A predictive simulation framework for personalised in silico gait retraining in knee osteoarthritis.

Abstract

Gait retraining is typically performed with either a single, or small set of generic instructions which target a specific angle such as trunk lean or foot progression angle. These generic instructions may limit the ability of these interventions to effectively reduce knee joint loading magnitude, the ultimate goal of gait retraining. This aim of this study was to introduce and demonstrate a proof-of-concept predictive simulation framework for personalised gait retraining with an explicit implementation of reducing knee joint loading. An existing open-source framework was adapted by first updating the musculoskeletal model to include medial and lateral contact points, and a detailed ligament structure. Second, the objective function was updated to (1) explicitly minimise the magnitude of medial and lateral compressive knee joint loading and (2) allow tracking of patient specific habitual gait kinematics. Through a series of exemplar simulations the framework has been shown to be able to loosely track sagittal plane lower limb kinematics, reduce the magnitude of estimate knee joint contact loading verified through inverse simulations, and is sensitive to tibiofemoral alignment. The adapted framework is capable of generating novel movement patterns which have been shown in silico to reduce the magnitude of knee joint loading and may be suitable for personalised gait retraining. Future studies will include assessment of feasibility of implementing generated gait patterns for gait retraining as well as potential longitudinal effectiveness of such approaches.