Sleep Monitoring Evaluation Algorithm Research And System Design Based On The EEG-ECG

Sleep disorder is a common disease in sub-healthy people,which is manifested as poor sleep quality,insomnia,etc.,and may induce various cardiovascular and cerebrovascular diseases.Sleep stage is the research premise of sleep disorder,according to which the sleep quality can be analyzed and the direction can be provided for solving the sleep problem.Traditional sleep stages are manually marked by professional sleep physicians according to Polysomnography(PSG).However,the collection process of PSG is complicated,with strict requirements on the environment,and manual labeling is inefficient,time-consuming and arduous.In order to solve these defects of existing sleep monitoring systems,this paper designed a sleep monitoring system based on eeg-ecg signals,and studied the automatic sleep stage algorithm in detail based on the single-channel frontal EEG and single-channel ECG signals.First,PSG system was used to obtain the sleep data of 15 healthy people clinically,and the sleep stages were marked by sleep physicians.Features were extracted from different modes based on the single channel frontal EEG and single channel ECG signals.In the electroencephalogram,13 characteristic values were extracted according to the frequency domain analysis and nonlinear dynamics analysis of EEG signals.In electrocardiogram mode,13 characteristic values were extracted according to the ECG signal and the timedomain and frequency-domain characteristics of heart rate variability.In the coupled mode,based on the improved coupling algorithm,feature values were extracted from the perspectives of cardiopulmonary information coupling and cardiopulmonary information coupling,and a total of 14 feature values were extracted from the perspective of cardiopulmonary information coupling.A total of 18 kinds of eigenvalues were extracted from the Angle of the coupling of cardiac and cerebral information.Then,according to the correlation between the characteristic values and each sleep stage,the four sample sets were sorted out and recombined,which were input into BP neural network,random forest,GA-SVM and PSO-SVM classifier for automatic sleep staging,and the staging results were analyzed in detail.Results show that the GA-SVM classification algorithm of sleep staging effect is better,and the modal of sample set based on ECG sleep staging accuracy of about 69% on average,based on single channel ECG signal sample set of sleep staging accuracy of about 73% on average,modal of sample set based on EEG sleep staging accuracy of about 88% on average,based on EEG,ECG signal sample set of sleep staging accuracy of about 91% on average.Finally,a portable sleep monitoring system was designed based on EEG and ECG signals during sleep.The hardware part of the system USES the embedded STM32F103 as the kernel for data acquisition,packaging and transmission,and the software part USES the Android system for feature extraction of EEG and ECG signals,data storage and real-time visual display of data,and the data can be transmitted to the server for offline analysis,so as to achieve the purpose of home use and portability of the system.