Research On Non-invasive Continuous Blood Pressure Prediction Algorithm Based On PPG Morphology

Blood pressure is one of the important physiological indicators in health monitoring,and it has great medical significance for its continuous monitoring,which can help prevent hypertension and assist in the treatment of hypertension.Although the accuracy of the current continuous blood pressure detection technology can reach an acceptable range,most of the technology is based on pulse wave conduction time.This method requires multiple sensors and is not suitable for long-term blood pressure monitoring.This article aims to design a method for predicting blood pressure values(diastolic blood pressure,systolic blood pressure)using only the photoelectric volume pulse wave signal.The main research contents are as follows:(1)Establish a blood pressure database containing photoelectric volume pulse wave signals of different forms through an international public database;(2)Design a robust method that can accurately and effectively extract photoelectric volume pulse wave Morphological characteristic value and blood pressure gold standard value extraction algorithm;(3)Use photoelectric volume pulse wave characteristic value parameter and blood pressure gold standard for modeling,and study a variety of mathematical models based on photoelectric volume pulse wave morphology;(4)Predict blood pressure through different models,including linear model,non-linear model,and mixed model of linear and non-linear,using the test set to evaluate the effectiveness of the algorithm.By comparing various models,the accuracy of the mixed linear and nonlinear models is the highest,followed by the accuracy of the nonlinear models.The accuracy of the diastolic blood pressure in the hybrid model is 0.1613 ± 3.8704 mm Hg,and the accuracy of the systolic blood pressure is 0.2940 ± 6.8573 mm Hg.The accuracy of the systolic blood pressure model and the diastolic blood pressure model have reached the blood pressure measurement standard set by the American AAMI organization.The experimental results show that this paper implements a high-precision morphological characteristic blood pressure prediction algorithm based on photoelectric volume pulse wave,which can achieve continuous non-invasive measurement of blood pressure,which lays a certain algorithm foundation for the research of wearable smart blood pressure watches.