Cardiovascular Health Assess-Ment Based On Bcg Collected From Intelligent Mattress

With the advent of the "Internet +" era,the traditional mode of health and medical care has undergone a change from a passive medical treatment mode to an active health care mode.Relying on cloud computing technology and mobile communications technology,intelligent health products continue to innovate and widely used,and people’s lives have entered a new way to fully quantify healthy living.Sleep Health Monitoring mattress as a low power consumption,interference-free,unobtrusive real-time health monitoring instruments,came into being and quickly developed into a popular nighttime means health care in this trend.In this paper,an embedded mattress based on a piezoelectric ceramic sensor array has been developed.The research of this paper focuses on the mattress.The mattress collects mixed signals that reflect the human body’s action on the mattress,including the vital signs and activity signals of the user,such as heartbeat,respiration and other small signals and body movements,turning over and other large action signals.Among them,the signal that can reflect the heartbeat are ballistocardiogram(BCG)signal that can be extracted from the mixed signals collected by the mattress.The object of this article is the BCG signal.In view of the characteristics of the BCG signal extracted by the mattress equipment in this paper,a J-peak detection algorithm based on blind segmentation and an extended multi-sample learning algorithm are proposed to calculate the cardiac cycle of BCG signal.Heartbeat waveforms in BCG signals vary from person to person,which can have a significant impact on the accuracy of the cardiac cycle calculation.Therefore,this paper presents an extended multi-instance learning approach.The MIT-BIH database provides data on arrhythmia of ECG signals.To use the MIT-BIH database in the study of BCG arrhythmias,it is first verified that some of the parameters of the BCG and ECG heart rate variability(HRV)analyzes are relevant and the differences are not statistically significant.Next,using the HRV index as a feature,a Support Vector Machine(SVM)and a Random Forest(RF)classifier are used to classify the arrhythmia signals.The results show that the RF classifier has a higher accuracy of discriminating between abnormal and non-abnormal heart rate signals.In addition,another application of BCG signal is studied in this paper.According to different characteristics of BCG signals under different sleeping positions,BCG signals of three different sleeping positions are classified by RF classifier,and the current sleeping position is determined through the judgment.This method provides a new idea for sleeping position detection.