In-lab versus at-home activity recognition in ambulatory subjects with incomplete spinal cord injury. Issue 1 (December 2017)

• Record Type: Journal Article
• Authors: Albert, Mark; Azeze, Yohannes; Courtois, Michael; Jayaraman, Arun
• Is Part Of: Journal of neuroengineering and rehabilitation. Volume 14:Issue 1(2017)
• Page Start: 1
• Page End: 6
• Publication Date: 2017-12
• Subjects: Activity recognition -- Activity tracking -- Incomplete spinal cord injury -- At-home -- Machine learning; Nervous system -- Diseases -- Patients -- Rehabilitation -- Periodicals; Nervous system -- Wounds and injuries -- Rehabilitation -- Periodicals; Biomedical engineering; 616.8043005
• DOI: 10.1186/s12984-017-0222-5
• Languages: English
• ISSNs: 1743-0003

Abstract

Abstract Background Although commercially available activity trackers can aid in tracking therapy and recovery of patients, most devices perform poorly for patients with irregular movement patterns. Standard machine learning techniques can be applied on recorded accelerometer signals in order to classify the activities of ambulatory subjects with incomplete spinal cord injury in a way that is specific to this population and the location of the recording—at home or in the clinic. Methods Subjects were instructed to perform a standardized set of movements while wearing a waist-worn accelerometer in the clinic and at-home. Activities included lying, sitting, standing, walking, wheeling, and stair climbing. Multiple classifiers and validation methods were used to quantify the ability of the machine learning techniques to distinguish the activities recorded in-lab or at-home. Results In the lab, classifiers trained and tested using within-subject cross-validation provided an accuracy of 91.6%. When the classifier was trained on data collected in the lab but tested on at home data, the accuracy fell to 54.6% indicating distinct movement patterns between locations. However, the accuracy of the at-home classifications, when training the classifier with at-home data, improved to 85.9%. Conclusion Individuals with unique movement patterns can benefit from using tailored activity recognition algorithms easily implemented using modern machine learning methods on collected movement data.