Abstract

This work addresses three objectives related to the design and modeling of an instrumented knee sleeve and its high deflection strain gauges for at-home knee rehabilitation. First, a multi-objective Bayesian approach was developed to economically optimize a high deflection strain gauge's mechanical and electrical behavior. Four objective functions were simultaneously optimized with two design variables in seven iterations; 90% faster than a grid search approach. The presented algorithm adjusts variables in the manufacturing process of high deflection strain gauges and produces a sensor with optimized properties for a predetermined purpose. Second, an algorithm was created to accurately approximate the dynamic stress state of a given area of skin. The magnitude and orientation of local skin strain were calculated using linear strain theory throughout continuous motion to find candidate strain-sensor locations with the highest, repeatable variance. Candidate locations were identified for twenty-five participants performing common knee rehabilitation exercises. The cohort was grouped into three groups by skin motion similarity to determine what biometric measures most likely influence skin movement during joint motion. The procedure presented here can be adopted to other parts of the body and constitutes a necessary step to maximize the effectiveness of skin-mounted, strain-based sensors. Third, a knee sleeve instrumented with a network of high deflection strain gauges was designed, and its ability to estimate knee angles was analyzed. Using data from an instrumented knee sleeve and motion capture from 18 participants, subject-specific adaptive boosting random forest models were created to estimate knee angles from sixteen sensors on the knee sleeve. It was found that the sixteen sensors were able to model the flexion/ extension and internal/ external rotation of the knee with an average root--mean--square errors of 7.6 and 1.8 degrees, respectively. The device developed in this work presents a cost-efficient alternative to measure two important degrees of freedom for natural joint mechanics. We propose that wearable devices will become a key enabling technology for at-home rehabilitation, but sensors and garments must be optimized to increase the device's accuracy. The device presented here will allow physical therapists to better help patients in their homes. This work also provides enabling data and tools for engineers to design optimized sensor networks for implementation in wearable devices.

College and Department

Mechanical Engineering

Rights

https://lib.byu.edu/about/copyright/

Date Submitted

2022-01-24

Document Type

Dissertation

Handle

http://hdl.lib.byu.edu/1877/etd12658

Keywords

piezoresponsive, high deflection strain gauge, rehabilitation, wearables, skin mechanics, multiobjective optimization

Language

english

Download

Included in

Engineering Commons

Share

COinS