Learning Cardiac Features: ECG Biometrics Across Time and~Exercise
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.
Related
- Cardiopulmonary Exercise Testing in Patients with Cardiac Pacemakers: From Device Optimization to Individualized Cardiac Rehabilitation
- Dynamic graph vector autoregression for characterizing cardiac electromechanical dynamics during incremental exercise from the joint phase space of electrocardiogram and phonocardiogram
- The Hidden Danger of Aortic Stenosis-Exercise-Induced Cardiac Arrest in a Young Man with Bicuspid Aortic Valve.
- Exercise‐based cardiac rehabilitation for coronary heart disease
- Echocardiographic characteristics and clinical features of cardiac involvement in Behçet’s disease: a retrospective single-center case series
BibTeX
@misc{thiebaud2026learning,
title = {{Learning Cardiac Features: ECG Biometrics Across Time and~Exercise}},
author = {Luca Thiebaud and Paul Chauchat and Mustapha Ouladsine and Stéphane Delliaux},
howpublished = {arXiv},
year = {2026},
month = sep,
doi = {10.48550/arxiv.2609.21962},
eprint = {2609.21962},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2609.21962},
}
RIS
TY - UNPB
TI - Learning Cardiac Features: ECG Biometrics Across Time and~Exercise
AU - Luca Thiebaud
AU - Paul Chauchat
AU - Mustapha Ouladsine
AU - Stéphane Delliaux
PY - 2026
DA - 2026/09/18
PB - arXiv
DO - 10.48550/arxiv.2609.21962
UR - https://arxiv.org/abs/2609.21962
AB - 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.
ER -