocardiograms (ECGs) carry subject-specific patterns enabling reliable individual discrimination, forming the basis of ECG biometrics. Beyond authentication, this paradigm holds significant potential to secure sensitive cardiac data and to serve as a pretext task in self-supervised learning. Yet, most studies remain confined to singlesession, resting data, leaving robustness to temporal and physiological variations largely untested. We address this gap by evaluating ECG biometrics under realistic conditions involving exercise-induced stress and cross-session variability. A Siamese ResNet with late multi-lead fusion strategy is trained on a large ECG dataset extracted from cardiopulmonary exercise tests and evaluated with a exercise-and time-aware protocol, as well as on public benchmarks. This first extensive assessment of ECG biometrics under combined physiological and temporal variability achieves an intra-session rest-to-peak EER of 1.7% and stateof-the-art 3.9% on the CYBHi dataset. Findings support the presence of an intrinsic cardiac signature resilient to physiological and temporal drift.