CN-122004926-A - Noninvasive blood pressure measurement method and system based on multi-feature fusion

CN122004926ACN 122004926 ACN122004926 ACN 122004926ACN-122004926-A

Abstract

The invention discloses a noninvasive blood pressure measurement method and system based on multi-feature fusion, wherein the method comprises the steps of transmitting ultrasonic signals to a user through an ultrasonic probe and acquiring echo signals; the method comprises the steps of acquiring echo signals, extracting a plurality of blood vessel parameter characteristics based on a characteristic extraction algorithm according to the echo signals, correcting and determining characteristic baseline data corresponding to a user according to the plurality of blood vessel parameter characteristics, and determining blood pressure data corresponding to the user according to fusion analysis of the plurality of blood vessel parameter characteristics and the characteristic baseline data based on a prediction model. Therefore, the invention can realize multi-mode noninvasive blood pressure continuous monitoring based on ultrasonic echo, improve the accuracy and instantaneity of blood pressure measurement, and reduce the risk of blood pressure evaluation deviation caused by the intermittence of traditional cuff measurement or discomfort of users.

Inventors

CAI WEIZHONG
LIANG CHENGMIN
XU PEIDA
LIN XIFENG

Assignees

广州索诺星信息科技有限公司

Dates

Publication Date: 20260512
Application Date: 20260129

Claims (10)

1. A method for noninvasive blood pressure measurement based on multi-feature fusion, the method comprising: Transmitting ultrasonic signals to a user through an ultrasonic probe and acquiring echo signals; Extracting a plurality of vascular parameter features based on a feature extraction algorithm according to the echo signals; correcting and determining characteristic baseline data corresponding to the user according to the plurality of blood vessel parameter characteristics; based on a prediction model, determining blood pressure data corresponding to the user according to fusion analysis of the plurality of blood vessel parameter characteristics and the characteristic baseline data.
2. The method for non-invasive blood pressure measurement based on multi-feature fusion according to claim 1, wherein the extracting a plurality of vascular parameter features based on a feature extraction algorithm according to the echo signal comprises: synchronously acquiring electrocardiosignals of the user through an electrocardio sensing device; Determining a plurality of response time points corresponding to the user based on the corresponding relation between the electrocardiosignal and the echo signal; And determining a plurality of vascular parameter characteristics according to the echo signals corresponding to the response time points.
3. The method for noninvasive blood pressure measurement based on multi-feature fusion of claim 2, wherein the determining a plurality of response time points corresponding to the user based on the correspondence between the electrocardiograph signal and the echo signal comprises: Carrying out frequency domain analysis on the electrocardiosignals to obtain electrocardiosignal frequencies corresponding to a plurality of preset time points; performing frequency domain analysis on the echo signals to obtain echo signal frequencies corresponding to a plurality of preset time points; for each preset time point, calculating a signal difference vector between the electrocardiosignal and the echo signal corresponding to the preset time point; calculating a frequency difference vector between the electrocardiosignal frequency and the echo signal frequency corresponding to the preset time point; inputting the signal difference vector and the frequency difference vector into a trained matching prediction model to obtain the signal matching degree corresponding to the output preset time point, wherein the matching prediction model is obtained by training a training data set comprising a plurality of training difference vector pairs and corresponding signal response matching labels; And screening out the preset time points with the signal matching degree larger than a preset matching degree threshold value to obtain a plurality of response time points.
4. The method for noninvasive blood pressure measurement based on multi-feature fusion of claim 2, wherein determining a plurality of vascular parameter features from the echo signals corresponding to the response time points comprises: For each blood vessel parameter type, determining a response time rule and a calculation model corresponding to the blood vessel parameter type based on a historical calculation record of the blood vessel parameter type, wherein the response time rule is used for limiting at least one of a time point type, an adjacent time point distance and a time point position of a response time point; Screening a plurality of matching time points from all the response time points according to the response time rule; and inputting the echo signals corresponding to all the matching time points into the calculation model to obtain the vascular parameter characteristics corresponding to the vascular parameter types.
5. The method of claim 1, wherein the vascular parameter is characterized by a vascular morphology parameter, a hemodynamic parameter, a vascular diameter rate of change parameter, a pulse wave velocity parameter, or a vascular elastic modulus parameter.
6. The method for non-invasive blood pressure measurement based on multi-feature fusion according to claim 1, wherein the correcting the determining of the feature baseline data corresponding to the user according to the plurality of blood vessel parameter features comprises: For each blood vessel parameter type, acquiring reference data corresponding to the blood vessel parameter type corresponding to the user from a reference database; calculating the data similarity between the reference data corresponding to the blood vessel parameter type and any other reference data corresponding to the blood vessel parameter type; determining other blood vessel parameter types with the data similarity larger than a preset similarity threshold as associated parameter types; based on a mathematical fitting algorithm, fitting and determining a polynomial expression between each associated parameter type and the blood vessel parameter type according to the blood vessel parameter characteristics corresponding to each associated parameter type and the blood vessel parameter characteristics corresponding to the blood vessel parameter type; and correcting the reference data according to the polynomial expression to obtain characteristic baseline data corresponding to the blood vessel parameter type.
7. The method for non-invasive blood pressure measurement based on multi-feature fusion according to claim 6, wherein the correcting the reference data according to the polynomial expression to obtain the feature baseline data corresponding to the blood vessel parameter type comprises: substituting all the reference data corresponding to the associated parameter types into the polynomial expression to calculate associated reference data corresponding to the blood vessel parameter types; calculating a data difference value between the reference data corresponding to the blood vessel parameter type and the associated reference data; Calculating a correction value proportional to the data difference; And calculating the difference between the reference data and the correction value to obtain the characteristic baseline data corresponding to the blood vessel parameter type.
8. The method for non-invasive blood pressure measurement based on multi-feature fusion according to claim 1, wherein the determining the blood pressure data corresponding to the user based on the prediction model according to the fusion analysis of the plurality of blood vessel parameter features and the feature baseline data comprises: Calculating a data difference between each of the vessel parameter features and the corresponding feature baseline data; combining all the vascular parameter characteristics and the corresponding data differences into a data difference vector; the data difference vector is input into a trained blood pressure prediction model to obtain output blood pressure data corresponding to the user, and the blood pressure prediction model is obtained through training of a training data set comprising a plurality of training blood vessel parameter feature sets and corresponding baseline data difference labels and blood pressure labels.
9. A multi-feature fusion-based noninvasive blood pressure measurement system, the system comprising: the acquisition module is used for transmitting ultrasonic signals to a user through the ultrasonic probe and acquiring echo signals; The extraction module is used for extracting a plurality of vascular parameter characteristics based on a characteristic extraction algorithm according to the echo signals; The correction module is used for correcting and determining characteristic baseline data corresponding to the user according to the plurality of blood vessel parameter characteristics; and the determining module is used for determining the blood pressure data corresponding to the user according to fusion analysis of the plurality of blood vessel parameter characteristics and the characteristic baseline data based on a prediction model.
10. A multi-feature fusion-based noninvasive blood pressure measurement system, the system comprising: a memory storing executable program code; a processor coupled to the memory; the processor invokes the executable program code stored in the memory to perform the non-invasive blood pressure measurement method based on multi-feature fusion as claimed in any of claims 1-8.

Description

Noninvasive blood pressure measurement method and system based on multi-feature fusion Technical Field The invention relates to the technical field of data processing, in particular to a noninvasive blood pressure measurement method and system based on multi-feature fusion. Background With the rapid growth of continuous cardiovascular health monitoring demands, medical institutions and users increasingly pay attention to achieving high-precision and real-time blood pressure assessment in a noninvasive manner, wherein how to improve the accuracy and continuity of blood pressure measurement so as to reduce the risk of assessment deviation becomes a key technical problem. In the prior art, cuff intermittent measurement or a continuous estimation method based on photoelectric volume pulse waves is generally adopted to acquire blood pressure data, and a fixed algorithm or a simple threshold correction processing signal is relied on to support daily blood pressure monitoring. The existing solution is difficult to realize high-precision continuous prediction of blood pressure due to lack of feature extraction of echo signals of an ultrasonic probe, comprehensive analysis of a plurality of blood vessel parameters and real-time correction fusion of feature baseline data, and the problems of user discomfort caused by cuff compression, dynamic information loss caused by measurement intervals and individual baseline drift cannot be overcome by a common intermittent cuff measurement or single-mode estimation strategy, so that the blood pressure estimation precision and instantaneity are insufficient, estimation deviation is easily caused by measurement discontinuity or user discomfort, and the reliability and clinical value of noninvasive blood pressure monitoring in early warning and long-term management of cardiovascular diseases are limited. It can be seen that the prior art has defects and needs to be solved. Disclosure of Invention The technical problem to be solved by the invention is to provide a noninvasive blood pressure measurement method and a noninvasive blood pressure measurement system based on multi-feature fusion, which can realize multi-mode noninvasive blood pressure continuous monitoring based on ultrasonic echo, improve the accuracy and instantaneity of blood pressure measurement and reduce the risk of blood pressure evaluation deviation caused by the discontinuity of traditional cuff type measurement or user discomfort. In order to solve the technical problems, the first aspect of the invention discloses a noninvasive blood pressure measurement method based on multi-feature fusion, which comprises the following steps: Transmitting ultrasonic signals to a user through an ultrasonic probe and acquiring echo signals; Extracting a plurality of vascular parameter features based on a feature extraction algorithm according to the echo signals; correcting and determining characteristic baseline data corresponding to the user according to the plurality of blood vessel parameter characteristics; based on a prediction model, determining blood pressure data corresponding to the user according to fusion analysis of the plurality of blood vessel parameter characteristics and the characteristic baseline data. As an optional implementation manner, in a first aspect of the present invention, the extracting a plurality of vascular parameter features according to the echo signal based on a feature extraction algorithm includes: synchronously acquiring electrocardiosignals of the user through an electrocardio sensing device; Determining a plurality of response time points corresponding to the user based on the corresponding relation between the electrocardiosignal and the echo signal; And determining a plurality of vascular parameter characteristics according to the echo signals corresponding to the response time points. As an optional implementation manner, in the first aspect of the present invention, the determining, based on a correspondence between the electrocardiograph signal and the echo signal, a plurality of response time points corresponding to the user includes: Carrying out frequency domain analysis on the electrocardiosignals to obtain electrocardiosignal frequencies corresponding to a plurality of preset time points; performing frequency domain analysis on the echo signals to obtain echo signal frequencies corresponding to a plurality of preset time points; for each preset time point, calculating a signal difference vector between the electrocardiosignal and the echo signal corresponding to the preset time point; calculating a frequency difference vector between the electrocardiosignal frequency and the echo signal frequency corresponding to the preset time point; inputting the signal difference vector and the frequency difference vector into a trained matching prediction model to obtain the signal matching degree corresponding to the output preset time point, wherein the matching prediction model is obt