Solving musculoskeletal biomechanics with machine learning

Abstract

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.

Artificial physics engine for real-time inverse dynamics of arm and hand movement

“…Yet, we achieve similar torque prediction performance (less than 0.15 Nm errors with high level of noise) as in these low-dimensional systems (Çallar and Böttger 2022). Furthermore, we propose this approach as the extension for the full musculoskeletal dynamics that details moment arm and muscle length relationships with posture (Smirnov et al 2021). This combination of physically explainable ANN models can, thus, provide decoding of motor intent or the control of wearable powered robotics using muscle-level resolution of biomechanical state.…”

“…In the context of the hand and arm, a realistic musculoskeletal model consists of about 23 degrees of freedom (DOF) actuated by 52 force-generating musculotendon units scaled by limb segment geometry. This complexity is a challenge for real-time simulations, for example, for brain-computer interfaces (BCI), and can benefit from approximations of musculoskeletal transformations (Sobinov et al 2020; Smirnov et al 2021) and limb physics. Artificial neural networks (ANNs) were previously applied to solve this problem by approximating the input-output transformations from neural signals to the intended movement (Hochberg et al 2006; Donoghue et al 2007; Collinger et al 2013; Wodlinger et al 2015).…”

Does joint impedance improve dynamic leg simulations with explicit and implicit solvers?

“…The simulation frequency needs to be minimized with acceptable level of errors for closed-loop control applications. We chose the error thresholds (<1% kinematic and <5% kinetic) based on our previous evaluation of errors in the approximation of musculoskeletal dynamics (21,22). We found forward and inverse computations to be real-time accurate at frequencies as low as 200 Hz (Fig.…”

An investigation of musculoskeletal discomforts among mining truck drivers with respect to human vibration and awkward body posture using random forest algorithm

“…HTV among the mining truck drivers was not considered an influencing factor for musculoskeletal symptoms since exposure levels recorded were below the ELV according to the ISO recommendation. It should be noted that in a (Smirnov et al, 2021).…”