Estimation of Apnea-Hypopnea Index in Sleep Breathing Disorders with the Use of Artificial Neural Networks

Sleep Apnea Syndrome (SAS) becomes an important medical and social problem of contemporary societies. It is burdensome, it can be dangerous to health and even cause of death. The most efficient way to detect this syndrome is polysomnography. It gives good results but it is expensive and not commonly available. Main aim of this study is to present another, easier and cheaper way to detect SAS. Proposed method is based on prediction of sleep state using only oximetry and heart rate. The Artificial Neural Network (ANN) algorithm to predict time series was introduced. These networks were used to detect apneas and hypopneas to support diagnose of SAS and to detect whether patient sleeps or not. All data needed to train and test ANN were collected in sleep laboratory for a group of five considered patients with diagnosed SAS. The presented in this work results show that it is possible to predict apneas during sleep with high rate of accuracy, just with use of information about heart rate and blood oxygen saturation. It means that presented method could be effective to diagnose this disease using only simple device with implemented ANN.