Integrating musculoskeletal simulation and machine learning: a hybrid approach for personalized ankle-foot exoskeleton assistance strategies.

Introduction: Lower limb exoskeletons have shown considerable potential in assisting human walking, particularly by reducing metabolic cost (MC), leading to a surge of interest in this field in recent years. However, owing to significant individual differences and the uncertainty of movements, challenges still exist in the personalized design and control of exoskeletons in human-robot interactions. Methods: In this study, we propose a hybrid data-driven approach that integrates musculoskeletal simulation with machine learning technology to customize personalized assistance strategies efficiently and adaptively for ankle-foot exoskeletons. First, optimal assistance strategies that can theoretically minimize MC, were derived from forward muscle-driven simulations on an open-source dataset. Then, a neural network was utilized to explore the relationships among different individuals, movements, and optimal strategies, thus developing a predictive model. Results: With respect to transfer learning, our approach exhibited effectiveness and adaptability when faced with new individuals and movements. The simulation results further indicated that our approach successfully reduced the MC of calf muscles by approximately 20% compared to normal walking conditions. Discussion: This hybrid approach offers an alternative for personalizing assistance strategy that may further guide exoskeleton design.

Wearable Robot Design Optimization Using Closed-Form Human-Robot Dynamic Interaction Model.

Wearable robots are emerging as a viable and effective solution for assisting and enabling people who suffer from balance and mobility disorders. Virtual prototyping is a powerful tool to design robots, preventing the costly iterative physical prototyping and testing. Design of wearable robots through modelling, however, often involves computationally expensive and error-prone multi-body simulations wrapped in an optimization framework to simulate human-robot-environment interactions. This paper proposes a framework to make the human-robot link segment system statically determinate, allowing for the closed-form inverse dynamics formulation of the link-segment model to be solved directly in order to simulate human-robot dynamic interactions. The paper also uses a technique developed by the authors to estimate the walking ground reactions from reference kinematic data, avoiding the need to measure them. The proposed framework is (a) computationally efficient and (b) transparent and easy to interpret, and (c) eliminates the need for optimization, detailed musculoskeletal modelling and measuring ground reaction forces for normal walking simulations. It is used to optimise the position of hip and ankle joints and the actuator torque-velocity requirements for a seven segments of a lower-limb wearable robot that is attached to the user at the shoes and pelvis. Gait measurements were carried out on six healthy subjects, and the data were used for design optimization and validation. The new technique promises to offer a significant advance in the way in which wearable robots can be designed.

Calibrationless monocular vision musculoskeletal simulation during gait.

With computer vision technology and prediction of ground reaction forces (GRF), a previous study performed markerless motion capture and musculoskeletal simulation with two smartphones (OpenCap). A recent approach can reconstruct 3D human motion from a single video without calibration and it may further simplify the motion capture process. However it has not been combined with musculoskeletal simulation and the validity is unclear. Therefore, the purpose of this study was to determine the validity of the musculoskeletal simulation using a monocular vision approach. An open-source dataset that contains motion capture and video data during gait from 10 healthy participants was used. Human motion reconstruction with the skinned human (SMPL) model was performed on each video. Virtual marker data was generated by extracting the position data from the SMPL skin vertices. Inverse kinematics, GRF prediction (only for monocular vision approach), inverse dynamics and static optimization were performed using a musculoskeletal model for experimental motion capture data and the generated virtual markers from videos. Mean absolute errors (MAE) between motion capture based and monocular vision based simulation outcomes were calculated. The MAE were 8.4° for joint angles, 5.0 % bodyweight for GRF, 1.1 % bodyweight*height for joint moments and 0.11 for estimated muscle activations from 16 muscles. The entire MAE was larger but some were comparable to OpenCap. Using the monocular vision approach, motion capture and musculoskeletal simulation can be done with no preparations and is beneficial for clinicians to quantify the daily gait assessment.

Minimization of metabolic cost of transport predicts changes in gait mechanics over a range of ankle-foot orthosis stiffnesses in individuals with bilateral plantar flexor weakness.

Neuromuscular disorders often lead to ankle plantar flexor muscle weakness, which impairs ankle push-off power and forward propulsion during gait. To improve walking speed and reduce metabolic cost of transport (mCoT), patients with plantar flexor weakness are provided dorsal-leaf spring ankle-foot orthoses (AFOs). It is widely believed that mCoT during gait depends on the AFO stiffness and an optimal AFO stiffness that minimizes mCoT exists. The biomechanics behind why and how an optimal stiffness exists and benefits individuals with plantar flexor weakness are not well understood. We hypothesized that the AFO would reduce the required support moment and, hence, metabolic cost contributions of the ankle plantar flexor and knee extensor muscles during stance, and reduce hip flexor metabolic cost to initiate swing. To test these hypotheses, we generated neuromusculoskeletal simulations to represent gait of an individual with bilateral plantar flexor weakness wearing an AFO with varying stiffness. Predictions were based on the objective of minimizing mCoT, loading rates at impact and head accelerations at each stiffness level, and the motor patterns were determined via dynamic optimization. The predictive gait simulation results were compared to experimental data from subjects with bilateral plantar flexor weakness walking with varying AFO stiffness. Our simulations demonstrated that reductions in mCoT with increasing stiffness were attributed to reductions in quadriceps metabolic cost during midstance. Increases in mCoT above optimum stiffness were attributed to the increasing metabolic cost of both hip flexor and hamstrings muscles. The insights gained from our predictive gait simulations could inform clinicians on the prescription of personalized AFOs. With further model individualization, simulations based on mCoT minimization may sufficiently predict adaptations to an AFO in individuals with plantar flexor weakness.

Evaluation of trunk muscle coactivation predictions in multi-body models.

Musculoskeletal simulations with muscle optimization aim to minimize muscle effort, hence are considered unable to predict the activation of antagonistic muscles. However, activation of antagonistic muscles might be necessary to satisfy the dynamic equilibrium. This study aims to elucidate under which conditions coactivation can be predicted, to evaluate factors modulating it, and to compare the antagonistic activations predicted by the lumbar spine model with literature data. Simple 2D and 3D models, comprising of 2 or 3 rigid bodies, with simple or multi-joint muscles, were created to study conditions under which muscle coactivity is predicted. An existing musculoskeletal model of the lumbar spine developed in AnyBody was used to investigate the effects of modeling intra-abdominal pressure (IAP), linear/cubic and load/activity-based muscle recruitment criterion on predicted coactivation during forward flexion and lateral bending. The predicted antagonist activations were compared to reported EMG data. Muscle coactivity was predicted with simplified models when multi-joint muscles were present or the model was three-dimensional. During forward flexion and lateral bending, the coactivation ratio predicted by the model showed good agreement with experimental values. Predicted coactivation was negligibly influenced by IAP but substantially reduced with a force-based recruitment criterion. The conditions needed in multi-body models to predict coactivity are: three-dimensionality or multi-joint muscles, unless perfect antagonists. The antagonist activations are required to balance 3D moments but do not reflect other physiological phenomena, which might explain the discrepancies between model predictions and experimental data. Nevertheless, the findings confirm the ability of the multi-body trunk models to predict muscle coactivity and suggest their overall validity.

Selective dorsal rhizotomy and its effect on muscle force during walking: A comprehensive study.

Selective dorsal rhizotomy (SDR) is commonly used to permanently reduce spasticity in children with cerebral palsy (CP). However, studies have yielded varying results regarding muscle strength and activity after SDR. Some studies indicate weakness or no changes, while a recent study using NMSK simulations demonstrates improvements in muscle forces during walking. These findings suggest that SDR may alleviate spasticity, reducing dynamic muscle constraints and enhancing muscle force without altering muscle activity during walking in children with CP. In this study, we combined NMSK simulations with physical examinations to assess children with CP who underwent SDR, comparing them to well-matched peers who did not undergo the procedure. Each group (SDR and No-SDR) included 81 children, with pre- and post-SDR assessments. Both groups were well-matched in terms of demographics, clinical characteristics, and gait parameters. The results of the physical examination indicate that SDR significantly reduces spasticity without impacting muscle strength. Furthermore, our findings show no significant differences in gait deviation index improvements and walking speed between the two groups. Additionally, there were no statistically significant changes in muscle activity during walking before and after SDR for both groups. NMSK results demonstrate a significant increase in muscle force in the semimembranosus and calf muscles during walking, compared to children with CP who received other clinical treatments. Our findings confirm that although SDR does not significantly impact muscle strength compared to other treatments, it creates a more favorable dynamic environment for suboptimal muscle force production, which is essential for walking.

Designing customized temporomandibular fossa prosthesis based on envelope surface of condyle movement: validation via in silico musculoskeletal simulation.

Objective: This study presents an innovative articular fossa prosthesis generated by the envelope surface of condyle movement, and compares its mandible movements, muscle activities, and joint reaction forces with two temporomandibular joint (TMJ) prostheses using multibody musculoskeletal simulation. Methods: A healthy 23-year-old female was recruited for this study. Cone-beam computed tomographic (CBCT) was performed to reconstruct the mandibular bone geometry. A customized TMJ fossa prosthesis was designed based on the subject-specific envelope surface of condyle movement (ESCM). Mandibular kinematics and jaw-closing muscle electromyography (EMG) were simultaneously recorded during maximum jaw opening-closing movements. To validate our prosthesis design, a mandibular musculoskeletal model was established using flexible multibody dynamics and the obtained kinematics and EMG data. The Biomet fossa prosthesis and the ellipsoidal fossa prosthesis designed by imitating the lower limb prostheses were used for comparison. Simulations were performed to analyze the effects of different fossa prostheses on jaw opening-closing motions, mandibular muscle activation, and contact forces. Results: The maximum opening displacement for the envelope-based fossa prosthesis was greater than those for Biomet and ellipsoidal prostheses (36 mm, 35 mm, and 33 mm, respectively). The mandibular musculoskeletal model with ellipsoidal prosthesis led to dislocation near maximal jaw opening. Compared to Biomet, the envelope-based fossa reduced the digastric and lateral pterygoid activation at maximal jaw opening. It also reduced the maximal resistance to condylar sliding on the intact side by 63.2 N. Conclusion: A customized TMJ fossa prosthesis was successfully developed using the ESCM concept. Our study of musculoskeletal multibody modeling has highlighted its advantages and potential. The artificial fossa design successfully achieved a wider condylar range of motion. It also reduced the activation of jaw opening muscles on the affected side and resistance on the intact side. This study showed that an ESCM-based approach may be useful for optimizing TMJ fossa prostheses design.

Knee kinematics are primarily determined by implant alignment but knee kinetics are mainly influenced by muscle coordination strategy.

Implant malalignment has been reported to be a primary reason for revision total knee arthroplasty (TKA). In addition, altered muscle coordination patterns are commonly observed in TKA patients, which is thought to alter knee contact loads. A comprehensive understanding of the influence of surgical implantation and muscle recruitment strategies on joint contact mechanics is crucial to improve surgical techniques, increase implant longevity, and inform rehabilitation protocols. In this study, a detailed musculoskeletal model with a 12 degrees of freedom knee was developed to represent a TKA subject from the CAMS-Knee datasets. Using motion capture and ground reaction force data, a level walking cycle was simulated and the joint movement and loading patterns were estimated using a novel technique for concurrent optimization of muscle activations and joint kinematics. In addition, over 12'000 Monte Carlo simulations were performed to predict knee contact mechanics during walking, considering numerous combinations of implant alignment and muscle activation scenarios. Validation of our baseline simulation showed good agreement between the model kinematics and loading patterns against the in vivo data. Our analyses reveal a considerable impact of implant alignment on the joint kinematics, while variation in muscle activation strategies mainly affects knee contact loading. Moreover, our results indicate that high knee compressive forces do not necessarily originate from extreme kinematics and vice versa. This study provides an improved understanding of the complex inter-relationships between loading and movement patterns resulting from different surgical implantation and muscle coordination strategies and presents a validated framework towards population-based modelling in TKA.

The Determination of Assistance-as-Needed Support by an Ankle-Foot Orthosis for Patients with Foot Drop.

Patients who suffer from foot drop have impaired gait pattern functions and a higher risk of stumbling and falling. Therefore, they are usually treated with an assistive device, a so-called ankle-foot orthosis. The support of the orthosis should be in accordance with the motor requirements of the patient and should only be provided when needed, which is referred to as assistance-as-needed. Thus, in this publication, an approach is presented to determine the assistance-as-needed support using musculoskeletal human models. Based on motion capture recordings of multiple subjects performing gaits at different speeds, a parameter study varying the optimal force of a reserve actuator representing the ankle-foot orthosis added in the musculoskeletal simulation is conducted. The results show the dependency of the simulation results on the selected optimal force of the reserve actuator but with a possible identification of the assistance-as-needed support required from the ankle-foot orthosis. The required increase in support due to the increasing severity of foot drop is especially demonstrated with the approach. With this approach, information for the required support of individual subjects can be gathered, which can further be used to derive the design of an ankle-foot orthosis that optimally assists the subjects.

Musculoskeletal simulation of professional ski jumpers during take-off considering aerodynamic forces.

Introduction: Musculoskeletal simulation has been widely used to analyze athletes' movements in various competitive sports, but never in ski jumping. Aerodynamic forces during ski jumping take-off have been difficult to account for in dynamic simulation. The purpose of this study was to establish an efficient approach of musculoskeletal simulation of ski jumping take-off considering aerodynamic forces and to analyze the muscle function and activity. Methods: Camera-based marker-less motion capture was implemented to measure the take-off kinematics of eight professional jumpers. A suitable full-body musculoskeletal model was constructed for the simulation. A method based on inverse dynamics iteration was developed and validated to estimate the take-off ground reaction force. The aerodynamic forces, which were calculated based on body kinematics and computational fluid dynamics simulations, were exerted on the musculoskeletal model as external forces. The activation and joint torque contributions of lower extremity muscles were calculated through static optimization. Results: The estimated take-off ground reaction forces show similar trend with the results from past studies. Although overall inconsistencies between simulated muscle activation and EMG from previous studies were observed, it is worth noting that the activation of the tibialis anterior, gluteus maximus, and long head of the biceps femoris was similar to specific EMG results. Among lower extremity extensors, soleus, vastus lateralis, biceps femoris long head, gluteus maximus, and semimembranosus showed high levels of activation and joint extension torque contribution. Discussion: Results of this study advanced the understanding of muscle action during ski jumping take-off. The simulation approach we developed may help guide the physical training of jumpers for improved take-off performance and can also be extended to other phases of ski jumping.

Mechanical upside of PAO mainstream fixations: co-simulation based on early postoperative gait characteristics of DDH patients.

Purpose: To investigate the early postoperative gait characteristics of patients who underwent periacetabular osteotomy (PAO) and predict the biomechanical performance of two commonly used PAO fixation methods: iliac screw (IS) and transverse screw (TS). Methods: A total of 12 patients with unilateral developmental dysplasia of the hip (DDH) (mean age 27.81 ± 4.64 years, 42% male) that were scheduled to undergo PAO surgery were included in this study. Their preoperative CT images and pre- and postoperative gait data were used to create subject-specific musculoskeletal models and complete the inverse dynamics analysis (IDA). Two patients with typical gait characteristics were selected using clustering analysis, and their IDA data were incorporated into finite element (FE) models of IS and TS fixations. Failure simulation was performed by applying iterative steps with increasing gait load to predict yield load. Stress results and yield loads were calculated for each FE model at different phases of the gait cycle. Results: Postoperative gait showed improvement compared to preoperative gait but remained inferior to that of healthy individuals. Postoperative gait was characterized by a lower hip range of motion, lower peri-ilium muscle forces, particularly in the abductors, and a sharper initial peak and flatter second peak of hip joint reaction force (HRF). Finite element analysis (FEA) showed a trend of increasing stress during the second-fourth phases of the gait cycle, with lower stress levels in other phases. At high-stress gait phases, the mean stress of maximum p¯100 differed significantly between IS and TS (p < 0.05) and between coupled and uncoupled muscle forces (p < 0.05). Failure analysis predicted a slightly larger yield load for TS configurations (6.21*BW) than that for IS (6.16*BW), but both were well above the gait load. Coupled and uncoupled groups showed similar results, but uncoupled groups had lower yield loads (5.9*BW). Conclusion: PAO early postoperative gait shows a normalized trend, but abnormalities persist. IS and TS are both capable of resisting mechanical strain failure, with no significant mechanical advantage found for transverse screw fixation during PAO early postoperative gait. Additionally, it is important to note that the TS may have a higher risk of cyclic fatigue failure due to the localized greater stress concentration. Furthermore, the most medial screw is crucial for pelvic stability.

Simulation of passive exotendon assistive device for agricultural harvesting task.

This study proposes and investigates the feasibility of the passive assistive device to assist agricultural harvesting task and reduce the Musculoskeletal Disorder (MSD) risk of harvesters using computational musculoskeletal modelling and simulations. Several passive assistive devices comprised of elastic exotendon, which acts in parallel with different back muscles (rectus abdominis, longissimus, and iliocostalis), were designed and modelled. These passive assistive devices were integrated individually into the musculoskeletal model to provide passive support for the harvesting task. The muscle activation, muscle force, and joint moment were computed with biomechanical simulations for unassisted and assisted motions. The simulation results demonstrated that passive assistive devices reduced muscle activation, muscle force, and joint moment, particularly when the devices were attached to the iliocostalis and rectus abdominis. It was also discovered that assisting the longissimus muscle can alleviate the workload by distributing a portion of it to the rectus abdominis. The findings in this study support the feasibility of adopting passive assistive devices to reduce the MSD risk of the harvesters during agricultural harvesting. These findings can provide valuable insights to the engineers and designers of physical assistive devices on which muscle(s) to assist during agricultural harvesting.

Increase in lateral contact force in the tibiotalar joint during walking in flatfoot patients with reduced stiffness of the spring ligament.

Foot deformities in patients with flexible flatfeet, such as the flattened medial arch and hindfoot valgus, affect the force distribution around the tibiotalar joint during walking and increase the risk of secondary injuries. In this study, we developed a multi-segment foot model that could calculate the dynamics around the tibiotalar joint and investigated the difference in the kinetics between normal feet and feet with flatfoot. Ten participants with normal feet and ten with flexible flatfoot were enrolled in the study. The body kinematics, ground reaction force, and foot pressure of the participants were recorded during walking. A five-segment foot model was developed to calculate contact forces in the tibiotalar joint. A flatfoot model was developed by modifying the stiffness of the spring ligaments of a normal foot model. Ground reaction force was applied to the plantar surface of the foot models. The foot models were attached to a full-body musculoskeletal model to conduct inverse dynamic simulations of walking. Participants with flatfoot had significantly greater lateral contact force (1.19 BW vs. 0.80 BW) and more posteriorly located center of pressure (33.7 % vs. 46.6 %) in the tibiotalar joint than those with normal feet (p < 0.05). The average and peak posterior tibialis muscle forces were significantly larger in participants with flatfoot than in those with normal feet (3.06 BW vs. 2.22 BW; 4.52 BW vs. 3.33 BW). The altered mechanics may influence the risk of arthritis.

A review of the efforts to develop muscle and musculoskeletal models for biomechanics in the last 50 years.

Both the Hill and the Huxley muscle models had already been described by the time the International Society of Biomechanics was founded 50 years ago, but had seen little use before the 1970s due to the lack of computing. As computers and computational methods became available in the 1970s, the field of musculoskeletal modeling developed and Hill type muscle models were adopted by biomechanists due to their relative computational simplicity as compared to Huxley type muscle models. Muscle forces computed by Hill type muscle models provide good agreement in conditions similar to the initial studies, i.e. for small muscles contracting under steady and controlled conditions. However, more recent validation studies have identified that Hill type muscle models are least accurate for natural in vivo locomotor behaviours at submaximal activations, fast speeds and for larger muscles, and thus need to be improved for their use in understanding human movements. Developments in muscle modelling have tackled these shortcomings. However, over the last 50 years musculoskeletal simulations have been largely based on traditional Hill type muscle models or even simplifications of this model that neglected the interaction of the muscle with a compliant tendon. The introduction of direct collocation in musculoskeletal simulations about 15 years ago along with further improvements in computational power and numerical methods enabled the use of more complex muscle models in simulations of whole-body movement. Whereas Hill type models are still the norm, we may finally be ready to adopt more complex muscle models into musculoskeletal simulations of human movement.

Ten steps to becoming a musculoskeletal simulation expert: A half-century of progress and outlook for the future.

Over the past half-century, musculoskeletal simulations have deepened our knowledge of human and animal movement. This article outlines ten steps to becoming a musculoskeletal simulation expert so you can contribute to the next half-century of technical innovation and scientific discovery. We advocate looking to the past, present, and future to harness the power of simulations that seek to understand and improve mobility. Instead of presenting a comprehensive literature review, we articulate a set of ideas intended to help researchers use simulations effectively and responsibly by understanding the work on which today's musculoskeletal simulations are built, following established modeling and simulation principles, and branching out in new directions.

Gait Asymmetry Variation in Kinematics, Kinetics, and Muscle Force along with the Severity Levels of Knee Osteoarthritis.

OBJECTIVE: Knee osteoarthritis (OA) patients exhibit greater gait asymmetry than healthy controls. However, gait asymmetry in kinematics, kinetics and muscle forces across patients with different severity levels of knee OA is still unknown. The study aimed to investigate the changes of gait asymmetry in lower limb kinematics, kinetics, and muscle force across patients with different severity levels of knee OA. METHODS: This is a cross-sectional study. From January 2020 to January 2021, 118 patients with symptomatic and radiographic medial knee OA were categorized into three groups using the Kellgren and Lawrence scale (mild: grade 1 and 2, n = 37; moderate: grade 3, n = 31; severe: grade 4, n = 50). During self-paced walking, marker trajectories and ground reaction forces data were recorded. Musculoskeletal simulations were used to determine gait kinematics, kinetics, and muscle force. One-way analysis of variance with Tukey's post-hoc test was used to evaluate group difference. Paired-sample t-test was used to compared the between-limb difference. RESULTS: In the Severe group, significantly greater asymmetry index in knee flexion/extension range of motion (45%) was observed with a greater value on the contralateral side (p < 0.01), compared to the Mild (15%) and Moderate (15%) groups. Significantly higher peak hip contact force (JCF) on the contralateral side was found in the Mild (more affected side: 3.80 ± 0.67 BW, contralateral side: 4.01 ± 0.58 BW), Moderate (more affected side: 3.67 ± 0.56 BW, contralateral side: 4.07 ± 0.81 BW), and Severe groups (more affected side: 3.66 ± 0.79 BW, contralateral side: 3.94 ± 0.64 BW) (p < 0.05). Significantly greater gluteus medius muscle force on the contralateral side was found in Mild (more affected side: 0.48 ± 0.09 BW, contralateral side: 0.52 ± 0.12 BW), Moderate (more affected side: 0.45 ± 0.10 BW, contralateral side: 0.51 ± 0.15 BW), and Severe groups (more affected side: 0.42 ± 0.15 BW, contralateral side: 0.47 ± 0.12 BW) (p < 0.05). The contralateral side showing significantly higher peak knee adduction moment and medial knee JCF was only observed in the Mild group (p < 0.05). CONCLUSIONS: Gait asymmetry in kinematics and muscle forces increased from mild to severe knee OA. Asymmetrical gait pattern tends to transfer loads from the more affected side to the contralateral side. Peak hip JCF and gluteus medius muscle force can be used to detect this asymmetrical gait pattern in patients with knee OA, regardless of severity levels.

Can static optimization detect changes in peak medial knee contact forces induced by gait modifications?

Medial knee contact force (MCF) is related to the pathomechanics of medial knee osteoarthritis. However, MCF cannot be directly measured in the native knee, making it difficult for therapeutic gait modifications to target this metric. Static optimization, a musculoskeletal simulation technique, can estimate MCF, but there has been little work validating its ability to detect changes in MCF induced by gait modifications. In this study, we quantified the error in MCF estimates from static optimization compared to measurements from instrumented knee replacements during normal walking and seven different gait modifications. We then identified minimum magnitudes of simulated MCF changes for which static optimization correctly identified the direction of change (i.e., whether MCF increased or decreased) at least 70% of the time. A full-body musculoskeletal model with a multi-compartment knee and static optimization was used to estimate MCF. Simulations were evaluated using experimental data from three subjects with instrumented knee replacements who walked with various gait modifications for a total of 115 steps. Static optimization underpredicted the first peak (mean absolute error = 0.16 bodyweights) and overpredicted the second peak (mean absolute error = 0.31 bodyweights) of MCF. Average root mean square error in MCF over stance phase was 0.32 bodyweights. Static optimization detected the direction of change with at least 70% accuracy for early-stance reductions, late-stance reductions, and early-stance increases in peak MCF of at least 0.10 bodyweights. These results suggest that a static optimization approach accurately detects the direction of change in early-stance medial knee loading, potentially making it a valuable tool for evaluating the biomechanical efficacy of gait modifications for knee osteoarthritis.

Intra- and inter-subject variability of femoral growth plate stresses in typically developing children and children with cerebral palsy.

Little is known about the influence of mechanical loading on growth plate stresses and femoral growth. A multi-scale workflow based on musculoskeletal simulations and mechanobiological finite element (FE) analysis can be used to estimate growth plate loading and femoral growth trends. Personalizing the model in this workflow is time-consuming and therefore previous studies included small sample sizes (N < 4) or generic finite element models. The aim of this study was to develop a semi-automated toolbox to perform this workflow and to quantify intra-subject variability in growth plate stresses in 13 typically developing (TD) children and 12 children with cerebral palsy (CP). Additionally, we investigated the influence of the musculoskeletal model and the chosen material properties on the simulation results. Intra-subject variability in growth plate stresses was higher in cerebral palsy than in typically developing children. The highest osteogenic index (OI) was observed in the posterior region in 62% of the TD femurs while in children with CP the lateral region was the most common (50%). A representative reference osteogenic index distribution heatmap generated from data of 26 TD children's femurs showed a ring shape with low values in the center region and high values at the border of the growth plate. Our simulation results can be used as reference values for further investigations. Furthermore, the code of the developed GP-Tool ("Growth Prediction-Tool") is freely available on GitHub (https://github.com/WilliKoller/GP-Tool) to enable peers to conduct mechanobiological growth studies with larger sample sizes to improve our understanding of femoral growth and to support clinical decision making in the near future.

Inertia-Constrained Reinforcement Learning to Enhance Human Motor Control Modeling.

Locomotor impairment is a highly prevalent and significant source of disability and significantly impacts the quality of life of a large portion of the population. Despite decades of research on human locomotion, challenges remain in simulating human movement to study the features of musculoskeletal drivers and clinical conditions. Most recent efforts to utilize reinforcement learning (RL) techniques are promising in the simulation of human locomotion and reveal musculoskeletal drives. However, these simulations often fail to mimic natural human locomotion because most reinforcement strategies have yet to consider any reference data regarding human movement. To address these challenges, in this study, we designed a reward function based on the trajectory optimization rewards (TOR) and bio-inspired rewards, which includes the rewards obtained from reference motion data captured by a single Inertial Moment Unit (IMU) sensor. The sensor was equipped on the participants' pelvis to capture reference motion data. We also adapted the reward function by leveraging previous research on walking simulations for TOR. The experimental results showed that the simulated agents with the modified reward function performed better in mimicking the collected IMU data from participants, which means that the simulated human locomotion was more realistic. As a bio-inspired defined cost, IMU data enhanced the agent's capacity to converge during the training process. As a result, the models' convergence was faster than those developed without reference motion data. Consequently, human locomotion can be simulated more quickly and in a broader range of environments, with a better simulation performance.

Effects of Obesity on Medial Tibiofemoral Cartilage Mechanics in Females-An Exploration Using Musculoskeletal Simulation and Probabilistic Cartilage Failure Modelling.

This study examined the effects of obesity on cartilage mechanics and longitudinal failure probability at the medial tibiofemoral compartment, using combined musculoskeletal simulation and probabilistic failure modelling approaches. The current investigation examined twenty obese females (BMI > 30.0 kg/m2) and 20 healthy weight (BMI < 25.0 kg/m2) females. Walking kinematics were obtained via an 8-camera optoelectric system, and a force plate was used to collect ground reaction forces. Musculoskeletal simulation and probabilistic failure modelling were utilized to explore medial tibiofemoral forces and cartilage probability. Comparisons between groups were undertaken using linear mixed-effects models. Net peak cartilage forces, stress and strain were significantly larger in the obese group (force = 2013.92 N, stress = 3.03 MPa & strain = 0.25), compared to health weight (force = 1493.21 N, stress 2.26 MPa & strain = 0.19). In addition, medial tibiofemoral cartilage failure probability was also significantly larger in the obese group (42.98%) compared to healthy weight (11.63%). The findings from the current investigation show that obesity has a profoundly negative influence on longitudinal medial knee cartilage health and strongly advocates for the implementation of effective weight management programs into long-term musculoskeletal management strategies.