Towards remote monitoring of dynamic arm supports for individuals with Duchenne muscular dystrophy using 3D accelerometry. (1st November 2022)

• Record Type: Journal Article
• Authors: Hartman, Amy; Elkhadrawi, Mahmoud; McKendry, Sarah; Akcakaya, Murat; Bendixen, Roxanna M.
• Is Part Of: Expert systems with applications. Volume 205(2022)
• Publication Date: 2022-11-01
• Subjects: 3D Accelerometry -- Dynamic arm supports -- Duchenne muscular dystrophy -- Classification -- Feature selection; Expert systems (Computer science) -- Periodicals; Systèmes experts (Informatique) -- Périodiques; Electronic journals; 006.33
• DOI: 10.1016/j.eswa.2022.117712
• Languages: English
• ISSNs: 0957-4174

Abstract

Highlights: Dynamic arm supports promote independence in users with Duchenne muscular dystrophy. Accelerometry data during a device trial offers clear insight on usage, performance. In this study we use feature extraction and SVM classifiers with accelerometry data. We can identify use of a dynamic arm support and success of standardized movements. The proposed methods are crucial first steps towards remote monitoring of device use. Abstract: Objective: Actively actuated dynamic arm supports, such as the KINOVA O540, provide support for increased independence for individuals with upper limb weaknesses like young men with Duchenne muscular dystrophy. However, there is a significant need for a standardized method to evaluate the impact that dynamic arm supports have on functional activity within the community setting. Approach: This study uses accelerometry and motor performance data collected during a dynamic arm support device trial with young men with Duchenne muscular dystrophy. Using the accelerometry data during each task of the Performance of Upper Limb test, we used the Support Vector Machine (SVM) learning method to test device sensitivity (identification of use of the device) and success sensitivity (identification of success with each task). Main Results: Through this method, we demonstrate high levels of accuracy, sensitivity and specificity when identifying the use or non-use of the device within the data. Further, in the times of device use, this method categorizes the data into "success" and "failure" groups with high accuracy. Significance: This study provides a novel application of accelerometry in tandem with the use of a dynamic arm support device. Utilizing standardized movement items, we present an important first step towards remote monitoring of a dynamic arm support device in the natural setting. Through these results, we push the state of the science forward and aim in future work to use remote monitoring to capture change in goal areas and daily tasks during device trials.