DE-PADA: Personalized Augmentation and Domain Adaptation for ECG Biometrics Across Physiological States

TL;DR

This work tackles the challenge of ECG-based biometrics under varying physiological states, particularly post-exercise heart-rate increases. It introduces DE-PADA, a Dual Expert model that separately processes PQRS and ST intervals, augmented by subject-specific ST augmentation and domain adaptation using an Auxiliary subject cohort. The combination yields substantial gains in post-exercise identification (e.g., Ex_P2 +$26.75 ext{%}$, Ex_P1 +$11.72 ext{%}$) while preserving resting-state performance (Sit IDR ≈ $98.12 ext{%}$), demonstrating improved robustness to intra-subject variability. The approach offers practical improvements for real-world ECG biometrics by enabling reliable identification across diverse states without requiring exercise data in training.

Abstract

Electrocardiogram (ECG)-based biometrics offer a promising method for user identification, combining intrinsic liveness detection with morphological uniqueness. However, elevated heart rates introduce significant physiological variability, posing challenges to pattern recognition systems and leading to a notable performance gap between resting and post-exercise conditions. Addressing this gap is critical for advancing ECG-based biometric systems for real-world applications. We propose DE-PADA, a Dual Expert model with Personalized Augmentation and Domain Adaptation, designed to enhance robustness across diverse physiological states. The model is trained primarily on resting-state data from the evaluation dataset, without direct exposure to their exercise data. To address variability, DE-PADA incorporates ECG-specific innovations, including heartbeat segmentation into the PQRS interval, known for its relative temporal consistency, and the heart rate-sensitive ST interval, enabling targeted feature extraction tailored to each region's unique characteristics. Personalized augmentation simulates subject-specific T-wave variability across heart rates using individual T-wave peak predictions to adapt augmentation ranges. Domain adaptation further improves generalization by leveraging auxiliary data from supplementary subjects used exclusively for training, including both resting and exercise conditions. Experiments on the University of Toronto ECG Database demonstrate the model's effectiveness. DE-PADA achieves relative improvements in post-exercise identification rates of 26.75% in the initial recovery phase and 11.72% in the late recovery phase, while maintaining a 98.12% identification rate in the sitting position. These results highlight DE-PADA's ability to address intra-subject variability and enhance the robustness of ECG-based biometric systems across diverse physiological states.

Figures (9)

• Figure 1: Overview of the preprocessing flow, with an average size of $W=3$ used for illustration.
• Figure 1: Personalized Augmentation Algorithm
• Figure 1: The input signal is split to PQRS and ST intervals for the Dual Expert model with a 50 ms overlap.
• Figure 1: (a) The architecture of the Standard CNN. The CNN backbone consists of 1D convolutional layers (in_ch$\times$out_ch$\times$kernel/size/stride/padding), batch normalization, MaxPooling, and ReLU activations. The feature maps are flattened and passed to the MLP classifier, which includes Fully Connected (FC) layers, ReLU activation, dropout, and a final softmax layer for classification. (b) The DE model includes the pre-trained PQRS and ST backbones, which share the architecture with the Standard model backbone. The MLP classifiers of the Standard and DE models differ only by the input dimension.
• Figure 1: Classifier output dimensions during domain adaptation. (a) During training the output dimension includes subjects from Target and Auxiliary sets. (b) After training the classes corresponding to the Auxiliary set are removed.