Application Research Of Multi-layer Complex Network In Classification Schizophrenia

In recent years,brain disease research based on complex networks has become a hotspot.The use of EEG data to construct complex networks is an important method for traditional research on human brain diseases.Previous studies have focused on single-layer complex networks,focusing on a certain frequency band and a certain stage.There are differences in attributes,but the traditional research paradigm has unavoidable flaws.The various stages of the entire frequency band cannot be considered as a whole.The frequency bands of each stage of the experiment are also discussed separately,and the network characteristics of the single-layer complex network also make some main information.It is hidden or cannot be expressed,so that the significant difference attribute obtained by it is less,the classification accuracy rate is low,and the classification correctness is insufficient.Based on multilayer complex network theory,it can contain many different types of objects and relationships at the same time,which is richer than traditional complex network representation,and better support modeling and analysis and prediction tasks.This paper explores the construction and analysis methods of multilayer brain function network from different angles,and proposes two methods for constructing multi-layer brain function network.Based on the EEG data of patients with schizophrenia,the multi-layer brain network is constructed by fully utilizing the theory of multi-layer complex networks.By analyzing and comparing the global and local attributes of the patient's multi-layer brain network,the significant attributes of the differences are found,and the SVM classifier is used for classification and comparison.The classification accuracy rate is up to 90%.According to the comprehensive analysis,the brain network attribute obtained by the multi-layer complex network theory can be used as the judgment index of schizophrenia disease,and applied to the research and clinical diagnosis of schizophrenia.main tasks as follows:(1)Construct a multi-layer brain function network.First,the scalp node is used as the node,the node key is connected to determine the edge and the threshold is determined,and then the phase and frequency are set.Finally,the three-layer complex and the five-band are combined to construct a three-layer complex network and a five-layer complex network.(2)Calculate the local and global attributes of the three-layer brain function network transformed by time series.In each frequency band,compare the three-layer brain network attributes of the schizophrenia patient group and the normal person group,extract significant difference attribute values,and perform classification analysis.Find out the nodes in the working memory experiment where the patient group is abnormal in each frequency band and analyze its function on brain regions.The results show that the local attribute synthesis is better than the global attribute.In the frequency band comparison,the indicators of the ? band have different performances and can be considered as important indicators for distinguishing patients with schizophrenia and normal people.(3)Calculate the local and global attributes of the five-layer brain function transformed by the frequency domain.In each of the three stages,calculate the attribute values of the two groups of five-layer brain function networks,and use machine learning to classify and compare the significant differences..Observe the changes of the significant differences in the properties of each attribute in the working memory experiment,and analyze the changes in brain function.The results showed that there were a large number of significant differences between the schizophrenia group and the normal group in each frequency band,which were distributed in various brain regions,which proved that schizophrenia is a whole brain disease rather than just a local brain region.Lesion.In a word,this paper focuses on the construction method and classification analysis of brain function network based on multi-layer complex network,explores the basic topological properties of multi-layer brain function network,and provides new ideas and perspectives for brain network under EEG structure.