Application Of Deep Neural Network To Study The Sleep Stage Scoring

Traditionally,automatic sleep staging is a very challenging laborious and time-consuming task because most existing automatic sleep staging methods are based on single-channel electroencephalography(EEG)data.Since sleep is an essential mechanism of human self-healing,for this derived a lot of research related to sleep mechanism.However,these methods ignore the physician's overall observation of multiple channel EEG and EOG signals for the sleep stage scoring process.To resolve this problem,we optimized data structures for the physician carried out a detailed study and modeling of the scoring process.Finally,considered the rapid developments in the field of artificial intelligence and Artificial Neural Networks that possess powerful nonlinear fitting capability.We propose an automatic sleep scoring method based on multi-channel EEG.We introduce the use of a deep convolutional neural network(CNN)on raw EEG samples for supervised learning of sleep stage prediction.The network has 11 layers,the 30s epoch to be classified,the predicted sample and the first two periods as input data,and does not require any signal preprocessing or feature extraction.Just need to execute simple data enhancement to the data class that is not balanced,the deletion of specific data and Non-universality of data.We used our EEG and EOG expert assessment of polysomnography(PSG)data from the Third People's Hospital of Longyan City,Fujian Province to train and evaluate our system.Experimental results display that our system performance is comparable to that of human experts.Experiments proved that combination of neural network and medical time series data is entirely feasible,and EOG data should not be ignored in research of automatic sleep stage.We will also explore the use of methods that use other medical time series data to solve other problems,such as disease detection and health analysis.The combination of experiment and theory can better benefit human society.