Autor(s):

Galya Georgieva-Tsaneva, Institute of Robotics, Bulgarian Academy of Sciences, Bulgaria, galitsaneva@abv.bg

Krasimir Cheshmedzhiev, Institute of Robotics, Bulgarian Academy of Sciences, Bulgaria, Krasimir Cheshmedzhiev

https://doi.org/10.55630/STEM.2025.0725

Abstract:

    ECG/PPG signal processing is a cornerstone of modern cardiovascular diagnostics. While artificial intelligence has already enhanced ECG analysis through accurate detection, classification, and prediction of cardiac events, its integration into mobile platforms enables continuous, ubiquitous monitoring. This paper introduces a novel framework that couples state-of-the-art AI methodologies with the Digital Twin paradigm to create a personalized, real-time virtual replica of a patient’s cardiac function. We survey deep learning and hybrid wavelet–neural approaches for QRS complex detection, arrhythmia classification, and heartbeat segmentation, and propose methods for incremental on-device learning to address data imbalance and inter-subject variability. Annotated datasets such as MIT-BIH are extended with synthetic augmentation to populate and calibrate the digital twin models, enabling generalization across heterogeneous populations. The proposed architecture emphasizes low-latency inference, energy-aware computation, and secure data flows suitable for mobile and wearable devices. By embedding interpretability layers and adaptive feedback loops, the system closes the gap between passive ECG monitoring and actionable, individualized cardiac care. Our results demonstrate that AI-driven ECG digital twins can significantly outperform traditional algorithms in accuracy and adaptability, filling a critical scientific gap and opening new pathways for predictive, preventive, and personalized cardiovascular healthcare.

Keywords:

Artificial Intelligence; ECG; PPG; Digital Twin; Mobile and Wearable Platforms; Deep Learning; Wavelet–Neural Networks; Incremental On-Device Learning; Interpretability; Personalized Cardiac Monitoring;

Received:

28-10-2025

Accepted:

22-12-2025

Published:

29-12-2025

Cite (APA style):

Georgieva-Tsaneva, G., Cheshmedzhiev, K. (2025). Artificial Intelligence for EGG/PPG Signal Processing on Mobile Platforms, Science Series "Innovative STEM Education", volume 07, ISSN: 2683-1333, Institute of Mathematics and Informatics – Bulgarian Academy of Sciences, pp. 272-290, DOI: https://doi.org/10.55630/STEM.2025.0725

PDF file address:

http://www.math.bas.bg/vt/stemedu/books/07/STEM.2025.0725.pdf