Sleep order detection model using support vector machines and features extracted from brain ECG signals

Out of many sleep disorders exist, sleep apnea is the most serious disorder in detection and cure. This disorder occurs when breathing of the person is disrupted or delay during their sleep. The untreated people will stop breathing regularly during sleep, which means that the whole body and the brain will not get the sufficient oxygen. Pause in breathing can have a particular domain in frequency and event. One of the common types of the sleep apnea is Obstructive sleep apnea (OSA), polysomnography (PSG) is used to examine the Obstructive sleep apnea (OSA) in sleep labs. The test of Obstructive sleep apnea (OSA) is too expensive and difficult as an expert is required to observe the patient overnight. So nowadays with the collaboration of bioengineers, new techniques are developed to classify and detect sleep apnea with higher accuracy. This paper mainly focuses on the computerized classification of OSA subject which is measured by very short length epochs of the electrocardiogram (ECG) data. Here we have implemented our model to classify sleep apnea recordings data with different classifiers like Naive Bayes, KNN, Random forest, support vector machines (SVM), C4.5, LVQ, Quadratic, and Bagging, but the best result is obtained from the SVM classifier with the highest accuracy of 94.32%.