Recognition Of Alcohol Use Disorders Based On EEG

Alcohol use disorder(AUD)is a kind of chronic and recurrent mental disorder caused by many factors such as alcoholism,heredity,environment and so on.The accuracy of traditional screening methods will be affected by experience and subjective factors.Through the comparative study of electroencephalogram(EEG)between AUD patients and normal people,we can find that there are significant differences between the EEG of alcohol use disorder patients and normal people.Therefore,this paper analyzes and classifies the electroencephalogram signals of the patients with AUD and the normal control group by the method of combining signal processing and machine learning,so as to improve the accuracy of AUD recognition.In this paper,the artifacts of the original electroencephalogram were firstly processed,and an improved Hilbert-Huang transforms(HHT)algorithm was proposed based on complete ensemble empirical mode decomposition with adaptive noise(CEEMDAN),and the improved Hilbert-Huang transforms was combined with independent component analysis to remove the artifacts in electroencephalogram signals.Compared with other methods,the method proposed in this paper can identify artifact components more effectively,and the root mean square error(RMSE)of electroencephalogram signals after artifact removal is reasonable,and the SNR is higher.This method provides more effective information for signal feature extraction and analysis,and can effectively improve the accuracy of recognition results.Then the feature extraction and analysis of EEG signal after artifact removal were carried out.Butterworth bandpass filter is utilized to extract EEG signals of different frequency rhythm wave,the study found that the alcohol use disorder experimental group and normal control group the gamma power spectral density has significant difference,the characteristics of the two electroencephalogram signals can be used as identification,with a total space pattern filtering characteristics of multichannel filtering,the multi-channel feature fusion to generate new feature matrix,improve the identification accuracy.Finally,the support vector machine(SVM)model was selected to complete the task of EEG recognition.In order to obtain the optimal classification model,the adaptive differential evolution(ADE)algorithm is used to optimize the support vector machine algorithm model.Adaptive differential evolution solves the problem that the standard differential evolution is easy to fall into the local optimal solution and the convergence speed is too fast.The simulation results show that the recognition accuracy of ADE-SVM classifier model can reach 97.50% in a certain experimental environment.Compared with other models,the average recognition accuracy is higher and the system model is more stable.