Predicting Knee Joint Kinematics from Wearable Sensor Data in People with Knee Osteoarthritis and Clinical Considerations for Future Machine Learning Models.

Jay-Shian Tan,Peter Kent,Paul Davey,Peter O’Sullivan,Sawitchaya Tippaya,Tara Binnie,Anne Smith,Amity Campbell,J P Caneiro,Kathryn Napier

doi:10.3390/s22020446

Abstract

Deep learning models developed to predict knee joint kinematics are usually trained on inertial measurement unit (IMU) data from healthy people and only for the activity of walking. Yet, people with knee osteoarthritis have difficulties with other activities and there are a lack of studies using IMU training data from this population. Our objective was to conduct a proof-of-concept study to determine the feasibility of using IMU training data from people with knee osteoarthritis performing multiple clinically important activities to predict knee joint sagittal plane kinematics using a deep learning approach. We trained a bidirectional long short-term memory model on IMU data from 17 participants with knee osteoarthritis to estimate knee joint flexion kinematics for phases of walking, transitioning to and from a chair, and negotiating stairs. We tested two models, a double-leg model (four IMUs) and a single-leg model (two IMUs). The single-leg model demonstrated less prediction error compared to the double-leg model. Across the different activity phases, RMSE (SD) ranged from 7.04° (2.6) to 11.78° (6.04), MAE (SD) from 5.99° (2.34) to 10.37° (5.44), and Pearson’s R from 0.85 to 0.99 using leave-one-subject-out cross-validation. This study demonstrates the feasibility of using IMU training data from people who have knee osteoarthritis for the prediction of kinematics for multiple clinically relevant activities.

Highlights

People who have knee osteoarthritis commonly report pain and physical limitation performing functional activities such as walking, transitioning from a chair and negotiating stairs [1]
The difference between the double-leg and the single-leg model was small, with an root mean square error (RMSE) difference ranging from 0.11◦ to 1.96◦ and mean absolute error (MAE) from 0.01◦ to 1.46◦ for timeseries predictions
We developed a bidirectional LSTM (BiLSTM) kinematic prediction model on inertial measurement unit (IMU) training data that included walking, negotiating stairs and transitioning to/from a chair for people who have knee osteoarthritis

Summary

Introduction

People who have knee osteoarthritis commonly report pain and physical limitation performing functional activities such as walking, transitioning from a chair and negotiating stairs [1]. During these activities they use less sagittal plane range of movement (knee flexion) during particular phases of activities (e.g., stance phase of walking) compared to people who do not have osteoarthritis [2,3,4,5]. A person may have difficulty descending stairs because they do not use available knee flexion movement during the stance phase Interventions such as exercise [7] and total knee replacement [8] have demonstrated the ability to improve knee flexion angle during walking in people who have knee osteoarthritis. There are currently several limitations to clinicians being able to accurately quantify sagittal plane knee range of movement during functional activities in both clinical and free-living environments (e.g., patient’s home or work, or during recreation)

Objectives

Results

Discussion

Conclusion