Solving musculoskeletal biomechanics with machine learning.
Solving musculoskeletal biomechanics with machine learning.
复制标题
DOI：
10.7717/peerj-cs.663
复制
发表时间：
2021
期刊：
影响因子：
--
通讯作者：
Yakovenko S
中科院分区：
文献类型：
--
作者：
Smirnov Y;Smirnov D;Popov A;Yakovenko S
深度学习是一种用于描述肌肉骨骼动力学的相对较新的计算技术。不同姿势下肌肉几何形状的实验关系是可以通过相对简单的函数近似的高维空间变换，这为机器学习（ML）应用提供了机会。在这项研究中，我们挑战了一般ML算法的问题，近似的姿势依赖的力臂和肌肉长度的关系，人类手臂和手部肌肉。我们使用了两种类型的算法，光梯度增强机（LGB）和全连接人工神经网络（ANN）解决了33个肌肉的缠绕运动学，每个肌肉的自由度（DOF）为18个DOF的手臂和手模型。输入-输出训练和测试数据集，其中关节角度是输入，肌肉长度和力矩臂是输出，由我们以前的现象学模型基于自动生成的多项式结构生成。两种模型的误差水平相似：神经网络模型的肌肉长度误差为0.08 ± 0.05%，力臂误差为0.53 ± 0.29%，LGB模型的误差相似，分别为0.18 ± 0.06%和0.13 ± 0.07%。LGB模型仅用103个样本就达到了训练目标，而ANN需要106个样本;然而，LGB模型在评估中比ANN模型慢约39倍。所开发模型的足够性能证明了ML未来在各种应用（例如先进动力假肢）中用于肌肉骨骼转化的适用性。
Deep learning is a relatively new computational technique for the description of the musculoskeletal dynamics. The experimental relationships of muscle geometry in different postures are the high-dimensional spatial transformations that can be approximated by relatively simple functions, which opens the opportunity for machine learning (ML) applications. In this study, we challenged general ML algorithms with the problem of approximating the posture-dependent moment arm and muscle length relationships of the human arm and hand muscles. We used two types of algorithms, light gradient boosting machine (LGB) and fully connected artificial neural network (ANN) solving the wrapping kinematics of 33 muscles spanning up to six degrees of freedom (DOF) each for the arm and hand model with 18 DOFs. The input-output training and testing datasets, where joint angles were the input and the muscle length and moment arms were the output, were generated by our previous phenomenological model based on the autogenerated polynomial structures. Both models achieved a similar level of errors: ANN model errors were 0.08 ± 0.05% for muscle lengths and 0.53 ± 0.29% for moment arms, and LGB model made similar errors—0.18 ± 0.06% and 0.13 ± 0.07%, respectively. LGB model reached the training goal with only 103 samples, while ANN required 106 samples; however, LGB models were about 39 times slower than ANN models in the evaluation. The sufficient performance of developed models demonstrates the future applicability of ML for musculoskeletal transformations in a variety of applications, such as in advanced powered prosthetics.