Research On Continuous Prediction Method Of Respiratory Signal Based On Human Physiological Data

Respiratory signal monitoring plays a very important role in clinical and family daily monitoring.The traditional respiratory airflow method and chest impedance method have limitations and are not suitable for long-term continuous monitoring.Some researchers use electrocardiogram or photoplethysmographic to fit respiratory signals,which has poor accuracy and single characteristic attribute.Therefore,this paper studies the prediction method of respiratory signal in resting state and active state.The main work is as follows.Firstly,the respiratory related physiological signals and physiological parameters are analyzed,and a method of respiratory signal feature extraction based on electrocardiogram and photoplethysmographic is proposed.It provides the basis for the later respiratory signal prediction algorithm.Secondly,a prediction method based on multi parameters is proposed,in which oxygen saturation,heart rate,pulse rate and other physiological parameters are used as the related characteristics of respiratory prediction.Aiming at the problem of noise interference in the original signal,wavelet threshold de-noising method is utilized to denoise the original signal.Then,the Extreme Gradient Boosting Decision Tree model is constructed,the greedy algorithm is used to find the optimal segmentation point,and the pruning strategy is used to optimize the model,so as to realize the prediction of respiratory signal waveform at rest.Thirdly,a method of respiratory signal prediction based on rhythm is proposed.From the perspective of electrocardiogram and photoplethysmographic signal waveform,the peak detection method is used to determine the reference point,extract multiple respiratory related rhythm features and carry out correlation analysis.Based on the data combination of different rhythm characteristics,the classification regression tree model is established,and the prediction of respiratory signal waveform under active state is realized by pruning optimization model.Finally,the above two respiratory signal prediction methods are implemented by coding,which are based on the BIDMC and PPG?Da Li A data sets shows the effectiveness of the proposed method,and comparative experiments and results analysis are carried out.