Abstract

Total knee arthroplasty (TKA) procedures are increasingly common in the US and globally. However, rehabilitation outcomes may be negatively affected by limited access to physical therapy and patients' tendency to not fully adhere to recommended exercises between therapy sessions. This study aims to develop a remote system to aid rehabilitation, building upon recent sensor development at Brigham Young University, providing physical therapy-like biofeedback during unsupervised at-home exercises. The system developed by this research consists of three primary components: (1) a sensor array for measuring leg/knee biomechanics, (2) a machine learning algorithm to predict knee flexion angle from the sensor data, and (3) a system to deliver actionable biofeedback to the user. Several modes of biofeedback were considered to relay knee angle predictions back to the user via a smart phone, using visual, audio and haptic feedback. A study on the effectiveness and user preferences of several biofeedback methods was conducted using a knee flexion exercise as the basis for the study. This revealed that while the effectiveness of the feedback methods was almost identical, user preference varied; users typically favored visual cues over audio and haptic cues. This knowledge informed the design of the mobile application used to relay biofeedback to users using the data collected from the sensor-embedded knee sleeve as input. The rehabilitation aid system utilizes an array of 16 large deflection strain gauges composed of conductive nanocomposite materials. Using this array of sensors mounted on an elastic knee sleeve, a random forest model was developed to predict maximum knee flexion angles during rehabilitation exercises. A method for model calibration / generation in an at-home setting was also developed to account for changes in position and sensor resistance between data collection sessions. A printed circuit board (PCB) was custom built to transfer data from the sleeve to a smartphone application, which uploaded the data to the cloud. Once in the cloud, the data was run through a session-specific machine learning model to predict the maximum knee flexion achieved across the repetition. This prediction was then sent back to the smartphone application, where it was relayed to the user using biofeedback. This process was repeated for each repetition in a set of 10 knee flexion exercises. A system validation was carried out with a small cohort of recent TKA recipients with both in-lab and at-home use of the system. The results of this study showed an improvement in maximum knee flexion, as well as positive feedback from participants. The findings recorded in this thesis have led to the design, creation, and validation of a novel system for aiding TKA rehabilitation that may be transferable to other commonly injured joints, offering additional exciting applications of this new technology.

Degree

MS

College and Department

Ira A. Fulton College of Engineering; Mechanical Engineering

Rights

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

Date Submitted

2024-08-07

Document Type

Thesis

Handle

http://hdl.lib.byu.edu/1877/etd13400

Keywords

nanocomposite, strain gauge, large deflection, total knee replacement, biofeedback application, random forest modeling, knee rehabilitation device

Language

english

Download

Included in

Engineering Commons

Share

COinS