阅读说明：本技术 非接触式血压变异性实时测量系统、计算机设备和存储介质 (Non-contact blood pressure variability real-time measurement system, computer device and storage medium ) 是由 洪弘 奚梦婷 李彧晟 顾陈 孙理 朱晓华 于 2019-08-30 设计创作，主要内容包括：本发明公开了一种非接触式血压变异性实时测量系统、计算机设备和存储介质,包括：脉搏波采集模块,利用生命体征监测雷达实时采集主动脉脉搏波波形信号；脉搏波实时传导时间获取模块,用于对主动脉脉搏波波形信号进行特征点提取,根据特征点求取脉搏波实时传导时间；血压测量模块,用于建立基于血压与脉搏波传导时间的高斯过程回归算法模型,并由该模型获取脉搏波实时传导时间对应的实时血压值；血压变异性测量模块,用于根据血压测量模块测得的当前时刻的血压和上一时刻的血压,求取受试者的血压变异性。相比传统的血压变异性测量系统,本发明的系统只需单个射频传感器即可实现非接触、无创、准确地测量血压变异性值,有效可行,性能可靠,且精度更高。(The invention discloses a non-contact blood pressure variability real-time measuring system, a computer device and a storage medium, comprising: the pulse wave acquisition module is used for acquiring an aortic pulse wave waveform signal in real time by using a vital sign monitoring radar; the pulse wave real-time conduction time acquisition module is used for extracting characteristic points of the aortic pulse wave waveform signals and solving the pulse wave real-time conduction time according to the characteristic points; the blood pressure measuring module is used for establishing a Gaussian process regression algorithm model based on blood pressure and pulse wave conduction time, and acquiring a real-time blood pressure value corresponding to the pulse wave real-time conduction time by the model; and the blood pressure variability measuring module is used for solving the blood pressure variability of the subject according to the blood pressure at the current moment and the blood pressure at the previous moment measured by the blood pressure measuring module. Compared with the traditional blood pressure variability measuring system, the system can realize non-contact, non-invasive and accurate measurement of the blood pressure variability value only by a single radio frequency sensor, and has the advantages of effectiveness, feasibility, reliable performance and higher precision.)

1. A non-contact real-time blood pressure variability measurement system, comprising:

the pulse wave acquisition module is used for acquiring an aortic pulse wave waveform signal of the subject in real time by using a vital sign monitoring radar;

the pulse wave real-time conduction time acquisition module is used for extracting characteristic points of the aortic pulse wave waveform signals and solving the pulse wave real-time conduction time according to the characteristic points;

the blood pressure measuring module is used for establishing a Gaussian process regression algorithm model based on blood pressure and pulse wave conduction time, and acquiring a real-time blood pressure value corresponding to the pulse wave real-time conduction time according to the model, wherein the blood pressure comprises systolic pressure and diastolic pressure;

and the blood pressure variability measuring module is used for solving the blood pressure variability of the subject according to the blood pressure at the current moment and the blood pressure at the previous moment measured by the blood pressure measuring module.

2. The system for real-time measurement of non-contact blood pressure variability according to claim 1, wherein the pulse wave acquisition module acquires the aortic pulse wave waveform signal of the subject in real time by using a vital sign monitoring radar, specifically:

the subject sits or lies down, a single vital sign monitoring radar is placed to aim at the abdomen or the back of the subject, vital sign signals are continuously recorded, the subject breathes normally firstly in the process, then holds breath for x seconds, and then breathes normally, and the value of x is 20-35 s in the circulation; the vital sign signals recorded by the vital sign monitoring radar comprise pulse waves, heartbeats and clutter; the vital sign monitoring radar is a continuous wave radar;

intercepting the vital sign signals measured within x seconds of breath holding, demodulating the vital sign signals of the x seconds, and then removing heartbeat and clutter through a band-pass filter to obtain the aortic pulse waveform signals of the subject.

3. The system according to claim 1, wherein the module for acquiring real-time conduction time of pulse wave is configured to extract feature points of the waveform signal of aortic pulse wave, and to solve the real-time conduction time of pulse wave according to the feature points, specifically:

the method comprises the steps of extracting characteristic points of an aorta pulse wave waveform signal acquired by a vital sign monitoring radar, and obtaining each main characteristic point of the aorta pulse wave waveform signal, wherein the main characteristic points comprise a pulse wave trough, a first pulse wave contraction peak, a second pulse wave contraction peak and a pulse wave descending channel.

Solving the pulse wave conduction time by the first contraction peak of the aortic pulse wave and the stopping point of the ejection period, namely two characteristic points of the pulse wave descent isthmus, specifically:

extracting the time corresponding to the first contraction peak of the aortic pulse wave, namely the time corresponding to the first maximum value of the aortic pulse wave waveform, and recording as T_S(ii) a Extracting the time corresponding to the descending isthmus of the pulse wave, namely the time corresponding to the second inflection point of the waveform of the aortic pulse wave, and recording the time as T_DThen, the pulse transit time PTT is:

4. the system according to claim 1, wherein the blood pressure measurement module is configured to establish a gaussian process regression algorithm model based on blood pressure and pulse wave transit time, and obtain a real-time blood pressure value corresponding to the pulse wave real-time transit time according to the model, the blood pressure includes systolic pressure and diastolic pressure, and specifically:

(1) measuring the blood pressure observed value BP of the subject by using a sphygmomanometer₁,BP₂,…,BP_nTaking all the observed values as a training set, wherein n is the number of training samples, and each observed value is taken as a point sampled in multi-dimensional Gaussian distribution; PTT₁,PTT₂,…,PTT_nAre each BP₁,BP₂，…,BP_nThe corresponding pulse transit time;

modeling each blood pressure observation BP as some implicit function f (PTT) plus one coincidence mean value of 0 and variance ofIs recorded as the independent Gaussian distribution noise εNamely:

where PTT is the input vector, f (PTT) assumes that a gaussian process is given a priori, i.e.:

f(PTT)～GP(0,K)

obtaining a blood pressure observation value BP and a blood pressure predicted value BP according to Bayes law and the mapping from low dimension to high dimension of an independent variable PTT through a kernel function_*Joint prior distribution of (c):

where K ═ K (PTT ) is an n × n order symmetric positive definite covariance matrix, and the elements in the matrix are used to measure PTT_*With PTT_*The correlation between them; k (PTT )_*)＝K(PTT_*,PTT)^TTesting value PTT for pulse wave conduction time_*An n x 1 order covariance matrix between the input given pulse wave transit time observation value PTT of the training set; k (PTT)_*,PTT_*) Is PTT_*(ii) its own covariance; i is_nIs an n-dimensional identity matrix, σ_nIs white gaussian noise;

(2) blood pressure prediction value BP_*Obeying high-dimensional Gaussian distribution, and further deducing to obtain a blood pressure predicted value BP by a Bayesian formula_*The posterior distribution of the model is a Gaussian process regression algorithm model of blood pressure and pulse wave conduction time:

in the formula, BP_*For output, PTT, BP, PTT_*Is input;

wherein the content of the first and second substances,

the abbreviation is:

in the formula, k (PTT)_*,PTT_*) Is PTT_*The covariance function of itself is then determined,predicting BP for blood pressure_*Mean value of, V (BP)_*)＝cov(BP_*) Predicting BP for blood pressure_*Variance of (4), blood pressure predicted value BP_*In accordance with a mean value ofVariance is V (BP)_*) (ii) a gaussian distribution of; wherein the predicted systolic blood pressure value SBP of the blood pressure_*Obey mean value ofVariance is V (SBP)_*) (ii) a gaussian distribution of; predicted diastolic blood pressure value DBP_*Obey mean value ofVariance is V (DBP)_*) (ii) a gaussian distribution of;

(3) obtaining a corresponding real-time blood pressure value according to the pulse wave conduction time PTT:

and substituting the PTT into the Gaussian distributions corresponding to the SBP and the DBP to obtain the predicted values of the corresponding SBP and the DBP.

5. The system according to claim 1, wherein the blood pressure variability measurement module is configured to determine the blood pressure variability of the subject according to the blood pressure at the current time and the blood pressure at the previous time, and specifically:

using a blood pressure measuring module at t₀Measuring the real-time blood pressure of the subject at time t₀Measuring the real-time blood pressure of the subject at the moment + t, wherein the value of t is one-half of the sampling frequency, and dividing t₀Blood pressure at time + t and t₀Subtracting the blood pressure at the moment to obtain the blood pressure variability BPV of the subject:

BPV(t₀)＝BP(t₀+t)-BP(t₀)。

6. a computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements the functions of the system module according to any one of claims 1 to 5 when executing the computer program by:

extracting characteristic points of the aortic pulse wave waveform signals, and solving the real-time conduction time of the pulse waves according to the characteristic points;

establishing a Gaussian process regression algorithm model based on blood pressure and pulse wave conduction time, and acquiring a real-time blood pressure value corresponding to the pulse wave real-time conduction time according to the model;

and calculating the blood pressure variability of the subject according to the blood pressure at the current moment and the blood pressure at the previous moment measured by the blood pressure measuring module.

7. The computer device according to claim 6, wherein the processor, when executing the computer program, implements feature point extraction on the aortic pulse wave waveform signal, and finds the pulse wave real-time transit time according to the feature points, specifically:

the method comprises the steps of extracting characteristic points of an aorta pulse wave waveform signal acquired by a vital sign monitoring radar, and obtaining each main characteristic point of the aorta pulse wave waveform signal, wherein the main characteristic points comprise a pulse wave trough, a first pulse wave contraction peak, a second pulse wave contraction peak and a pulse wave descending channel.

Solving the pulse wave conduction time by the first contraction peak of the aortic pulse wave and the stopping point of the ejection period, namely two characteristic points of the pulse wave descent isthmus, specifically:

extracting the time corresponding to the first contraction peak of the aortic pulse wave, namely the time corresponding to the first maximum value of the aortic pulse wave waveform, and recording as T_S(ii) a Extracting the time corresponding to the descending isthmus of the pulse wave, namely the time corresponding to the second inflection point of the waveform of the aortic pulse wave, and recording the time as T_DThen, the pulse transit time PTT is:

8. the computer device according to claim 6, wherein the processor, when executing the computer program, implements establishing a gaussian process regression algorithm model based on blood pressure and pulse wave transit time, and obtains a real-time blood pressure value corresponding to the pulse wave real-time transit time according to the model, specifically:

(1) measuring the blood pressure observed value BP of the subject by using a sphygmomanometer₁,BP₂,…,BP_nTaking all the observed values as a training set, wherein n is the number of training samples, and each observed value is taken as a point sampled in multi-dimensional Gaussian distribution; PTT₁,PTT₂,…,PTT_nAre each BP₁,BP₂，…,BP_nThe corresponding pulse transit time;

modeling each blood pressure observation BP as some implicit function f (PTT) plus one coincidence mean value of 0 and variance ofIs recorded as the independent Gaussian distribution noise εNamely:

where PTT is the input vector, f (PTT) assumes that a gaussian process is given a priori, i.e.:

f(PTT)～GP(0,K)

obtaining a blood pressure observation value BP and a blood pressure predicted value BP according to Bayes law and the mapping from low dimension to high dimension of an independent variable PTT through a kernel function_*Joint prior distribution of (c):

where K ═ K (PTT ) is an n × n order symmetric positive definite covariance matrix, and the elements in the matrix are used to measure PTT_*With PTT_*The correlation between them; k (PTT )_*)＝K(PTT_*,PTT)^TTesting value PTT for pulse wave conduction time_*An n x 1 order covariance matrix between the input given pulse wave transit time observation value PTT of the training set; k (PTT)_*,PTT_*) Is PTT_*(ii) its own covariance; i is_nIs an n-dimensional identity matrix, σ_nIs white gaussian noise;

(2) blood pressure prediction value BP_*Obeying high-dimensional Gaussian distribution, and further deducing to obtain a blood pressure predicted value BP by a Bayesian formula_*The posterior distribution of the model is a Gaussian process regression algorithm model of blood pressure and pulse wave conduction time:

in the formula, BP_*For output, PTT, BP, PTT_*Is input;

wherein the content of the first and second substances,

the abbreviation is:

in the formula, k (PTT)_*,PTT_*) Is PTT_*The covariance function of itself is then determined,predicting BP for blood pressure_*Mean value of, V (BP)_*)＝cov(BP_*) Predicting BP for blood pressure_*Variance of (4), blood pressure predicted value BP_*In accordance with a mean value ofVariance is V (BP)_*) (ii) a gaussian distribution of; wherein the predicted systolic blood pressure value SBP of the blood pressure_*Obey mean value ofVariance is V (SBP)_*) (ii) a gaussian distribution of; predicted diastolic blood pressure value DBP_*Obey mean value ofVariance is V (DBP)_*) (ii) a gaussian distribution of;

(3) obtaining a corresponding real-time blood pressure value according to the pulse wave conduction time PTT:

and substituting the PTT into the Gaussian distributions corresponding to the SBP and the DBP to obtain the predicted values of the corresponding SBP and the DBP.

9. A computer-readable storage medium, on which a computer program is stored, wherein the computer program, when executed by a processor, implements the functions of the system module according to any one of claims 1 to 5 by:

extracting characteristic points of the aortic pulse wave waveform signals, and solving the real-time conduction time of the pulse waves according to the characteristic points;

establishing a Gaussian process regression algorithm model based on blood pressure and pulse wave conduction time, and acquiring a real-time blood pressure value corresponding to the pulse wave real-time conduction time according to the model;

and calculating the blood pressure variability of the subject according to the blood pressure at the current moment and the blood pressure at the previous moment measured by the blood pressure measuring module.

Technical Field

The invention belongs to the field of medical detection systems, and particularly relates to a non-contact real-time blood pressure variability measurement system, computer equipment and a storage medium.

Background

According to research, the blood pressure has fluctuation within a period of time, and the blood pressure variability represents the fluctuation degree of the blood pressure within a certain period of time and is a non-invasive index for evaluating the cardiovascular autonomic nervous activity. It is known that blood pressure and heart rate vary in time, such as the physiological variation of blood pressure variation, i.e. "two peaks and one valley" of diurnal variation.

The blood pressure variability direct measurement equipment known in the market at present is invasive, needs to be monitored in an arterial cannula, has the most accurate measured blood pressure result, needs to insert a pressure sensor into a human aorta to detect a blood pressure signal, but has the disadvantages of great harm to a human body, expensive equipment and long time consumption. The blood pressure variability measurement, which is generally used clinically, is expressed in terms of Standard Deviation (SD) or coefficient of variation (standard deviation/average, CV) or the like of blood pressure readings measured for a specific period of time. Although the conventional dynamic blood pressure monitor can acquire the blood pressure fluctuation condition of a patient within 24 hours, the blood pressure value of a certain time period is reflected, the real-time continuity of monitoring is not guaranteed, and the daily life of the patient is seriously interfered because the cuff is inflated for each measurement. Meanwhile, contact monitoring cannot directly contact some individuals, such as: large area burn patients, infectious disease patients, dermatologic patients, newborn babies, etc., which limits the range of applications of contact monitoring. In addition, contact blood pressure variability monitoring often requires professional medical personnel, and it takes nearly an hour to prepare before each monitoring, which is cumbersome to operate, labor and financial intensive, and does not allow real-time monitoring.

Disclosure of Invention

The invention aims to provide a non-contact type blood pressure variability real-time measuring system, a computer device and a storage medium, which can realize continuous real-time non-invasive measurement of blood pressure variability.

The technical solution for realizing the purpose of the invention is as follows: a non-contact real-time measurement system of blood pressure variability, comprising:

the pulse wave acquisition module is used for acquiring an aortic pulse wave waveform signal of the subject in real time by using a vital sign monitoring radar;

the pulse wave real-time conduction time acquisition module is used for extracting characteristic points of the aortic pulse wave waveform signals and solving the pulse wave real-time conduction time according to the characteristic points;

the blood pressure measuring module is used for establishing a Gaussian process regression algorithm model based on blood pressure and pulse wave conduction time, and acquiring a real-time blood pressure value corresponding to the pulse wave real-time conduction time according to the model, wherein the blood pressure comprises systolic pressure and diastolic pressure;

and the blood pressure variability measuring module is used for solving the blood pressure variability of the subject according to the blood pressure at the current moment and the blood pressure at the previous moment measured by the blood pressure measuring module.

A computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing the functions of each module of the system when executing the computer program.

A computer-readable storage medium, on which a computer program is stored which, when executed by a processor, implements the functionality of each of the modules of the above-described system.

Compared with the prior art, the invention has the following remarkable advantages: 1) the non-contact measurement of blood pressure variability can be realized by using a single vital sign monitoring radar, and the non-contact measurement can penetrate through obstacles such as clothes, bedding and the like, so that compared with the traditional contact monitoring, the operation is more convenient, the discomfort of a human body can be reduced, and a plurality of contact limitations can be overcome; 2) when the system acquires blood pressure, a Gaussian process regression algorithm model of the blood pressure and the pulse wave conduction time is established, the Gaussian process synthesizes a Bayesian algorithm, a nonparametric probability model can be obtained, compared with methods such as a support vector machine and a random forest, the needed model parameters are less, but the prediction effect is not inferior to that of the methods such as the support vector machine and the random forest, the probability significance explanation can be made on the output result, and the good result is achieved in practical application; 3) monitoring by using a single continuous wave radar, measuring an aortic pulse waveform signal aiming at the abdomen or back of a subject, establishing Gaussian process regression to obtain a blood pressure value, and realizing non-invasive measurement of ultra-short time blood pressure variability; compared with the traditional blood pressure variability measurement, the method can be calculated without professional medical staff and a plurality of sensors for measuring a plurality of arteries; compared with the traditional measurement, the system needs fewer instruments, is simple to operate, does not need to contact the human body, and has a wider application prospect; 4) the system is simple and effective, reliable in performance and convenient to implement.

The present invention is described in further detail below with reference to the attached drawing figures.

Drawings

FIG. 1 is a schematic diagram of the non-contact real-time blood pressure variability measuring system according to the present invention.

FIG. 2 is a schematic diagram of a Gaussian process regression algorithm model curve of blood pressure and pulse wave transit time established by the blood pressure measurement module of the present invention, wherein (a) is a schematic diagram of a PTT-SBP Gaussian regression prediction model curve, and (b) is a schematic diagram of a DBP-DBP Gaussian regression prediction model curve.

FIG. 3 is a long-term blood pressure variability measurement plot of a subject in an embodiment of the present invention.

FIG. 4 is a graph of the short-term blood pressure variability measurements of a subject in an embodiment of the invention.

FIG. 5 is a diagram of pulse waveform measurement of a subject according to an embodiment of the present invention.

Fig. 6 is an ultrashort blood pressure variability measurement chart of the subject, i.e. beat-to-beat blood pressure measurement chart in the embodiment of the present invention.

Detailed Description

Referring to fig. 1, the present invention provides a non-contact real-time blood pressure variability measuring system, which comprises:

the pulse wave acquisition module is used for acquiring an aortic pulse wave waveform signal of the subject in real time by using a vital sign monitoring radar;

the pulse wave real-time conduction time acquisition module is used for extracting characteristic points of the aortic pulse wave waveform signals and solving the pulse wave real-time conduction time according to the characteristic points;

the blood pressure measuring module is used for establishing a Gaussian process regression algorithm model based on blood pressure and pulse wave conduction time, and acquiring a real-time blood pressure value corresponding to the pulse wave real-time conduction time according to the model, wherein the blood pressure comprises systolic pressure and diastolic pressure;

and the blood pressure variability measuring module is used for solving the blood pressure variability of the subject according to the blood pressure at the current moment and the blood pressure at the previous moment measured by the blood pressure measuring module.

Further, the pulse wave acquisition module utilizes the vital sign monitoring radar to acquire the aortic pulse wave waveform signal of the subject in real time, and specifically comprises:

the subject sits or lies down, a single vital sign monitoring radar is placed to aim at the abdomen or the back of the subject, vital sign signals are continuously recorded, the subject breathes normally firstly in the process, then holds breath for x seconds, and then breathes normally, and the value of x is 20-35 s in the circulation; the vital sign signals recorded by the vital sign monitoring radar comprise pulse waves, heartbeats and clutter; the vital sign monitoring radar is a continuous wave radar;

intercepting vital sign signals measured within x seconds of breath holding, demodulating the vital sign signals within x seconds, and then removing heartbeat and clutter through a band-pass filter to obtain an aortic pulse waveform signal of a subject;

further, the real-time conduction time of pulse wave obtains the module for carry out the characteristic point to aortic pulse wave waveform signal and draw, solve the real-time conduction time of pulse wave according to the characteristic point, specifically do:

the method comprises the steps of extracting characteristic points of an aorta pulse wave waveform signal acquired by a vital sign monitoring radar, and obtaining each main characteristic point of the aorta pulse wave waveform signal, wherein the main characteristic points comprise a pulse wave trough, a first pulse wave contraction peak, a second pulse wave contraction peak and a pulse wave descending channel.

Solving the pulse wave conduction time by the first contraction peak of the aortic pulse wave and the stopping point of the ejection period, namely two characteristic points of the pulse wave descent isthmus, specifically:

extracting the time corresponding to the first systolic peak of the aortic pulse wave,namely the time corresponding to the first maximum value of the aortic pulse wave waveform, which is recorded as T_S(ii) a Extracting the time corresponding to the descending isthmus of the pulse wave, namely the time corresponding to the second inflection point of the waveform of the aortic pulse wave, and recording the time as T_DThen, the pulse transit time PTT is:

further, the blood pressure measuring module is used for establishing a gaussian process regression algorithm model based on blood pressure and pulse wave conduction time, and acquiring a real-time blood pressure value corresponding to the pulse wave real-time conduction time according to the model, wherein the blood pressure comprises systolic pressure and diastolic pressure, and the specific steps are as follows:

(1) measuring the blood pressure observed value BP of the subject by using a sphygmomanometer₁,BP₂,…,BP_nTaking all the observed values as a training set, wherein n is the number of training samples, and each observed value is taken as a point sampled in multi-dimensional Gaussian distribution; PTT₁,PTT₂,…,PTT_nAre each BP₁,BP₂，…,BP_nThe corresponding pulse transit time;

modeling each blood pressure observation BP as some implicit function f (PTT) plus one coincidence mean value of 0 and variance ofIs recorded as the independent Gaussian distribution noise εNamely:

where PTT is the input vector, f (PTT) assumes that a gaussian process is given a priori, i.e.:

f(PTT)～GP(0,K)

according to Bayes law and the mapping of independent variable PTT from low dimension to high dimension through kernel functionBlood pressure observed value BP and blood pressure predicted value BP_*Joint prior distribution of (c):

where K ═ K (PTT ) is an n × n order symmetric positive definite covariance matrix, and the elements in the matrix are used to measure PTT_*With PTT_*The correlation between them; k (PTT )_*)＝K(PTT_*,PTT)^TTesting value PTT for pulse wave conduction time_*An n x 1 order covariance matrix between the input given pulse wave transit time observation value PTT of the training set; k (PTT)_*,PTT_*) Is PTT_*(ii) its own covariance; i is_nIs an n-dimensional identity matrix, σ_nIs white gaussian noise;

(2) blood pressure prediction value BP_*Obeying high-dimensional Gaussian distribution, and further deducing to obtain a blood pressure predicted value BP by a Bayesian formula_*The posterior distribution of the model is a Gaussian process regression algorithm model of blood pressure and pulse wave conduction time:

in the formula, BP_*For output, PTT, BP, PTT_*Is input;

wherein the content of the first and second substances,

the abbreviation is:

in the formula, k (PTT)_*,PTT_*) Is PTT_*The covariance function of itself is then determined,predicting BP for blood pressure_*Mean value of, V (BP)_*)＝cov(BP_*) Predicting BP for blood pressure_*Variance of (4), blood pressure predicted value BP_*In accordance with a mean value ofVariance is V (BP)_*) (ii) a gaussian distribution of; wherein the predicted systolic blood pressure value SBP of the blood pressure_*Obey mean value ofVariance is V (SBP)_*) (ii) a gaussian distribution of; predicted diastolic blood pressure value DBP_*Obey mean value ofVariance is V (DBP)_*) (ii) a gaussian distribution of;

(3) obtaining a corresponding real-time blood pressure value according to the pulse wave conduction time PTT:

substituting PTT into the corresponding Gaussian distributions of SBP and DBP as shown in FIG. 2, the predicted values of SBP and DBP can be obtained.

Further, the blood pressure variability measurement module is configured to calculate the blood pressure variability of the subject according to the blood pressure at the current time and the blood pressure at the previous time, specifically:

using a blood pressure measuring module at t₀Measuring the real-time blood pressure of the subject at time t₀Measuring the real-time blood pressure of the subject at the moment + t, wherein the value of t is one-half of the sampling frequency, and dividing t₀Blood pressure at time + t and t₀Subtracting the blood pressure at the moment to obtain the blood pressure variability BPV of the subject:

BPV(t₀)＝BP(t₀+t)-BP(t₀)。

the blood pressure variability is mainly classified into long-time blood pressure variability, short-time blood pressure variability and ultra-short-time blood pressure variability. Long-term blood pressure variability is mainly exponential in variation over a day or several weeks; the short-time blood pressure variability refers to 24-hour dynamic blood pressure, namely blood pressure variation in a small time, and the measurement method is generally to measure every 1-2 hours; the ultrashort blood pressure variability refers to the variation of blood pressure from beat to beat, and the blood pressure caused by each heart beat has a slight difference. The continuous wave radar adopted by the invention can realize long-time continuous measurement, can feed back data in real time, calculates the pulse wave conduction time corresponding to each waveform, and can obtain the change of blood pressure per minute and per second, namely ultrashort-time blood pressure variability.

A computer device comprises a memory, a processor and a computer program which is stored on the memory and can run on the processor, wherein the processor realizes the functions of the system module when executing the computer program, and the specific process is as follows:

the method comprises the following steps of utilizing a vital sign monitoring radar to collect aortic pulse waveform signals of a subject in real time, and specifically comprising the following steps: the subject sits or lies down, a single vital sign monitoring radar is placed to aim at the abdomen or the back of the subject, vital sign signals are continuously recorded, the subject breathes normally firstly in the process, then holds breath for x seconds, and then breathes normally, and the value of x is 20-35 s in the circulation; the vital sign signals recorded by the vital sign monitoring radar comprise pulse waves, heartbeats and clutter; the vital sign monitoring radar is a continuous wave radar;

intercepting the vital sign signals measured within x seconds of breath holding, demodulating the vital sign signals of the x seconds, and then removing heartbeat and clutter through a band-pass filter to obtain the aortic pulse waveform signals of the subject.

Carry out the characteristic point to aortic pulse wave waveform signal and draw, ask the real-time conduction time of pulse wave according to the characteristic point, specifically do:

the method comprises the steps of extracting characteristic points of an aorta pulse wave waveform signal acquired by a vital sign monitoring radar, and obtaining each main characteristic point of the aorta pulse wave waveform signal, wherein the main characteristic points comprise a pulse wave trough, a first pulse wave contraction peak, a second pulse wave contraction peak and a pulse wave descending channel.

Solving the pulse wave conduction time by the first contraction peak of the aortic pulse wave and the stopping point of the ejection period, namely two characteristic points of the pulse wave descent isthmus, specifically:

extracting the time corresponding to the first contraction peak of the aortic pulse wave, namely the time corresponding to the first maximum value of the aortic pulse wave waveform, and recording as T_S(ii) a Extracting the time corresponding to the descending isthmus of the pulse wave, namely the time corresponding to the second inflection point of the waveform of the aortic pulse wave, and recording the time as T_DThen, the pulse transit time PTT is:

establishing a Gaussian process regression algorithm model based on blood pressure and pulse wave conduction time, and acquiring a real-time blood pressure value corresponding to the pulse wave real-time conduction time according to the model, wherein the blood pressure comprises systolic pressure and diastolic pressure; the method specifically comprises the following steps:

(1) measuring the blood pressure observed value BP of the subject by using a sphygmomanometer₁,BP₂,…,BP_nTaking all the observed values as a training set, wherein n is the number of training samples, and each observed value is taken as a point sampled in multi-dimensional Gaussian distribution; PTT₁,PTT₂,…,PTT_nAre each BP₁,BP₂，…,BP_nThe corresponding pulse transit time;

modeling each blood pressure observation BP as some implicit function f (PTT) plus one coincidence mean value of 0 and variance ofIs recorded as the independent Gaussian distribution noise εNamely:

where PTT is the input vector, f (PTT) assumes that a gaussian process is given a priori, i.e.:

f(PTT)～GP(0,K)

obtaining a blood pressure observation value BP and a blood pressure predicted value BP according to Bayes law and the mapping from low dimension to high dimension of an independent variable PTT through a kernel function_*Joint prior distribution of (c):

where K ═ K (PTT ) is an n × n order symmetric positive definite covariance matrix, and the elements in the matrix are used to measure PTT_*With PTT_*The correlation between them; k (PTT )_*)＝K(PTT_*,PTT)^TTesting value PTT for pulse wave conduction time_*An n x 1 order covariance matrix between the input given pulse wave transit time observation value PTT of the training set; k (PTT)_*,PTT_*) Is PTT_*(ii) its own covariance; i is_nIs an n-dimensional identity matrix, σ_nIs white gaussian noise;

(2) blood pressure prediction value BP_*Obeying high-dimensional Gaussian distribution, and further deducing to obtain a blood pressure predicted value BP by a Bayesian formula_*The posterior distribution of the model is a Gaussian process regression algorithm model of blood pressure and pulse wave conduction time:

in the formula, BP_*For output, PTT, BP, PTT_*Is input;

wherein the content of the first and second substances,

the abbreviation is:

in the formula, k (PTT)_*,PTT_*) Is PTT_*The covariance function of itself is then determined,predicting BP for blood pressure_*Mean value of, V (BP)_*)＝cov(BP_*) Predicting BP for blood pressure_*Variance of (4), blood pressure predicted value BP_*In accordance with a mean value ofVariance is V (BP)_*) (ii) a gaussian distribution of; wherein the predicted systolic blood pressure value SBP of the blood pressure_*Obey mean value ofVariance is V (SBP)_*) (ii) a gaussian distribution of; predicted diastolic blood pressure value DBP_*Obey mean value ofVariance is V (DBP)_*) (ii) a gaussian distribution of;

(3) obtaining a corresponding real-time blood pressure value according to the pulse wave conduction time PTT:

and substituting the PTT into the Gaussian distributions corresponding to the SBP and the DBP to obtain the predicted values of the corresponding SBP and the DBP.

According to the blood pressure at the current moment and the blood pressure at the previous moment measured by the blood pressure measuring module, the blood pressure variability of the subject is obtained, and the method specifically comprises the following steps:

using bloodPressure measurement module at t₀Measuring the real-time blood pressure of the subject at time t₀Measuring the real-time blood pressure of the subject at the moment + t, wherein the value of t is one-half of the sampling frequency, and dividing t₀Blood pressure at time + t and t₀Subtracting the blood pressure at the moment to obtain the blood pressure variability BPV of the subject:

BPV(t₀)＝BP(t₀+t)-BP(t₀)。

a computer-readable storage medium, on which a computer program is stored, the computer program, when being executed by a processor, implementing the functions of the system modules as described above, the specific process is as follows:

the method comprises the following steps of utilizing a vital sign monitoring radar to collect aortic pulse waveform signals of a subject in real time, and specifically comprising the following steps: the subject sits or lies down, a single vital sign monitoring radar is placed to aim at the abdomen or the back of the subject, vital sign signals are continuously recorded, the subject breathes normally firstly in the process, then holds breath for x seconds, and then breathes normally, and the value of x is 20-35 s in the circulation; the vital sign signals recorded by the vital sign monitoring radar comprise pulse waves, heartbeats and clutter; the vital sign monitoring radar is a continuous wave radar;

intercepting the vital sign signals measured within x seconds of breath holding, demodulating the vital sign signals of the x seconds, and then removing heartbeat and clutter through a band-pass filter to obtain the aortic pulse waveform signals of the subject.

Carry out the characteristic point to aortic pulse wave waveform signal and draw, ask the real-time conduction time of pulse wave according to the characteristic point, specifically do:

the method comprises the steps of extracting characteristic points of an aorta pulse wave waveform signal acquired by a vital sign monitoring radar, and obtaining each main characteristic point of the aorta pulse wave waveform signal, wherein the main characteristic points comprise a pulse wave trough, a first pulse wave contraction peak, a second pulse wave contraction peak and a pulse wave descending channel.

Solving the pulse wave conduction time by the first contraction peak of the aortic pulse wave and the stopping point of the ejection period, namely two characteristic points of the pulse wave descent isthmus, specifically:

extracting first systolic wave of aortic pulse waveThe time corresponding to the peak, i.e. the time corresponding to the first maximum value of the aortic pulse wave waveform, is recorded as T_S(ii) a Extracting the time corresponding to the descending isthmus of the pulse wave, namely the time corresponding to the second inflection point of the waveform of the aortic pulse wave, and recording the time as T_DThen, the pulse transit time PTT is:

establishing a Gaussian process regression algorithm model based on blood pressure and pulse wave conduction time, and acquiring a real-time blood pressure value corresponding to the pulse wave real-time conduction time according to the model, wherein the blood pressure comprises systolic pressure and diastolic pressure; the method specifically comprises the following steps:

(1) measuring the blood pressure observed value BP of the subject by using a sphygmomanometer₁,BP₂,…,BP_nTaking all the observed values as a training set, wherein n is the number of training samples, and each observed value is taken as a point sampled in multi-dimensional Gaussian distribution; PTT₁,PTT₂,…,PTT_nAre each BP₁,BP₂，…,BP_nThe corresponding pulse transit time;

modeling each blood pressure observation BP as some implicit function f (PTT) plus one coincidence mean value of 0 and variance ofIs recorded as the independent Gaussian distribution noise εNamely:

where PTT is the input vector, f (PTT) assumes that a gaussian process is given a priori, i.e.:

f(PTT)～GP(0,K)

obtaining a blood pressure observation value BP according to Bayes law and the mapping from low dimension to high dimension of an independent variable PTT through a kernel functionAnd blood pressure prediction value BP_*Joint prior distribution of (c):

where K ═ K (PTT ) is an n × n order symmetric positive definite covariance matrix, and the elements in the matrix are used to measure PTT_*With PTT_*The correlation between them; k (PTT )_*)＝K(PTT_*,PTT)^TTesting value PTT for pulse wave conduction time_*An n x 1 order covariance matrix between the input given pulse wave transit time observation value PTT of the training set; k (PTT)_*,PTT_*) Is PTT_*(ii) its own covariance; i is_nIs an n-dimensional identity matrix, σ_nIs white gaussian noise;

(2) blood pressure prediction value BP_*Obeying high-dimensional Gaussian distribution, and further deducing to obtain a blood pressure predicted value BP by a Bayesian formula_*The posterior distribution of the model is a Gaussian process regression algorithm model of blood pressure and pulse wave conduction time:

in the formula, BP_*For output, PTT, BP, PTT_*Is input;

wherein the content of the first and second substances,

the abbreviation is:

in the formula, k (PTT)_*,PTT_*) Is PTT_*The covariance function of itself is then determined,predicting BP for blood pressure_*Mean value of, V (BP)_*)＝cov(BP_*) Predicting BP for blood pressure_*Variance of (4), blood pressure predicted value BP_*In accordance with a mean value ofVariance is V (BP)_*) (ii) a gaussian distribution of; wherein the predicted systolic blood pressure value SBP of the blood pressure_*Obey mean value ofVariance is V (SBP)_*) (ii) a gaussian distribution of; predicted diastolic blood pressure value DBP_*Obey mean value ofVariance is V (DBP)_*) (ii) a gaussian distribution of;

(3) obtaining a corresponding real-time blood pressure value according to the pulse wave conduction time PTT:

and substituting the PTT into the Gaussian distributions corresponding to the SBP and the DBP to obtain the predicted values of the corresponding SBP and the DBP.

According to the blood pressure at the current moment and the blood pressure at the previous moment measured by the blood pressure measuring module, the blood pressure variability of the subject is obtained, and the method specifically comprises the following steps:

using a blood pressure measuring module at t₀Measuring the real-time blood pressure of the subject at time t₀Measuring the real-time blood pressure of the subject at the moment + t, wherein the value of t is one-half of the sampling frequency, and dividing t₀Blood pressure at time + t and t₀Subtracting the blood pressure at the moment to obtain the blood pressure variability BPV of the subject:

BPV(t₀)＝BP(t₀+t)-BP(t₀)。

the present invention will be described in further detail with reference to examples.

Examples

In this embodiment, the system of the present invention and the dedicated blood pressure monitor are respectively used to test the blood pressure variability of 10 subjects.

The system of the invention is used for testing and comprises the following contents:

1. the aortic pulse wave signals of 10 subjects are acquired by a pulse wave acquisition module in the system.

2. The aorta pulse wave waveform characteristic points of the 10 subjects are obtained by a pulse wave real-time conduction time obtaining module in the system, and the pulse wave real-time conduction time of the 10 persons is solved according to each main characteristic point.

3. A Gaussian process regression algorithm model of blood pressure and pulse wave conduction time is established by a blood pressure measuring module in the system, the pulse wave real-time conduction time of 10 persons is input into a waveform analysis device, and then the real-time blood pressure value of each person is obtained according to the pulse wave conduction time.

4. The blood pressure variability of the subject is obtained by a blood pressure variability measuring module in the system, the blood pressure value comprises systolic pressure and diastolic pressure, and the systolic pressure is subtracted from the diastolic pressure of 10 persons to obtain the blood pressure variability.

The detection results of the system and the special blood pressure detector are shown in the following table 1:

TABLE 1 results of blood pressure value calculation by Gaussian regression prediction model

As can be seen from the above table 1, compared with the measurement result of the special blood pressure detector, the measurement result of the system of the invention has the error kept within 5mmHg, which shows that the measurement result of the invention is accurate. One subject was selected from the subjects, and the subjects were first asked to measure blood pressure at the same time every day for six consecutive days, and the changes in SBP and DBP, i.e. long-term blood pressure variability, were obtained as shown in fig. 3.

After the experiment, the subjects were asked to measure blood pressure once every two hours from eight am to six pm within one day by using continuous wave radar to obtain changes in SBP and DBP, i.e. to obtain short-term blood pressure variability as shown in fig. 4.

As can be seen from FIGS. 3 and 4, the system of the present invention can effectively realize the measurement of long-term blood pressure variability and short-term blood pressure variability, and has a certain research value. On the basis, the continuous wave radar is adopted, so that long-time continuous measurement can be realized, data can be fed back in real time, the pulse wave conduction time corresponding to each waveform is calculated, and the change of blood pressure per minute and per second can be obtained, namely the continuous wave radar can realize the measurement of blood pressure beat by beat. Still taking the subject as an example, the pulse wave waveform obtained by the continuous wave radar measurement is shown in fig. 5.

As can be seen from fig. 5, the continuous wave radar can measure and feed back the pulse wave waveforms in real time, and each pulse wave waveform can be solved to obtain the corresponding cardiovascular characteristic parameters. The characteristic parameters corresponding to each waveform obtained are input into a gaussian process regression model to obtain real-time blood pressure parameters, and finally, SBP is obtained, and the beat-to-beat blood pressure of DBP, that is, the ultrashort-time blood pressure variability, is shown in fig. 6.

In summary, the non-contact real-time blood pressure variability measurement system, the computer device and the storage medium of the invention can realize the measurement of three conditions of long-time blood pressure variability, short-time blood pressure variability and ultra-short-time blood pressure variability. Compared with the traditional blood pressure variability measuring device, the measuring system can realize non-contact, non-invasive and accurate measurement of the blood pressure variability value only by a single radio frequency sensor, and has the advantages of effectiveness, feasibility, reliable performance and higher precision.