Evaluation of Machine Learning Models for Classifying Upper Extremity Exercises Using Inertial Measurement Unit-Based Kinematic Data.

Manuel E Hernandez,Pratik Chaudhari,Nicole Johnson,Andrew Hua,David Buchner,Joshua Quinton,Bruce Schatz

doi:10.1109/jbhi.2020.2999902

Abstract

The amount of home-based exercise prescribed by a physical therapist is difficult to monitor. However, the integration of wearable inertial measurement unit (IMU) devices can aid in monitoring home exercise by analyzing exercise biomechanics. The objective of this study is to evaluate machine learning models for classifying nine different upper extremity exercises, based upon kinematic data captured from an IMU-based device. Fifty participants performed one compound and eight isolation exercises with their right arm. Each exercise was performed ten times for a total of 4500 trials. Joint angles were calculated using IMUs that were placed on the hand, forearm, upper arm, and torso. Various machine learning models were developed with different algorithms and train-test splits. Random forest models with flattened kinematic data as a feature had the greatest accuracy (98.6%). Using triaxial joint range of motion as the feature set resulted in decreased accuracy (91.9%) with faster speeds. Accuracy did not decrease below 90% until training size was decreased to 5% from 50%. Accuracy decreased (88.7%) when splitting data by participant. Upper extremity exercises can be classified accurately using kinematic data from a wearable IMU device. A random forest classification model was developed that quickly and accurately classified exercises. Sampling frequency and lower training splits had a modest effect on performance. When the data were split by subject stratification, larger training sizes were required for acceptable algorithm performance. These findings set the basis for more objective and accurate measurements of home-based exercise using emerging healthcare technologies.

Highlights

E XERCISE regimens are prescribed by physical therapists for a variety of purposes, including treatment of chronic diseases, treatment of acute joint injuries, and prevention of injuries [1], [2]
Multilayer Perceptron (MLP) models took at least two orders of magnitude longer to train in comparison to Random Forest (RF) and 3-NN; the testing time for MLP was similar to RF and faster than 3-NN
This study found that machine learning methods could accurately classify a variety of upper extremity exercises using biomechanical data from an inertial measurement unit (IMU)-based device

Summary

Introduction

E XERCISE regimens are prescribed by physical therapists for a variety of purposes, including treatment of chronic diseases (e.g. rheumatoid arthritis), treatment of acute joint injuries (e.g. tears of rotator cuff muscles at the shoulder), and prevention of injuries (e.g. fall injuries in older adults) [1], [2]. Exercise can be prescribed on most or all days of the week, making supervision by therapists expensive and logistically difficult. One study reported that nearly half of patients assigned self-monitored physical therapy (PT) had to be switched to supervised PT after six weeks, due to muscle atrophy, reduced range of motion (ROM), and low compliance [3]. Given that higher adherence to PT exercise is associated with greater physical function, self-perceived effect, and decreased pain [4], it is concerning that other studies have reported only 35-72% of participants had complete adherence to prescribed PT exercise [5], [6]. While exercise logs are commonly used to monitor home exercise, logs have limitations due to over-reporting and memory errors by patients and due to the inherent difficulty of measuring exercise quality by self-report

Objectives

Methods

Results

Discussion

Conclusion