Detection of daily postures and walking modalities using a single chest-mounted tri-axial accelerometer. (July 2018)

• Record Type: Journal Article
• Authors: Nazarahari, Milad; Rouhani, Hossein
• Is Part Of: Medical engineering & physics. Volume 57(2018)
• Page Start: 75
• Page End: 81
• Publication Date: 2018-07
• Subjects: Daily activity analysis -- Postural transition -- Inertial sensor -- Trunk inclination -- Stair ascending/descending; Biomedical engineering -- Periodicals; Biomedical Engineering -- Periodicals; Physics -- Periodicals; Génie biomédical -- Périodiques; Biomedical engineering; Electronic journals; Periodicals; 610.28
• DOI: 10.1016/j.medengphy.2018.04.008
• Languages: English
• ISSNs: 1350-4533

Abstract

Highlights: A number of postural transitions were detected using a chest-mounted accelerometer. Walking with different speeds was detected using the chest-mounted accelerometer. Level walking, walking upstairs and downstairs were classified by the same sensor. Transitions and walking were classified with high sensitivity and specificity. Accuracy of activity detections were compared to those presented in the literature. Abstract: This study presents a novel method for the detection and classification of a wide range of physical activities, including standing, sitting, lying, level walking, and walking upstairs and downstairs using a single chest-mounted accelerometer. The trunk inclination angle and variation of the gravitational component of the accelerometer recording were used for detection and classification of postural transitions and walking modalities. In addition, biomechanical features of each transition were used to reject false detections. To validate the accuracy of the presented method, two studies were performed, first in the (1) laboratory environment, where a motion capture system was the reference system (ten healthy subjects), and second (2) in the free-living environment where a handheld camera was the reference system (ten healthy subjects). The first study showed that the proposed method obtained higher accuracy, sensitivity, and specificity in detection of postural transitions and walking modalities compared to other methods in the literature when implemented on the same dataset. The second study obtained (1) the sensitivity and specificity of 100% for detection of sit-to-lie, lie-to-sit, and stand-to-sit, and 100% and 97%, respectively, for detection of sit-to-stand, and (2) the accuracy of 99%, 99%, and 95% for detection of slow, normal, and fast level walking, and 97% and 96% for detection of walking upstairs and downstairs. The proposed method enabled detection and classification of postural transitions and walking modalities with high sensitivity and specificity using only one chest-mounted accelerometer. This approach can be used for convenient and reliable assessment of physical activities in long-term.