Research On Physiological Indexes Of Exercise And Risk Prediction Of Autonomic Activity

This thesis analyzed the heart rate during exercise,in the cool-down duration of jogging and at the rest status in three months’ jogging experiment.The influence of long-term jogging on the physical condition and autonomic control system of the heart has been found through statistical analysis and machine learning.The work of this thesis includes the following three aspects:Firstly,it analyzed the trend and complexity of heart rate under different exercise loads.Heart rate can reflect the activity of the cardiac autonomic control system,and it is commonly used in health monitoring,exercise intensity assessment and physiological arousal measurement.The mean value of heart rate,the range of local Hurst exponents and the relative fluctuation coefficient were weighted and fused to form a new arousal index.As a traditional arousal index,mean heart rate can reflect the trend of cardiac autonomic activity,but cannot reflect the complex competitive relationship of sympathetic and parasympathetic activity in the process of physical exercise.The resting state of sympathetic nerve activity and parasympathetic nerve activity is a dynamic balance.However,the greater the intensity of exercise,the stronger the sympathetic activity and the inhibition of parasympathetic activity.The range of local Hurst exponents and the relative fluctuation coefficient can reflect the fluctuation of heart rate and describe the changes and long-range correlations of heart rate.Therefore,they were used as new indexes of heart rate to reveal the competitive relationship of cardiac autonomic activity.Secondly,under the condition of long-term aerobic exercise,it analyzed the changes of rest-status heart rate and other physical condition indicators.The experimental results show that,during the 3-month jogging experiment,the subjects reported better sleep quality,mental status and other physical condition indicators.Statistical analysis of the blood pressure shows that long-term jogging made the blood pressure better than that before the jogging experiment.Specifically,long term jogging not only decreased the blood pressure of the subjects whose blood pressure before the experiment is at or above the upper limit of normal blood pressure,but also increased the blood pressure of the subjects whose blood pressure before the experiment is at or below the lower limit of normal blood pressure.Body weight,body mass index and body fat percentage were also significantly decreased,revealing the fat reduction effect of long term jogging.Therefore,long-term aerobic exercise can promote people’s health.In the analysis of 3-month rest-status heart rate,it is found that,compared with that before the experiment,the rest-status heart rate slowed down in the early stage of aerobic exercise(the first month),but the rest-status heart rate showed an upward trend in the second and third months.1/26 subject had developed frequently premature heartbeats after long-term aerobic exercise.Thirdly,it analyzed the heart rate time series in the cool-down duration of aerobic exercise.6/28 subjects showed very slow autonomic recovery,and 5/28 subjects had frequently premature heartbeats,suggesting that long-term aerobic exercise is accompanied with the risk of abnormal autonomic activity.In the first minute after aerobic exercise,i.e.the rapid recovery period,the parasympathetic activity became strong,fighting for the control of the heartbeats and slowing down the heart rate.In the 1-6 minutes after aerobic exercise,i.e.the slow recovery period,the autonomic activity was still trying to regain the dynamic balance between sympathetic system and parasympathetic system.After 6 minutes of recovery,the dynamic balance of the two autonomic systems was reconstructed.This paper acquired and created two ECG data sets.One dataset is for the cool-down ECG data with poor autonomic recovery,and the other dataset is for the cool-down ECG data with normal autonomic recovery.219 original features were extracted from the RR interval time series of the ECG data sets,and backward selection algorithm was applied for feature selection.Three classifiers,namely,K-Nearest Neighbor(KNN),Na?ve Bayes(NB)and Support Vector Machine(SVM),were compared for their abilities to distinguish heartbeat data of poor autonomic recovery from those of normal autonomic recovery.NB classifier obtained a true positive rate of 68.49% and a true negative rate of 82.33%,KNN classifier obtained a true positive rate of 72.66% and a true negative rate of 82.65%,and SVM classifier obtained a true positive rate of 92.09% and a true negative rate of 82.65%.By this means,a risk early warning system of abnormal autonomic activity is established,and it provides a technical support for the monitoring of poor autonomic recovery after aerobic exercises.