Learning Cardiac Features: ECG Biometrics Across Time and~Exercise
A Siamese ResNet identifies people from ECG under exercise stress and across sessions at low error rates.
The paper tests ECG biometrics beyond single-session resting recordings, measuring robustness to exercise stress and change across sessions. A Siamese ResNet with late multi-lead fusion, trained on cardiopulmonary exercise-test ECGs, achieves a 1.7% intra-session rest-to-peak equal error rate and 3.9% EER on the public CYBHi dataset. The authors conclude that a subject-specific cardiac signature can persist despite physiological and temporal drift, with possible use for authentication and self-supervised pretraining.
- Siamese ResNet with late multi-lead fusion identifies people from ECG.
- Intra-session rest-to-peak equal error rate is 1.7%.
- Reports 3.9% equal error rate on the public CYBHi dataset.
- Evaluation covers exercise stress and cross-session variability.
Full article147 words · extracted from arxiv.org · click to collapse
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.
Text extracted automatically; images, tables and formatting may be missing. Original: https://arxiv.org/abs/2609.21962