| The autonomic nervous system in humans and animals is regulated by a dynamic balance between its sympathetic nerves and its parasympathetic nerves,and most of our organs including heart are regulated by them.Autonomic imbalance can be achieved by sympathetic overactivity,decreased parasympathetic tone,or both.Many epidemics share a common but rarely treated mechanism,sustained autonomic imbalance.Physiological studies have found that the autonomic nervous system plays an important role in controlling blood pressure.In a large number of people,the occurrence of the autonomic imbalance precedes that of hypertension,and promotes its development.The analysis of autonomic nervous pattern in hypertension is significant to block the destructive effect caused by autonomic imbalance on cardiovascular system in advance.There is no literature that proposes rapid and effective quantitative indicators on autonomic nervous pattern in hypertension.Therefore,it is the research issue explored in this thesis.This thesis,based on machine learning algorithms,has analyzed short-term heart rate variability to determine differences in autonomic nervous function between hypertensive patients and normal population.The electrocardiogram(ECG)of hypertensive patients are 137 ECG recordings provided by Smart Health for Assessing the Risk of Events via ECG(shareedb).The data of healthy subjects include the ECG of 18 subjects from the MIT-BIH Normal Sinus Rhythm Database(nsrdb)and RR intervals of 54 subjects from the Normal Sinus Rhythm RR Interval Database(nsr2db).Based on the published literature on autonomic neurophysiology,17 HRV features were extracted to distinguish the rhythm of the hypertension from that of the health,and to quantify the difference between the pattern of autonomic nervous system in the hypertension and that in the health.Kolmogorov-Smirnov statistical test and sequential backward selection(SBS)were applied to get the best HRV features combination to distinguish hypertensive ECG from normal ECG.In addition,support vector machine(SVM),k-nearest neighbor(KNN)and random forest(RF)were applied as classifiers in the study.The performance of each classifier was evaluated independently using the leave-one-subject-out validation method.The main research results and conclusions are as follows:(1)Among the 17 initial HRV features,15 features show the significant difference between hypertensive group and normal group.This result shows that the autonomic nervous regulation in hypertensive patients is significantly different from that in healthy people.According to the VLF,LF and LF/HF of HRV,the HRV index of hypertension group is significantly higher than that of healthy group.The result shows sympathetic overactivity in hypertensive patients,and it is a sign of impaired cardiovascular autonomic function.LF/HF reflects the competition between sympathetic nerves and parasympathetic nerves.In hypertensive group,sympathetic nerve plays a leading role in the autonomic nervous regulation.(2)RF with five HRV features is the best predictive model for hypertensive autonomic nervous pattern.After SBS process and model optimization,the best five HRV features are: sample entropy(SampEn),VLF,root mean square of successful differences(RMSSD),LF/HF and vector angle index(VAI).In the binary classification of hypertension group/normal group,the area under curve(AUC)obtained by RF with these five quantitative indicators is 0.9075,and the accuracy(ACC)is 86.44%.The sensitivity of the classifier is 93.43%,which is meaningful in the daily monitoring of HRV abnormality caused by hypertension.(3)In the SBS process,all classifiers selected SampEn as the best feature.This result shows that SampEn is more suitable to predict the hypertensive risk than VLF.In the hypertensive group,the SampEn is significantly reduced,which indicates that the reduction in the complexity of autonomic nervous activity is an important trace in the formation and development of hypertension. |
PDF Full Text Request