A Sleep Staging Method Using Multi-parameters

Abstract

Sleep staging is an important basis for the diagnosis of sleep disorders and other related diseases. An auto sleep staging method based on multi-parameter features fusion was proposed. First, using the Discrete Wavelet Transform (DWT) to filter the raw signals, obtain 8 features of sleep staging. Then the parameters of features were inputted to Support Vector Machine (SVM) for training and classification. Finally, using the cross-validation to model different samples for verifying the generalization ability. The results show that the accuracy of the method reach 93.08%, which is 9.39% higher than based EEG, and reached 85.25% with cross-validation, that indicated a better generalization ability.