Sleep apnea detection for prephase diagnosis using third level holter recording device

Sleep Apnea Syndrome (SAS) is generally analysed by an expensive medical routine including polysomnographic overnight recordings. Relatively low cost Holter device has been designed to record some significant physiological signals for sleep apnea at home. The device is capable of recording ECG, respiratory effort, oronasal airflow and oxygen saturation data on a high capacity memory card over a long period of time simultaneously for prephase sleep apnea monitoring. Under the scope of this study, algorithms based on processing solely ECG signal and based on artificial neural network using signals from three channels have been developed for prephase sleep apnea diagnosis. The ECG based detection algorithm uses the variation on Power Spectral Densities (PSDs) of Heart Rate Variability (HRV) and RR interval (RRI) signals obtained from ECGs. In neural network approach, oxygen desaturation, oronasal signal parameters along with heart rate variation are processed as inputs to distributed Time Delayed Neural Network (TDNN).