SVM based dynamic classifier for sleep disorder monitoring wearable device

An accelerometer embedded wrist-worn device is widely used for sleep assessment. However, conventional methods determine a state of user to "sleep" or "wakefulness" according to whether the accelerometer value of individual epoch exceeds a certain threshold or not. As a result, h...

Full description

Saved in:
Bibliographic Details
Main Authors: Park, Jaihyun, Kim, Daehun, Yang, Cheoljong, Ko, Hanseok
Format: Conference Proceeding
Language:English
Subjects:
Accelerometers
Classification algorithms
Classifiers
Consumer electronics
Devices
Dynamics
Electronics
Kernel
Movements
Sleep
Support vector machines
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:An accelerometer embedded wrist-worn device is widely used for sleep assessment. However, conventional methods determine a state of user to "sleep" or "wakefulness" according to whether the accelerometer value of individual epoch exceeds a certain threshold or not. As a result, high miss-classification rate is observed due to user's small intermittent movements while sleeping and short term movements while awake. In this paper, a novel approach is proposed that mitigates such problems by employing a dynamic classifier which analyzes similarity between the neighboring data scores obtained from support vector machine classifier. Performance of the proposed method is evaluated using 50 real data sets and its superiority is verified.
ISSN:2158-4001
DOI:10.1109/ICCE.2016.7430624