Analysis Of Sleep Eeg Signals Based On Multiscale Jensen-shannon Divergence And Non-linear Granger Causality

As an essential link in the process of life activity,the sleep is usually used to restore the body vitality,keep the mental state,and act as an important part of the memory capacity increasing.Recently,a large number of scholars who are interested has took part in the study actively,at the same time,the study has important practical significance.The nonlinear dynamics analysis is one of the popular research methods which was used to study the sleep physiological signal.This article mainly adopted a classic algorithms,namely multi-scale Jenson-Shannon Divergence to analyze the differences between the sleep EEG signals in two kinds of states,the awake period and the NREM-I period.In addition,this paper also uses the nonlinear granger causality test method to verify the direction of inflution between the EEG signals and the ECG signals,the related conclusion hoped can give some help for sleep research.The main contents of this paper are as follows:First,in this paper,based on the theory of nonlinear dynamics knowledge,JSD and multi-scale JSD algorithm are proposed to research the differences between the EEG signals both in the states of conscious period and NREM-I period,then using the SPSS statistical software to validate the veracity and the reliability of the experiments.With the error bar graph method to show the differences between the EEG signals of both two kinds of states.The research results show that the JSD algorithm and the improved multi-scale JSD algorithm can effectively distinguish the EEG signals between the awake period and NREM-I stage.We hope that the algorithm we proposed can be further used in the study of sleep EEG in installment,can provide all kinds of diseases diagnosis and treatment of sleep with effective auxillary function,the research has important practical significance.Second,this article uses the multi-scale JSD algorithm to analyze the differences between the alpha EEG signals and the theta EEG signals of the conscious period and the NREM-I period.Then using the SPSS statistical software to verify the the result of the alpha and theta EEG signals.The two group of experiments got a good result from both in accuracy and reliability.With the error bar graph,this paper also analysed two different states of alpha sleep EEG signals and theta sleep EEG signals,the result show that the improved multi-scale JSD algorithm can effectively distinguish thealpha EEG signals and theta EEG signals between the awake period and NREM-I period,shows that these two situations exist significant differences between the alpha EEG signals and theta EEG signals,which is unified with the scientific basis,namely the alpha waves in eeg signals are more than the other waves when people are awaked,however,there are more theta waves in eeg signals when people are in state of NREM sleep stage I(shallow sleep phases),proved that the EEG signals in two states exist significant differences,which also tell me that the algorithm we proposed can be further used in the study of sleep EEG in installment,which can be provided to help all kinds of disease diagnosis and treatment of sleep with effective auxiliary function,the research has an important practical significance.Third,this article introduces two kinds of granger causality algorithm based on kernel function to research the direction of the interaction between the sleep EEG signals and the sleep ECG signals.The experimental results show that both algorithm can be used to explore the diretion of interaction between the sleep physiological signal,and the index of the granger causality of the direction from the sleep EEG signals to the sleep ECG signals is greater than the index of the opposite direction.It is important to study the direction of interactions between physiological signals.