WO2019127623A1

WO2019127623A1 - Method and device for constructing model for estimating arterial blood vessel age

Info

Publication number: WO2019127623A1
Application number: PCT/CN2018/070071
Authority: WO
Inventors: 李烨; 刘记奎; 苗芬; 闻博; 刘增丁
Original assignee: 中国科学院深圳先进技术研究院
Priority date: 2017-12-27
Filing date: 2018-01-02
Publication date: 2019-07-04
Also published as: CN108185996A; CN108185996B

Abstract

A method and device for constructing a model for estimating arterial blood vessel age, which relates to the technical field of cardiovascular disease monitoring, the method comprising: collecting reference arterial blood vessel ages, physiological signals, and individualized information of sample users; executing a weighted feature acquisition operation according to gender for each sample user respectively; extracting features according to the physiological signals and the individualized information; normalizing each feature to obtain a normalized feature; calculating a weight coefficient of each feature respectively according to a correlation coefficient between each feature and the reference arterial blood vessel age; calculating a weighted feature of each feature according to the normalized feature and the weight coefficient; training a neural network by using the weighted features of each sample user and the reference arterial blood vessel age as sample data to obtain a model for estimating arterial blood vessel age. The device and method estimate accurately and are suitable for household medical treatment, and may provide more useful physiological parameters for home healthcare.

Description

Method and device for constructing arterial blood vessel age estimation model

Cross-reference to related applications

The present application claims priority to Chinese Patent Application No. 201711453635.7, entitled "Method and Apparatus for Constructing Arterial Vascular Age Estimation Model", filed on December 27, 2017, the entire contents of which is incorporated herein by reference. in.

Technical field

The invention relates to the technical field of cardiovascular disease monitoring, in particular to a method and a device for constructing an arterial blood vessel age estimation model.

Background technique

Cardiovascular disease (CVD) has the characteristics of high morbidity and high disability, and has been the leading cause of death in China since 1990. Therefore, it is of great clinical significance to achieve screening and primary prevention of high-risk populations of asymptomatic CVD. Arteriosclerosis and changes in the elasticity of the arterial lumen play an important role in the development of CVD and are prerequisites for the occurrence of cardiovascular disease. Therefore, monitoring of vascular sclerosis and elastic changes is more predictive of cardiovascular events. The criteria for judging arterial elasticity and age mainly include the following three types: Framingham score, pulse wave velocity, and carotid intima-media thickness.

The Framingham score is a good indicator of the age of arterial aging, giving the risk of cardiovascular and cerebrovascular disease within ten years. However, this method does not give the corresponding blood vessel age, and the blood vessel age is more intuitive than the risk of the disease, and the psychological feeling is more able to urge the patient to adopt a more reasonable lifestyle; in addition, the implementation of the above technology can only be carried out in the hospital, many patients In the case of no onset, it is rare to go to the hospital for examination, thus greatly reducing the prevention of cardiovascular disease.

Summary of the invention

In view of this, the object of the present invention is to provide a method and a device for constructing an arterial blood vessel age estimation model, which can realize the estimation of blood vessel age through portable or wearable devices, and is accurate and suitable for home medical treatment, and improves cardiovascular disease. Preventive effects.

In a first aspect, an embodiment of the present invention provides a method for constructing an arterial blood vessel age estimation model, comprising: collecting a reference arterial blood vessel age, a physiological signal, and individualized information of a sample user; the physiological signal includes a synchronized PPG signal and an ECG signal; The information includes gender and wingspan; for each sample user, weighted feature acquisition operations are performed separately according to gender: feature extraction is performed according to physiological signals and individualized information; the above features include: pulse wave velocity, normalized heavy wave Time delay with the main wave, normalized pulse wave rise time and BMI index; normalization of each feature to obtain normalized features; respectively, according to the correlation coefficient between each feature and the reference arterial age Weighting coefficient; calculating the weighted features of each feature according to the normalized feature and the weighting coefficient; training the neural network with the weighted features of each sample user and the reference arterial age as sample data, and obtaining an arterial blood vessel age estimation model.

Further, the step of collecting physiological signals of a plurality of users of the same sex comprises: measuring upper arm blood pressure, PPG signal and ECG signal of a plurality of users of the same sex; measuring order of upper arm blood pressure, PPG signal and ECG signal is: first upper arm Blood pressure measurement, PPG signal and ECG signal synchronization measurement, second upper arm blood pressure measurement; when the difference between the measured values of the upper arm blood pressure measurement is not greater than the preset deviation threshold, the PPG signal and the ECG signal are determined as the user's body information.

Further, the step of extracting features according to the physiological signal and the individualized information comprises: performing R wave peak point detection on the ECG signal, and performing pulse wave start point A, main wave B, tidal wave C, and notch D on the PPG signal and The doubling wave E detection; dividing the pulse conduction distance by the average time delay of the pulse wave starting point A and the R wave peak point to obtain the pulse wave conduction velocity; dividing the average time delay of the dicrotic wave E and the main wave B by the average heartbeat period The time delay of normalizing the heavy wave and the main wave; dividing the average time delay of the main wave B and the starting pulse wave A by the average heartbeat cycle to obtain the normalized pulse wave rising branch time.

Further, the step of calculating the weight coefficients of the respective features according to the correlation coefficients of the respective features and the reference arterial age includes: respectively calculating the correlation coefficient L _{i of the} feature and the reference arterial blood vessel age by correlation coefficient analysis; Calculate the weighting factor c _{i of} each feature:

Where n is the number of features.

Further, the step of calculating the weighting features of the respective features according to the normalized features and the weighting coefficients comprises: multiplying the normalized features by the corresponding weighting coefficients respectively to obtain weighting features of the respective features.

Further, before the step of extracting features according to the body information, the method further comprises: performing denoising preprocessing on the PPG signal and the ECG signal.

Further, the personalized information includes one or more of the following: natural age, height, weight, and smoking age;

The step of normalizing each feature to obtain a normalized feature includes: normalizing each feature in combination with natural age and smoke age to obtain a normalized feature.

Further, the method further includes: collecting a physiological signal and individualized information of the target user; performing a weighted feature acquisition operation on the target user; and inputting a weighted feature of the target user into an arterial blood vessel age estimation model corresponding to the gender of the target user to obtain a target The user's arterial age.

In a second aspect, an embodiment of the present invention further provides an apparatus for constructing an arterial blood vessel age estimation model, comprising: a sample collection module, configured to collect reference arterial blood vessel age, physiological signals, and individualized information of a sample user; the physiological signal includes a synchronized PPG. The signal and the ECG signal, the individualized information includes a gender and a span; the weighted feature acquisition module is configured to perform a weighted feature acquisition operation according to the gender for each sample user: extracting features according to the physiological signal and the individualized information; : pulse wave conduction velocity, time delay of normalized heavy wave and main wave, normalized pulse wave rising branch time and BMI index; normalization of each feature to obtain normalized features; according to various features and references The correlation coefficient of arterial age is used to calculate the weight coefficients of each feature respectively; the weighted features of each feature are calculated according to the normalized features and the weight coefficients; the training module is used to take the weighted features of each sample user and the reference arterial age as samples Data training neural network, get the arterial age estimation model type.

Further, the method further includes: a target acquisition module, configured to collect physiological signals and individualized information of the target user; a target weighted feature acquisition module, configured to perform a weighted feature acquisition operation on the target user; and an age estimation module, configured to target the user The weighted feature inputs an arterial blood vessel age estimation model corresponding to the gender of the target user, and obtains the arterial blood vessel age of the target user.

The embodiments of the present invention bring about the following beneficial effects: the method and device for constructing an arterial blood vessel age estimation model provided by the embodiments of the present invention, the non-linear fitting of the blood vessel age by the reference user's reference arterial age and body information, the artery is constructed The blood vessel age estimation model can improve the estimation accuracy of arterial blood vessel age, and can estimate the blood vessel age through portable or wearable devices. The estimation is accurate and suitable for home medical treatment, and can provide more useful physiological parameters for family health.

Other features and advantages of the present disclosure will be set forth in the description which follows.

The above described objects, features, and advantages of the present invention will become more apparent from the description of the appended claims.

DRAWINGS

In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the specific embodiments or the description of the prior art will be briefly described below, and obviously, the attached in the following description The drawings are some embodiments of the present invention, and those skilled in the art can obtain other drawings based on these drawings without any creative work.

1 is a schematic flow chart of a method for constructing an arterial blood vessel age estimation model according to an embodiment of the present invention;

2 is a schematic diagram of a PPG and an ECG according to an embodiment of the present invention;

3 is a schematic flow chart of a method for constructing an arterial blood vessel age estimation model according to an embodiment of the present invention;

4 is a schematic structural diagram of a method for constructing an arterial blood vessel age estimation model according to an embodiment of the present invention;

FIG. 5 is a schematic structural diagram of a method for constructing an arterial blood vessel age estimation model according to an embodiment of the present invention; FIG.

FIG. 6 is a schematic structural diagram of a monitoring device according to an embodiment of the present invention.

Detailed ways

The embodiments of the present invention will be clearly and completely described in detail with reference to the accompanying drawings. An embodiment. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative efforts are within the scope of the present invention.

The above criteria for jugular elasticity and age are as follows:

(1) Framingham score: The American Framingham study conducted a cardiovascular disease risk score (Flemingham score) based on cholesterol and non-cholesterol factors, including non-cholesterol factors including diabetes (high risk factors), age (male) >45, female>55), smoking, hypertension, high-density lipoprotein, a history of coronary heart disease in first-degree relatives. The study scores each factor by a large amount of data that is tracked cumulatively, and then judges the risk based on the overall score, which can reflect the vascular aging according to the risk score.

(2) Pulse wave velocity: The pulse wave velocity (PWV) can reflect the softness and hardness of the blood vessel. For example, the harder the elastic, the weaker the blood vessel PWV, and the slower the elastic PWV. Therefore, in some studies, pulse rate is often used as a measure of blood vessel hardness. For example, the arteriosclerosis measuring instrument developed by Omron reflects the degree of vascular sclerosis by measuring the pulse wave velocity cfPWV of the neck-femoral artery, and then converting the arterial age with other measured indexes.

(3) Carotid intima-media thickness: In arteries, the thicker the intima and media, the more easily the arterial vessels are blocked by arteriosclerotic plaques. The doctor can easily measure the intima-media thickness of the carotid artery by ultrasound, and this measurement can be used to estimate the age of the blood vessel.

The above three methods may not accurately reflect the aging of the arteries, or may not directly give the age of the arterial vessels, and are not suitable for use in home medical measurement, which is not conducive to the prevention and treatment of cardiovascular diseases. Based on this, the method and device for constructing an arterial blood vessel age estimation model provided by the embodiments of the present invention can realize the estimation of the blood vessel age through a portable or wearable device, and the estimation is accurate and suitable for home medical treatment.

The specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

Embodiment 1

The embodiment of the invention provides a method for constructing an arterial blood vessel age estimation model, and a flow chart of the method for constructing an arterial blood vessel age estimation model shown in FIG. 1 , the method comprising the following steps:

In step S102, the reference arterial blood vessel age, physiological signal and individualized information of the sample user are collected.

Because of the large physiological differences among users with different genders, in the construction of the arterial blood vessel age estimation model, considering the accuracy of the estimation model, an estimation model will be constructed for men and women respectively. Among them, different gender models use the same kind of data, and feature extraction and model training also use the same method. The physiological signals include synchronized PPG (photoplethysmograph) signals and ECG (electrocardiogram) signals, and individualized information includes gender and abduction. Among them, the arm extension is used to calculate the pulse wave conduction velocity.

In the present embodiment, the reference arterial blood vessel age of the sample user can be measured by using an OMRON arteriosclerosis apparatus. The reference arterial blood vessel age can also be obtained by other means, which is not limited in this embodiment. In this embodiment, the estimation of the age of the individual arterial vessels can be performed by the home medical device (electronic sphygmomanometer) and the wearable device (measuring the fingertip photoplethysmographic pulse wave and the electrocardiogram), so that the measurement can be independently performed by the user in the home medical treatment.

In step S104, for each sample user, the weighted feature acquisition operation is performed separately according to the gender.

The weighted feature acquisition operation may include: extracting features according to physiological signals and individualized information; normalizing each feature to obtain normalized features; calculating respective features according to correlation coefficients of respective features and reference arterial ages Weighting coefficient; calculating the weighting characteristics of each feature according to the normalized feature and the weighting coefficient.

(1) Feature extraction based on the above physiological signals and individualized information, including feature extraction of PPG and ECG. Key points are detected on PPG and ECG before extraction. The detection of key points of ECG includes R wave peak detection, which is used as the starting time reference for pulse wave conduction; PPG key point detection mainly detects pulse wave starting point A , main wave B, tidal wave C, notch D and heavy wave wave E. Referring to the schematic diagrams of PPG and ECG shown in Figure 2, the above various key points are shown.

The above extracted features may include: pulse wave velocity, time delay of normalized heavy wave and main wave, normalized pulse wave rise time, and BMI index. Referring to Figure 2, the above features are calculated as follows:

Pulse wave conduction velocity (PWV): First, the pulse transit time (PTT) is calculated from the time delay between the PPG start point A and the R wave of the ECG, and then calculated by the following formula: PWV = pulse conduction distance (L/2) / PTT. Among them, PTT is the average pulse wave time calculated by multiple times, and L is the arm extension (the arms are straight and shoulder flat, and the distance between the two middle finger ends is measured).

Time delay of normalized heavy wave and main wave (T1): measure the time delay Tbe between main wave B and heavy wave E, and then calculate T1, which is solved by T1=Tbe/RR, where Tbe is average Time delay, RR is the measured average heartbeat period, that is, the average time difference between two adjacent R waves.

Normalized pulse wave rising branch time (T2): Measure the time delay Tab between the main wave B and the starting point A, and then calculate T2, which is calculated as: T2=Tab/RR, where Tab is the average time delay.

BMI index: The BMI index is calculated by the following formula: BMI index = kilograms of body weight / square of height meters. It should be noted that when collecting the individualized information of the sample user, the height and weight of the sample user can also be collected, and then the BMI index can be calculated by height and weight, or if the sample user knows his own BMI index. The BMI index of the sample user can be directly collected, which is not limited in this embodiment.

(2) Normalize each feature to obtain a normalized feature.

The above PWV, T1, T2, and BMI are normalized to the interval [a, b], respectively. The normalization method can use the Z-score standard method, and its conversion function is x ^* = x - μ / σ, where μ and σ are the mean and standard deviation of all the features of the sample set (the normalization of the feature is for each The feature attributes are normalized separately). When estimating using the model constructed in this embodiment, the same μ and σ are used for feature normalization. The normalization method can also adopt other existing normalization methods, which is not limited in this embodiment.

(3) The weight coefficients of the respective features are respectively calculated according to the correlation coefficients of the respective features and the reference arterial age.

In this step, the correlation coefficient L _{i of the} feature and the reference arterial age can be separately solved by correlation coefficient analysis, and then the weight coefficient c _{i of} each feature is calculated according to the following formula:

Where n is the number of features. Three features PWV, T1, and T2 are used in this embodiment, and each correlation coefficient is represented by L ₁ , L ₂ , and L ₃ , respectively, and n=3.

(4) Calculate the weighting features of each feature based on the normalized features and the weight coefficients.

The weighted features of the respective features described above can be obtained by multiplying the normalized features obtained in the above step (2) by the corresponding weight coefficients obtained in the above step (3).

In step S106, the neural network is trained with the weighted feature of each sample user and the reference arterial age as sample data to obtain an arterial age estimation model.

In this embodiment, the neural network can be used to fit the blood vessel age, and the neural network is trained by the sample data of the sample set to obtain the optimal network parameters, thereby obtaining a nonlinear model with good blood vessel age prediction ability for the test sample. The input to the neural network is the above characteristics PWV, T1, T2, and BMI, and the output is the arterial age. The above neural network may be a multi-layer neural network, and the number of hidden layers and the number of nodes per layer may be any number that achieves a superior result.

The method for constructing an arterial blood vessel age estimation model provided by the embodiment of the present invention nonlinearly fits the blood vessel age by the reference user's reference arterial age and body information, and constructs an arterial blood vessel age estimation model, which can improve the estimation accuracy of the arterial blood vessel age. And the estimation of blood vessel age can be achieved by portable or wearable devices, which is accurate and suitable for home medical treatment, and can provide more useful physiological parameters for family health.

In order to reduce the measurement error when collecting the user's body information, it can be done as follows:

(1) Measure the user's upper arm blood pressure, PPG signal, and ECG signal. The measurement order of the upper arm blood pressure, the PPG signal and the ECG signal is: the first upper arm blood pressure measurement, the PPG signal and the ECG signal synchronous measurement, and the second upper arm blood pressure measurement. By taking two blood pressure measurements and then averaging, the error caused by the measurement can be reduced, and the average value of the blood pressure measurement according to the above sequence can better reflect the blood pressure value when measuring PPG and ECG. The collection position of the PPG is located at the finger end, and the ECG measurement can select the limb lead mode to facilitate user operation. The PPG signal and the ECG signal are measured simultaneously to facilitate the calculation of the features in subsequent steps.

(2) When the difference between the measured values of the upper arm blood pressure measurement is not greater than the preset deviation threshold, it is determined that the PPG signal and the ECG signal are the user's body information.

The preset deviation threshold may be determined according to the stability requirement of the blood pressure, for example, may be determined to be 15 mmHg. When the error of the two measurements is greater than 15 mmHg, the sample user may be in an unstable state or the measurement is in a problem, and the PPG signal of the measurement is discarded. And the ECG signal, measured again in the above order.

Considering that there is noise in the collected PPG signal and the ECG signal, in order to improve the accuracy of feature extraction, the performing weighted feature acquisition operation further includes the steps of performing denoising preprocessing on the PPG signal and the ECG signal. Denoising pre-processing of the original PPG and ECG signals collected may include de-baseline drift, power frequency interference, and other noise. For PPG, the noise is mainly the baseline drift. The high-pass filter with a cutoff frequency of 0.3HZ can be used to remove the baseline drift. For the ECG signal, the baseline drift can be removed by wavelet technique and then denoised by wavelet and Butterworth filter. The method removes other interference noise from the ECG.

In order to improve the estimation accuracy of the arterial blood vessel age model, more parameters may be added for model fitting. The individualized information of the sample user may also include one or more of the following: natural age, height, weight, and age of smoke. Among them, height and weight can be used to calculate the BMI index. The above personalized information can be obtained by user input. After adding the individualized information, the step of normalizing the above features to obtain a normalized feature may further include: normalizing each feature according to the natural age and the smoke age to obtain a normalized feature, that is, increasing Characteristics: The natural age and the smoke age are also normalized, and the subsequent steps of calculating the weight coefficient and the corresponding weighting feature are performed with all the features after adding the above features, and the model is fitted using the weighted features of all the above features. .

Through the above-mentioned natural age, PWV, T1, T2, BMI index, and smoking age, arterial age fitting can be performed according to various models, such as a one-dimensional linear model, multiple linear regression, multiple nonlinear regression, and the like.

After obtaining the arterial blood vessel age estimation model by the method of the present embodiment, it can be used to estimate the arterial blood vessel age of the target user, and the weighted feature is input to the arterial blood vessel age estimation model to obtain the arterial blood vessel age output of the target user. The above method of the embodiment further includes the steps of: collecting body information of the target user; performing a weighted feature acquisition operation on the target user; and inputting the weighted feature of the target user into an arterial blood vessel age estimation model corresponding to the gender of the target user to obtain the target user. The age of the arteries.

The steps of the acquisition step and the weighting feature acquisition step are the same as those in the process of constructing the above-mentioned arterial blood vessel age estimation model, and are not described herein again.

The method for constructing an arterial blood vessel age estimation model provided by the embodiment of the present invention can estimate the arterial blood vessel age of the target user after constructing the arterial blood vessel age estimation model, and the estimation is accurate and suitable for family medical treatment, and can provide more useful for family health. Physiological parameters.

Embodiment 2

The embodiment of the invention provides a method for constructing an arterial blood vessel age estimation model, which is described by taking the user's body information including age, PWV, T1, T2, BMI index and smoke age as an example. In the method provided by the embodiment, the selected parameters are reasonably designed according to the existing medical and mathematical theories, and all of them have a significant influence on the age of the user's arteries, and the detailed analysis is as follows:

(1) The age of blood vessels generally increases with the increase of natural age, but due to individual factors (such as smoking, diet, obesity, etc.), the increase of the two is not linear; (2) PWV is a direct response to arterial elasticity. Important indicators, the better the vascular elasticity, the slower the PWV, the harder the blood vessel is, the faster the PWV is. (3) The time delay between the main wave and the tremor wave in the pulse wave is directly related to the elasticity of the blood vessel. The better the elasticity, the time delay of the blood vessel T1 (4) The pulse wave rise time T2 reflects the time from the start of the blood to the maximum blood pressure. The longer the time, the greater the peripheral resistance, which indirectly indicates the greater the degree of vascular aging; (5) The higher the BMI index It is said that the more obese people are, the obesity is an important factor of cardiovascular disease; (6) Framingham study proves that smoking has serious harm to cardiovascular disease, and the greater the age of smoke, the greater the harm.

The method for constructing an arterial blood vessel age estimation model provided by this embodiment, see the flow diagram shown in FIG. 3, includes the following steps:

Step S302, recording basic data of the sample user.

The basic data includes age, gender, height, weight, arm span (both arms straight and shoulder level, measuring the distance L between the two middle finger tips) and the age of the smoke. In this embodiment, the male and female models are independently modeled, and the age of the samples is evenly distributed between 30 and 70 years old.

Step S304, acquiring the upper arm blood pressure, the PPG signal and the ECG signal of the sample user.

Step S306, performing denoising preprocessing on the PPG signal and the ECG signal.

Step S308, performing ECG and PPG key point detection. After obtaining the ECG and PPG after denoising preprocessing, key points are detected on the waveforms of the two.

Step S310, performing feature extraction on the ECG and the PPG. This feature includes the above PWV, T1, and T2.

In step S312, correlation coefficients L1, L2, L3, L4, L5, and L6 between blood vessel age and age, PWV, T1, T2, BMI index, and smoke age are automatically analyzed by correlation coefficients.

Among them, the correlation coefficient is automatically calculated by a pre-written program. L1～L6 are the correlation coefficient between blood vessel age and age, PWV, T1, T2, BMI index and tobacco age. The correlation coefficient L1 is taken as an example. The correlation coefficient between arterial age and age indicates the two. Correlation degree (0 ≤ L1 ≤ 1), when L1 = 0, there is no relationship between blood vessel age and age. When L1 = 1, it indicates that the two are completely related.

In step S314, the age, PWV, T1, T2, BMI index, and smoke age are respectively normalized. In the present embodiment, a, p, t1, t2, bmi and s are respectively used to represent the feature vector normalized by the above feature.

In step S316, the feature weights of age, PWV, T1, T2, BMI index, and smoke age are calculated. In the present embodiment, c _i (i = 1, ..., 6) is used to express the contribution weights of the age, PWV, T1, T2, BMI index, and smoke age for the blood vessel age estimation model.

Step S318, calculating a weighting feature. The normalized feature obtained in step S314 is multiplied by the feature weight obtained in step S316 to obtain a weighted feature vector. The weighted feature vector can be expressed as [a*c1, p*c2, t1*c3, t2*c4, bmi*c5, s*c6].

In step S320, the nonlinear model is fitted to obtain an artery blood vessel age estimation model.

In this embodiment, a multi-layer neural network can be used to fit the blood vessel age. The input characteristics of the multi-layer neural network are: age, PWV, T1, T2, BMI index, and smoke age, and the output is the estimated arterial age.

Step S322, input the weighted feature of the natural age, PWV, T1, T2, BMI index and the age of the target user to the above estimated model to obtain the estimated arterial age of the target user.

Embodiment 3

An embodiment of the present invention provides a device for constructing an arterial blood vessel age estimation model. Referring to the structure of the arterial blood vessel age estimation model construction device shown in FIG. 4, the sample collection module 10, the weighted feature acquisition module 20, and the training module 30 are provided. The functions of each module are as follows:

The sample collection module 10 is configured to collect reference arterial blood vessel age, physiological signals and individualized information of the sample user; the physiological signals include synchronized PPG signals and ECG signals, and the individualized information includes gender and abduction;

The weighted feature obtaining module 20 is configured to perform a weighted feature acquisition operation according to the gender for each sample user: extracting features according to the physiological signal and the individualized information; the features include: pulse wave conduction velocity, normalized heavy wave and The time delay of the main wave, the normalized pulse wave rise time and the BMI index; normalize each feature to obtain a normalized feature; calculate the weight of each feature according to the correlation coefficient between each feature and the reference arterial age Coefficient; calculating a weighted feature of each feature based on the normalized feature and the weighting coefficient;

The training module 30 is configured to train the neural network with the weighted features of each sample user and the reference arterial age as sample data to obtain an arterial age estimation model.

Referring to the structural diagram of the apparatus for constructing an arterial age estimation model shown in FIG. 5, the apparatus further includes:

The target collection module 40 is configured to collect physiological signals and individualized information of the target user;

The target weighted feature obtaining module 50 is configured to perform a weighted feature obtaining operation on the target user;

The age estimation module 60 is configured to input a weighted feature of the target user into an arterial blood age estimation model corresponding to the gender of the target user, to obtain an arterial blood vessel age of the target user.

The apparatus for constructing an arterial blood vessel age estimation model provided by the embodiment of the present invention nonlinearly fits the blood vessel age by using the reference arterial blood vessel age and body information of the sample user, and constructing an arterial blood vessel age estimation model can improve the estimation accuracy of the arterial blood vessel age. And the estimation of blood vessel age can be achieved by portable or wearable devices, which is accurate and suitable for home medical treatment, and can provide more useful physiological parameters for family health.

Embodiments of the present invention also provide a monitoring device that can be separately provided, and with a home medical device (such as an electronic sphygmomanometer) and a wearable device (for example, a wristband that can measure the fingertip photoelectric volume pulse wave and the electrocardiogram, exercise) A communication connection, such as a watch, may also be provided inside the above-mentioned home medical device or wearable device, thereby utilizing the hardware conditions of the above-described home medical device or wearable device. The monitoring device may include the arterial blood vessel age estimation model constructing device provided by the above embodiments.

Referring to the structural diagram of a monitoring device shown in FIG. 6, a processor 600 and a machine readable storage medium 601 are stored. The machine readable storage medium 601 stores machine executable instructions executable by the processor 600, and the processor 600 executes The machine executable instructions to implement the methods provided by the above embodiments.

The monitoring device shown in FIG. 6 further includes a bus 602 and a communication interface 603, and the processor 600, the communication interface 603, and the machine readable storage medium 601 are connected by a bus 602. The communication interface 603 can be communicably connected with the above-mentioned home medical device and the wearable device, and acquire physiological signals or individualized information, such as PPG, ECG, and the like collected by the home medical device and the wearable device.

The machine readable storage medium 601 may include a high speed random access memory (RAM), and may also include a non-volatile memory, such as at least one disk storage. The communication connection between the system network element and the at least one other network element may be implemented by using at least one communication interface 603 (which may be wired or wireless), and may use an Internet, a wide area network, a local network, a metropolitan area network, or the like. The bus 602 can be an ISA bus, a PCI bus, or an EISA bus. The bus can be divided into an address bus, a data bus, a control bus, and the like. For ease of representation, only one double-headed arrow is shown in Figure 6, but it does not mean that there is only one bus or one type of bus.

Processor 600 may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method may be completed by an integrated logic circuit of hardware in the processor 600 or an instruction in a form of software. The processor 600 may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP processor, etc.), or a digital signal processor (DSP). ), Application Specific Integrated Circuit (ASIC), Field-Programmable Gate Array (FPGA) or other programmable logic device, discrete gate or transistor logic device, discrete hardware component. The methods, steps, and logical block diagrams disclosed in the embodiments of the invention may be implemented or carried out. The general purpose processor may be a microprocessor or the processor or any conventional processor or the like. The steps of the method disclosed in connection with the embodiments of the present disclosure may be directly embodied by the hardware decoding processor, or by a combination of hardware and software modules in the decoding processor. The software module can be located in a conventional storage medium such as random access memory, flash memory, read only memory, programmable read only memory or electrically erasable programmable memory, registers, and the like. The storage medium is located in a machine readable storage medium 601, and the processor 600 reads information in the machine readable storage medium 601, in combination with hardware thereof, to perform the steps of the method of the foregoing embodiments, including construction of an arterial blood vessel age estimation model and use of the model Estimate the arterial age of the target user.

Embodiments of the present invention also provide a machine readable storage medium storing machine executable instructions that, when invoked and executed by a processor, cause the processor to implement the above The method of the embodiment.

A person skilled in the art can clearly understand that for the convenience and brevity of the description, the specific working process of the system and the device described above can refer to the corresponding process in the foregoing method embodiments, and details are not described herein again.

The above functions, if implemented in the form of software functional units and sold or used as stand-alone products, may be stored in a computer readable storage medium. Based on such understanding, the portion of the technical solution of the present disclosure that contributes in essence or to the prior art or the portion of the technical solution may be embodied in the form of a software product stored in a storage medium, including The instructions are used to cause a computer device (which may be a personal computer, server, or network device, etc.) to perform all or part of the steps of the above-described methods of various embodiments of the present disclosure. The foregoing storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, and the like. .

It should be noted that the above embodiments are merely specific embodiments of the present disclosure, and are not intended to limit the technical solutions of the present disclosure, and the scope of protection of the present disclosure is not limited thereto, although reference is made to the foregoing embodiments. The present disclosure has been described in detail, and those skilled in the art should understand that any one skilled in the art can still modify the technical solutions described in the foregoing embodiments or can easily think of the technical solutions disclosed in the foregoing embodiments. Variations, or equivalents to some of the technical features; and the modifications, variations, or substitutions of the present invention are not to be construed as departing from the spirit and scope of the embodiments of the present disclosure. Inside. Therefore, the scope of protection of the disclosure should be determined by the scope of the claims.

Claims

A method for constructing an arterial blood vessel age estimation model, comprising:

Collecting reference arterial age, physiological signals, and individualized information of the sample user; the physiological signals include synchronized PPG signals and ECG signals, the individualized information including gender and abduction;

For each of the sample users, performing a weighted feature acquisition operation according to gender: extracting features according to the physiological signal and the personalized information; the features include: pulse wave conduction velocity, normalized heavy wave and Time delay of the main wave, normalized pulse wave rise time and BMI index; normalization of each of the features to obtain a normalized feature; correlation coefficient according to each of the characteristics and the reference arterial age Calculating weight coefficients of each of the features separately; calculating weighting features of each of the features according to the normalized features and the weight coefficients;

The neural network is trained with the weighting feature of each of the sample users and the reference arterial age as sample data to obtain an arterial age estimation model.
The method according to claim 1, wherein the step of collecting physiological signals of a plurality of users of the same gender comprises:

Measuring an upper arm blood pressure, a PPG signal, and an ECG signal of a plurality of users of the same sex; the upper arm blood pressure, the PPG signal, and the ECG signal are measured in the following order: first upper arm blood pressure measurement, PPG signal and ECG signal simultaneous measurement, And the second upper arm blood pressure measurement;

When the difference between the measured values of the two upper arm blood pressure measurements is not greater than the preset deviation threshold, it is determined that the PPG signal and the ECG signal are user's body information.
The method according to claim 1, wherein the step of extracting features according to the physiological signal and the personalized information comprises:

Performing R wave peak point detection on the ECG signal, and performing pulse wave start point A, main wave B, tidal wave C, notch D, and heavy wave wave E detection on the PPG signal;

Dividing the pulse conduction distance by the average time delay of the pulse wave start point A and the R wave peak point to obtain the pulse wave conduction velocity;

Dividing the average time delay of the heavy wave E and the main wave B by an average heartbeat period to obtain a time delay of the normalized dicrotic wave and the main wave;

Dividing the average time delay of the main wave B and the starting pulse wave A by the average heartbeat period results in the normalized pulse wave rising branch time.
The method according to claim 1, wherein said calculating a weighting coefficient of each of said features according to a correlation coefficient of each of said characteristics and said reference arterial age comprises:

Correlation coefficient L i of the feature and the reference arterial age is respectively solved by correlation coefficient analysis;

The weighting coefficients c i of each of the features are respectively calculated according to the following formula:

Where n is the number of features.
The method according to claim 1, wherein the step of calculating a weighting feature of each of the features according to the normalized feature and the weighting coefficient comprises:

The normalized features are respectively multiplied by the corresponding weight coefficients to obtain weighted features of each of the features.
The method according to claim 1, wherein before the step of extracting features according to the body information, the method further comprises:

Denoising preprocessing is performed on the PPG signal and the ECG signal.
The method according to claim 1, wherein the personalized information further comprises one or more of the following: natural age, height, weight, and age of smoke;

The step of normalizing each of the features to obtain a normalized feature comprises: normalizing each of the features in combination with a natural age and a smoke age to obtain a normalized feature.
The method of claim 1 further comprising:

Collecting physiological signals and individualized information of the target user;

Performing the weighted feature acquisition operation on the target user;

The weighted feature of the target user is input to the arterial blood vessel age estimation model corresponding to the gender of the target user to obtain an arterial blood vessel age of the target user.
An apparatus for constructing an arterial blood vessel age estimation model, comprising:

a sample collection module, configured to collect reference arterial age, physiological signals, and individualized information of the sample user; the physiological signal includes a synchronized PPG signal and an ECG signal, the individualized information including gender and abduction;

a weighted feature acquisition module, configured to perform a weighted feature acquisition operation according to a gender for each of the sample users: extracting features according to the physiological signal and the personalized information; the features include: pulse wave conduction speed, Normalizing the time delay of the bobo wave and the main wave, normalizing the pulse wave rising branch time, and the BMI index; normalizing each of the features to obtain a normalized feature; according to each of the features and the reference Calculating a weight coefficient of each of the features by calculating a correlation coefficient of an arterial age; calculating a weighting feature of each of the features according to the normalized feature and the weighting coefficient;

And a training module, configured to train the neural network with the weighted feature of each of the sample users and the reference arterial age as sample data to obtain an arterial age estimation model.
The device according to claim 9, wherein the device further comprises:

a target acquisition module, configured to collect physiological signals and individualized information of the target user;

a target weighted feature obtaining module, configured to perform the weighted feature obtaining operation on the target user;

And an age estimation module, configured to input the weighted feature of the target user into the arterial blood vessel age estimation model corresponding to the gender of the target user, to obtain an arterial blood vessel age of the target user.