Research On Spatial Cognitive Training And Eeg Analysis Of Community Elderly Based On BCI-VR

At present,the enhancement of spatial cognition through game training is a hot topic in the field of cognitive science and computer science and technology in China and abroad,However,there are few researches on spatial cognition training for the elderly in the community,and the current method of evaluating spatial cognitive training is relatively simple.Therefore,based on the theory of spatial cognitive psychology and the living environment of the elderly,a reasonable spatial cognitive training game is designed,and an in-depth study of the training effect is carried out with the help of the analysis of EEG signals.Firstly,this paper combines spatial cognitive training theory and the living environment of the elderly to design a spatial cognitive training experiment based on virtual reality.The experiment introduces experimental group game,control group game and test game as the carriers for the training and evaluation of the subjects.The experimental group game is "Virtual Community".The game is based on the theoretical principles of spatial positioning,navigation and memory;The group game is a 2D desktop game "Angry Bird";the test game is "Virtual City Walking",and uses the Brain-Computer Interface-Virtual Realtiy(BCI-VR)mode.VR provides immersive test tasks,passive brain-computer interface to detect the brain state of the subject.This article statistically analyze the spatial test scale,behavior data and functional brain network attribute characteristics.Secondly,in order to further verify the effectiveness of the spatial cognitive training of the experimental group,this paper intends to use the Common Spatial Pattern(CSP)method for feature extraction of EEG signals before and after spatial training.However,the traditional common spatial pattern requires strict assuming linearity,this paper proposes an EEG signal feature extraction method based on permutation condition mutual information common spatial pattern(PCMICSP).This method replaces the covariance matrix in the original CSP algorithm with the help of the permutation condition mutual information matrix,so that the CSP can simultaneously construct the spatial domain filter by the linear and nonlinear correlation of the EEG signals.In this paper,the EEG signals of the two test links before and after the spatial cognitive training of the experimental group are used as data sets.The PCMICSP features in different time and frequency domains are combined,and the performance of PCMICSP is verified by the CNN classification method.Finally,this experiment recruite 13 community elderly volunteers to participate in this training,of which 7 subjects are randomly assigned to the experimental group,6subjects are randomly assigned to the control group,they participate in the 28-day period training and testing.And we verify the effectiveness of spatial cognitive training and quantitative evaluation of experimental design from many aspects.In addition,this paper also verifies the superiority of the PCMICSP feature extraction algorithm in the spatial cognitive assessment of EEG signal analysis.