Doctoral Dissertations

Date of Award

12-2021

Degree Type

Dissertation

Degree Name

Doctor of Philosophy

Major

Biomedical Engineering

Major Professor

Eric R. Espinoza-Wade

Committee Members

Eric R. Espinoza-Wade, Dustin L. Crouch, Jeff Reinbolt, Xiaopeng Zhao, Amir Sadovnik

Abstract

Nearly 800,000 people in the United States suffer stroke annually. Following the onset of stroke, survivors will exhibit deficits, such as hemiplegia, which will limit their function and ability to perform activities of daily living (ADLs). In order to regain independence, many stroke survivors will employ maladaptive compensatory strategies to help with the completion of tasks. Compensation is generally defined as any performance of a task that is different than the way it may have been performed before the onset of a neurodegenerative disorder. While for some severely impaired individuals, compensation may be necessary, for most these maladaptive strategies ultimately lead to a decreased quality of life and lifespan. Compensatory behaviors are most often developed in the ambient setting when an individual is unmonitored by a health care professional. Thus, it is necessary to develop a system which can autonomously detect compensatory behavior and then administer feedback or an intervention based on the compensation measured.

To develop this system, we first performed a scoping review to determine the specific, quantitative definitions of feedback described in the literature. The results of this analysis showed that compensation is most often defined in the context of a segment and a task being performed. Given that specific definitions of compensation are task specific, we used minimally intrusive inertial measurement units (IMUs) to evaluate the significant differences between commonly performed tasks (unimanual, bimanual symmetric, and bimanual asymmetric) and between groups (control and post-stroke). We then tested the capability of various machine learning methods to differentiate between groups and task types given features derived from the time-series human motion data. Ultimately, we created an echo state neural network (ESNN) which could differentiate between healthy and unhealthy task performance with an acceptable level of accuracy. We then developed an application which could use the ESNN in real time to detect compensatory behaviors. Using language specifically designed to enhance recovery, participants were given feedback on their performance in a two week pilot trial of the system, and the results of their recovery were reported.

The goal of this research was to aid in the long term recovery and decrease the use of compensatory strategies for persons post-stroke.

Files over 3MB may be slow to open. For best results, right-click and select "save as..."

Share

COinS