In this paper, we present a deep learning framework ‘Rehab-Net’ for effectively classifying three upper limb movements of the human arm, involving extension, flexion and rotation of the forearm which over the time could provide a measure of rehabilitation progress. The proposed framework, Rehab-Net is formulated with a personalized, light weight and low complex, customized CNN model, using 2-layers of Convolutional neural network (CNN), interleaved with pooling layers, followed by a fully-connected layer that classifies the three movements from tri-axial acceleration input data collected from the wrist. The proposed Rehab-net framework was validated on sensor data collected in two situations-a) seminaturalistic environment involving an archetypal activity of ‘making-tea’ with 4 stroke survivors and b) natural environment, where 10 stroke survivors were free to perform any desired arm movement for a duration of 120 minutes. We achieve an overall accuracy of 97.89% on semi-naturalistic data and 88.87% on naturalistic data which exceeded state-of-the-art learning algorithms namely, Linear Discriminant Analysis, Support Vector Machines, and k-means clustering with an average accuracy of 48.89%, 44.14% and 27.64%. Subsequently, a computational complexity analysis of the proposed model has been discussed with an eye towards hardware implementation. The clinical significance of this study is to accurately monitor the clinical progress of the rehabilitated subjects under the ambulatory settings.
Assistive Technologies (ATs), defined as “electrical or mechanical devices designed to help people recover movement”, demonstrate clinical benefits in upper limb stroke rehabilitation; however translation into clinical practice is poor. Uptake is dependent on a complex relationship between all stakeholders. Our aim was to understand patients’, carers’ (P&Cs) and healthcare professionals’ (HCPs) experience and views of upper limb rehabilitation and ATs, to identify barriers and opportunities critical to the effective translation of ATs into clinical practice. This work was conducted in the UK, which has a state funded healthcare system, but the findings have relevance to all healthcare systems.
28 May 2020 in IEEE/ASME Transactions on Mechatronics
Inertial sensing suites now permeate all forms of smart automation, yet a plateau exists in real-world derivation of global orientation. Magnetic field fluctuations and inefficient sensor fusion still inhibit deployment. We introduce a new algorithm, an Extended Complementary Filter (ECF), to derive 3D rigid body orientation from inertial sensing suites addressing these challenges. The ECF combines computational efficiency of classic complementary filters with improved accuracy compared to popular optimization filters. We present a complete formulation of the algorithm, including an extension to address the challenge of orientation accuracy in the presence of fluctuating magnetic fields. Performance is tested under a variety of conditions and benchmarked against the commonly used gradient decent (GDA) inertial sensor fusion algorithm. Results demonstrate improved efficiency, with the ECF achieving convergence 30% faster than standard alternatives. We further demonstrate an improved robustness to sources of magnetic interference in pitch and roll and to fast changes of orientation in the yaw direction. The ECF has been implemented at the core of a wearable rehabilitation system tracking movement of stroke patients for home telehealth. The ECF and accompanying magnetic disturbance rejection algorithm enables previously unachievable real-time patient movement feedback in the form of a full virtual human (avatar), even in the presence of magnetic disturbance. Algorithm efficiency and accuracy have also spawned an entire commercial product line released by the company x-io. We believe the ECF and accompanying magnetic disturbance routines are key enablers for future widespread use of wearable systems with the capacity for global orientation tracking
1 July 2017 in Journal of Neurologic Physical Therapy
Background and Purpose: Stroke, predominantly a condition of older age, is a major cause of acquired disability in the global population and puts an increasing burden on health care resources. Clear evidence for the importance of intensity of therapy in optimizing functional outcomes is found in animal models, supported by neuroimaging and behavioral research, and strengthened by recent meta-analyses from multiple clinical trials. However, providing intensive therapy using conventional treatment paradigms is expensive and sometimes not feasible because of social and environmental factors. This article addresses the need for cost-effective increased intensity of practice and suggests potential benefits of telehealth (TH) as an innovative model of care in physical therapy. Summary of Key Points: We provide an overview of TH and present evidence that a web-supported program, used in conjunction with constraint-induced therapy (CIT), can increase intensity and adherence to a rehabilitation regimen. The design and feasibility testing of this web-based program, “LifeCIT,” is presented. We describe how wearable sensors can monitor activity and provide feedback to patients and therapists. The methodology for the development of a wearable device with embedded inertial and mechanomyographic sensors, algorithms to classify functional movement, and a graphical user interface to present meaningful data to patients to support a home exercise program is explained. Recommendations for Clinical Practice: We propose that wearable sensor technologies and TH programs have the potential to provide most-effective, intensive, home-based stroke rehabilitation.
5 September 2008 in Neurorehabilitation and Neural Repair
Background. Objective assessment of impairments after stroke is vital for evidence-based therapy and progress monitoring. Objective. This study determines the utility of outcome measures obtained from an instrumented wrist rig for future rehabilitation trials. The tests undertaken were evaluated in terms of sensitivity to detect differences between normal and impaired participants, test-retest repeatability (repeatability coefficient and intraclass correlation coefficient [ICC]), and interrater agreement (Bland and Altman limits of agreement). Methods . Twelve participants with chronic poststroke hemiparesis (mean 5.6 years); and 12 unimpaired volunteers performed a series of tasks in the rig. The hemiparetic arm (impaired group) and dominant arm (unimpaired group) were tested in 3 sessions on the same day by 2 assessors. Signals were analyzed to derive a tracking index (motor control), stretch index (spasticity), flexor modulation index (FMI) (muscle activation), force angle index (FAI) (stiffness), range of movement, and isometric force. Results and Conclusions. The means of all tests differed between impaired and unimpaired participants except for range of movement into flexion, the FAI, and the FMI. Repeatability coefficients for each test are presented as benchmark values for use in future trials in which the wrist rig tests may be used to detect change. Test-retest reliability was excellent in the impaired group (ICC = 0.88-0.98) and poor to excellent in the unimpaired group (ICC = 0.06-0.89). The Bland-Altman ranges showed no bias between assessors, and that the interassessor variability was similar to that between repeats by the same assessor for most tests.
1 January 2020 in Journal of Rehabilitation and Assistive Technologies Engineering
Introduction: The Mechanical Muscle Activity with Real-time Kinematics project aims to develop a device incorporating wearable sensors for arm rehabilitation following stroke. These will record kinematic activity using inertial measurement units and mechanical muscle activity. The gold standard for measuring muscle activity is electromyography; however, mechanomyography offers an appropriate alterative for our home-based rehabilitation device. We have patent filed a new laboratory-tested device that combines an inertial measurement unit with mechanomyography. We report on the validity and reliability of the mechanomyography against electromyography sensors. Methods: In 18 healthy adults (27–82 years), mechanomyography and electromyography recordings were taken from the forearm flexor and extensor muscles during voluntary contractions. Isometric contractions were performed at different percentages of maximal force to examine the validity of mechanomyography. Root-mean-square of mechanomyography and electromyography was measured during 1 s epocs of isometric flexion and extension. Dynamic contractions were recorded during a tracking task on two days, one week apart, to examine reliability of muscle onset timing. Results: Reliability of mechanomyography onset was high (intraclass correlation coefficient = 0.78) and was comparable with electromyography (intraclass correlation coefficient = 0.79). The correlation between force and mechanomyography was high (R2 = 0.94). Conclusion: The mechanomyography device records valid and reliable signals of mechanical muscle activity on different days.