Üçüncü Düzey Taşınabilir Cihaz Kayıtları İçin Dinamik Sinir Ağları Kullanarak Uyku Apnesi Tespiti

Abstract

In this study, dynamic neural networks based procedures have been developed to perform detection of Sleep Apnea Syndrome (SAS) for the data recorded by holter device at home and hospital sleep room. Previous studies mainly concentrated on collecting similar data simultaneously from both developed portable device and Polysomnography (PSG), however studies of detecting apnea episodes for the portable device has not been completed before. In the framework of analyzing and improving the quality of data obtained from portable monitoring devices, records have been discussed. Data recorded in this study are subjected to various noises. First denoising process was performed on data using wavelet transform, then signal pruning procedure was applied for cleaning unusual R-R segments of electrocardiography (ECG) signals. For the noise eliminated ECG signals recorded from three data channel, data set for performing the QRS detection and heart rate variability (HRV) was prepared for apnea diagnostic procedure.