Sleep-related fall monitoring among elderly using non-invasive wireless bio-sensors
Loading...
Files
Date
2014
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
University of New Brunswick
Abstract
Human cognitive function decreases with aging thereby increasing the risk of fall. A fall can cause severe injury, long hospitalization time, and often affects an individual's quality of life. Fall data obtained from a nursing home in New Brunswick shows that 50% of fall incidents happen during night sleep. In this thesis, a fall detection and prediction system is developed in which sleep brain activity is captured and analyzed in real time. A fall classification method for the brain signals captured as Electroencephalography is developed using Support Vector Machine (SVM) and Time-Frequency Kernels. In this fall-prediction system, a patient wears a hat with a light weight wireless biosensor device to capture EEG signals, and then sent wirelessly to a back-end server for real-time analysis of the data sets. Over the supervised training period, the server gets enough data from a subject and starts to learn the threshold value between normal and abnormal EEG for the subject. When the system is trained with signature data, it gives more accurate detection result.