Muscle Synergies Facilitate Prediction of Subject-specific Walking Motions

This project provides Matlab code and experimental walking data for generating dynamically consistent full-body walking predictions for a high functioning stroke subject. The predictions are generated using a patient-specific full-body OpenSim/Matlab neuromusculoskeletal model and the direct collocation optimal control software GPOPS-II, which functions within the Matlab programming environment. The neuromusculoskeletal model is calibrated to the subject's walking data collected at his self-selected speed of 0.5 m/s to produce a patient-specific lower-body kinematic model, patient-specific foot-ground contact models, patient-specific EMG-driven muscle models, and patient-specific neural control models based on the subject's muscle synergy structure. The calibrated model is used to predict how the subject will walk at 0.5 and 0.8 m/s using either joint torque controls (5 per leg), muscle activation controls (35 per leg), or muscle synergy controls (5 per leg) applied to the lower body joints. For now, this project is a placeholder pending acceptance of the associated journal manuscript by Frontiers in Bioengineering and Biotechnology. <br/><br/>This project includes the following software/data packages: <br/> <ul> <li> <a href="https://simtk.org/frs?group_id=1093#pack_1899">Optimal Control Code </a> : This zip folder contains all the code for running optimal control simulations.IMPORTANT NOTE: Due to our use of MPI as a parallelization tool, this code will not function correctly on other computers without a significant amount of work recompiling code. We are very sorry for the inconvenience and hope to correct this issue in the future.Code for creating the EMG to moment model used in these simulations will be available at https://simtk.org/projects/emgdrivenmodel. </li> </ul>

本项研究提供了适用于生成动态一致性全身行走预测的 Matlab 代码和实验行走数据，针对功能健全的卒中患者。预测通过患者专用的全身 OpenSim/Matlab 神经肌肉骨骼模型以及直接拟合最优控制软件 GPOPS-II 实现，该软件在 Matlab 编程环境中运行。神经肌肉骨骼模型经过校准，以适应患者自选速度 0.5 m/s 的行走数据，从而生成患者特有的下肢运动学模型、患者特有的足底地面接触模型、患者特有的肌电驱动肌肉模型，以及基于患者肌肉协同结构的患者特有的神经控制模型。校准后的模型用于预测患者以 0.5 m/s 和 0.8 m/s 的速度行走时，通过关节扭矩控制（每腿 5 个）、肌肉激活控制（每腿 35 个）或肌肉协同控制（每腿 5 个）作用于下肢关节的情况。目前，该项目仅为等待《生物工程与生物技术前沿》期刊接受相关论文而暂时保留。该项目包括以下软件/数据包：
<ul>
<li><a href="https://simtk.org/frs?group_id=1093#pack_1899">最优控制代码</a>：此 zip 文件夹包含运行最优控制模拟的所有代码。重要提示：由于我们使用了 MPI 作为并行化工具，此代码在其他计算机上可能无法正确运行，除非进行大量工作重新编译代码。我们对此造成的不便深表歉意，并希望在将来修正这一问题。用于创建这些模拟中使用的肌电到力矩模型的代码将可在 https://simtk.org/projects/emgdrivenmodel 获取。</li>
</ul>