Causality Analysis Of Bivariate Time Series In The Magnetoencephalography

Depression is a serious emotional disorder,especially in recent years,as the pace of life has accelerated,all walks of life have frequently seen the problem of depression caused by the high pressure of young people.The pathogenesis of depression is relatively complex.The purpose of this paper is to study the causal problem between the bivariate time series in the magnetoencephalography of the depressed group and the healthy control group,and compare these two group,so as to distinguish the depressed patients from the normal healthy people.This paper has the following three aspects of research content:Firstly,a linear granger causality method was used to analyze the causality between the left and right symmetrical channels of the same brain areas in the magnetoencephalography.It was found that under the stimulus of negative and neutral emotional pictures,the causality coefficient of the depression group was generally larger than that of the healthy control group,while under the stimulus of positive emotional pictures,the causality coefficient was generally smaller than that of the healthy control group.By analyzing the causality of different brain regions and observing the results between the frontal and occipital regions,we can also find that under the negative and neutral emotional picture stimulation,the causality coefficient of the depression group is greater than that of the healthy control group.Secondly,a nonlinear granger causality method of phase space reconstruction was adopted,and the result was better than that of linear granger causality analysis method.In the study of causality between the left and right symmetrical channels of the same brain region,it was found that the causality coefficient of the depressed group was higher than that of the healthy control group when under the stimulation of three different emotional pictures,especially in the frontal,occipital,and temporal region.In addition,when comparing the causal of different directions of symmetrical channels of the same brain region in the same group,it was found that the causal coefficient of the depression group was asymmetrical.To study the bivariate causality between different brain regions,we observe the results between frontal and occipital regions,it was concluded that the causality coefficient of depression group is greater than that of healthy control group.Finally,the method of kernel function granger causality was adopted,by the theory of reconstructing Hilbert space,the causality analysis was promoted to a higher dimensional space,we precisely adopted polynomial kernel function and gaussian kernel function which were two kinds of commonly used kernel function.The result of these two method was the same and were consistent with the result of linear granger causality,but the effect is better.When analyzing the bivariate causality between the left and right symmetrical channels of the same brain region,it was found that under the negative and neutral emotional picture stimulation,the depression group had greater causality than the healthy control group,and the difference was not so obvious under the positive emotional picture stimulation.When analyzing the causality between channels in different brain regions,we observed the causality between the bivariate channels in the occipital and frontal regions,and it was also found that under the stimulus of negative and neutral emotional pictures,the causality in the depressed patients group was larger than that in the healthy control group.