Learning Cardiac Features: ECG Biometrics Across Time and~Exercise
Summary
The paper evaluates ECG biometrics under realistic conditions involving exercise-induced stress and cross-session variability using a Siamese ResNet model, achieving state-of-the-art performance.
View Cached Full Text
Cached at: 09/21/26, 09:36 AM
# Learning Cardiac Features: ECG Biometrics Across Time and~Exercise Source: [https://arxiv.org/abs/2609.21962](https://arxiv.org/abs/2609.21962) Authors:[Luca Thiebaud](https://arxiv.org/search/cs?searchtype=author&query=Thiebaud,+L)\(AMU, AMU SCI, DIAPRO, LIS\),[Paul Chauchat](https://arxiv.org/search/cs?searchtype=author&query=Chauchat,+P)\(AMU SCI, AMU, LIS, DIAPRO\),[Mustapha Ouladsine](https://arxiv.org/search/cs?searchtype=author&query=Ouladsine,+M)\(AMU SCI, AMU, LIS, DIAPRO\),[Stéphane Delliaux](https://arxiv.org/search/cs?searchtype=author&query=Delliaux,+S)\(AMU, APHM, C2VN\) [View PDF](https://arxiv.org/pdf/2609.21962) > 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\. ## Submission history From: Luca Thiebaud \[[view email](https://arxiv.org/show-email/e743631b/2609.21962)\] \[via CCSD proxy\] **\[v1\]**Fri, 18 Sep 2026 16:22:31 UTC \(4,821 KB\)
Similar Articles
ECG-InterpBench: Benchmarking the Interpretability of ECG Foundation Models with Matched-Scale Sparse Autoencoders
ECG-InterpBench is a new benchmark that systematically evaluates the interpretability of ECG foundation model representations using matched-scale sparse autoencoders, covering reconstruction fidelity, clinical concept accessibility, and reproducibility across 450 cells.
ECG Mirage: Revealing and Mitigating the Underutilisation of ECGs in Vision-Language Models for Clinical Prediction
The study identifies 'ECG Mirage' in vision-language models for clinical prediction, where models underutilize ECG data despite apparent multimodal capability, and proposes visual prompt tuning as an efficient mitigation strategy.
Graph-Based Pseudo-multimodal Contrastive Learning for 12-Lead ECG Representations
This paper proposes Graph-CMMC, a graph-based pseudo-multimodal contrastive learning framework for 12-lead ECG representations, which effectively models inter-lead dependencies and achieves competitive performance in cardiac disease classification.
Analysis of Respiratory Sinus Arrhythmia with Neural Networks
This paper introduces a neural network-based approach for estimating respiratory rate from ECG signals by leveraging Respiratory Sinus Arrhythmia. It evaluates different deep learning architectures to automatically extract features, offering a scalable solution for non-invasive monitoring.
Holtercare-Bench: A Multimodal Benchmark for Evaluating Long-Term Dynamic ECG Analysis
Holtercare-Bench is a multimodal benchmark introduced to evaluate long-term dynamic ECG analysis using the Holtercare-23K dataset, revealing performance gaps in current MLLMs and providing improvements through fine-tuning for clinical applications.