Analysis Of Autonomic Nervous Pattern Of Sleep Deprivation

Sleep is an indispensable physiological state of the human body,adequate sleep is conducive to the recovery of mental and physical strength.However,with the rapid development of society and the improvement of economic level,people’s life stress also increases.The rapid pace of life,the use of electronic products and night shift work lead to the frequent occurrence of SD.SD has become an important reason to affect people’s health,thus causing widespread concern from all walks of life.SD causes imbalances in the heart rate and ANS,which can lead to an increased incidence of cardiovascular diseases.The human ANS adjusts the normal function of various organs through the dynamic balance between the SNS and PNS.Under normal circumstances,the sympathetic nerve activity is enhanced during the day,and the parasympathetic nerve is dominant at night.During SD,the autonomic nerve is out of balance,the sympathetic nerve is excited at night,and the parasympathetic nerve is inhibited.Analyzing the autonomic nervous pattern of SD is helpful to prevent the occurrence of cardiovascular diseases caused by SD.However,machine learning methods and quantitative indicators have not been used to study the patterns of autonomic nervous activity in SD.Therefore,this paper takes this issue as a research problem.HRV was analyzed based on a machine learning approach to determine the difference in autonomic function between SD data and BL physiological data.Original ECG data were collected from 60 undergraduate and graduate students in Southwest University,including 30 males and 30 females.Data from 50 subjects(25 males and 25females)were used to train the classifier,and physiological data from 10 subjects(5males and 5 females)who did not participate in the training and feature selection were used to validate the generalization performance of the trained classifier and the selected feature subset.In this study,25 HRV features were extracted to distinguish between SD data and BL data,and to quantify the differences between the autonomic nerve activity patterns in SD and BL data.Statistical test and sequential backward selection algorithm were applied to obtain the optimal HRV feature subset to distinguish between SD and BL data.In addition,SVM,KNN and LDA were used as the classifiers to distinguish between SD ECG data and BL ECG data,and the performance of the classifiers was evaluated by the leave-one-subject-out cross-test method during the feature selection process.The main research contents and results of this paper are as follows.(1)In this paper,we designed an experimental paradigm of SD physiological data collection and BL physiological data collection.Through 4 hours of real data collection,we obtained physiological data of 60 subjects.T-test was performed for BL data and 1:30 A.M.SD data.The results showed that 18 HRV features were significantly different between SD data and BL data.SVM classifier was used to distinguish SD data and BL data.The recognition rate of 81.82% was obtained on the validation dataset by using a three-dimension feature subset.(2)The BL data of 22 P.M.and SD data of every half hour from 23 P.M.to 2 A.M.in the morning were compared by t-test.It is found that there were significant differences between BL data and subsequent SD data of every half hour,and the numbers of significantly different features were 4,8,16,17,13 and 18,respectively.This shows that with the extension of SD time,more different HRV features appears,revealing more obvious effect of SD on ANS.(3)The low frequency power feature of three-dimension feature subset is a feature selected by multiple classifiers.Therefore,the feature is chosen as a quantitative autonomic nervous activity index of SD.The time evolution analysis is carried out in the data set every half hour.It is found that the index has an increasing trend with the extension of SD time.Considering that the number of features having between-group differences increased with the prolongation of SD,and HRV low-frequency power increased with the prolongation of SD,the following conclusions were drawn.With the prolongation of SD,the specific changes of autonomic nervous activity became more obvious.Sympathetic nervous activation was dominant,and the activity was increasing with prolonged SD time.