Research On EEG Data Recognition Method For Mild Cognitive Impairment Diagnosis

As the early stage of Alzheimers disease(AD),Mild Cognitive Impairment(MCI)has attracted more and more attention from all walks of life.Among them,the coupling characteristics of electroencephalogram(EEG)data have become important biomarkers for the early identification of AD and MCI.However,most of the current methods only focus on the coupling relationship between the two brain regions,and less consider the influence of other channels on this coupling.Based on this,this paper proposes a multi-dimensional EEG signal coupling feature extraction method and finds the most suitable classifier for this feature classification through experiments,the fast learning network(FLN)method.Firstly,the fast learning network(FLN)classification method with the best combination of sorting conditions and mutual information is studied.Based on the sorting condition mutual information method,the coupling feature extraction of EEG signals was combined with a variety of mainstream classifiers to construct the classification function of EEG signals,through amnestic Mild Cognitive Impairment(aMCI).Type 2 diabetes mellitus(T2DM)patients were compared with PCMI characteristics extracted from EEG signals in different brain regions in different T2 DM patients without aMCI.The classification of PCMI and various classifiers was compared.EEG signal recognition performance and.The experimental results show that under the combination of multiple brain regions such as alpha1 and alpha2,the coupling characteristics based on PCMI extraction can be significantly better than other classifiers and PCMI after classification by FLN classifier.Secondly,the multivariate permutation conditional mutual information(MPCMI)method is proposed in consideration of the influence of other channels on coupling relationship between two channels.In this paper,the multi-channel simulation EEG data generated by the multi-channel coupled neuronal group model and the true multi-channel resting state EEG data of a2 M patients with T2 DM and NC patients without aMCI were used to verify the MPCMI method in extracting the coupling strength.Features and superiority of EEG signal recognition in aMCI patients.The results show that the MPCMI coupling feature extraction method outperforms the other three coupling feature extraction methods in real EEG signal analysis.In summary,this paper studies the method of identifying MCI EEG data from the perspective of the classifier method and multidimensional coupling feature extraction method that are coupled with the coupling features.The MCI experimental data and simulation data verify that the FLN can be used as the best combination of classifier with PCMI in the field of EEG data recognition of MCI patients,and MPCMI can be used as the best multi-channel coupling feature extraction method and method.