Advanced musculoskeletal modeling techniques to complement ergonomic decision making in the workplace

Lower back and muscular pain in the upper limbs are reported in three out of five workers in the EU. In Belgium, 69% of the yearly-reported work-related disorders were musculoskeletal system disorders. Despite ongoing efforts to improve the working conditions, work-related musculoskeletal disorder (WMSDs) prevalence is still increasing. The origin of WMSDs has proven to be multifactorial, i.e., organizational, psychosocial, and individual risk factors. Additionally, research has shown that excessive physical workloads and forces during repetitive movements, non-neutral postures, and heavy material handlings can also play an important role. Indeed, recent evidence suggests that all biomaterials, i.e., muscles, articular cartilage, intervertebral discs, are susceptible to damage via the process of fatigue failure. This process suggests that magnitude of forces such as joint contact and muscle forces combined with repetition frequency should be considered when evaluating the risk of WMSDs development. However, these forces are often neglected by previous ergonomic risk assessment scales or estimated using joint-moment based proxies to analyze musculoskeletal loading during occupational tasks. In addition, musculoskeletal loading is often only analyzed in a single joint (e.g. the L5S1), neglecting potential overloading in neighboring joints. An assistive device that potentially can reduce these joint contact, muscle forces or prevent overloading, is a passive shoulder exoskeleton. It has been shown to reduce muscle loading during quasi-static tasks but its effect on joint contact forces and potential loading shifts, also during more dynamic occupational tasks, is not investigated yet.  

The overarching aim of this Ph.D. is to investigate if advanced musculoskeletal modeling techniques can provide new relevant insights into full-body musculoskeletal loading topography in the workplace and the potential role of exoskeletons as musculoskeletal loading reducing devices. The work is divided into three main objectives to achieve this overarching goal: (1) Assess full-body musculoskeletal loading during occupational tasks using a dedicated musculoskeletal modeling workflow. (2) Investigate the effect of a passive shoulder exoskeleton on musculoskeletal loading topography – using a lab-based and in-field evaluation. (3) Develop and demonstrate the added value of an innovative assessment tool that combines musculoskeletal modeling with insights from the fatigue failure theory to evaluate full-body musculoskeletal loading to support ergonomic advice in the workplace.

Firstly, to substantiate ergonomic advice on standard lifting techniques i.e. stoop and squat, we need to analyze the full-body musculoskeletal loading topography in the joints of the lower and upper extremities and the lower back using advanced biomechanical measures (study 1). To do so, joint moments and joint power of ten participants performing the two standard lifting techniques were quantified using 3D full-body motion capture data as input for the full-body musculoskeletal modeling workflow combined with the analyses of muscle activations of several superficial muscles. Peak moments and peak joint power in L5-S1 were not different between the squat and the stoop. In addition, moment impulse in L5-S1 was higher during stoop lifting. However, higher peak moments and peak power in the hip, knee, elbow and shoulder were found during squat lifting. This is reflected in higher peak electromyography (EMG) but lower muscle effort in the analyzed muscles during the squat. We concluded that neither stoop nor squat lifting could be favored as the moment impulse in the back was lower during squat, however higher musculoskeletal loading in the extremities was observed. This suggests that ergonomic advice should shift from a more regional approach to considering overall, full-body musculoskeletal loading.

In the second study, full-body musculoskeletal loading topography was documented during 10 occupational tasks. A validated musculoskeletal modeling workflow with a detailed spine was used to evaluate musculoskeletal loading for 20 participants performing a range of occupational tasks as well as the association between proxy variables (i.e. joint moments) and joint contact forces. The loading topography confirmed previous ergonomic guidelines advising to lift heavy loads around elbow height as it imposed the lowest L5-S1 loading. However, an ergonomist should be cautious when suggesting this lifting technique as shoulder and elbow loading are higher than during the other occupational tasks analyzed. The modeling-based loading topography can be used as the basis for a job rotation schedule aiming to reduce the full-body musculoskeletal loading. This is highly relevant as current ergonomic guidelines often do not consider other joints, i.e. knee, hip, shoulder, and elbow. In addition, the joint moment-based proxy estimations were confirmed to be only reliable estimates during occupational tasks with an upright, standing posture, except for the thoracic spinal region.

Based on studies 1 and 2, we could conclude that ergonomic risk analyses and therefore ergonomic advice would improve if full-body musculoskeletal loading in terms of joint contact forces is accurately quantified compared to the proxies used on the work floor.

Next, the potential effects of a newly developed shoulder exoskeleton on joint contact forces and potential loading shifts were quantified during 4 occupational tasks i.e. a low-, ergonomic-, high lift, and an overhead wiring task in study 3. A full-body musculoskeletal modeling workflow was used to calculate musculoskeletal loading in the elbow, shoulder, L5, hip and knee. The exoskeleton’s usability was quantified using the system usability scale. When wearing the Exo4Work exoskeleton, shoulder and elbow musculoskeletal loading decreased during the high lift and overhead wiring task, without shifting the musculoskeletal load to the back, hip, and knee. In contrast, musculoskeletal loading in the back and shoulder, as well as hip and knee increased during the ergonomic lift and low lift. When wearing the exoskeleton, muscle activity of the Trapezius descendens, Deltoideus anterior, Deltoideus medius and Biceps brachii were only significantly reduced during the high lift and the exoskeleton had no effect on muscle fatigue. Therefore, we advise using the Exo4Work exoskeleton only for tasks above shoulder height in order to avoid overloading in neighboring joints.

In the fourth study, the same Exo4Work exoskeleton was evaluated using EMG and inertial measurement units (IMU’s) while 6 participants performed their daily working tasks. Full-body kinematics were estimated using the Xsens software and used as input for the musculoskeletal modeling workflow in OpenSim. Despite promising system usability scores observed for the Exo4Work exoskeleton, we did not observe any functionally relevant effects on musculoskeletal loading across all 6 workers. Nevertheless, the exoskeleton provided close to maximal assistance during the material handlings in all participants This is because the provided assistance of the passive shoulder exoskeletons is related to the shoulder flexion angle and the average and maximal flexion angle observed during the material handlings were 80 degrees and 100 degrees shoulder flexion respectively. The discrepancy between lab-based and in-field evaluation could result from the fact that tasks analyzed in a laboratory are standardized and oversimplified, and therefore these tasks do not mimic real-life working situations. Therefore, the effects of a passive shoulder exoskeleton observed in the laboratory could be overestimated and evaluating exoskeletons on the work floor is mandatory. Nevertheless, this is the first study to investigate the effect of a shoulder exoskeleton on full-body musculoskeletal loading and the current study showed that musculoskeletal modeling workflow can be used to investigate the effects of an ergonomic intervention on the work floor.

Based on the results of study 3, we would advise using the Exo4Work exoskeleton for occupational tasks above shoulder height. However, study 4 showed that more research is needed in the workplace to fully understand its effect on musculoskeletal loading.

In the last study, we used the musculoskeletal modeling workflow and insights from the fatigue failure theory i.e. considering magnitude of joint contact and muscle forces in combination with loading frequency, to develop a new ergonomic risk assessment called the MATE. Three case studies highlighting the added value of the MATE are described and compared with previously developed risk assessment scales: (1) First, risk on WMSDs of an occupational task is estimated based on an IMU-based approach to demonstrate the added value of the MATE on the work floor. After that, we use the MATE to evaluate the risk on the WMSDs-reducing effect of (2) a shoulder and (3) back exoskeleton, demonstrating its added value in exoskeleton assistance tuning. The MATE showed similar or better risk estimation accuracy than previously developed risk assessment scales and can be used during occupational tasks with higher lumbar flexion during which previously developed risk assessments scales lack joint contact force estimation accuracy. Thereafter, the case studies showed that the MATE can be used to evaluate the full-body musculoskeletal loading and differentiate risk on WMSDs between anatomical regions. In addition, the effects of shoulder and back exoskeletons on the work floor can be evaluated and used for assistance tuning. Therefore, we believe that a risk assessment that uses a more complex musculoskeletal modelling approach in combination with the fatigue failure theory such as the MATE will improve future ergonomic advice.

Overall, this Ph.D. contributed to better and relevant insights into full-body musculoskeletal loading topography in the workplace and the potential role of exoskeletons as musculoskeletal loading reducing devices