Research On Automatic Sleep Stage Classification Methods Based On Spiking Neural Networks And System Implementation

Sleep quality directly affects the health of the human body,and sleep staging is the basis and core issue for assessing sleep quality.At present,most work of sleep staging is based on the multi-channel physiological signals provided by polysomnography(PSG),which are manually interpreted by professional sleep physicians.This process requires a lot of time and manpower,and its interpretation accuracy is not high.Therefore,the study of the automatic sleep staging model has extremely high research value and clinical application value.With the continuous development of the third-generation neural network,spiking neural networks,its powerful computing power and the design advantages of biomimicry are more obvious than the deep learning neural network.Hence,the research goal of this paper is to build an automatic sleep staging model based on spiking neural networks,and to implement an automatic sleep staging display system based on this proposed model.This paper mainly builds an automatic sleep staging model based on spiking neural networks.Firstly,the original sleep data is preprocessed to obtain sleep staging samples;Secondly,the sleep staging samples are subjected to Ben spiking coding to obtain discrete spiking sequences;Finally,the spiking sequence is input into a spiking neural networks constructed with leaky integrate and fire model and improved recurrent neural networks for automatic classification.The original data set used in the experiments of this paper is the data set 3 from the ISRUC sleep data set provided by the Institute of Systems and Robotics of the University of Coimbra,Portugal.It is composed of 10 healthy people’s sleep physiological signal data without sleep disorders.The experimental results show that the accuracy of the automatic sleep staging method based on the spiking neural networks is higher,and the model is more bio-interpretable in the design process,which proves the feasibility and advantages of automatic sleep staging using spiking neural networks.According to the actual needs of clinical sleep staging work,this paper implements an automatic sleep staging display system.An automatic sleep staging model based on spiking neural networks proposed in this paper is applied to the system,which mainly realizes the display of original physiological information,the display of sleep staging results,sleep quality assessment and other related functions.After actual testing and use,the automatic sleep staging display system has achieved better expected results.