Study On The Detection Methods Of Noninvasive Continuous Blood Pressure Monitor Based On Pulse Wave

Blood pressure is a very important physiological parameter for human.It can reflect the abnormalities in human heart and blood vessels,and it is also an important basis for disease prediction,diagnosis and treatment effectiveness evaluation.Due to the fact that blood pressure is a potential manifestation of various cardiovascular diseases,and short-term changes in blood pressure may also predict the deterioration and outbreak of these diseases,it is very important to monitor blood pressure continuously.However,the existing noninvasive continuous blood pressure monitoring methods are not suitable for long-term continuous monitoring because of their large equipment,high requirements on testers,and long measurement time.To date,there is not yet a satisfying solution for continuous monitoring of blood pressure.In view of this situation,this paper presents a noninvasive continuous blood pressure measurement method based on pulse wave.This method includes four modules:pulse wave signal detection,pulse wave feature extraction,blood pressure prediction model establishment and blood pressure prediction model evaluation.The main research work of this thesis includes the following aspects;(1)Selection and extraction of pulse wave characteristic parameters.According to the formation mechanism of pulse wave and its propagation process in arteries,the influencing factors of pulse wave and blood pressure are analyzed,and the features related to blood pressure are selected and normalized(2)A multivariate linear model based on the features of pulse wave was established by using the data from Multiparameter Intelligent Monitoring in Intensive Care(MIMIC II).(3)Real-time dynamic estimation of blood pressure was performed using improved stepwise regression and elastic-net regression,and the results based on the two methods were compared.(4)The pulse wave and blood pressure data were collected synchronously by multi-channel physiological acquisition system to test the proposed dynamic model.The results showed that the average accuracy of the predicted values of diastolic blood pressure and systolic blood pressure in the MIMIC II database were 98%and 96%,respectively.The error of diastolic blood pressure prediction was about 0.96 mmHg and that of the systolic blood pressure prediction was about 1.61 mmHg.Predictions based on healthy adults indicated that the predicted results also met the AAMI standard(5 ± 8 mmHg).It is verified that the blood pressure detection method based on pulse wave features can be used for noninvasive continuous blood pressure detection.The novelties of this thesis are:(1)Fourteen pulse wave parameters are proposed for blood pressure prediction.In particular,a dynamic thresholding method was advanced to extract features and improved the accuracy of parameter extraction.(2)A method of dynamic prediction of blood pressure is proposed.The human body is a dynamic system,and the dynamic prediction of blood pressure is more suitable for descripting the ever-changing blood pressure.To date,the studies on dynamic blood pressure predictions are limited.(3)The elastic-net algorithm was applied to the field of blood pressure prediction,and satisfying results were obtained.