Classification of Sleep Apnea Types Using Clustering with SVM Classifier

Classification of Sleep Apnea Types Using Clustering with SVM Classifier

Faiza Charfi, Ali Kraiem
Copyright: © 2012 |Pages: 10
DOI: 10.4018/ijbce.2012010103
(Individual Articles)
No Current Special Offers


A new automated approach for the polysomnography (PSG) characterization and classification with the combination of FastICA, clustering and support vector machines (SVM) is presented in this paper. The method is based on two key steps. In the first step, the authors adopt the Principal Component Analysis (PCA) and Fast Independent Component Analysis (FastICA) approaches to separate and transform the original inputs into uncorrelated and mutually independent new features. In the second step, they utilize the K_Means clustering combined with Support Vector Machine (SVM) to build the proposed classifier. Multiple SVM kernels such as the linear, quadratic, polynomial, and radial basic functions are used for the classification of central and obstructive sleep apnea. Their results suggest the high reliability and high classification accuracy of polynomial kernel.
Article Preview

Most researches on sleep related disorders rely strongly on Polysomnography. The PSG standard comprises continuous recordings of several channels of physiological signals which can be utilized for monitoring nocturnal sleep. The histories of the polysomnography and sleep technology are closely linked to the first successful experiments to monitor brain activity based on the signal processing (Berger, 1933). Aserinsky and Kleitman noticed the presence of Rapid Eye Movements (REMs) during the sleep (Aserinsky & Kleitman, 2003). The significance of the electromyography for the analysis of sleep stages is associated with the discovery of muscular atonia during REM sleep phase (Jouvet & Michel, 1959).

Accurate assessment of sleep quality is achievable by analyzing PSG signal recordings. Several automatic techniques have been developed by researchers in order to detect and characterize sleep disturbance patterns, in particular Fuzzy and Neural Network Algorithm (Ceylan et al., 2009), machine learning methods (Michie et al., 1994), pattern-recognition methods like k-nearest neighbors and Bayesian classifiers (Aha et al., 1991; Dasarathy, 1990), and expert systems (Hu, 1997). These approaches either involve all information sources available in the PSG recordings or use one or more data channels in the PSG recording (Bsoul, 2010). In Varady et al. (2002) the authors have used the whole PSG recording and four Artificial Neural Networks in order to classify data into phases of normal breathing, hypoapnea and apnea. Among approaches that process one or more channel of PSG data, the majority of them utilized electrical activity of the heart during PSG.

Complete Article List

Search this Journal:
Volume 12: 1 Issue (2024): Forthcoming, Available for Pre-Order
Volume 11: 2 Issues (2022): 1 Released, 1 Forthcoming
Volume 10: 2 Issues (2021)
Volume 9: 2 Issues (2020)
Volume 8: 2 Issues (2019)
Volume 7: 2 Issues (2018)
Volume 6: 2 Issues (2017)
Volume 5: 2 Issues (2016)
Volume 4: 2 Issues (2015)
Volume 3: 2 Issues (2014)
Volume 2: 2 Issues (2013)
Volume 1: 2 Issues (2012)
View Complete Journal Contents Listing