Wearable Sensor Data to Track Subject-Specific Movement Patterns Related to Clinical Outcomes Using a Machine Learning Approach

Abstract

Wearable sensors can provide detailed information on human movement but the clinical impact of this information remains limited. We propose a machine learning approach, using wearable sensor data, to identify subject-specific changes in gait patterns related to improvements in clinical outcomes. Eight patients with knee osteoarthritis (OA) completed two gait trials before and one following an exercise intervention. Wearable sensor data (e.g., 3-dimensional (3D) linear accelerations) were collected from a sensor located near the lower back, lateral thigh and lateral shank during level treadmill walking at a preferred speed. Wearable sensor data from the 2 pre-intervention gait trials were used to define each individual’s typical movement pattern using a one-class support vector machine (OCSVM). The percentage of strides defined as outliers, based on the pre-intervention gait data and the OCSVM, were used to define the overall change in an individual’s movement pattern. The correlation between the change in movement patterns following the intervention (i.e., percentage of outliers) and improvement in self-reported clinical outcomes (e.g., pain and function) was assessed using a Spearman rank correlation. The number of outliers observed post-intervention exhibited a large association (ρ = 0.78) with improvements in self-reported clinical outcomes. These findings demonstrate a proof-of-concept and a novel methodological approach for integrating machine learning and wearable sensor data. This approach provides an objective and evidence-informed way to understand clinically important changes in human movement patterns in response to exercise therapy.

A Validation Study of Freezing of Gait (FoG) Detection and Machine-Learning-Based FoG Prediction Using Estimated Gait Characteristics with a Wearable Accelerometer

“…Principal Component Analysis (PCA) of the original data, consisting of 14 features, resulted in 14 principal components. The principal components are described in terms of total variance in the original data, which should be at least 95% [ 41 , 42 ]. In the proposed study, all 14 principal components provide a variance of more than 95%, and hence all 14 principal components were selected as features.…”

Estimation of Gait Mechanics Based on Simulated and Measured IMU Data Using an Artificial Neural Network

“…Systems that are capable of determining motion kinematics and kinetics without expensive equipment and with less expert knowledge required will drastically increase the availability of motion analysis to a wider range of people. By providing wearable easy-to-use systems in daily life, risky motion patterns (e.g., in gait) might be identified before a major injury occurs or the onset of gait related diseases (Kobsar and Ferber, 2018;Majumder et al, 2019).…”

Progress in wearable electronics/photonics—Moving toward the era of artificial intelligence and internet of things

“…For decades, different from conventional machine‐learning techniques that are limited by their relatively weak ability to handle natural data sets, deep learning can extract much higher‐level and more meaningful features by training an end‐to‐end neural network 49,52,54,56 . In addition, deep learning as a new subfield of machine learning provides an efficient way to adaptively learn representative features from collected raw signals especially on unsupervised and incremental learning, which has made great achievements in image processing, speech recognition, human activity recognition, and so on 7,11‐13 .…”

“…Besides, AIoT (AI + IoT) based on the collective integration of AI and IoT has also emerged and been considered as the state‐of‐the‐art technology to enable intelligent ecosystems in broad IoT applications 45‐48 . When combining wearable electronics/photonics with AI technology, the resultant wearable systems are able to perform a more complicated and comprehensive analysis on the acquired data sets (training sets) beyond the capability of conventional approaches 49,50 . Then this trained model can be used to predict the classification of the new incoming data, acting as the conditioning to trigger an intended event.…”

“…The accuracy of prediction can be improved through choosing suitable algorithms, 51 tuning the parameter of algorithms, 52 and fusing different types of data from diversified sensors 53,54 . Fundamentally, the intelligent systems can change the way of sensing and interaction, with a wide range of applications in advanced identity recognition, personalized healthcare monitoring and treatment, smart home/office/building, smart IoT, encrypted interactions in virtual reality (VR) and augmented reality (AR) environment, and so on 49‐55 …”