Power, Control, and Data Processing Systems

Power, Control, and Data Processing Systems

Determining the Optimal Number of User Access Modifications in ECG-Based Authentication Systems Using a Transfer Learning Approach

Document Type : Original Research

Authors
Narges Eshaghi 1
Mohammad Habibi 2
Nasour Bagheri 3
Ali Khatibi 4
1 Department of Mathematics, Tafresh University, Tafresh 39518-79611, Iran
2 Department of Mathematics, Tafresh University, Tafresh, Iran
3 CPS2 Lab, Department of Communication, Faculty of Electrical Engineering, Shahid Rajaee Teacher Training University, Iran; Electronics Research Institute, Sharif University of Technology, Iran.
4 Assistant Professor, Department of Mechanical Engineering, Tafresh University, Tafresh 39518-79611, Iran
10.30511/pcdp.2026.2084603.1064
Abstract
A key challenge in processing electrocardiogram (ECG) signals for personalized biometric authentication systems is the efficient retraining of deep learning models when new user data become available. This study employs a transfer learning approach combined with the VGG16 architecture to investigate the permissible threshold for repeatedly integrating data from a single user—assigned with varying access labels—into a pre-trained base model. The base model was initially trained on 46 users from the MIT-BIH database. For each target user, data were simulated across 100 iterations with random label assignments and incrementally incorporated via gradual fine-tuning. The results reveal that the final model's performance improves with the addition of up to approximately 44 to 46 target models—a quantity comparable to the number of initial users—after which performance declines. The optimal range for adding new data from a single user was identified as 11 to 16 iterations. Within this interval, the model achieves an accuracy exceeding 99% without exhibiting catastrophic forgetting. These findings confirm that while transfer learning offers an efficient mechanism for frequent updates in ECG-based biometric systems, it possesses a finite capacity for accommodating repeated modifications. Determining this threshold provides valuable practical guidance for the development of security and health monitoring systems that must balance functional flexibility with overall model stability.
Keywords
Transfer Learning
Electrocardiogram (ECG) Signal
Catastrophic Forgetting
Biometric Authentication
Subjects
[1]    D. Bartusik-Aebisher, K. Rogóż, and D. Aebisher, “Artificial Intelligence and ECG: A New Frontier in Cardiac Diagnostics and Prevention,” 2025. doi: 10.3390/biomedicines13071685.
[2]    Wikimedia, 2026.  https://commons.wikimedia.org/wiki/File:SinusRhythmLabels.svg.
[3]    J. Ding, Q. Liu, B. W.-K. Ling, and W. Li, “Heart rate estimation based on joint convolutional neural network and conditional generative adversarial network via heart rate variabilities and other features extracted from photoplethysmograms,” Biomed. Signal Process. Control, vol. 112, p. 108831, 2026, doi: 10.1016/j.bspc.2025.108831.
[4]    V. A. Ardeti, V. R. Kolluru, G. T. Varghese, and R. K. Patjoshi, “An overview on state-of-the-art electrocardiogram signal processing methods: Traditional to AI-based approaches,” Expert Syst. Appl., vol. 217, p. 119561, 2023, doi: 10.1016/j.eswa.2023.119561.
[5]    D. Flores-Martin, F. Díaz-Barrancas, P. J. Pardo, J. Berrocal, and J. M. Murillo, “ECGTwinMentor: A digital twin-based system for personalized ECG learning and support,” SoftwareX, vol. 32, p. 102398, 2025, doi: 10.1016/j.softx.2025.102398.
[6]    Y. Yang, M. Li, and J. Liu, “Generative Diffusion-Based Task Incremental Learning Method for Decoding Motor Imagery EEG,” 2025. doi: 10.3390/brainsci15020098.
[7]    C. V Nguyen and C. D. Do, “Transfer learning in ECG diagnosis: Is it effective?,” PLoS One, vol. 20, no. 5, p. e0316043, May 2025, doi: 10.1371/journal.pone.0316043.
[8]    Ayeesha Soudagar et al., “Transfer Learning Strategies for Cardiovascular Disease Detection in ECG Imagery,” Biomed. Eng. Comput. Biol., vol. 16, p. 11795972251397812, Dec. 2025, doi: 10.1177/11795972251397812.
[9]    G. Monachino, B. Zanchi, M. Wand, G. Conte, A. Tzovara, and F. D. Faraci, “Overcoming data scarcity in life-threatening arrhythmia detection through transfer learning,” Commun. Med., vol. 5, no. 1, p. 248, 2025, doi: 10.1038/s43856-025-00982-9.
[10]    G. Petmezas et al., “State-of-the-Art Deep Learning Methods on Electrocardiogram Data: Systematic Review,” JMIR Med Inf., vol. 10, no. 8, p. e38454, 2022, doi: 10.2196/38454.
[11]    F. Hajamohideen et al., “Four-way classification of Alzheimer’s disease using deep Siamese convolutional neural network with triplet-loss function,” Brain Informatics, vol. 10, no. 1, p. 5, 2023, doi: 10.1186/s40708-023-00184-w.
[12]    M. Ezz, “Deep Learning-Driven Single-Lead ECG Classification: A Rapid Approach for Comprehensive Cardiac Diagnostics,” 2025. doi: 10.3390/diagnostics15030384.
[13]    H. Narotamo, M. Dias, R. Santos, A. V Carreiro, H. Gamboa, and M. Silveira, “Deep learning for ECG classification: A comparative study of 1D and 2D representations and multimodal fusion approaches,” Biomed. Signal Process. Control, vol. 93, p. 106141, 2024, doi: 10.1016/j.bspc.2024.106141.
[14]    S. Arif, M. M. Siddiqui, N. Alqahtani, K. Shah, M. A. Khan, and N. Kraiem, “Advanced Signal Processing Algorithm-Based State Estimation and Denoising of ECG Signals for Biomedical Applications,” IEEE Access, vol. 13, pp. 137240–137254, 2025, doi: 10.1109/ACCESS.2025.3595453.
[15]    E. G. Addisu, T. G. Yirga, H. G. Yirga, and A. D. Yehuala, “Transfer learning-based hybrid VGG16-machine learning approach for heart disease detection with explainable artificial intelligence,” Front. Artif. Intell., vol. Volume 8-, 2025, doi: 10.3389/frai.2025.1504281.
[16]    M. F. Abrar et al., “Automated heart disease detection using Swin Transformer and ECG signal processing: a high-accuracy approach,” Sci. Rep., vol. 15, no. 1, p. 39338, 2025, doi: 10.1038/s41598-025-23097-6.
[17]    G. B. Moody and R. G. Mark, “The impact of the MIT-BIH Arrhythmia Database,” IEEE Eng. Med. Biol. Mag., vol. 20, no. 3, pp. 45–50, 2001, doi: 10.1109/51.932724.
Power, Control, and Data Processing Systems
Volume 3, Issue 3
Summer 2026
Pages 10-17

  • Receive Date 08 February 2026
  • Revise Date 23 May 2026
  • Accept Date 24 May 2026
  • First Publish Date 24 May 2026
  • Publish Date 01 September 2026