Entropy Study On EEG Signals Of Mild Cognitive Impairment In Diabetic Patients

Mild Cognitive Impairment(MCI)is an early stage of transition from normal aging to Alzheimer's disease(AD).The main feature are memory impairment and no significant decline in daily living capacity.Mild cognitive dysfunction which is induced by diabetes is an important factor in the development of dementia,and the frequency of type 2 diabetes conversion to dementia is extremely high.With the prevalence of increased year by year,type 2 diabetes seriously affected the quality of life of patients.So the study of diabetes and cognitive function related to the EEG signal can deepen the understanding of the mechanism of cognitive impairment of diabetes,and it is of great significance in cognitive impairment early prevention.Firstly,the basic knowledge of EEG signal is briefly studied,including the mechanism and characteristics of EEG signal,the classification of artifacts,and the common preprocessing method.This paper analyzes the latest developments and development trends of EEG research based on entropy algorithm at home and abroad.There is a brief analysis about EEG data collection,diabetes MCI inclusion and exclusion criteria and cognitive function assessment of the commonly used scale.Secondly,the bispectral entropy algorithm based on AR model estimation is studyed which is based on the bispectrum research.The nonlinear dynamic characteristics of Logistic map model are studied.Logistic mapping model is used to simulate the bispectral entropy algorithm.At the same time,anti-noise and data length are used as the standard to analyze the performance of the algorithm.The bispectrum entropy algorithm has good anti-noise characteristics and insensitive to data length selection.Therefore,the algorithm has wide practicability and can be used to analyze the actual EEG data.The algorithm was applied to the analysis of nonlinear changes of the EEG signals in the MCI group and the control group,and the Pearson linear correlation between the bispectral entropy and the neuropsychological test was explored.Finally,the algorithm of multiscale sample entropy which is based on the study of sample entropy is studyed.At the same time,the Logistic mapping model is used to simulate the performance of the multiscale sample entropy algorithm.The effects of different data lengths on the sample entropy are studied.The influence of the noise intensity on the sample entropy in different scale regions are studied,and the change of the sample entropy values under different scale conditions are studied.The multi-scale sample entropy algorithm and sample entropy algorithm were applied to the analysis of EEG signals in patients with actual diabetes mellitus,and the Pearson linear correlation analysis between multi-scale sample entropy and neuropsychological test was performed in both short-scale and long-scale regions.In contrast to the diabetic control group,the differences in the multi-scale complexity of diabetes were explored in both long and short scale regions.