CDS&E: A Next-Generation Computation Framework for Predicting Optimal Walking Motion

点击翻译按钮获取中文摘要

CBET-1404767FreglyCommon clinical examples of neuromusculoskeletal impairments include osteoarthritis, stroke, and Parkinson's disease, which together affect roughly 15% of the U.S. adult population. Such impairments result in reduced mobility, an increased risk of associated health conditions (e.g., heart disease, diabetes, high blood pressure, obesity), and a decreased quality of life. Because extent and characteristics of impairment vary from individual to individual, customized approaches are needed to address this important societal problem. However, current approaches tend to be highly subjective and follow a "one size fits all" approach, resulting in limited restoration of walking function for individuals afflicted with these impairments.The long-term goal of this research is to use computer models to design novel walking function approaches for individuals affected by neuromusculoskeletal disorders. The objective of this project is to develop and distribute fast and easy-to-use computer simulation technology that can predict individual walking changes resulting from a proposed treatment. If successful, the project could have wide-reaching benefits to the field, society, and education. For the field, neuromusculoskeletal modeling researchers who are not familiar with the proposed technology or do not possess strong programming skills will be able to develop predictive walking simulations with relative ease. In addition, researchers will be exposed to and have the chance to interact with the new technology through planned workshops at national and international conferences, as well as through broad distribution via the web. For society, researchers will be able to generate customized rehabilitation strategies. For example, customized walking predictions could be used to identify new ways to minimize knee contact forces for individuals with knee osteoarthritis or maximize walking speed and symmetry for individuals who have had a stroke or have Parkinson's disease. For education, "at risk" high school students from underrepresented groups will be exposed to ways that technology is being used to improve human health. This project proposes to develop novel optimal control technology tailored to the unique needs of predictive human walking simulations. Optimal control is a branch of engineering theory that predicts a control strategy that will produce the best-possible performance of a specified dynamical system (for example, determine how to fire rocket thrusters such that a rocket reaches a desired orbit with minimum fuel expenditure). Although optimal control theory has been used extensively to solve aerospace problems, its capabilities have not been exploited for human movement applications. This project will integrate the two traditionally unrelated fields of neuromusculoskeletal modeling and optimal control. The integrated technology will make it easy to perform complex three-dimensional walking simulations that reproduce and predict heterogeneous walking data sets. The technology will be custom tailored to the unique challenges of walking simulations (e.g., intermittent contact between the feet and the ground) and will be able to solve three-dimensional walking problems that are currently intractable or extremely time consuming. The primary development challenge will be to use the known structure of the optimal control problem formulation to improve dramatically the computational speed and robustness of the solution process for walking problems. The primary utilization challenge will be to integrate neuromusculoskeletal models with diverse types of walking data so that models and data are consistent with one another. The technology will use the Matlab programming environment and will be based on the freely-available OpenSim musculoskeletal modeling software developed by researchers at Stanford University. A suite of three benchmark problems involving complex three-dimensional walking problems will be used to evaluate the technology. The technology and benchmark problems will be broadly distributed to the research community via the web and conferences to help advance the entire field. The ability to calibrate individual-specific neuromusculoskeletal walking models and predict the corresponding walking motions in minutes rather than hours or days of CPU time would be an engineering breakthrough that has the potential to transform the way musculoskeletal modeling researchers perform large-scale human moment simulations.