Real Time Sleep Apnea Event Detection with Deep Neural Network

Sleep apnea is a condition where repeatedly the nasal airway stops off while the respiratory effort continues during sleep. Quite a number of researches have been conducted on exploring alternative methods to classify apnea patients and healthy subjects. However, not much work has been done on real time apnea event detection. The purpose of this research is to propose an efficient technique to detect apnea events in real time sequence using Deep Neural Networks (DNN). The network is trained using electroencephalogram (EEG) signals collected from seven individual apnea patients. The trained model is used to test EEG signals of other subjects, which is a very challenging task and rarely attempted. Then the model is tested on other apnea subjects. The performance of the proposed method is extensively tested and results are compared with those obtained by some conventional methods.