Research On Diagnosis Method Of Mild Cognitive Impairment Based On EEG Feature Analysis

Mild cognitive impairment(MCI)is the early stage of Alzheimer’s disease(AD).The incidence rate of MCI patients transforming into AD is 10 times that of normal persons,but this stage is reversible and is the key stage of diagnosis and treatment.The implementation of diagnostic identification for MCI stage can help reduce the risk of dementia caused by AD,which has important theoretical research significance and clinical application value.This topic explores the EEG marker characteristics that can fully characterize the brain functional state of MCI,to help realize the diagnosis of MCI.The research analyzes the brain function marker characteristics of MCI patients from the two aspects of EEG entropy characteristics and brain function network characteristics and constructs the MCI diagnosis model based on support vector machine to realize the diagnosis of MCI.Firstly,from the perspective of MCI brain function complexity evaluation,to solve the problems of signal jitter and distortion under multi-scale entropy and difficulty to mine the details of EEG signals,an optimization algorithm of EEG feature extraction based on multi-scale entropy is introduced.The algorithm constructs multi-scale sequences,and then evaluates the correlation between sequences,to realize the maximum degree of information mining.The results showed that there were significant differences in the prefrontal lobe,anterior temporal lobe and middle temporal lobe in MCI group(P < 0.01).Further,taking this feature as the classifier input,the classification accuracy of the three brain regions is 83.3%,86.7%and 73.3% respectively,of which the AUC values of the anterior temporal lobe brain regions are 0.753 and 0.733 respectively.Further,from the perspective of MCI brain function connectivity evaluation,two brain function network connection models,namely phase synchronization index network and directional transfer function network,are constructed to evaluate the cognitive function status of MCI patients and the evaluation effect difference of network model.The results show that the classification accuracy and AUC of the two network models are 66.6% and0.748,80.0% and 0.783 respectively.Finally,given the insufficient representation of the existing features on the global transmission ability of brain function,a new small world attribute feature,the efficiency density feature,is introduced.This feature more comprehensively describes the effect of information transmission on brain function networks,and its classification accuracy and AUC value are improved to 86.6% and 0.830 respectively.Based on PyQt5,SQL and other environments,combined with the research content of this subject,the MCI diagnosis and evaluation system is designed and implemented.The system includes modules such as data acquisition,rhythm extraction,network model and diagnosis and evaluation,which can effectively realize the evaluation and diagnosis of MCI patients.