Li Lunjian, Liu Xuanchi, Zhang Lihai
Department of Infrastructure Engineering, The University of Melbourne, Parkville, VIC, Australia.
J Neuroeng Rehabil. 2025 Jun 24;22(1):140. doi: 10.1186/s12984-025-01669-x.
Distal radius fractures (DRFs) are common fracture types and elderly patients often struggle to achieve functional recovery, which could be overcome by precise rehabilitation. This study aims to develop an innovative approach for acquiring patient-specific musculoskeletal models to provide guidelines for therapists to tailor rehabilitation plans individually.
A wearable EMG detector (Myo armband) and a dynamometer (KDG grip strength tester, EH101) were used to collect EMG signals and grip forces from 20 volunteers at 0, 30, 50, 70, and 100 N, which were considered low-level gripping. The collected data was used to train neural networks to predict maximum grip force from low-level grip data only. Based on a novel scaling function, personalized models were scaled from a standard musculoskeletal model and were validated by comparing their results with experiments. Sequentially, the musculoskeletal forces of two volunteers with different muscle strengths (one strong in muscle strength and the other is weak, compared to baseline) were simulated under extension exercises to investigate the impact of individual muscle strengths on rehabilitation outcomes.
The trained model predicts the maximum grip force by EMG signals well. Based on the scaling function, the corresponding personalized musculoskeletal models can simulate grip forces that align well with experiment observations. The muscle loadings were also scaled proportionally to their scaling coefficients. However, the contact forces are not linear to the scaling coefficients. The healing outcome of weak individuals shows satisfactory improvement while that of strong individuals performs ordinarily.
This study has successfully developed a convenient approach to detect the maximum grip strength of patients and verified the feasibility of scaling the musculoskeletal models. The non-linear relationship of contract forces to the scaling coefficients indicates the complexity of the musculoskeletal system. The healing outcomes from the case studies suggest that while adequate mechanical stimuli are beneficial, excessive or inappropriate stimuli can impede the healing process.
桡骨远端骨折（DRF）是常见的骨折类型，老年患者往往难以实现功能恢复，而精确的康复治疗可以克服这一问题。本研究旨在开发一种创新方法，获取患者特异性的肌肉骨骼模型，为治疗师制定个性化康复计划提供指导。
使用可穿戴式肌电图探测器（Myo臂带）和测力计（KDG握力测试仪，EH101）从20名志愿者处收集肌电信号和0、30、50、70和100牛的握力，这些被视为低水平抓握力。收集的数据用于训练神经网络，仅根据低水平抓握数据预测最大握力。基于一种新颖的缩放函数，从标准肌肉骨骼模型缩放得到个性化模型，并通过将其结果与实验结果进行比较来验证。随后，在伸展运动下模拟两名肌肉力量不同（与基线相比，一名肌肉力量强，另一名弱）的志愿者的肌肉骨骼力，以研究个体肌肉力量对康复结果的影响。
训练后的模型能很好地通过肌电信号预测最大握力。基于缩放函数，相应的个性化肌肉骨骼模型可以模拟出与实验观察结果吻合良好的握力。肌肉负荷也与其缩放系数成比例缩放。然而，接触力与缩放系数并非呈线性关系。肌肉力量较弱个体的愈合结果显示出令人满意的改善，而肌肉力量较强个体的愈合情况则一般。
本研究成功开发了一种便捷方法来检测患者的最大握力，并验证了缩放肌肉骨骼模型的可行性。接触力与缩放系数的非线性关系表明了肌肉骨骼系统的复杂性。案例研究的愈合结果表明，虽然适当的机械刺激有益，但过度或不适当的刺激可能会阻碍愈合过程。
