Intelligent blood pressure measurement method and system based on pulse wave method

Abstract

The present application relates to an intelligent blood pressure measurement method and system based on the pulse wave method, which belongs to the field of medical detection technology. The blood pressure measurement method includes: obtaining the pulse wave signal collected from the subject's body surface by the pulse wave signal acquisition device in real time; filtering and denoising the pulse wave signal to obtain the pre-processed pulse wave signal; extracting multiple characteristic parameters from the pre-processed pulse wave signal to obtain a characteristic parameter data set; analyzing the correlation between the characteristic parameters in the characteristic parameter data set and the blood pressure based on the characteristic value method, and extracting a sensitive characteristic value data set; based on a pre-established proportional relationship model, calculating a preliminary blood pressure value according to the sensitive characteristic value data set; inputting the preliminary blood pressure value and the sensitive characteristic value data set into a pre-trained neural network model to obtain an optimized final blood pressure value; and outputting and displaying the final blood pressure value through a display module. The present application can improve the accuracy and stability of blood pressure measurement.

Description

Technical Field

[0001] The present application relates to the field of medical detection technology, and in particular to an intelligent blood pressure measurement method and system based on a pulse wave method. Background Art

[0002] In the existing medical testing field, blood pressure measurement is one of the key parameters for evaluating human vital signs. Traditional blood pressure measurement methods, such as auscultation and mercury sphygmomanometer, rely on manual operation by professionals. Although these methods are widely used in clinical applications, they have the disadvantages of complex operation, strong subjectivity, and great interference from environmental factors. In addition, since these methods require professional operation, their application in home and portable monitoring is limited, and it is difficult to achieve autonomous measurement and long-term monitoring.

[0003] With the development of sensor technology and signal processing technology, the blood pressure measurement method based on pulse wave has gradually become a hot topic in research and application. This method uses photoelectric sensors to collect pulse wave signals and calculates blood pressure values ​​by analyzing the waveform and time characteristics of the pulse wave. The advantages of this method are its non-invasiveness and portability, which makes it suitable for daily health monitoring and home medical equipment.

[0004] However, the pulse wave signal is affected by many factors, such as environmental noise, body movement, and light source stability, which leads to significant differences in pulse wave characteristics between different individuals and in different scenarios, making it difficult to accurately measure a wide range of people. Therefore, how to ensure the accuracy and stability of blood pressure measurement to meet the application needs of different individuals and environments is an urgent problem to be solved. Summary of the invention

[0005] In order to improve the accuracy and stability of blood pressure measurement, the present application provides an intelligent blood pressure measurement method and system based on the pulse wave method.

[0006] In the first aspect, the present application provides an intelligent blood pressure measurement method based on the pulse wave method, which adopts the following technical solutions:

[0007] An intelligent blood pressure measurement method based on a pulse wave method, the blood pressure measurement method comprising:

[0008] Acquiring in real time the pulse wave signal collected from the subject's body surface by the pulse wave signal collection device;

[0009] Filtering and denoising the pulse wave signal to obtain a preprocessed pulse wave signal;

[0010] Extracting multiple characteristic parameters from the preprocessed pulse wave signal to obtain a characteristic parameter data set;

[0011] Analyze the correlation between the characteristic parameters in the characteristic parameter data set and the blood pressure based on the characteristic value method, and extract a sensitive characteristic value data set;

[0012] Based on a pre-established proportional relationship model, a preliminary blood pressure value is calculated according to the sensitive characteristic value data set;

[0013] Inputting the preliminary blood pressure value and the sensitive characteristic value data set into a pre-trained neural network model, optimizing the preliminary blood pressure value to obtain an optimized final blood pressure value;

[0014] The final blood pressure value is output and displayed through a display module.

[0015] By adopting the above technical solutions, the pulse wave signal is acquired in real time and filtered to ensure the high quality of the signal; the pertinence of the blood pressure calculation model is enhanced by extracting characteristic parameters and identifying sensitive characteristic values; the precise optimization of blood pressure values ​​is achieved by combining the proportional relationship model with the neural network model, which can provide accurate, real-time and non-invasive blood pressure measurement, and provide users with a convenient and reliable means of health monitoring.

[0016] Optionally, the step of extracting multiple characteristic parameters from the preprocessed pulse wave signal to obtain a characteristic parameter data set includes:

[0017] The preprocessed pulse wave signal is processed in segments according to a preset time window;

[0018] In each preset time window, the key waveform feature points of the pulse wave signal are detected; wherein the key waveform feature points include peak points and valley points;

[0019] Based on the peak points and valley points in the key waveform feature points, the waveform amplitude, rise time and fall time of the pulse wave signal are calculated;

[0020] The waveform amplitude, rise time and fall time of the pulse wave signal are standardized to obtain a characteristic parameter data set.

[0021] By adopting the above technical scheme, the pre-processed pulse wave signal is segmented into time windows, key feature points are detected, characteristic parameters are calculated and standardized, thereby forming a structured and unified characteristic parameter data set, ensuring the accuracy and consistency of the characteristic parameters, providing high-quality data input for the subsequent blood pressure calculation model, and significantly improving the accuracy and reliability of blood pressure measurement.

[0022] Optionally, based on a pre-established proportional relationship model, the step of calculating a preliminary blood pressure value according to the sensitive characteristic value data set includes:

[0023] Loading a pre-established proportional relationship model from system storage; wherein the proportional relationship model is obtained by regression analysis based on clinical experimental data;

[0024] The sensitive characteristic value data set is input into the proportional relationship model, and a preliminary blood pressure value is obtained by calculation based on the proportional coefficient and bias term of the proportional relationship model; wherein the preliminary blood pressure value includes a preliminary calculated systolic pressure value and diastolic pressure value.

[0025] By adopting the above technical solution, the pulse wave signal can be processed efficiently and accurately and the blood pressure estimation result can be output, which provides a good foundation for subsequent optimization and model adjustment. Through this step, the system can quickly perform preliminary blood pressure measurement while ensuring a certain accuracy, providing a convenient and reliable solution for practical applications.

[0026] Optionally, the blood pressure measurement method further includes a training step of the neural network model, and the training step includes:

[0027] Acquire a sample data set and perform data preprocessing; wherein the sample data set includes sample characteristic value data of the pulse wave signal, preliminary blood pressure sample values, and corresponding actual blood pressure measurement values, wherein the actual blood pressure measurement values ​​include the actually measured systolic pressure and diastolic pressure;

[0028] Dividing the sample data set into a training set, a validation set and a test set;

[0029] The training set is input into a pre-built neural network model for training, the gradient is calculated by a back propagation algorithm, and the weight and bias of the neural network model are updated; wherein the neural network model adopts a wavelet neural network model, the wavelet neural network model includes an input layer, a hidden layer and an output layer, and the activation function of the hidden layer adopts a wavelet basis function;

[0030] Validating the neural network model based on the validation set, evaluating the performance of the neural network model and adjusting model hyperparameters according to the validation results;

[0031] The prediction ability of the adjusted neural network model is tested based on the test set to obtain the trained neural network model.

[0032] By adopting the above technical solution and selecting a wavelet neural network model, it is possible to extract richer multi-scale features from the pulse wave signal and capture the complex nonlinear relationship between the pulse wave features and blood pressure; during the model training process, the network parameters are adjusted through the back propagation algorithm to optimize the error in the preliminary blood pressure value, and finally an accurate blood pressure prediction value is output, which not only improves the accuracy of blood pressure measurement, but also enhances the sensitivity and stability of the model to the input signal, and has broad application prospects.

[0033] Optionally, after the step of outputting and displaying the final blood pressure value through a display module, the method further includes:

[0034] Determining whether the final blood pressure value exceeds a normal blood pressure range;

[0035] If so, a blood pressure warning prompt is generated according to the final blood pressure value and sent to the corresponding user terminal.

[0036] By adopting the above technical solution, from blood pressure measurement, data display, abnormal blood pressure judgment and early warning prompts, to the transmission of early warning information, comprehensive monitoring and timely response to user health are achieved. This not only provides accurate blood pressure measurement, but also can automatically judge abnormal blood pressure and generate early warning information, greatly improving the intelligence of the blood pressure measurement system, effectively protecting the health and safety of users, and improving medical response efficiency.

[0037] In the second aspect, the present application provides an intelligent blood pressure measurement system based on the pulse wave method, which adopts the following technical solutions:

[0038] An intelligent blood pressure measurement system based on pulse wave method, the blood pressure measurement system comprising:

[0039] A signal acquisition module, used for acquiring in real time the pulse wave signal collected from the subject's body surface by the pulse wave signal acquisition device;

[0040] A signal preprocessing module, used for filtering and denoising the pulse wave signal to obtain a preprocessed pulse wave signal;

[0041] A feature extraction module is used to extract multiple feature parameters from the preprocessed pulse wave signal to obtain a feature parameter data set;

[0042] A sensitive feature data set generation module, used to analyze the correlation between the feature parameters in the feature parameter data set and the blood pressure based on the feature value method, and extract a sensitive feature value data set;

[0043] A preliminary blood pressure value calculation module, used to calculate the preliminary blood pressure value according to the sensitive characteristic value data set based on a pre-established proportional relationship model;

[0044] A final blood pressure value calculation module, used to input the preliminary blood pressure value and the sensitive characteristic value data set into a pre-trained neural network model, optimize the preliminary blood pressure value, and obtain an optimized final blood pressure value;

[0045] The output display module is used to output and display the final blood pressure value through a display module.

[0046] Optionally, the feature extraction module includes:

[0047] A segmentation processing unit, used for segmenting the preprocessed pulse wave signal according to a preset time window;

[0048] A waveform feature point detection unit, used to detect key waveform feature points of the pulse wave signal in each preset time window; wherein the key waveform feature points include peak points and valley points;

[0049] A characteristic parameter calculation unit, used for calculating the waveform amplitude, rise time and fall time of the pulse wave signal based on the peak points and valley points in the key waveform characteristic points;

[0050] The standardization processing unit is used to standardize the waveform amplitude, rise time and fall time of the pulse wave signal to obtain a characteristic parameter data set.

[0051] Optionally, the preliminary blood pressure value calculation module includes:

[0052] A proportional relationship model loading unit, used to load a pre-established proportional relationship model from a system storage; wherein the proportional relationship model is obtained by regression analysis based on clinical experimental data;

[0053] The preliminary blood pressure value calculation unit is used to input the sensitive characteristic value data set into the proportional relationship model, and calculate according to the proportional coefficient and bias term of the proportional relationship model to obtain a preliminary blood pressure value; wherein the preliminary blood pressure value includes the preliminary calculated systolic pressure value and diastolic pressure value.

[0054] In a third aspect, the present application provides a computer device, which adopts the following technical solution:

[0055] A computer device comprises a memory, a processor and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the method as described in the first aspect.

[0056] In a fourth aspect, the present application provides a computer-readable storage medium, which adopts the following technical solution:

[0057] A computer-readable storage medium stores a computer program that can be loaded by a processor and execute any one of the methods in the first aspect.

[0058] In summary, the present application includes at least one of the following beneficial technical effects: real-time acquisition of pulse wave signals and filtering processing to ensure high signal quality; enhanced pertinence of the blood pressure calculation model by extracting characteristic parameters and identifying sensitive characteristic values; and precise optimization of blood pressure values ​​by combining proportional relationship models with neural network models, thereby providing accurate, real-time and non-invasive blood pressure measurement, and providing users with a convenient and reliable means of health monitoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 This is a first flow chart of an intelligent blood pressure measurement method based on the pulse wave method according to one of the embodiments of the present application.

[0060] Figure 2 This is a second flow chart of an intelligent blood pressure measurement method based on the pulse wave method according to one of the embodiments of the present application.

[0061] Figure 3 This is a third flow chart of an intelligent blood pressure measurement method based on the pulse wave method according to one of the embodiments of the present application.

[0062] Figure 4 This is a fourth flow chart of an intelligent blood pressure measurement method based on the pulse wave method according to one of the embodiments of the present application.

[0063] Figure 5 This is a fifth flow chart of an intelligent blood pressure measurement method based on the pulse wave method according to one of the embodiments of the present application. DETAILED DESCRIPTION

[0064] In order to make the purpose, technical solutions and advantages of this application more clear, the following Figure 1-5 It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0065] As the demand for health monitoring increases in modern society, accurate and non-invasive blood pressure measurement methods are becoming increasingly important. Traditional blood pressure measurement methods, such as cuff-type sphygmomanometers, are commonly used in medical environments, but are inconvenient to use in daily life and may not provide continuous monitoring capabilities. Blood pressure measurement technology based on pulse wave signals has gradually become a research hotspot due to its non-invasive, continuous and convenient characteristics. However, pulse wave signals are susceptible to noise and environmental interference, and have a complex relationship with blood pressure, which makes the accuracy and stability of blood pressure measurement easily affected.

[0066] Based on this, an embodiment of the present application discloses an intelligent blood pressure measurement method based on the pulse wave method.

[0067] Reference Figure 1, an intelligent blood pressure measurement method based on pulse wave method, the blood pressure measurement method includes:

[0068] Step S101, acquiring in real time a pulse wave signal collected from the subject's body surface by a pulse wave signal collection device;

[0069] In one embodiment of the present application, the pulse wave signal acquisition device may use a photoelectric sensor to generate a pulse wave signal by detecting the blood flow on the subject's body surface. The pulse wave signal usually changes with the fluctuation of the heartbeat, and thus can reflect the dynamic characteristics of blood circulation in real time.

[0070] Specifically, photoplethysmography (PPG) can be used. As a common photoelectric sensor technology, pulse wave signals can be obtained by emitting light to the skin and detecting changes in the intensity of the reflected light.

[0071] Step S102, filtering and denoising the pulse wave signal to obtain a pre-processed pulse wave signal;

[0072] Among them, the pulse wave signal may be interfered by the environment (such as electromagnetic waves or body movement) during the acquisition process, so it is necessary to remove the noise through filtering and denoising processing (such as high-pass filtering, low-pass filtering and band-pass filtering) to retain the signal part that is meaningful to blood pressure analysis, providing a high-quality data foundation for subsequent feature extraction. For example, if there is low-frequency noise in the pulse wave signal (such as signal changes caused by breathing), these low-frequency components can be removed by a high-pass filter, thereby highlighting the main fluctuations of the pulse wave.

[0073] Step S103, extracting a plurality of characteristic parameters from the preprocessed pulse wave signal to obtain a characteristic parameter data set;

[0074] Among them, the pulse wave signal contains various information about blood pressure, and these key information can be structured through feature extraction. Specifically, the feature parameter data set includes waveform features and time features, such as the amplitude, rise time, and fall time of the waveform. These features directly or indirectly reflect the changes in blood pressure. For example, the peak time and valley time of the pulse wave can be extracted, and the difference between them can be calculated as a feature parameter reflecting the blood pressure fluctuation rate.

[0075] Step S104, analyzing the correlation between the characteristic parameters in the characteristic parameter data set and the blood pressure based on the characteristic value method, and extracting a sensitive characteristic value data set;

[0076] Among them, the eigenvalue method is to determine which characteristic parameters are most sensitive to changes in blood pressure through statistical analysis. By analyzing the correlation between characteristic parameters and blood pressure, those characteristic values ​​that can significantly reflect changes in blood pressure are selected to form a sensitive characteristic value data set to ensure that the input data of the model has a high correlation, thereby improving the accuracy of the blood pressure measurement model.

[0077] In some embodiments, principal component analysis (PCA) can be used to extract the principal component features that can most significantly represent the waveform characteristics from the multidimensional characteristic parameters of the pulse wave signal. The specific steps are: first extract multidimensional characteristic parameters from the pulse wave signal, including waveform amplitude, rise time, and fall time, and then standardize these characteristic parameters to eliminate the influence of different characteristic dimension differences on the analysis results; after standardization, calculate the covariance matrix of the standardized data matrix to evaluate the correlation between the features; next, by solving the eigenvalues ​​and eigenvectors of the covariance matrix, the variance contribution rate of each principal component is obtained; the principal components can be sorted according to the size of the eigenvalues, and the number of principal components to be retained can be determined by the cumulative variance contribution rate, and the first few principal components that can explain more than 90% of the variance are usually selected; then the original data is projected into these selected principal component spaces, thereby reducing the dimension of the data from the original multidimensional features to fewer principal component dimensions, and preliminarily determining the most representative, large-amplitude, or most sensitive characteristic parameters in the pulse wave signal.

[0078] It is understandable that the data after dimension reduction by principal component analysis still retains most of the data information, but reduces the influence of redundant features. Therefore, by using the principal components after dimension reduction as the input of the subsequent model, the computational efficiency and prediction accuracy of the model can be further optimized after verification, ensuring that the model can capture the most critical feature changes in the pulse wave signal when predicting blood pressure.

[0079] Step S105, based on the pre-established proportional relationship model, a preliminary blood pressure value is calculated according to the sensitive characteristic value data set;

[0080] Specifically, the proportional relationship model established by actual clinical data can be used to preliminarily calculate the blood pressure value based on the sensitive characteristic value data set. The proportional relationship model is an estimate based on the empirical relationship between the sensitive characteristic value and the blood pressure. For example, if it is known that the peak amplitude (A) is proportional to the systolic pressure and the rise time (T) is inversely proportional to the diastolic pressure, the preliminary blood pressure value can be estimated by the linear combination of these two parameters, so that the subsequent optimization process has a clear starting point.

[0081] Step S106, inputting the preliminary blood pressure value and the sensitive characteristic value data set into the pre-trained neural network model, optimizing the preliminary blood pressure value, and obtaining the optimized final blood pressure value;

[0082] Among them, the neural network model is trained through a large amount of training data and optimization algorithms. The model can identify the complex nonlinear relationship between pulse wave signal characteristics and blood pressure. By inputting preliminary blood pressure values ​​and sensitive characteristic value data sets, the preliminary blood pressure values ​​are finely optimized to improve the measurement accuracy. By correcting the errors in the preliminary blood pressure values, the subtle changes in the characteristic values ​​are used to improve the accuracy and stability of blood pressure measurement.

[0083] In some embodiments, a wavelet neural network (WNN) model can be used to take preliminary blood pressure values ​​and characteristic parameter data sets as input, and optimize the output of the final blood pressure values, including the final calculated systolic and diastolic blood pressure values, through the model's nonlinear mapping and multi-layer hidden layer analysis.

[0084] Step S107, outputting and displaying the final blood pressure value through a display module.

[0085] Among them, the optimized blood pressure value can be output through the display module, which is convenient for users to view their blood pressure status in real time. By real-time feedback of measurement results, the user experience is improved, making blood pressure monitoring more efficient and convenient; in some embodiments, the display module can be an electronic screen or transmitted to a smart device via Bluetooth / WiFi, which is convenient for data management and remote monitoring.

[0086] In the above implementation, the pulse wave signal is acquired in real time and filtered to ensure the high quality of the signal; the pertinence of the blood pressure calculation model is enhanced by extracting characteristic parameters and identifying sensitive characteristic values; the proportional relationship model and the neural network model are combined to achieve precise optimization of the blood pressure value, which can provide accurate, real-time and non-invasive blood pressure measurement, and provide users with a convenient and reliable health monitoring method.

[0087] Reference Figure 2 As an implementation of step S103, the step of extracting multiple characteristic parameters from the preprocessed pulse wave signal to obtain a characteristic parameter data set includes:

[0088] Step S201, segmenting the preprocessed pulse wave signal according to a preset time window;

[0089] The pulse wave signal is segmented according to a fixed time window, that is, divided into multiple periodic segments to ensure that each window contains a complete pulse wave cycle or multiple cycles, so as to facilitate capturing the changing characteristics of the signal within the time range. For example, assuming that the time window is set to 2 seconds, each segment of the signal contains two complete pulse wave cycles.

[0090] It can be understood that by segmenting the signal, it can ensure that the signal in each window is comparable, and the periodic characteristics of the pulse wave can be more effectively identified, providing a basis for subsequent feature point detection and calculation.

[0091] Step S202, detecting key waveform feature points of the pulse wave signal within each preset time window;

[0092] Among them, in each segmented time window, the key waveform feature points of the pulse wave are detected, including the peak point (wave crest) and the valley point (wave trough). The peak point usually represents the highest point of the pulse wave, and the valley point represents the lowest point. These feature points can reflect the amplitude and periodic changes of the pulse wave signal.

[0093] Illustratively, within a 2-second time window, the algorithm detects the highest and lowest values ​​of each pulse wave. For example, the peak point is at 1.2 seconds, and the valley point is at 0.8 seconds. By detecting the peak and valley points, the shape and basic characteristics of the waveform can be determined, providing an accurate benchmark for the subsequent calculation of feature parameters, thereby ensuring the accuracy and stability of feature extraction.

[0094] Step S203, based on the peak points and valley points in the key waveform feature points, the waveform amplitude, rise time and fall time of the pulse wave signal are calculated;

[0095] Specifically, using the peak points and valley points detected in the previous step, multiple key parameters of the pulse wave can be calculated. These key parameters are closely related to blood pressure and can provide accurate data support for subsequent blood pressure measurements.

[0096] Among them, the waveform amplitude is the difference between the peak point and the valley point, which indicates the amplitude of the pulse wave, and may include the peak amplitude and the valley amplitude. In the pulse wave signal, the peak amplitude corresponds to the peak of the pulse wave, which usually reflects the maximum blood pressure during heart contraction, and the valley amplitude corresponds to the valley of the pulse wave, which usually reflects the lowest blood pressure during heart diastole; the rise time is the time interval from the valley point to the peak point, which reflects the rise speed of the pulse wave; the fall time is the time interval from the peak point to the next valley point, which indicates the fall speed of the pulse wave.

[0097] For example, when the peak point is detected to be 1.2 seconds and the valley point is 0.8 seconds, the waveform amplitude is calculated as the difference between the peak and valley points (such as 120-80=40); the rise time is 1.2-0.8=0.4 seconds, which means the time required for the waveform to go from the lowest point to the highest point; the fall time is the time from the next valley point to the current peak point.

[0098] Step S204, normalizing the waveform amplitude, rise time and fall time of the pulse wave signal to obtain a characteristic parameter data set.

[0099] Among them, since the pulse wave signals of different individuals or different measurement environments may have dimension differences or numerical range differences, in order to make these characteristic parameters comparable and consistent, these characteristics need to be standardized. Standardization usually converts the data to a fixed range (such as 0 to 1) to ensure that subsequent calculations and analyses are not affected by dimension differences.

[0100] For example, assuming that the waveform amplitude has a value of 40, the rise time is 0.4 seconds, and the fall time is 0.5 seconds; through normalization, these values ​​can be converted into standardized values ​​within the range of 0.8, 0.6, and 0.75 to ensure that these parameters have a uniform scale in a neural network or other algorithm.

[0101] In the above implementation, the preprocessed pulse wave signal is subjected to time window segmentation, key feature point detection, feature parameter calculation and standardization, thereby forming a structured and unified feature parameter data set, ensuring the accuracy and consistency of the feature parameters, providing high-quality data input for the subsequent blood pressure calculation model, and significantly improving the accuracy and reliability of blood pressure measurement.

[0102] Reference Figure 3 As an implementation of step S105, the step of calculating the preliminary blood pressure value according to the sensitive characteristic value data set based on the pre-established proportional relationship model includes:

[0103] Step S301, loading a pre-established proportional relationship model from system storage;

[0104] Specifically, a proportional relationship model based on clinical experimental data is loaded from the system storage module. The model uses a regression analysis method to fit the pulse wave signal data of a large number of clinical samples with the actual blood pressure values ​​to obtain the empirical proportional relationship between the pulse wave characteristic parameters and blood pressure; the model contains multiple proportional coefficients and bias terms, which represent the relationship between different characteristic parameters and blood pressure.

[0105] Among them, regression analysis is used to describe the relationship between two or more variables. By analyzing the relationship between the pulse wave characteristics (such as peak amplitude and rise time) in the clinical data and the blood pressure values, the linear or nonlinear relationship model between each characteristic parameter and the blood pressure is obtained, and finally the proportional coefficient and bias term are obtained, which can be used to estimate the blood pressure of unseen pulse wave signals.

[0106] In some embodiments, the calculation formula of the proportional relationship model is, for example:

[0107] Systolic blood pressure = k1*A+b1, diastolic blood pressure = k2*1 / T+b2;

[0108] In the above formula, k1 and k2 are proportional coefficients, b1 and b2 are bias terms, A is the peak amplitude, and T is the rise time.

[0109] Step S302, input the sensitive characteristic value data set into the proportional relationship model, calculate according to the proportional coefficient and bias term of the proportional relationship model, and obtain the preliminary blood pressure value; wherein the preliminary blood pressure value includes the preliminary calculated systolic pressure value and diastolic pressure value.

[0110] Among them, the sensitive eigenvalue data set may include multiple parameters extracted from the pulse wave signal. According to the established model, only the key eigenvalues ​​that have a greater impact on blood pressure changes (such as peak amplitude and rise time) need to be input. These eigenvalues ​​are multiplied by the corresponding proportional coefficients in the model and adjusted by the bias term to finally output the preliminary blood pressure value. It should be noted that systolic and diastolic pressures are calculated by different combinations of eigenvalues ​​and proportional coefficients.

[0111] Specifically, a simple linear combination of key features such as peak amplitude and rise time is used, the proportional coefficient is used to adjust the influence of the characteristic parameters on blood pressure, and the bias term is used to offset the overall calculation to improve the model fit. Through this calculation step, the system can obtain preliminary systolic blood pressure (SBP) and diastolic blood pressure (DBP) values.

[0112] In the above implementation, by efficiently and accurately processing the pulse wave signal and outputting the blood pressure estimation result, a good foundation is provided for subsequent optimization and model adjustment. The system can quickly perform preliminary measurement of blood pressure while ensuring a certain accuracy, providing a convenient and reliable solution for practical applications.

[0113] Reference Figure 4 As an implementation of step S106, the blood pressure measurement method further includes a training step of a neural network model, and the training step includes:

[0114] Step S401, obtaining a sample data set and performing data preprocessing;

[0115] The sample data set includes sample characteristic value data of the pulse wave signal, preliminary blood pressure sample values, and corresponding actual blood pressure measurement values, and the actual blood pressure measurement values ​​include the actually measured systolic pressure and diastolic pressure;

[0116] Specifically, the sample data set is usually composed of clinical data, including a large number of historical pulse wave signals and the corresponding actual measured blood pressure values. Among them, the sample characteristic value data is the waveform amplitude, rise time, fall time, etc. extracted from the historical pulse wave signal, and the preliminary blood pressure sample value is the preliminary estimated blood pressure value calculated by the sample characteristic value data through the proportional model.

[0117] It is understandable that in order for the model to correctly learn the nonlinear relationship between pulse wave features and actual blood pressure, these data must undergo preprocessing steps such as denoising, standardization or normalization to enhance the input of the neural network.

[0118] Step S402, dividing the sample data set into a training set, a validation set and a test set;

[0119] The training set is used to learn the model parameters, the validation set is used to tune the hyperparameters and prevent overfitting during the training process, and the test set is used to finally evaluate the performance of the model. In some embodiments, the training set, validation set, and test set can be divided into 70%, 15%, and 15% ratios.

[0120] Step S403, inputting the training set into the pre-built neural network model for training, calculating the gradient through the back propagation algorithm, and updating the weights and biases of the neural network model;

[0121] The neural network model adopts a wavelet neural network model, which includes an input layer, a hidden layer and an output layer, and the activation function of the hidden layer adopts a wavelet basis function.

[0122] In the embodiment of the present application, the input layer is composed of the extracted sensitive feature value data set (such as waveform amplitude, rise time, fall time) and the preliminary blood pressure value. The hidden layer processes the characteristics of the pulse wave signal through multi-layer wavelet transform, gradually extracts the nonlinear characteristics sensitive to blood pressure, and the output layer is the final optimized systolic and diastolic pressure values. Among them, the use of wavelet basis function as the activation function in the hidden layer can perform multi-scale analysis on the input signal, process the nonlinear relationship between local features and global features, and enhance the network's ability to process complex signals.

[0123] In addition, during the model training process, the error between the predicted value and the actual blood pressure value is calculated through the back propagation algorithm. The gradient is calculated layer by layer through chain derivation, and the weight and bias of each layer are adjusted to continuously optimize the predictive ability of the model.

[0124] Step S404, verifying the neural network model based on the verification set, evaluating the performance of the neural network model and adjusting the model hyperparameters according to the verification results;

[0125] Specifically, during the training process, the validation set data is regularly input into the neural network model to evaluate the performance of the model. Through the validation set, the model's hyperparameters (such as learning rate, number of hidden layers, regularization parameters, etc.) can be adjusted in real time to give the model better generalization capabilities. In addition, the model's hyperparameters can be optimized through feedback from the validation set. If the loss on the validation set increases or the model's accuracy cannot be improved, the learning rate, number of nodes in the hidden layer, or regularization term can be adjusted to ensure that the model performs well on unseen data.

[0126] Step S405, testing the prediction ability of the adjusted neural network model based on the test set to obtain a trained neural network model.

[0127] Specifically, the pulse wave signal features and preliminary blood pressure values ​​in the test set are input into the neural network model, and the error between the blood pressure value output by the model and the actual blood pressure value is calculated. By evaluating the model output, it is determined whether the model has sufficient prediction accuracy in the actual application scenario.

[0128] It is understandable that after training and validation, the test set is used to evaluate the final performance of the model. The data in the test set is not involved in training and validation, so it can truly reflect the performance of the model on unseen data. The prediction ability of the model can be objectively measured through the evaluation indicators of the test set (such as mean square error MSE, mean absolute error MAE, etc.).

[0129] In the above implementation, by selecting a wavelet neural network model, richer multi-scale features can be extracted from the pulse wave signal, and the complex nonlinear relationship between the pulse wave features and blood pressure can be captured; during the model training process, the network parameters are adjusted through the back propagation algorithm to optimize the error in the preliminary blood pressure value, and finally an accurate blood pressure prediction value is output, which not only improves the accuracy of blood pressure measurement, but also enhances the sensitivity and stability of the model to the input signal, and has broad application prospects.

[0130] Reference Figure 5 As a further implementation of the intelligent blood pressure measurement method, after the step of outputting the final blood pressure value through the display module in step S107, the method further includes:

[0131] Step S501, determine whether the final blood pressure value exceeds the normal blood pressure range; if so, jump to step S502; if not, do not perform any operation.

[0132] The normal range of blood pressure is set according to medical standards. Generally, the systolic blood pressure of normal adults is in the range of 90-120 mmHg, and the diastolic blood pressure is in the range of 60-80 mmHg. In the embodiment of the present application, the system will automatically determine whether the displayed final blood pressure value exceeds these normal ranges, or whether it is at a critical value (such as high blood pressure or low blood pressure stage), and make corresponding processing.

[0133] Step S502: Generate a blood pressure warning prompt based on the final blood pressure value and send it to the corresponding user terminal.

[0134] Specifically, when the system detects that the final blood pressure value is outside the normal range, it will automatically generate an early warning and send it to the user terminal. This early warning can not only include basic warning information, but also provide detailed health advice based on the current measurement data, such as "Your systolic blood pressure is higher than the normal range, it is recommended to see a doctor immediately."

[0135] In the above implementation, from blood pressure measurement, data display, abnormal blood pressure judgment and early warning prompts, to the transmission of early warning information, comprehensive monitoring and timely response to the user's health are achieved, which not only can provide accurate blood pressure measurement, but also can automatically judge abnormal blood pressure and generate early warning information, greatly improving the intelligence of the blood pressure measurement system, effectively protecting the health and safety of users, and improving the efficiency of medical response.

[0136] The embodiment of the present application also discloses an intelligent blood pressure measurement system based on the pulse wave method.

[0137] An intelligent blood pressure measurement system based on pulse wave method, the blood pressure measurement system comprising:

[0138] A signal acquisition module, used for acquiring in real time the pulse wave signal collected from the subject's body surface by the pulse wave signal acquisition device;

[0139] A signal preprocessing module is used to filter and denoise the pulse wave signal to obtain a preprocessed pulse wave signal;

[0140] A feature extraction module is used to extract multiple feature parameters from the preprocessed pulse wave signal to obtain a feature parameter data set;

[0141] A sensitive feature data set generation module is used to analyze the correlation between the feature parameters in the feature parameter data set and the blood pressure based on the feature value method, and extract the sensitive feature value data set;

[0142] A preliminary blood pressure value calculation module, used to calculate the preliminary blood pressure value according to the sensitive characteristic value data set based on a pre-established proportional relationship model;

[0143] The final blood pressure value calculation module is used to input the preliminary blood pressure value and the sensitive characteristic value data set into the pre-trained neural network model, optimize the preliminary blood pressure value, and obtain the optimized final blood pressure value;

[0144] The output display module is used to output and display the final blood pressure value through the display module.

[0145] In the above implementation, a pulse wave signal is obtained from the subject's body surface, and after filtering, denoising and feature extraction, a sensitive characteristic value data set closely related to blood pressure is generated. Subsequently, the preliminarily estimated blood pressure value is optimized using a pre-established proportional relationship model in combination with a neural network algorithm to achieve accurate measurement of blood pressure. The intelligent blood pressure measurement system of the present application can not only effectively reduce noise interference and improve the accuracy and stability of blood pressure measurement, but also display the measurement results in real time through a display module, which is convenient for users to monitor their health and provides users with a convenient and reliable health management tool.

[0146] As an implementation of the feature extraction module, the feature extraction module includes:

[0147] A segmentation processing unit, used for segmenting the preprocessed pulse wave signal according to a preset time window;

[0148] A waveform feature point detection unit, used to detect key waveform feature points of the pulse wave signal in each preset time window; wherein the key waveform feature points include peak points and valley points;

[0149] A characteristic parameter calculation unit, used to calculate the waveform amplitude, rise time and fall time of the pulse wave signal based on the peak points and valley points in the key waveform characteristic points;

[0150] The standardization processing unit is used to standardize the waveform amplitude, rise time and fall time of the pulse wave signal to obtain a characteristic parameter data set.

[0151] In the above implementation, the preprocessed pulse wave signal is subjected to time window segmentation, key feature point detection, feature parameter calculation and standardization, thereby forming a structured and unified feature parameter data set, ensuring the accuracy and consistency of the feature parameters, providing high-quality data input for the subsequent blood pressure calculation model, and significantly improving the accuracy and reliability of blood pressure measurement.

[0152] As an implementation of the preliminary blood pressure value calculation module, the preliminary blood pressure value calculation module includes:

[0153] A proportional relationship model loading unit, used to load a pre-established proportional relationship model from system storage; wherein the proportional relationship model is obtained by regression analysis based on clinical experimental data;

[0154] The preliminary blood pressure value calculation unit is used to input the sensitive characteristic value data set into the proportional relationship model, and calculate according to the proportional coefficient and bias term of the proportional relationship model to obtain the preliminary blood pressure value; wherein the preliminary blood pressure value includes the preliminary calculated systolic pressure value and diastolic pressure value.

[0155] In the above implementation, the system can quickly perform preliminary measurement of blood pressure while ensuring a certain degree of accuracy, providing a convenient and reliable solution for practical applications.

[0156] An intelligent blood pressure measurement system based on the pulse wave method in an embodiment of the present application can implement any of the above-mentioned intelligent blood pressure measurement methods, and the specific working process of each module in the intelligent blood pressure measurement system can refer to the corresponding process in the above-mentioned method embodiment.

[0157] In the several embodiments provided in this application, it should be understood that the provided methods and systems can be implemented in other ways. For example, the system embodiments described above are only illustrative; for example, the division of a certain module is only a logical function division, and there may be other division methods in actual implementation, such as multiple modules can be combined or integrated into another system, or some features can be ignored or not executed.

[0158] The embodiment of the present application also discloses a computer device.

[0159] A computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, an intelligent blood pressure measurement method based on the pulse wave method as described above is implemented.

[0160] The embodiment of the present application also discloses a computer-readable storage medium.

[0161] A computer-readable storage medium stores a computer program that can be loaded by a processor and execute any one of the above-mentioned intelligent blood pressure measurement methods based on the pulse wave method.

[0162] Among them, computer-readable storage media can be any tangible medium that contains or stores a program, which can be used by or in conjunction with an instruction execution system, apparatus or device; the program code contained on the computer-readable medium can be transmitted using any appropriate medium, including but not limited to wireless, wire, optical cable, RF, etc., or any suitable combination of the above.

[0163] It should be noted that in the above embodiments, the description of each embodiment has different emphases, and for parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0164] The above are all preferred embodiments of the present application, and are not intended to limit the protection scope of the present application. Any feature disclosed in this specification (including the abstract and drawings), unless otherwise stated, can be replaced by other equivalent or alternative features with similar purposes. That is, unless otherwise stated, each feature is only an example of a series of equivalent or similar features.

Claims

1. An intelligent blood pressure measurement method based on pulse wave method, characterized in that: The blood pressure measurement method comprises: Acquiring in real time the pulse wave signal collected from the subject's body surface by the pulse wave signal collection device; Filtering and denoising the pulse wave signal to obtain a preprocessed pulse wave signal; Extracting multiple characteristic parameters from the preprocessed pulse wave signal to obtain a characteristic parameter data set; Analyze the correlation between the characteristic parameters in the characteristic parameter data set and the blood pressure based on the characteristic value method, and extract a sensitive characteristic value data set; Based on a pre-established proportional relationship model, a preliminary blood pressure value is calculated according to the sensitive characteristic value data set; Inputting the preliminary blood pressure value and the sensitive characteristic value data set into a pre-trained neural network model, optimizing the preliminary blood pressure value to obtain an optimized final blood pressure value; Outputting and displaying the final blood pressure value through a display module; Based on the pre-established proportional relationship model, the step of calculating the preliminary blood pressure value according to the sensitive characteristic value data set includes: Loading a pre-established proportional relationship model from system storage; wherein the proportional relationship model is obtained by regression analysis based on clinical experimental data; The sensitive characteristic value data set is input into the proportional relationship model, and a preliminary blood pressure value is obtained by calculation according to the proportional coefficient and the bias term of the proportional relationship model; wherein the preliminary blood pressure value includes a preliminary calculated systolic pressure value and a diastolic pressure value; the blood pressure measurement method further includes a training step of the neural network model, and the training step includes: Acquire a sample data set and perform data preprocessing; wherein the sample data set includes sample characteristic value data of the pulse wave signal, preliminary blood pressure sample values, and corresponding actual blood pressure measurement values, wherein the actual blood pressure measurement values ​​include the actually measured systolic pressure and diastolic pressure; Dividing the sample data set into a training set, a validation set and a test set; The training set is input into a pre-built neural network model for training, the error between the predicted value and the actual blood pressure value is calculated by a back propagation algorithm, the gradient is calculated layer by layer, and the weight and bias of the neural network model are updated; wherein the neural network model adopts a wavelet neural network model, the wavelet neural network model comprises an input layer, a hidden layer and an output layer, the input layer comprises sample feature value data and preliminary blood pressure sample values, and the activation function of the hidden layer adopts a wavelet basis function; Validating the neural network model based on the validation set, evaluating the performance of the neural network model and adjusting model hyperparameters according to the validation results; The prediction ability of the adjusted neural network model is tested based on the test set to obtain the trained neural network model.

2. The intelligent blood pressure measurement method based on pulse wave method according to claim 1, characterized in that: The steps of extracting multiple characteristic parameters from the preprocessed pulse wave signal to obtain a characteristic parameter data set include: The preprocessed pulse wave signal is processed in segments according to a preset time window; In each preset time window, the key waveform feature points of the pulse wave signal are detected; wherein the key waveform feature points include peak points and valley points; Based on the peak points and valley points in the key waveform feature points, the waveform amplitude, rise time and fall time of the pulse wave signal are calculated; The waveform amplitude, rise time and fall time of the pulse wave signal are standardized to obtain a characteristic parameter data set.

3. The intelligent blood pressure measurement method based on the pulse wave method according to any one of claims 1 to 2, characterized in that: After the step of outputting and displaying the final blood pressure value through a display module, the method further includes: Determining whether the final blood pressure value exceeds a normal blood pressure range; If so, a blood pressure warning prompt is generated according to the final blood pressure value and sent to the corresponding user terminal.

4. An intelligent blood pressure measurement system based on pulse wave method, characterized in that: Used to perform the intelligent blood pressure measurement method based on the pulse wave method according to any one of claims 1 to 3, the blood pressure measurement system comprising: A signal acquisition module, used for acquiring in real time the pulse wave signal collected from the subject's body surface by the pulse wave signal acquisition device; A signal preprocessing module, used for filtering and denoising the pulse wave signal to obtain a preprocessed pulse wave signal; A feature extraction module is used to extract multiple feature parameters from the preprocessed pulse wave signal to obtain a feature parameter data set; A sensitive feature data set generation module, used to analyze the correlation between the feature parameters in the feature parameter data set and the blood pressure based on the feature value method, and extract a sensitive feature value data set; A preliminary blood pressure value calculation module, used to calculate the preliminary blood pressure value according to the sensitive characteristic value data set based on a pre-established proportional relationship model; A final blood pressure value calculation module, used to input the preliminary blood pressure value and the sensitive characteristic value data set into a pre-trained neural network model, optimize the preliminary blood pressure value, and obtain an optimized final blood pressure value; The output display module is used to output and display the final blood pressure value through a display module.

5. The intelligent blood pressure measurement system based on pulse wave method according to claim 4, characterized in that: The feature extraction module comprises: A segmentation processing unit, used for segmenting the preprocessed pulse wave signal according to a preset time window; A waveform feature point detection unit, used to detect key waveform feature points of the pulse wave signal in each preset time window; wherein the key waveform feature points include peak points and valley points; A characteristic parameter calculation unit, used for calculating the waveform amplitude, rise time and fall time of the pulse wave signal based on the peak points and valley points in the key waveform characteristic points; The standardization processing unit is used to standardize the waveform amplitude, rise time and fall time of the pulse wave signal to obtain a characteristic parameter data set.

6. The intelligent blood pressure measurement system based on pulse wave method according to claim 4, characterized in that: The preliminary blood pressure value calculation module includes: A proportional relationship model loading unit, used to load a pre-established proportional relationship model from system storage; wherein the proportional relationship model is obtained by regression analysis based on clinical experimental data; The preliminary blood pressure value calculation unit is used to input the sensitive characteristic value data set into the proportional relationship model, and calculate according to the proportional coefficient and bias term of the proportional relationship model to obtain a preliminary blood pressure value; wherein the preliminary blood pressure value includes the preliminary calculated systolic pressure value and diastolic pressure value.

7. A computer device, characterized in that: The method comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the method according to any one of claims 1 to 3 when executing the program.

8. A computer-readable storage medium, characterized in that: A computer program is stored which can be loaded by a processor and execute the method according to any one of claims 1 to 3.

Images