Wearable Intelligent Machine Learning Rehabilitation Assessment for Stroke Patients Compared with Clinician Assessment

Guo, Liquan; Zhang, Bochao; Wang, Jiping; Wu, Qunqiang; Li, Xinming; Zhou, Linfu; Xiong, Daxi

doi:10.3390/jcm11247467

Cited by 11 publications

(7 citation statements)

References 30 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…To date, studies have largely focused on training ML models to directly predict clinical assessment scores (e.g., FMA, WMFT) or athletic level from kinematic data. These data are often acquired from the administration of the clinical assessments themselves [ 62 , 63 , 64 , 65 , 66 ], robotic platforms [ 67 ], or movement batteries [ 68 , 69 ], which require expert assessors, extra time, and/or specialized setups. Feature engineering must also be undertaken to support ML-based analysis.…”

Section: Discussionmentioning

confidence: 99%

Data-Driven Quantitation of Movement Abnormality after Stroke

Parnandi

Kaku²,

Venkatesan

et al. 2023

Bioengineering

View full text Add to dashboard Cite

Stroke commonly affects the ability of the upper extremities (UEs) to move normally. In clinical settings, identifying and measuring movement abnormality is challenging due to the imprecision and impracticality of available assessments. These challenges interfere with therapeutic tracking, communication, and treatment. We thus sought to develop an approach that blends precision and pragmatism, combining high-dimensional motion capture with out-of-distribution (OOD) detection. We used an array of wearable inertial measurement units to capture upper body motion in healthy and chronic stroke subjects performing a semi-structured, unconstrained 3D tabletop task. After data were labeled by human coders, we trained two deep learning models exclusively on healthy subject data to classify elemental movements (functional primitives). We tested these healthy subject-trained models on previously unseen healthy and stroke motion data. We found that model confidence, indexed by prediction probabilities, was generally high for healthy test data but significantly dropped when encountering OOD stroke data. Prediction probabilities worsened with more severe motor impairment categories and were directly correlated with individual impairment scores. Data inputs from the paretic UE, rather than trunk, most strongly influenced model confidence. We demonstrate for the first time that using OOD detection with high-dimensional motion data can reveal clinically meaningful movement abnormality in subjects with chronic stroke.

show abstract

Section: Discussionmentioning

confidence: 99%

Data-Driven Quantitation of Movement Abnormality after Stroke

Parnandi

Kaku²,

Venkatesan

et al. 2023

Bioengineering

View full text Add to dashboard Cite

show abstract

“…Researchers in [28] propose leveraging AI and patient data for personalized stroke outcome predictions. Recent stroke treatment advancements have reshaped care.…”

Section: Related Workmentioning

confidence: 99%

A Comparative Study Investigating Machine Learning Methods for EMG Data Classification in Post-Stroke Rehabilitation

M. Abdulaziz,

Loqman

2024

IRJIET

View full text Add to dashboard Cite

Physiotherapy is essential for boosting recovery and enhancing quality of life after a stroke. Individualized therapies are required for stroke rehabilitation. This work explores machine learning techniques for electromyography (EMG) data classification in the context of post-stroke rehabilitation, which is important for comprehending and improving motor function. Our analysis covers a wide range of methods, such as Gradient Boosting (GB), Histogram-Based Gradient Boosting, Cat Boost, K-Nearest Neighbors (KNN), Decision Tree (DT), Random Forest (RF), Linear Discriminant Analysis (LDA), and LDA. We measure their performance using criteria like accuracy, precision, recall, and F1-score through a thorough evaluation procedure. Our results show that DT and RF perform consistently better than other algorithms, proving their reliability and effectiveness in the classification of EMG data. But KNN is equally promising, with relatively lesser precision than LDA. The work emphasizes how decision-based algorithms and ensemble methods may effectively classify EMG data, which has ramifications for improving stroke rehabilitation techniques. These discoveries can be used by scholars and professionals to further the creation of machine learning-based instruments for accurate gesture recognition in post-stroke rehabilitation.

show abstract

“…Similarly, in [ 115 ], the authors used wearable inertial sensors placed on different body parts, which gather the movement data and transmit them through Zigbee to a PC. Pre-processing methods were applied to enhance the signal quality before extracting movement features through the multi-sensor fusion technique.…”

Section: Overview Of Available Wearable Technologies For Body Motion ...mentioning

confidence: 99%

Wearable Sensors and Smart Devices to Monitor Rehabilitation Parameters and Sports Performance: An Overview

Fazio

Mastronardi

Vittorio

et al. 2023

Sensors

View full text Add to dashboard Cite

A quantitative evaluation of kinetic parameters, the joint’s range of motion, heart rate, and breathing rate, can be employed in sports performance tracking and rehabilitation monitoring following injuries or surgical operations. However, many of the current detection systems are expensive and designed for clinical use, requiring the presence of a physician and medical staff to assist users in the device’s positioning and measurements. The goal of wearable sensors is to overcome the limitations of current devices, enabling the acquisition of a user’s vital signs directly from the body in an accurate and non–invasive way. In sports activities, wearable sensors allow athletes to monitor performance and body movements objectively, going beyond the coach’s subjective evaluation limits. The main goal of this review paper is to provide a comprehensive overview of wearable technologies and sensing systems to detect and monitor the physiological parameters of patients during post–operative rehabilitation and athletes’ training, and to present evidence that supports the efficacy of this technology for healthcare applications. First, a classification of the human physiological parameters acquired from the human body by sensors attached to sensitive skin locations or worn as a part of garments is introduced, carrying important feedback on the user’s health status. Then, a detailed description of the electromechanical transduction mechanisms allows a comparison of the technologies used in wearable applications to monitor sports and rehabilitation activities. This paves the way for an analysis of wearable technologies, providing a comprehensive comparison of the current state of the art of available sensors and systems. Comparative and statistical analyses are provided to point out useful insights for defining the best technologies and solutions for monitoring body movements. Lastly, the presented review is compared with similar ones reported in the literature to highlight its strengths and novelties.

show abstract

Wearable Intelligent Machine Learning Rehabilitation Assessment for Stroke Patients Compared with Clinician Assessment

Cited by 11 publications

References 30 publications

Data-Driven Quantitation of Movement Abnormality after Stroke

Data-Driven Quantitation of Movement Abnormality after Stroke

A Comparative Study Investigating Machine Learning Methods for EMG Data Classification in Post-Stroke Rehabilitation

Wearable Sensors and Smart Devices to Monitor Rehabilitation Parameters and Sports Performance: An Overview

Contact Info

Product

Resources

About