Tools for the Biomechanical Analysis of Human Movement: Application to Knee Osteoarthritis

点击翻译按钮获取中文摘要

My research career has focused on the development of engineering tools for the analysis of human movement. Working with my students, my research has contributed significantly to the development of techniques for the analysis and classification of human movement data. A challenge of biomechanical data is the substantial variability that may be confounded by uncertainties in the experimental methods. Ensuring reliable and consistent data within laboratories for longitudinal studies, between laboratories for data comparison, or in multicentre studies is, therefore, a significant and worthy challenge. I am proposing to improve the robustness and sensitivity of our classification techniques by quantifying and removing this nuisance variation. One of the key features of this new approach is that it can be applied to data already collected instead of only prescribing how data should be collected in the future. We have applied our classification tools to knee joint function and its deterioration due to osteoarthritis (OA) as an exemplar mechanism and this has motivated our work in computational modelling. Biological joints, just like their mechanical counterparts, progressively deteriorate in response to excessive loading conditions. We cannot, however, measure internal knee loads or muscle forces without invasive and potentially confounding surgical procedures, Therefore, we rely on computational models to estimate these forces. Our research group and others, have obtained accurate joint contact force estimates for healthy gait. Modelling OA gait adds a substantial challenge that we propose to meet through an innovative combination of our classification techniques and musculoskeletal modelling. Muscle forces contribute greatly to joint contact forces, and the subjects with OA use their muscles differently from healthy subjects. To reflect this reality, the models must be informed with muscle activation patterns that are recorded using electromyogram (EMG). One way to do this is with an EMG-driven model where the EMG are used to actuate the muscles in a musculoskeletal model. However, EMG-driven models are limited due to, among other things, unsatisfactory solutions to the EMG-Force relation. We have developed an alternative approach, best characterized as “EMG-informed” optimization in which a subject-specific simulation is created, and then we perturb this simulation according to measured differences in EMG, based on our data analysis techniques. The value of this approach lies in our ability to create subject-specific, muscle activation patterns such as those observed in patients with musculoskeletal diseases like knee osteoarthritis. Our work could lead to important new tools for quantitative diagnosis, and evaluation of treatments.