Alzheimer’s disease detection using skeleton data recorded with Kinect camera

• Published in: Cluster computing Vol. 23; no. 2; pp. 1469 - 1481
• Main Authors: Seifallahi, Mahmoud, Soltanizadeh, Hadi, Hassani Mehraban, Afsoon, Khamseh, Fatemeh
• Format: Journal Article
• Language: English
• Published: New York Springer US 01.06.2020; Springer Nature B.V
• Subjects: Cameras; Classification; Cognitive ability; Computer Communication Networks; Computer Science; Disease; Exercise; Gait; Older people; Operating Systems; Processor Architectures; Sensors; Statistical analysis; Support vector machines; Three dimensional imaging; Walking; Alzheimer detection; Kinect camera; Support vector machine; Gait analysis; Classification
• ISSN: 1386-7857; 1573-7543
• DOI: 10.1007/s10586-019-03014-z

Alzheimer’s disease (AD) is a neurodegenerative disease that leads to defects in cognitive and functional abilities of elderly people. In this paper, a novel methodology is presented to detect Alzheimer’s disease using recorded skeleton data with a KinectV.2 camera from the subject’s gait. After clinical assessment, the single-task walking test done by subjects was recorded with the kinectV.2 camera. Then, some descriptive statistical analyses were performed on the extracted features of recorded gait to compare them between people with Alzheimer’s disease and people without any cognitive impairment as the healthy control (HC) group. Then, a support vector machine classifier with different kernels was designed to classify subjects to AD and HC groups. The results show that the proposed method has acceptable results in comparison to previous studies to detect AD. The proposed method in this article has the accuracy, sensitivity, precision, and specificity of 92.31%, 96.33%, 88.62%, and 90.81% respectively to classify subjects to AD and HC groups.