The application of artificial neural network in diagnosis of sleep apnea syndrome

Abstract

In this paper authors propose a method of sleep analysis based on the algorithm of artificial neural networks. Unlike polysomnography methods, that are used commonly in clinical practice, the described method does not require specialist equipment or a qualified technician to analyze biomedical signals. The results presented in this work show that the properly implemented neural network algorithm can determine incidents during sleep and recognize its phases. Main idea was tested on the base of data collected from sleep laboratory of eight patients. From many signals collected during clinical assessment only two were taken under further consideration: heart rate and blood saturation. As it was shown, these two parameters measured during sleep allows to determine incidents occurring during sleeping and even to recognize actual stage of sleep. It means, that it is possible to use simple device that measures only heart rate and blood saturation to identify sleep apnea syndrome. The method is very effective and can replace the existing ways to recognizing sleep problems especially, when sleep examination of patient is conducted in home conditions.