Automated Detecting and Classifying of Sleep Apnea Syndrome Based on Genetic- SVM

Sleep apnea (SA) is one of the common sleep disorders. It has several consequences that can affect daily life activities. The common diagnose procedure is carried out through an overnight sleep test. The test usually includes of several bio-signals recordings that are used to detect this syndrome. The conventional approach of detecting the sleep apnea uses a manual analysis of most bio-signals to achieve reasonable accuracy. The manual process of this test, is highly cost and time consuming. This paper presents a novel automatic system for detecting and classifying apnea events by using just a few of bio-signals that are related to breathe defect. This method uses only the air flow, thoracic and abdominal respiratory movement as inputs for the system. The proposed technique consists of four main parts which are; signal segmentation, feature generation, feature selection and data reduction based on genetic SVM, and classification. Statistical analyzes on the attained results show efficiency of this system and its superiority versus previous methods even with more bio-signals as input