Cluster-guided adaptive Transformer for muscle fatigue prediction_Journal of Biomedical Engineering

Authors：

FAN Bo ^1,2 ,  BAO Xueliang ^1,2 , DING Lei ³ , WU Jiao ⁴

1. School of Information Engineering, Ningxia University, Yinchuan 750021, P. R. China;
2. Key Laboratory of Artificial Intelligence and Information Security for Channeling Computing Resources from the East to the West of Ningxia, Yinchuan 750021, P. R. China;
3. School of Electronic and Electrical Engineering, Ningxia University, Yinchuan 750021, P. R. China;
4. People’s Hospital of Ningxia Hui Autonomous Region, Yinchuan 750004, P. R. China;

Corresponding?author：

BAO Xueliang, Email: xlbao@nxu.edu.cn

Keywords：

Muscle fatigue prediction; Surface electromyography; Neural architecture search; Gaussian distribution; Transformer

DOI：

10.7507/1001-5515.202601012

Video：

Export PDF Favorites Scan Get Citation

Abstract Full text Figures/Tables Video References Cited by

Due to the significant non-stationarity and feature distribution discrepancies in surface electromyography (sEMG) signals during muscle fatigue monitoring, traditional fixed-parameter Transformer models often struggle to accurately capture the complex evolution of time-frequency characteristics across different fatigue stages. To address this limitation, this paper proposes a K-means clustering-guided neural architecture search method (CG-NAS) to achieve adaptive optimization of Transformer architectures based on data distribution characteristics. The method first classified input EMG features using the K-means clustering algorithm and constructed Gaussian distributions characterized by mean and variance to quantify the complexity of each cluster. These distribution priors then guided the neural architecture search process, enabling dynamic alignment between the architecture search space and data characteristics: for low-complexity data clusters with small mean and variance, lightweight Transformer architectures were selected, whereas for high-complexity clusters, architectures with greater width and depth were allocated. Experimental results demonstrated the superior performance of CG-NAS in muscle fatigue index prediction tasks, achieving a mean absolute error of 0.098 2 and a coefficient of determination of 0.957 3, significantly outperforming multiple benchmark models. The study shows that CG-NAS effectively aligns with the nonlinear evolution of time-frequency features during the fatigue process and provides an efficient and robust solution for fatigue monitoring.

Citation： FAN Bo, BAO Xueliang, DING Lei, WU Jiao. Cluster-guided adaptive Transformer for muscle fatigue prediction. Journal of Biomedical Engineering, 2026, 43(2): 250-258. doi: 10.7507/1001-5515.202601012 Copy

1.	Battiston B, Titolo P, Ciclamini D, et al. Peripheral nerve defects: overviews of practice in Europe. Hand Clin, 2017, 33(3): 545-550.
2.	Popovic M R, Masani K, Milosevic M, et al. Functional electrical stimulation therapy: mechanisms for recovery of function following spinal cord injury and stroke// Reinkensmeyer D J, Dietz V. Neurorehabilitation technology. Berlin: Springer, 2022: 401-427.
3.	Al-Mulla M R, Sepulveda F, Colley M A, et al. Review of non-invasive techniques to detect and predict localised muscle fatigue. Sensors, 2011, 11(4): 3545-3594.
4.	Ma X, Chen S, Li Q, et al. Identification of lower limb muscle fatigue in basketball players based on sEMG signals. Front Physiol, 2025, 16: 1689324.
5.	Srinidhi G, Agrawal K N, Kumari S, et al. Muscle fatigue assessment using surface electromyography in farm operations performed in protected cultivation. Sci Rep, 2025, 15(1): 33758.
6.	Yu C, Zhan J, Xu L, et al. Motor control performance-related modulation of beta-band EEG-sEMG coherence differs between general and local muscular exercise-induced fatigue. Eur J Appl Physiol, 2025, 125(7): 1869-1879.
7.	Zhang X. Research on gesture definition and electrode placement in pattern recognition of hand gesture action sEMG// Zhang D. Proceedings of the International Conference on Medical Biometrics. Berlin: Springer, 2008: 33-40.
8.	Tkach D, Huang H, Kuiken T A, et al. Study of stability of time-domain features for electromyographic pattern recognition. J Neuroeng Rehabil, 2010, 7(1): 21.
9.	Phinyomark A, Quaine F, Charbonnier S, et al. Feature extraction of the first difference of EMG time series for EMG pattern recognition. Comput Methods Programs Biomed, 2014, 117(2): 247-256.
10.	Coraggio G, Cera M, Cirelli M, et al. Review and comparison of linear algorithms to quantify muscle fatigue based on sEMG signals. Ergonomics, 2024, 67(11): 1729-1747.
11.	Thongpanja S, Phinyomark A, Phukpattaranont P, et al. Mean and median frequency of EMG signal to determine muscle force based on time-dependent power spectrum. Electron Electr Eng, 2013, 19(3): 51-56.
12.	Ahmad I, Kim J Y. Assessment of whole body and local muscle fatigue using electromyography and a perceived exertion scale for squat lifting. Int J Environ Res Public Health, 2018, 15(4): 784.
13.	Karthick P A, Ghosh D M, Ramakrishnan S. Surface electromyography based muscle fatigue detection using high-resolution time-frequency methods and machine learning algorithms. Comput Methods Programs Biomed, 2018, 154: 45-56.
14.	Daniel N, Ma?achowski J, Hyla M, et al. Wavelet analysis of the EMG signal to assess muscle fatigue in the lower extremities during symmetric movement on a rowing ergometer. Acta Bioeng Biomech, 2023, 25(2): 15-27.
15.	Sowmya S, Banerjee S S, Swaminathan R, et al. Assessment of muscle fatigue using phase entropy of sEMG signals during dynamic contractions of biceps brachii// 2023 9th International Conference on Control, Decision and Information Technologies (CoDIT). Piscataway: IEEE, 2023: 2253-2256.
16.	Wang Z, Guan X, Li D, et al. Study on muscle fatigue classification for manual lifting by fusing sEMG and MMG signals. Sensors, 2025, 25(16): 5023.
17.	Schrunder A D F, Huang Y K, Rodriguez S, et al. A real-time muscle fatigue detection system based on multifrequency EIM and sEMG for effective NMES. IEEE Sens J, 2024, 24(14): 22553-22564.
18.	Wang J, Sun S, Sun Y. A muscle fatigue classification model based on LSTM and improved wavelet packet threshold. Sensors, 2021, 21(19): 6369.
19.	Li K, Sun Y, Li J, et al. Characterization of muscle fatigue degree in cyclical movements based on the high-frequency components of sEMG. Biomimetics, 2025, 10(5): 291.
20.	Gharehbaghi A, Linden M, Akan A. A deep machine learning method for classifying cyclic time series of biological signals using time-growing neural network. IEEE Trans Neural Netw Learn Syst, 2018, 29(9): 4102-4115.
21.	Zhang G, Yang B, Zan P, et al. Multilevel assessment of exercise fatigue utilizing multiple attention and convolution network (MACNet) based on surface electromyography. IEEE Trans Neural Syst Rehabil Eng, 2025, 33: 243-254.
22.	Bala S, Vishnu V Y, Joshi D, et al. MEFFNet: forecasting myoelectric indices of muscle fatigue in healthy and post-stroke during voluntary and FES-induced dynamic contractions. IEEE Trans Neural Syst Rehabil Eng, 2024, 32: 2598-2611.
23.	Ao J, Liang S, Yan T, et al. Overcoming the effect of muscle fatigue on gesture recognition based on sEMG via generative adversarial networks. Expert Syst Appl, 2024, 238: 122304.
24.	Li M, Li J, Shu M. Detection of muscle fatigue by fusion of agonist and synergistic muscle sEMG signals// 2020 IEEE 33rd International Symposium on Computer-Based Medical Systems (CBMS). Rochester: IEEE, 2020: 95-98.
25.	Zhang W, Bai Z, Yan P, et al. Recognition of human lower limb motion and muscle fatigue status using a wearable FES-sEMG system. Sensors, 2024, 24(7): 2377.
26.	Wen Q, Zhou T, Zhang C, et al. Transformers in time series: a survey// Proceedings of the 32nd International Joint Conference on Artificial Intelligence (IJCAI). San Francisco: Morgan Kaufmann, 2023: 6750-6760.
27.	Dang Y K, Liu Z T, Yang X X, et al. A fatigue assessment method based on attention mechanism and surface electromyography. Internet Things Cyber-Phys Syst, 2023, 3: 112-120.
28.	Bonakdar A, Disselhorst-Klug C, Beltran Martinez K, et al. Perceived fatigue progression tracking during manual handling tasks using sEMG recordings. J Neuroeng Rehabil, 2025, 22(1): 245.
29.	Yu J, Yang J, Zhu C, et al. Hybrid CNN-LSTM-Transformer model for robust muscle fatigue detection during rehabilitation using sEMG signals// 2024 International Conference on Advanced Robotics and Mechatronics (ICARM). Piscataway: IEEE, 2024: 1-6.
30.	Acharya S, Kisi K, Gautam S R, et al. A high-performance hybrid Transformer-LSTM-XGBoost model for sEMG-based fatigue detection in simulated roofing postures. Buildings, 2025, 15(17): 3005.
31.	Tu P, Li J, Wang H. A multitask deep learning approach for sEMG-based human motion intention and muscle fatigue levels recognition. IEEE Trans Instrum Meas, 2025, 74: 1-10.
32.	Zendehbad S A, Ghasemi J, Samsami Khodadad F. FatigueNet: a hybrid graph neural network and transformer framework for real-time multimodal fatigue detection. Sci Rep, 2025, 15: 33781.
33.	Xi K, Lv D, Yang C, et al. Adaptive bandwidth chirp mode decomposition for muscle fatigue analysis. Measurement, 2026, 257(B): 118679.
34.	Wang S, Tang H, Wang B, et al. A novel approach to detecting muscle fatigue based on sEMG by using neural architecture search framework. IEEE Trans Neural Netw Learn Syst, 2023, 34(8): 4932-4943.
35.	Edwards J J, Coleman D A, Ritti-Dias R M, et al. Isometric exercise training and arterial hypertension: an updated review. Sports Med, 2024, 54(6): 1515-1529.
36.	Oh B M, Han T S, Lee W H, et al. Reliability of surface electromyography from the lower-limb muscles during maximal and submaximal voluntary isometric contractions in in-bed healthy individuals and patients with subacute stroke. Brain Neurorehabil, 2024, 17(2): e14.
37.	Borg G A V. Psychophysical bases of perceived exertion. Med Sci Sports Exerc, 1982, 14(5): 377-381.
38.	Williams N. The Borg Rating of Perceived Exertion (RPE) scale. Occup Med (Lond), 2017, 67(5): 404-405.
39.	Hussain J, Sundaraj K, Subramaniam I D, et al. Analysis of fatigue in the three heads of the triceps brachii during isometric contractions at various effort levels. J Musculoskelet Neuronal Interact, 2019, 19(3): 276-285.
40.	Merletti R, Knaflitz M, De Luca C J, et al. Myoelectric manifestations of fatigue in voluntary and electrically elicited contractions. J Appl Physiol, 1990, 69(5): 1810-1820.

1. Battiston B, Titolo P, Ciclamini D, et al. Peripheral nerve defects: overviews of practice in Europe. Hand Clin, 2017, 33(3): 545-550.
2. Popovic M R, Masani K, Milosevic M, et al. Functional electrical stimulation therapy: mechanisms for recovery of function following spinal cord injury and stroke// Reinkensmeyer D J, Dietz V. Neurorehabilitation technology. Berlin: Springer, 2022: 401-427.
3. Al-Mulla M R, Sepulveda F, Colley M A, et al. Review of non-invasive techniques to detect and predict localised muscle fatigue. Sensors, 2011, 11(4): 3545-3594.
4. Ma X, Chen S, Li Q, et al. Identification of lower limb muscle fatigue in basketball players based on sEMG signals. Front Physiol, 2025, 16: 1689324.
5. Srinidhi G, Agrawal K N, Kumari S, et al. Muscle fatigue assessment using surface electromyography in farm operations performed in protected cultivation. Sci Rep, 2025, 15(1): 33758.
6. Yu C, Zhan J, Xu L, et al. Motor control performance-related modulation of beta-band EEG-sEMG coherence differs between general and local muscular exercise-induced fatigue. Eur J Appl Physiol, 2025, 125(7): 1869-1879.
7. Zhang X. Research on gesture definition and electrode placement in pattern recognition of hand gesture action sEMG// Zhang D. Proceedings of the International Conference on Medical Biometrics. Berlin: Springer, 2008: 33-40.
8. Tkach D, Huang H, Kuiken T A, et al. Study of stability of time-domain features for electromyographic pattern recognition. J Neuroeng Rehabil, 2010, 7(1): 21.
9. Phinyomark A, Quaine F, Charbonnier S, et al. Feature extraction of the first difference of EMG time series for EMG pattern recognition. Comput Methods Programs Biomed, 2014, 117(2): 247-256.
10. Coraggio G, Cera M, Cirelli M, et al. Review and comparison of linear algorithms to quantify muscle fatigue based on sEMG signals. Ergonomics, 2024, 67(11): 1729-1747.
11. Thongpanja S, Phinyomark A, Phukpattaranont P, et al. Mean and median frequency of EMG signal to determine muscle force based on time-dependent power spectrum. Electron Electr Eng, 2013, 19(3): 51-56.
12. Ahmad I, Kim J Y. Assessment of whole body and local muscle fatigue using electromyography and a perceived exertion scale for squat lifting. Int J Environ Res Public Health, 2018, 15(4): 784.
13. Karthick P A, Ghosh D M, Ramakrishnan S. Surface electromyography based muscle fatigue detection using high-resolution time-frequency methods and machine learning algorithms. Comput Methods Programs Biomed, 2018, 154: 45-56.
14. Daniel N, Ma?achowski J, Hyla M, et al. Wavelet analysis of the EMG signal to assess muscle fatigue in the lower extremities during symmetric movement on a rowing ergometer. Acta Bioeng Biomech, 2023, 25(2): 15-27.
15. Sowmya S, Banerjee S S, Swaminathan R, et al. Assessment of muscle fatigue using phase entropy of sEMG signals during dynamic contractions of biceps brachii// 2023 9th International Conference on Control, Decision and Information Technologies (CoDIT). Piscataway: IEEE, 2023: 2253-2256.
16. Wang Z, Guan X, Li D, et al. Study on muscle fatigue classification for manual lifting by fusing sEMG and MMG signals. Sensors, 2025, 25(16): 5023.
17. Schrunder A D F, Huang Y K, Rodriguez S, et al. A real-time muscle fatigue detection system based on multifrequency EIM and sEMG for effective NMES. IEEE Sens J, 2024, 24(14): 22553-22564.
18. Wang J, Sun S, Sun Y. A muscle fatigue classification model based on LSTM and improved wavelet packet threshold. Sensors, 2021, 21(19): 6369.
19. Li K, Sun Y, Li J, et al. Characterization of muscle fatigue degree in cyclical movements based on the high-frequency components of sEMG. Biomimetics, 2025, 10(5): 291.
20. Gharehbaghi A, Linden M, Akan A. A deep machine learning method for classifying cyclic time series of biological signals using time-growing neural network. IEEE Trans Neural Netw Learn Syst, 2018, 29(9): 4102-4115.
21. Zhang G, Yang B, Zan P, et al. Multilevel assessment of exercise fatigue utilizing multiple attention and convolution network (MACNet) based on surface electromyography. IEEE Trans Neural Syst Rehabil Eng, 2025, 33: 243-254.
22. Bala S, Vishnu V Y, Joshi D, et al. MEFFNet: forecasting myoelectric indices of muscle fatigue in healthy and post-stroke during voluntary and FES-induced dynamic contractions. IEEE Trans Neural Syst Rehabil Eng, 2024, 32: 2598-2611.
23. Ao J, Liang S, Yan T, et al. Overcoming the effect of muscle fatigue on gesture recognition based on sEMG via generative adversarial networks. Expert Syst Appl, 2024, 238: 122304.
24. Li M, Li J, Shu M. Detection of muscle fatigue by fusion of agonist and synergistic muscle sEMG signals// 2020 IEEE 33rd International Symposium on Computer-Based Medical Systems (CBMS). Rochester: IEEE, 2020: 95-98.
25. Zhang W, Bai Z, Yan P, et al. Recognition of human lower limb motion and muscle fatigue status using a wearable FES-sEMG system. Sensors, 2024, 24(7): 2377.
26. Wen Q, Zhou T, Zhang C, et al. Transformers in time series: a survey// Proceedings of the 32nd International Joint Conference on Artificial Intelligence (IJCAI). San Francisco: Morgan Kaufmann, 2023: 6750-6760.
27. Dang Y K, Liu Z T, Yang X X, et al. A fatigue assessment method based on attention mechanism and surface electromyography. Internet Things Cyber-Phys Syst, 2023, 3: 112-120.
28. Bonakdar A, Disselhorst-Klug C, Beltran Martinez K, et al. Perceived fatigue progression tracking during manual handling tasks using sEMG recordings. J Neuroeng Rehabil, 2025, 22(1): 245.
29. Yu J, Yang J, Zhu C, et al. Hybrid CNN-LSTM-Transformer model for robust muscle fatigue detection during rehabilitation using sEMG signals// 2024 International Conference on Advanced Robotics and Mechatronics (ICARM). Piscataway: IEEE, 2024: 1-6.
30. Acharya S, Kisi K, Gautam S R, et al. A high-performance hybrid Transformer-LSTM-XGBoost model for sEMG-based fatigue detection in simulated roofing postures. Buildings, 2025, 15(17): 3005.
31. Tu P, Li J, Wang H. A multitask deep learning approach for sEMG-based human motion intention and muscle fatigue levels recognition. IEEE Trans Instrum Meas, 2025, 74: 1-10.
32. Zendehbad S A, Ghasemi J, Samsami Khodadad F. FatigueNet: a hybrid graph neural network and transformer framework for real-time multimodal fatigue detection. Sci Rep, 2025, 15: 33781.
33. Xi K, Lv D, Yang C, et al. Adaptive bandwidth chirp mode decomposition for muscle fatigue analysis. Measurement, 2026, 257(B): 118679.
34. Wang S, Tang H, Wang B, et al. A novel approach to detecting muscle fatigue based on sEMG by using neural architecture search framework. IEEE Trans Neural Netw Learn Syst, 2023, 34(8): 4932-4943.
35. Edwards J J, Coleman D A, Ritti-Dias R M, et al. Isometric exercise training and arterial hypertension: an updated review. Sports Med, 2024, 54(6): 1515-1529.
36. Oh B M, Han T S, Lee W H, et al. Reliability of surface electromyography from the lower-limb muscles during maximal and submaximal voluntary isometric contractions in in-bed healthy individuals and patients with subacute stroke. Brain Neurorehabil, 2024, 17(2): e14.
37. Borg G A V. Psychophysical bases of perceived exertion. Med Sci Sports Exerc, 1982, 14(5): 377-381.
38. Williams N. The Borg Rating of Perceived Exertion (RPE) scale. Occup Med (Lond), 2017, 67(5): 404-405.
39. Hussain J, Sundaraj K, Subramaniam I D, et al. Analysis of fatigue in the three heads of the triceps brachii during isometric contractions at various effort levels. J Musculoskelet Neuronal Interact, 2019, 19(3): 276-285.
40. Merletti R, Knaflitz M, De Luca C J, et al. Myoelectric manifestations of fatigue in voluntary and electrically elicited contractions. J Appl Physiol, 1990, 69(5): 1810-1820.

Journal of Biomedical Engineering

Cluster-guided adaptive Transformer for muscle fatigue prediction

Abstract Full text Figures/Tables Video References Cited by

Previous Article

Next Article

Format

Content