Development of a Predictive Model for Crutch-Assisted Gait using Optimal Control Theory

Abstract

Crutches are an assistive device used all over the world, and have been around for thousands of years. However, not a lot of studies have been conducted in order to study the effects of its use from a biomechanical point of view. This project is focused on the development of a predictive model for crutch-assisted gait by using a predictive simulation based on an Optimal Control Problem (OCP). The generation of predictive simulations can be used as a tool to further understand how the musculoskeletal system works during crutch-assisted walking. The project is divided in two main objectives: determining if a muscle-driven predictive simulation is able to reliably predict crutch-assisted gait and studying the possibilities of reconstructing different gait patterns by tracking experimental ground reaction forces (GRFs) in the crutches. In order to develop this project, the simulation framework PredSim was used. This framework, which proved reliable for non-assisted walking prediction, is modified to incorporate the use of crutches, and the settings are explored in order to obtain accurate predictions of crutch-assisted gait patterns. The use of different models, capable of moving in 2 or 3 dimensions, was also studied. OpenSim was used all throughout the project for the visualization and analysis of the resulting motions. For the tracking of GRFs, the OCP formulation was modified to add a tracking term, and experimental measures of the orientation and the forces applied to the tip of instrumentalized crutches were used as reference (GRFs) and as a way of validating results (orientations). In general, the results from 2D simulations did not lead to consistent results and a good percentage of them were not natural motions. However, 3D simulations succeeded in predicting crutch-assisted gait patterns with relative consistency that coincide with some of the patterns defined by physiotherapists and physicians for crutches use. The number of trials is limited and some limitations were identified during the development of the project, reason why it is complicated to give definitive proof or draw definitive conclusions based on the work done. However, it was observed that prediction of crutch-assisted gait patterns was possible, despite all the limitations, using the PredSim simulation framework. It could also be observed that the OCP’s cost function did not need to be modified from the original version, meaning that the choosing of an optimal crutch-assisted gait follows the same rules as non-assisted walking. Regarding the second objective of the project, GRFs were accurately tracked. Regardless of that, probably due to limitations of the project, the resulting gait patterns did not reliably reenact the experimentally measured patterns. In general, further exploration of the simulations’ configuration should be done. However, this project is a valuable introductory exploration into predictive simulations for crutch-assisted gait