Research Of Continuous Blood Pressure Measurement Based On PPG

Blood pressure is vital indicator,it reflects the physical condition and blood-supply ability of heart and vessels.It possesses instructive meaning on disease diagnose,treatment effect evaluation,medicine evaluation.Nowadays,blood pressure is depended on single measurement,which is the instant value of that time.In medical view,continuous blood value is more beneficial in deciding the amount and frequency of medicine taking.And that will help patients to release hypertension scientifically.Continuous blood pressure measurement devices used in hospital are in batch,every 15 or 30 minutes blood pressure value is recorded,which is not real continuing measurement.Therefore,a set of strategies was proposed to calculate every heart beats blood pressure by the paper.Content was shown as follow:To solve the difficulty in wearing of two-point or multi-source signal based blood pressure device.The paper used single pulse wave as input data to study pulsed wave based continuing blood pressure measurement method.To tackle the defect of pulse wave fiducial deletion in patient and elderly people,the entire single pulse wave data was used as features to calculate blood pressure.Firstly,a filter was processed on pulse wave data.Secondly,dinar step-wise regression algorithm was used to build blood pressure measurement model.Thirdly,blood pressure of every heart beat was calculated.Finally,continuing blood pressure value,include systolic and diastolic were obtained.Experiments were done on self-collected data and MIMIC public data a MIT database,a promising results ±5% error were gotten.In the study,it was found that step-wise regression algorithm took a long time in converging,it depress the efficiency of calculation.At same time,in medical view,the changing of blood pressure value was no-linear,Linear method would produce a low accuracy in blood pressure measurement.On the basement of former study,BP neural network made was established to calculate blood pressure on oneself experience data,form that,continuing blood pressure value was gotten.Experiments results showed that,25 self-collected healthy volunteers obtained a little more accurate blood pressure value compared to linear model,and a large margin improving was gotten on 33 patients of MIT MIMIC public data.Another promising result was that the BP neural Network model was efficacy on healthy young,patient,and elder people in blood pressure calculation.