Research On Sleep Staging Algorithm Based On Support Vector Machine Classifier

In recent years,as the pace of life accelerated and the pressure increased,the problem of sleep disorders has become more and more serious,affecting the quality of people's life and reducing work efficiency.Therefore,sleep problems need to be solved urgently.Parently,polysomnography is widely used in the diagnosis of sleep disorders,but it has the disadvantages of complicated operation,large instrument,low portability,etc.,which cannot meet the daily household needs and has low universality.Therefore,the monitoring and evaluation of sleep quality is of great signiticance.Based on single-conductivity electroencephalogram,electrooculogram and electromyography,the thesis achieves a high-storage PC software system with high ac-curacy of sleep staging,including human-computer interaction software and sleep stag-ing algorithm.The research content of this thesis mainly includes:1.Based on the difference of precision of physiological signal acquisition instrument and the characteristics of sleep physiological signal,the normalization parameters such as coefficient of variation,energy ratio of brain electrical spectrum and nonlinear entropy were proposed.The proposed feature parameters are vectors that are reduced by t-test and are selected.Optimal parameter combination as input source for sleep staging classifier.2.By comparing three common multivariate classifiers,the support vector machine classifier was selected.After the sleep stage of the support vector machine classifier,the preliminary results were corrected by combining the sleep physiological characteristics.3.The algorithm was tested using 80 groups of data from the National Sleep Research Resource,and the results of sleep staging were evaluated using three indicators:sensitivity,specificity,and accuracy.4.The upper computer software system with functions of information input,signal playback and sleep data analysis were implemented.In this thesis,the hybrid signal sleep staging algorithm based on support vector machine and its host computer software are implemented,and the performance of the algorithm is preliminarily verified.It has the advantages of high accuracy and wide applicability.It has practical application value of both medical and clinical sleep staging diagnosis.