Enhancing Accuracy and Reliability of Spinal Load Estimation in Lifting/Lowering Tasks: Insights from Inverse Dynamics-Based Multibody Modelling in OpenSim

Abstract

This thesis aims to enhance the accuracy, reliability, and accessibility of estimating intervertebral forces during lifting/lowering tasks, with implications for improving back health in diverse settings, including workplaces. Four musculoskeletal multibody modelling studies using OpenSim were conducted to achieve these objectives. First, a novel Stability-Constrained Static Optimization framework (SCSO) was developed for OpenSim to improve reliability in estimating spinal loads during static lifting tasks. The muscle activities predicted by the SCSO framework demonstrated closer agreement with electromyography data compared to those obtained by the default optimization solver in OpenSim, making it more suitable for estimating muscle forces and spinal joint loads during static tasks. Next, the impact of five approaches for modelling hand–mass interaction on estimated spinal joint loads during dynamic lifting tasks was evaluated. Results indicated substantial differences in predicted spinal loads, particularly with increasing movement speed. A new fully articulated thoracolumbar spine (FATLS) model was developed for Studies 3 and 4. Study 3 assessed techniques for normalizing spinal forces, aiming to remove body weight variation effects on spinal force estimates. The discussions of this study facilitated better result comparisons against experimental data in the final study. The newly developed FATLS model's estimation of spinal forces was validated against experimental data during dynamic lifting/lowering tasks in Study 4. The results of Study 4 demonstrated accurate capture of changes in maximum forces across various dynamic lifting tasks as well as effectively predicting time-varying spinal forces using the FATLS model. Key recommendations of the present thesis include using the FATLS model for estimating intervertebral forces during lifting tasks due to its high biofidelity, accessibility, and validation to a greater extent than previous models. Additionally, employing the SCSO framework is advised for estimating muscle activation patterns during static lifting tasks. Lastly, it is suggested to report uncertainty of spinal force estimates associated with the employed modelling approach for external hand forces and moments. Adhering to these recommendations can empower researchers and practitioners in occupational biomechanics, enabling more effective use of musculoskeletal modelling and contributing to improved back health.