CRII: SCH: Applying Motor Control Theories for Ambulatory Monitoring of 3D Upper-Limb Movement

University Of Massachusetts Amherst, Amherst MA

Investigators

Abstract

Accurate monitoring of three dimensional (3D) upper limb posture and motion in a community setting has been of paramount importance in rehabilitation. This work is aimed at developing a system that can provide an objective assessment of individually-tailored therapeutic treatments. Specifically, in movement disorders with a long recovery time, such as stroke or traumatic brain injury, continuous monitoring of movement using a minimally-invasive sensing has been a goal to support long-term adherence. A wrist-worn inertial sensor has been the most commonly used wearable sensor due to its immediacy, ubiquity, and acceptance for sustained use. However, developing a precise understanding of upper limb movements based on a single wrist-worn device is challenging. Data from these devices tends to drift over time; that is, a small error in the sensor measurements grows rapidly as the data are integrated to estimate the movement. This project aims to establish a system that allows accurate monitoring of upper limb movement using a single wrist-worn inertial sensor by specifically addressing the issue of drift. The proposed effort will advance the state-of-the-art in ambulatory monitoring of upper limb motion by exploiting the unique kinematic properties of voluntary upper limb movements mediated by the human central nervous system (CNS) and the physical properties of the musculoskeletal structure. The project starts from prior knowledge regarding the unique kinematic characteristics of limb motion that would eliminate the need for the second integration and many of the drift errors. This model will provide unique opportunities to develop a novel computational algorithms for precise measurement of upper limb motion and kinematics. This study will address the following scientific challenges: 1) development of mathematical models and computational algorithms to estimate the dynamically changing body direction, 2) establishment of a new machine learning framework to estimate the 3D position trajectory of the sensing unit without double integration by leveraging the motor control theories, and 3) development of a sequential algorithm to estimate the most likely kinematic profiles of limb joints based on the human musculoskeletal properties. The success of this project will lead to a major breakthrough in precise gesture monitoring in the free-living setting, opening a new door leading to previously unexplored datasets and potentially new development of personalized disease management via unobtrusive monitoring of motor functions in movement disorders. This project will also embrace the integration of interdisciplinary research and undergraduate/graduate training among the areas of wearable computing, signal processing, data science, and smart health. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.