Siamese ResNet for Robust ECG Biometrics Under Exercise Stress
September 21, 2026
A Siamese ResNet with late multi-lead fusion was developed to maintain ECG biometric accuracy during exercise and across different time sessions. The model addresses physiological variability to improve subject discrimination in non-resting, real-world cardiac monitoring scenarios.
HOW THIS AFFECTS YOU
●
researcherYou can leverage these findings to move biometric research beyond static, resting-state datasets.
●
healthThis provides a path toward more secure, continuous cardiac identity verification in wearable devices.