A Research Of Real-time Sleep Stage Algorithm Based On Heart-Rate And Breath-Rate

Sleep is a necessary process in human life,and good sleep can promote the recovery of mental state,attention,emotional control and judgment.With the quickening of the pace of life,the pressure of life and work is increasing gradually.Insomnia has become another major problem affecting the health of the human body.People are eager to improve their sleep quality.Therefore,how to monitor sleep more conveniently and automatically divide the physiological data into sleep has become the main challenge of studying sleep at this stage.Traditional sleep stages are used to calibrate sleep state by judging the change rule of EEG.However,the acquisition of EEG needs to paste various electrodes,which will affect the normal sleep state of the human body.During sleep,the heart rate and respiratory rate of the human body will exhibit similar rhythmic changes in EEG with the change of sleep phase.Moreover,the acquisition of heart rate and respiratory rate signals is simpler than EEG signals,and has less invasive effect on human sleep.So we decided to use heart rate and respiration rate to study sleep stages.At present,some scholars use the characteristics extracted from heart rate and respiratory signal as input,using the hidden Markov model as a classifier to calculate the sleep stages,but the accuracy rate of the staging results is only about 60%.In order to improve the results of sleep stage judgment,this paper applies the hybrid algorithm of HMM and BP neural network model to sleep stage calculation for the first time,and uses the powerful pattern recognition ability of BP neural network model to train and memorize the error matching results of the HMM model.The accuracy of the high sleep stage calculation.In this paper,the standard heart rate respiration rate in the MIT-BIH database is used to model and test.The results show that the recognition rate of this algorithm can reach 79.07%,which is 12% higher than that of the HMM algorithm.Compared with the existing SVM and random forest sleep staging algorithm based on heart rate breathing signal,the results show that the recognition rate of this algorithm in the NREM period is better than the other two algorithms.The sleep stage algorithm is applied to the self-designed real-time sleep monitoring system to achieve the monitoring of sleep state in the home environment.The system uses a mattress embedded in a piezoelectric sensor to collect the BCG signal of human sleep.It separates the heart rate signal and the respiratory rate signal and transfers it to the data acquisition server,which is analyzed and processed by the server and completed the recognition of the sleep state.End users can see their real time heart rate,sleep state,and sleep quality all night through a mobile phone,helping people to understand and improve their sleep more quickly and conveniently.The result shows that the accuracy rate of the sleep stage is 76.33%,which proves that the sleep monitoring system is feasible in the practical family application and lays the foundation for the sleep monitoring under the family environment in the future.The foundation.