Abstract

Using a new method to detect sleep stages in medical applications for reducing the workload of physicians in the analysis of sleep data is one of the key issues in recent years. In this study, the PPG (Photoplethysmogram) signal is used for the detection of sleep stages. Using a new method (ANFIS) and a new signal (PPG) for sleep stage detection. The signal features extracted using the conventional methods and best features are selected using neighborhood component analysis for classification method. Finally, sleep steps are detected using valid methods: Neural network, linear discriminant analysis, KNN, support vector machine, and ANFIS. The accuracy of sleep stages detection was 57.72%, 70.75%, 69.72%, 93.35% and 99.48% respectively. The computation time is 3.79 sec, 1.21 sec, 1.43 sec, 21.15sec, and 36 sec, respectively. This study shows that the proposed method has separated the sleep stages with acceptable accuracy.