EMG pattern recognition for thumb muscle states using wearable sensing and adaptive neural network
Article Sidebar
Main Article Content
Abstract
Accurate classification of electrophysiological signals, particularly electromyographic (EMG) is essential for the development of advanced systems in sports medicine, biomechanical prosthetics, and neurocomputer interfaces. However, challenges such as signal noise, device portability, and real-time processing constraints limit the practical deployment of EMG-based interfaces. In this paper, we present a custom wearable device for EMG data acquisition and real-time classification of muscle activity. The device integrates an ESP32C6 microcontroller for wireless data transmission, an AD8232 analog sensor module for electrophysiological signal capture, and Ag/AgCl electrodes placed on antagonist muscles of the hand. EMG signals are sampled at 1000 Hz, preprocessed by normalization and filtering, and then classified using a two-layer feedforward neural network trained with the ADAM optimization algorithm. The dataset contains 4000 consecutive time series that reflect the dynamics of EMG signals across three thumb motor states: rest, abduction, and adduction. The neural network achieved a classification accuracy of 94% in real time, with high stability and minimal delay, demonstrating reliable detection of muscle activity patterns. The integration of low-cost hardware with an adaptive neural classifier enables efficient real-time EMG signal interpretation. The use of ADAM optimization ensures stable convergence and robustness to signal variability. This work contributes a compact and effective solution for real-time EMG classification, paving the way for its application in wearable rehabilitation systems, neuro-controlled prosthetics, and intelligent human–machine interfaces.
Article Details
Issue
Section
This work is licensed under a Creative Commons Attribution 4.0 International License.
References
1. Biloborodova, T., Skarga-Bandurova, I., Derkach, M., Matiuk, D., & Zagorodna, N. (2024). Identification of Salient Brain Regions for Anxiety Disorders Using Nonlinear EEG Feature Analysis. In Collaboration across Disciplines for the Health of People, Animals and Ecosystems (pp. 180-184). IOS Press.
2. Skarga-Bandurova, I., Biloborodova, T., Skarha-Bandurov, I., Boltov, Y., & Derkach, M. (2021). A Multilayer LSTM Auto-Encoder for Fetal ECG Anomaly Detection. In pHealth 2021 (pp. 147-152). IOS press.
3. He, J., Gao, F., Wang, J., Wu, Q., Zhang, Q., & Lin, W. (2022). A Method Combining Multi-Feature Fusion and Optimized Deep Belief Network for EMG-Based Human Gait Classification. Mathematics, 10(22), 4387.
4. Benalcázar, M. E., Valdivieso Caraguay, Á. L., & Barona López, L. I. (2020). A User-Specific Hand Gesture Recognition Model Based on Feed-Forward Neural Networks, EMGs, and Correction of Sensor Orientation. Applied Sciences, 10(23), 8604.
5. Mary Vasanthi S, Jayasree T, HAITER LENIN A et al. Electromyography Signal Based Hand Gesture Classification System Using Hilbert Huang Transform and Deep Neural Networks, 22 March 2023, PREPRINT (Version 1) available at Research Square.
6. Mary Vasanthi S, T Jayasree. Recognition of Hand Gestures using Wavelet Packet Transform and Cascaded Feed Forward Neural Networks, 18 January 2023, PREPRINT (Version 1) available at Research Square.
7. Hincapie JG, Kirsch RF. Feasibility of EMG-based neural network controller for an upper extremity neuroprosthesis. IEEE Trans Neural Syst Rehabil Eng. 2009 Feb;17(1):80-90.
8. Shim, Hm., Lee, S. Multi-channel electromyography pattern classification using deep belief networks for enhanced user experience. J. Cent. South Univ. 22, 1801–1808 (2015).
9. Dezhen Xiong, Daohui Zhang, Xingang Zhao and Yiwen Zhao, "Deep Learning for EMG-based Human-Machine Interaction: A Review," IEEE/CAA J. Autom. Sinica, vol. 8, no. 3, pp. 512-533, Mar. 2021.
10. Vanga Karunakar Reddy, Ravi Kumar Av. Multi-channel neuro signal classification using Adam-based coyote optimization enabled deep belief network. Biomedical Signal Processing and Control, 2022, 77:103774.
11. Kim, D. H., Lee, G., & Kim, S. H. (2023). An ECG Stitching Scheme for Driver Arrhythmia Classification Based on Deep Learning. Sensors, 23(6), 3257.
12. Al-Khazzar, A., Altaweel, Z., & Hussain, J. (2024). Using deep neural networks in classifying electromyography signals for hand gestures. IAES International Journal of Artificial Intelligence (IJ-AI), 13(1), 217-227.
13. Kharchenko, V., Kor, A. L., & Rucinski, A. (Eds.). (2022). Dependable IoT for Human and Industry: Modeling, Architecting, Implementation. CRC Press.
14. Babakov, R. M., Biloborodova, T. O., Brezhniev, E. V., Bojko, A. O., Bousher, V. V., Bykovyy, P. Y., ... & Waleed, A. A. (2019). Internet of Things for Industry and Human Application. Vol. 3.
15. Mishko O., Matiuk D., Derkach M. (2024) Security of remote iot system management by integrating firewall configuration into tunneled traffic. Scientific Journal of TNTU (Tern.), vol 115, no 3, pp. 122–129.