Improving dynamic endurance time predictions for shoulder fatigue: A comparative evaluation

Abstract

Work-related musculoskeletal disorders (WMSDs) are commonplace in industry and a host of qualitative and quantitative approaches have been used to assuage the problem, including wearable sensors and biomechanical endurance models, both of which were used in the present study. Six endurance models (consumed endurance, new improved consumed endurance and the exponential and power Frey Law and Avin general and shoulder models) with four alternative maximum torque (${Torque}_{\max }$) quantification methods, including a novel approach to generate ${Torque}_{\max }$, were compared. The proposed approach to quantify ${Torque}_{\max }$, in combination with the new improved consumed endurance model produced the lowest root mean square errors (RMSE), and indicated improved performance compared to the literature. The mean RMSE was reduced from 41.08s to 19.11s for all subjects, from 26.13s to 12.16s for males, and 51.28s to 24.45s for females using the proposed method. R² for 25% and 45% standardised intensity dynamic tasks were .459 and .314 respectively, P < .01. This research provided an optimised and individualised endurance prediction approach for loaded dynamic movements which can be applied to industry tasks and may lead to reduced upper-limb strains, and potentially WMSDs.