Coupling Network Analysis Of Physiological Electrical Signal Based On Improved Symbolic Transfer Entropy

The human body is a complex physiological system which showing complicated nonlinear andnon-stationary behavior under the regulation of nerve center. All the physiological functions areinteracting continuously and constituting an integrated network. Failure in one mechanism will leadto the collapse of the entire network. The development of physical network helps to study theinteraction of physiological system, network topology, dynamic information and relationshipbetween physiological function and disease. In view of the great significance of physiologicalcoupling network, this paper mainly focuses on the following work:Firstly, the thesis proposes improvement on the method of time series symbolization based onresearch of original symbolic transfer entropy (STE), adopting dynamic adaptive segmentationmethod. Traditional symbolic dynamics divide symbol area into static range which leading to theoriginal sequence lose some detail information. The result will be severely affected due to thenon-stationary of the series although the dynamic characteristic preserves. The experimental resultsshow that the coupling between electroencephalogram (EEG) and electrocardiogram (ECG) is moresignificant when using improved STE algorithm, which is better to capture the dynamic informationof the signal and the change of complexity of system dynamics.Secondly, we employ improved STE algorithm on the physiological signals of sleep stages toget the integrated coupling network of different periods. Compared with the result of traditionalmethod, the advantages are obvious. Experiments show that quantitative results provide more clearanalysis of interaction strength and change among different physiological systems during differentperiods.Thirdly, combined with wavelet decomposition, the paper decomposes EEG into four kinds ofbasic wave, then calculates STE values among these waves and ECG、electromyography (EMG), tomake the coupling network more abundant. Through quantizing dynamic information of couplingstrength and analysing interaction among different systems, we can track physiological stateevolution process of multiple systems, and also provide effective basis for clinical medicine.