Monitoring of Sitting Postures With Sensor Networks in Controlled and Free-living Environments: Systematic Review (Preprint)

Abstract

BACKGROUND Background: Prolonged sitting postures have been reported to increase the probability of developing low back pain. Moreover, the majority of employees in the industrial world work ninety percent of their time in a seated position. OBJECTIVE This review focuses on the technologies and algorithms that have been used to classify seating postures on a chair with respect to spine and limb movements. METHODS Three electronic literature databases have been surveyed to identify the studies classifying sitting posture in adults. Fourteen articles have been finally shortlisted. These articles were categorized into low, medium, and high quality. Most of the articles were categorized as medium quality (12/14). RESULTS The majority of the studies used pressure sensors (13/14) to classify sitting postures. Neural Networks were the most frequently (6/14) used approaches for classifying sitting postures. CONCLUSIONS Based on the current study the classification of sitting posture is still in the nascent stage and hence, we would suggest personalized sitting posture analysis. Furthermore, the review emphasizes identifying at least five basic postures along with different limb and spine movements in a free-living environment. It is essential to annotate the data set with ground truths for subsequent training of the classifier to solve the sitting posture classification problem.