Continuous Detection Of Mental Stress Based On Heart Rate Variability

Mental stress is a very common problem in daily life.It is a risk factor for hypertension,atherosclerosis,and even sudden death.At the same time,long-term mental stress can lead to depression.There are also many researches on mental stress,among which the research on mental stress based on physiological signals accounts for a large proportion.In these studies,some use multi-mode physiological signals to achieve a good mental stress recognition rate.But collecting more physiological signals means more experimental overhead.Some use single-mode physiological signals,in order to improve the stress detection effect,extract multiple features,and some even as many as tens to hundreds.Whether using single-mode or multi-mode research,there is less involvement in continuous detection of mental stress.Because this needs to consider the choice of time scale of stress detection,the longer time scale,cannot reflect the subtle changes of mental stress,the shorter time scale,and the loss of feature validity.On this basis,this paper uses the single-mode physiological signal to calculate the RR interval,intercepts 104 RR interval sequences,extracts fewer and effective heart rate variability features,and achieves continuous detection of mental stress based on heart rate variability.The accurate heart rate or RR interval sequence(the reciprocal of each other)is the basis for mental stress recognition.Sources of heart rate signals include electrocardiograph(ECG)signals and pulse wave signals.In the process of accurate heart rate calculation,the calculation methods of ECG heart rate and pulse wave heart rate are discussed.In this paper,through the self-adaptive sliding window algorithm of ECG signals,the false detection and missed detection of R wave peaks are effectively avoided,and the accurate calculation of heart rate is achieved.In theory,the pulse wave rate can be calculated by the same method.However,since the pulse wave waveform is simple and highly susceptible to motion interference,it is difficult to detect the peak of the pulse wave.Therefore,the heart rate of the pulse wave combined with the three-dimensional acceleration is used.The exploration was carried out to realize the calculation of the exercise heart rate using the pulse wave.Through the comparison of two heart rate algorithms,it is found that the pulse wave exercise heart rate cannot calculate the effective heart rate variability features,and cannot reflect the autonomic nerve activity.Therefore,the ECG signal is selected to calculate the heart rate,and then the heart rate variability is calculated to achieve continuous detection of mental stress.Mental stress can be induced by laboratory protocol such as mental arithmetic tasks and social stress tests,and can also induce mental stress in real situations,such as driving,exams and other scenarios.This paper mainly deals with the mental stress in the real scene,and obtains the data of the mental stress induced by the real driving scene through the open standard database.In this driving data,the ECG signal is selected to calculate the RR interval sequence,the effective heart rate variability features are extracted,the mental stress detection model is designed,and the high stress three classification recognition rate is obtained.In order to further verify the generalization performance of the above stress detection model,this paper designs and implements the experiment of mental stress induced by the final exam.In the experiment,the questionnaire was used to evaluate the induction of mental stress,and the subjects were asked to draw the mental stress selfevaluation curve during the whole examination.The ECG data of the subjects were collected,the heart rate sequence was calculated,and the effective heart rate variability features were extracted.Using the aforementioned stress detection model,the mental stress changes throughout the examination process are continuously detected,and the stress recognition curve of the mental stress model is obtained.Finally,the correlation analysis between the mental stress self-evaluation curve and the stress recognition curve of the mental stress detection model is carried out to verify the generalization performance of the model and achieve continuous detection of mental stress.In addition,a local Hurst index analysis was performed on the exam ECG signal data to further explore the changes in heartbeat dynamic complexity under mental stress.In this paper,it is found that continuous detection of mental stress based on heart rate variability is feasible.104 heart beats are selected as time scales,and five features of MRR,MAFN,pNN50,RLHE and LF are extracted.Using Tree classifier,98% of mental stress classification recognition rate is obtained.In the correlation analysis between the mental stress self-evaluation curve of the 20 final exam subjects and the stress recognition curve of the mental stress detection model,a higher average correlation coefficient of 0.81 was achieved.The generalization performance of the mental stress detection model is verified,and the ability of the model to capture the details of mental stress changes is verified,and the continuous detection of mental stress is realized.