Intelligent non-invasive real-time blood pressure monitoring device and method based on multiple modes

Abstract

The invention provides an intelligent non-invasive real-time blood pressure monitoring device and method based on multiple modalities, and relates to the field of intelligent sensors.The method comprises the steps that prediction is conducted according to a motion signal, a first error influence parameter and a contact influence parameter are obtained, error analysis is conducted according to skin color information and the contact influence parameter, and a second error influence parameter is obtained; performing blood pressure prediction according to the motion signal to obtain a first predicted blood pressure interval, performing user blood pressure prediction according to the PPG signal to obtain predicted blood pressure, and performing error compensation based on the first error influence parameter and the second error influence parameter to obtain a second predicted blood pressure interval; according to the first predicted blood pressure interval and the second predicted blood pressure interval, a blood pressure monitoring error coefficient is obtained through calculation, and predicted blood pressure is marked. The method aims at solving the problems that a traditional blood pressure monitoring method cannot effectively recognize interference of external factors, and consequently the comprehensiveness, accuracy and reliability of a blood pressure monitoring result are insufficient, and more accurate, comprehensive and reliable blood pressure monitoring data can be provided for a user.

Description

Technical Field

[0001] The present invention relates to the field of intelligent sensors, and particularly to an intelligent non-invasive blood pressure real-time monitoring device and method based on multi-modalities. Background Art

[0002] With the rapid development of intelligent wearable devices, especially the popularization of smart watches, new possibilities for blood pressure monitoring have been proposed based on non-contact and non-invasive sensing technologies.

[0003] As a non-invasive physiological signal measurement technology, the PPG sensor has been applied to blood pressure prediction. The PPG sensor obtains blood pressure-related information by detecting the blood flow changes in blood vessels on the skin surface. However, its signal is greatly interfered by external factors (such as movement, sweating, skin color, etc.). Traditional blood pressure monitoring methods fail to effectively identify these influencing factors, resulting in deviation and instability in data results and affecting the accuracy of blood pressure monitoring. Summary of the Invention

[0004] The purpose of the present invention is to provide an intelligent non-invasive blood pressure real-time monitoring device and method based on multi-modalities to solve the technical problem that traditional blood pressure monitoring methods cannot effectively identify external factor interferences, resulting in insufficient comprehensiveness, accuracy, and reliability of blood pressure monitoring results, including: In the first aspect, the present invention provides an intelligent non-invasive blood pressure real-time monitoring device based on multi-modalities. The device includes a PPG sensor and a motion sensor, and further includes: a data acquisition module for collecting PPG signals and motion signals from a target user through the PPG sensor and the motion sensor, obtaining the PPG signals and motion signals, and collecting the skin color information of the target user; an error influence analysis module for performing sensing error influence analysis and contact influence parameter prediction based on the motion signal to obtain a first sensing error influence parameter and a contact influence parameter, and performing sensing error influence analysis based on the skin color information and the contact influence parameter to obtain a second sensing error influence parameter; a user blood pressure prediction module for predicting the user's blood pressure based on the motion signal to obtain a first predicted blood pressure range, predicting the user's blood pressure based on the PPG signal to obtain a predicted blood pressure, and performing error compensation based on the first sensing error influence parameter and the second sensing error influence parameter to obtain a second predicted blood pressure range; an error coefficient annotation module for calculating a blood pressure monitoring error coefficient based on the first predicted blood pressure range and the second predicted blood pressure range, and annotating the predicted blood pressure as the blood pressure monitoring result.

[0005] Preferably, the intelligent non-invasive blood pressure real-time monitoring device based on multi-modalities further includes: collecting PPG signals and motion signals of a target user within a recent preset time window through a PPG sensor and a motion sensor to obtain PPG signals and motion signals, wherein the motion sensor is a gyroscope; and acquiring the skin color information of the target user collected and recorded within a historical time period.

[0006] Preferably, the intelligent non-invasive blood pressure real-time monitoring device based on multi-modalities further includes: collecting a set of sample motion signals according to the monitoring data within a historical time period, and collecting the error magnitudes between different sample motion signals and the blood pressure predicted based on the PPG signal in the non-motion state, which are labeled as sample first sensing error influence parameters to obtain a set of sample first sensing error influence parameters; collecting the skin sweating parameters of the user under different sample motion signals, which are labeled as sample contact influence parameters to obtain a set of sample contact influence parameters; constructing a motion sensing error influence prediction branch and a contact influence prediction branch by using machine learning; using the set of sample motion signals as input features, and respectively using the set of sample first sensing error influence parameters and the set of sample contact influence parameters as output features to perform supervised training and testing on the motion sensing error influence prediction branch and the contact influence prediction branch until convergence; and inputting the motion signals into the motion sensing error influence prediction branch and the contact influence prediction branch respectively to predict and output the first sensing error influence parameter and the contact influence parameter.

[0007] Preferably, the intelligent non-invasive blood pressure real-time monitoring device based on multi-modalities further includes: collecting a set of sample skin color information and a set of sample contact influence parameters according to the monitoring data within a historical time period; collecting the error magnitudes between different sample skin color information and sample contact influence parameters and the blood pressure predicted based on the PPG signal compared with the preset skin color and the preset contact influence parameter, which are labeled as sample second sensing error influence parameters to obtain a set of sample second sensing error influence parameters; constructing a skin color sensing error influence prediction branch based on machine learning; using the set of sample skin color information, the set of sample contact influence parameters and the set of sample second sensing error influence parameters to perform supervised training and testing on the skin color sensing error influence prediction branch until convergence; and inputting the skin color information and the contact influence parameter into the skin color sensing error influence prediction branch to predict and output the second sensing error influence parameter.

[0008] Preferably, the intelligent non-invasive blood pressure real-time monitoring device based on multi-modalities further includes: collecting a set of sample motion signals according to the blood pressure monitoring data of the target user within a historical time period, and collecting the blood pressure ranges of the target user under different sample motion signals, and obtaining a set of sample first predicted blood pressure ranges through annotation; constructing a motion blood pressure prediction branch based on machine learning; using the set of sample motion signals and the set of sample first predicted blood pressure ranges to perform supervised training and testing on the motion blood pressure prediction branch until convergence; inputting the motion signals into the motion blood pressure prediction branch, and predicting and outputting to obtain a first predicted blood pressure range.

[0009] Preferably, the intelligent non-invasive blood pressure real-time monitoring device based on multi-modalities further includes: inputting the PPG signal into a pre-trained optoelectronic blood pressure prediction branch, and predicting and outputting to obtain a predicted blood pressure, wherein the optoelectronic blood pressure prediction branch is constructed based on machine learning and is supervised and trained to convergence by using a set of sample PPG signals and a set of sample blood pressures of the target user within a historical time period; calculating a sensing error influence parameter according to the first sensing error influence parameter and the second sensing error influence parameter; using the sensing error influence parameter to perform error compensation on the predicted blood pressure to obtain a second predicted blood pressure range.

[0010] Preferably, the intelligent non-invasive blood pressure real-time monitoring device based on multi-modalities further includes: obtaining the intersection of the first predicted blood pressure range and the second predicted blood pressure range to obtain an intersection blood pressure range; respectively calculating the ratios of the intersection blood pressure range to the first predicted blood pressure range and the second predicted blood pressure range, and calculating to obtain a first blood pressure monitoring error coefficient and a second blood pressure monitoring error coefficient; calculating a blood pressure monitoring error coefficient according to the first blood pressure monitoring error coefficient and the second blood pressure monitoring error coefficient; using the blood pressure monitoring error coefficient to annotate the predicted blood pressure to obtain a blood pressure monitoring result.

[0011] Second aspect, the present invention also provides a multi-modal based intelligent non-invasive blood pressure real-time monitoring method, including: collecting PPG signals and motion signals of a target user through a PPG sensor and a motion sensor to obtain PPG signals and motion signals, and collecting skin color information of the target user; performing sensing error impact analysis and contact impact parameter prediction based on the motion signals to obtain a first sensing error impact parameter and a contact impact parameter, and performing sensing error impact analysis based on the skin color information and the contact impact parameter to obtain a second sensing error impact parameter; performing user blood pressure prediction based on the motion signals to obtain a first predicted blood pressure range, performing user blood pressure prediction based on the PPG signals to obtain a predicted blood pressure, and performing error compensation based on the first sensing error impact parameter and the second sensing error impact parameter to obtain a second predicted blood pressure range; calculating a blood pressure monitoring error coefficient based on the first predicted blood pressure range and the second predicted blood pressure range, and annotating the predicted blood pressure as a blood pressure monitoring result.

[0012] The embodiments of the present invention have the following advantages: By collecting PPG signals and motion signals of a target user to obtain PPG signals and motion signals, and collecting skin color information of the target user; then performing sensing error impact analysis and contact impact parameter prediction based on the motion signals to obtain a first sensing error impact parameter and a contact impact parameter, and performing sensing error impact analysis based on the skin color information and the contact impact parameter to obtain a second sensing error impact parameter; then performing user blood pressure prediction based on the motion signals to obtain a first predicted blood pressure range; on the other hand, performing user blood pressure prediction based on the PPG signals to obtain a predicted blood pressure, and performing error compensation based on the first sensing error impact parameter and the second sensing error impact parameter to obtain a second predicted blood pressure range; finally calculating a blood pressure monitoring error coefficient based on the first predicted blood pressure range and the second predicted blood pressure range, and annotating the predicted blood pressure as a blood pressure monitoring result. That is to say, by combining multi-modal sensor data for sensing error analysis, it is possible to accurately identify and quantify sensing errors caused by different factors, improving the accuracy and reliability of obtaining sensing errors; then annotating the predicted blood pressure according to the error analysis result can provide feedback information about the measurement reliability for the user, helping the user effectively identify the reliability of the predicted blood pressure, so as to be able to provide more accurate, comprehensive and reliable blood pressure monitoring data for the user. Description of the Drawings

[0013] Figure 1 It is a schematic structural diagram of an intelligent non-invasive blood pressure real-time monitoring device based on multi-modal of the present invention; Figure 2 It is a step flow chart of an intelligent non-invasive blood pressure real-time monitoring method based on multi-modal of the present invention.

[0014] Description of reference numerals: Data acquisition module 11, error impact analysis module 12, user blood pressure prediction module 13, error coefficient annotation module 14. Detailed implementation manners

[0015] By providing an intelligent non-invasive blood pressure real-time monitoring device and method based on multi-modalities, the present invention solves the technical problem that traditional blood pressure monitoring methods cannot effectively identify external factor interferences, resulting in insufficient comprehensiveness, accuracy and reliability of blood pressure monitoring results. By combining multi-modal sensor data for sensing error analysis, it is possible to accurately identify and quantify sensing errors caused by different factors, improving the accuracy and reliability of obtaining sensing errors; then, according to the error analysis results, the predicted blood pressure is annotated, which can provide feedback information on the measurement reliability for users, helping users effectively identify the reliability of the predicted blood pressure, so as to be able to provide users with more accurate, comprehensive and reliable blood pressure monitoring data.

[0016] Next, the technical solutions in the present invention will be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments of the present invention. It should be understood that the present invention is not limited by the exemplary embodiments described herein. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the scope of protection of the present invention. Additionally, it should be noted that for the sake of description, only parts related to the present invention are shown in the accompanying drawings rather than all of them.

[0017] Embodiment 1. Please refer to the attached Figure 1 drawings. The present invention provides an intelligent non-invasive blood pressure real-time monitoring device based on multi-modalities. The device includes a PPG sensor and a motion sensor, and further includes: A data acquisition module 11, configured to collect PPG signals and motion signals of a target user through the PPG sensor and the motion sensor, obtain the PPG signals and the motion signals, and collect the skin color information of the target user.

[0018] Furthermore, the data acquisition module 11 is further configured to: Collect PPG signals and motion signals of a target user within a recent preset time window through the PPG sensor and the motion sensor, obtain the PPG signals and the motion signals, wherein the motion sensor is a gyroscope; obtain the skin color information of the target user collected and recorded during a historical time.

[0019] Specifically, the present invention relates to a multi-modal intelligent non-invasive blood pressure real-time monitoring device, specifically a blood pressure monitoring device combining a PPG sensor and a motion sensor, which can be assembled on wearable devices such as smart watches, capable of real-time monitoring of the user's blood pressure, and through the analysis of multi-modal data, improving the accuracy, comprehensiveness and reliability of blood pressure monitoring; among them, the PPG (photoplethysmography) sensor is used to detect the blood flow changes in the blood vessels on the surface of the user's skin. By emitting a light source to irradiate the skin surface and receiving the reflected light signal, the PPG sensor can analyze the blood flow changes, so as to obtain the pulse wave information related to blood pressure, which is one of the core technologies of non-invasive blood pressure monitoring. The motion sensor is a gyroscope, which is used to real-time monitor the user's motion state. Among them, changes in the motion state, such as walking, running, standing still or other activities, will affect the blood pressure measurement results. By collecting motion data, the impact of motion on blood pressure prediction can be analyzed, thereby improving the accuracy of blood pressure prediction. The PPG sensor and the motion sensor are integrated in wearable devices such as smart watches, which can ensure that the device can be worn on the user at any time for non-invasive blood pressure monitoring, avoiding the many inconveniences and discomforts of the traditional measurement method based on an airbag sphygmomanometer.

[0020] First, through the PPG sensor and the motion sensor, collect the PPG signal and the motion signal of the target user within the recently preset time window (which can be set according to the actual monitoring scenario, such as within the last 1 minute), that is, collect the optoelectronic signal of the blood flow on the surface of the target user's skin through the PPG sensor (photoplethysmography sensor). This sensor irradiates the skin by emitting a light source and receives the reflected light signal, and analyzes the blood flow situation according to the change of the reflected light; collect the motion state of the target user through the motion sensor (such as a gyroscope). The gyroscope sensor can measure the angular velocity and direction change of the user and provide data related to the user's activity. These data can help analyze the impact of motion on blood pressure prediction, especially in the dynamic state. Obtain the PPG signal and the motion signal through data collection. Among them, the PPG signal is a signal that monitors the blood flow change in the blood vessel by an optical method, and the motion signal includes the user's exercise intensity, exercise mode (such as walking, running, etc.). Exercise will affect the heart rate and thus affect the accuracy of blood pressure prediction.

[0021] On the other hand, obtain the skin color information of the target user collected and recorded within the historical time (such as within the last three months), that is, obtain the skin color information of the target user for the last time, such as light skin color, medium skin color and dark skin color, etc. The skin image can be captured by a camera, and the RGB value of the skin area can be extracted for skin color identification. Among them, the skin color will affect the absorption rate of the PPG signal. Dark skin color may cause stronger light absorption, thus affecting the transmission of the optoelectronic signal and the accuracy of blood pressure prediction.

[0022] An error impact analysis module 12, configured to perform sensing error impact analysis and contact impact parameter prediction based on the motion signal, obtain a first sensing error impact parameter and a contact impact parameter, and perform sensing error impact analysis based on the skin color information and the contact impact parameter to obtain a second sensing error impact parameter.

[0023] Furthermore, the error impact analysis module 12 is further configured to: Collect a sample motion signal set according to the monitoring data within a historical time period, and collect the error magnitudes between different sample motion signals and the blood pressure predicted based on the PPG signal in the non-motion state, which are labeled as sample first sensing error impact parameters, to obtain a sample first sensing error impact parameter set; collect the skin sweating parameters of the user under different sample motion signals, which are labeled as sample contact impact parameters, to obtain a sample contact impact parameter set; use machine learning to construct a motion sensing error impact prediction branch and a contact impact prediction branch; use the sample motion signal set as input features, and use the sample first sensing error impact parameter set and the sample contact impact parameter set as output features respectively, to perform supervised training and testing on the motion sensing error impact prediction branch and the contact impact prediction branch until convergence; input the motion signal into the motion sensing error impact prediction branch and the contact impact prediction branch respectively, and predict and output to obtain a first sensing error impact parameter and a contact impact parameter.

[0024] Specifically, according to the monitoring data within a historical time period (such as within the most recent half year), collect a plurality of sample motion signals to construct a sample motion signal set, and these data include different types of motion types (such as walking, running, etc.) and motion intensities; then, collect the error magnitudes between different sample motion signals and the blood pressure predicted based on the PPG signal in the non-motion state, for example, obtain motion data through motion sensors (such as gyroscopes, accelerometers, etc.), and distinguish the motion state and the non-motion state according to the motion intensity or state; then respectively obtain the predicted blood pressure based on the PPG signal in the motion state and the non-motion state, and calculate the error magnitude with the predicted blood pressure based on the PPG signal in the non-motion state as the reference, where the error magnitude is the absolute value of the blood pressure difference between the predicted blood pressure in the motion state and the predicted blood pressure in the non-motion state divided by the predicted blood pressure in the non-motion state. For example, assume the predicted blood pressure in the motion state is 105 and the predicted blood pressure in the non-motion state is 100, then the error magnitude is (105 - 100) / 100 which is equal to 5%, and label the error magnitude as the sample first sensing error impact parameter to obtain a sample first sensing error impact parameter set, where the sample motion signal and the sample first sensing error impact parameter are in one-to-one correspondence.

[0025] On the other hand, collect the skin sweating parameters of the user under different sample motion signals. Use a sweat sensor to directly measure the flow rate or humidity of sweat, such as the sweat flow rate per unit time (e.g., the amount of sweat per minute, with the unit of milliliters per minute). For example, during walking, the sweat amount of a certain user is 0.2 milliliters per minute, and during running, it is 0.5 milliliters per minute, and label it as the sample contact influence parameter to obtain a set of sample contact influence parameters. Among them, sweating will change the contact quality between the skin surface and the PPG sensor. Especially when sweat accumulates, it will cause changes such as light scattering and absorption, thereby affecting the quality and accuracy of the PPG signal.

[0026] Next, use machine learning to construct a motion sensing error influence prediction branch and a contact influence prediction branch. The motion sensing error influence prediction branch is used to predict the sensing error caused by motion. By training the relationship between the motion signal and the blood pressure error, this branch can predict the error range of blood pressure prediction according to the user's motion state; the contact influence prediction branch is used to predict the skin sweating parameters according to the user's motion state. For example, construct the motion sensing error influence prediction branch and the contact influence prediction branch based on the BP neural network. Among them, the motion sensing error influence prediction branch and the contact influence prediction branch are BP neural network models in machine learning that can be iteratively optimized, including an input layer, multiple hidden layers, and an output layer. Among them, the input data of the input layer of the motion sensing error influence prediction branch is the motion signal, and the output data is the first sensing error influence parameter. The input data of the input layer of the contact influence prediction branch is the motion signal, and the output data is the contact influence parameter (skin sweating parameter).

[0027] Further, the sample motion signal set, the sample first sensing error influence parameter set, and the sample contact influence parameter set are used as training data and divided into a training set and a test set according to a predetermined ratio. Usually, the training set accounts for 85% and the test set accounts for 15%. Then, with the sample motion signal as the input feature, the sample first sensing error influence parameter as the supervision feature of the motion sensing error influence prediction branch, and the sample contact influence parameter as the supervision feature of the contact influence prediction branch, the training set and the test set are used to perform supervised training and testing on the motion sensing error influence prediction branch and the contact influence prediction branch. During the training process, each batch of training data will undergo forward propagation through the neural network to generate prediction results. Then, by calculating the loss function (such as mean square error), the error is backpropagated to each layer of the neural network to update the weight parameters in the neural network. The model parameters are continuously optimized through multiple iterations until the training error converges. During the test process, by evaluating on the test set, the generalization ability of the model is verified. When the training error tends to be stable and the error on the test set also reaches the predetermined tolerance range, the training process is considered to converge, and the trained motion sensing error influence prediction branch and contact influence prediction branch are obtained.

[0028] Finally, the motion signal is respectively input into the motion sensing error influence prediction branch and the contact influence prediction branch for prediction, and the first sensing error influence parameter and the contact influence parameter are output. By using machine learning to construct the motion sensing error influence prediction branch and the contact influence prediction branch for prediction, the intelligent level of the prediction can be improved, and thus the accuracy and efficiency of the prediction results can be improved.

[0029] Further, the error influence analysis module 12 is further configured to: According to the monitoring data within the historical time, collect the sample skin color information set and the sample contact influence parameter set; collect the error amplitude of predicting blood pressure based on the PPG signal compared with the preset skin color and the preset contact influence parameter for different sample skin color information and sample contact influence parameters, and label it as the sample second sensing error influence parameter to obtain the sample second sensing error influence parameter set; based on machine learning, construct the skin color sensing error influence prediction branch; use the sample skin color information set, the sample contact influence parameter set, and the sample second sensing error influence parameter set to perform supervised training and testing on the skin color sensing error influence prediction branch until convergence; input the skin color information and the contact influence parameter into the skin color sensing error influence prediction branch, and predict and output to obtain the second sensing error influence parameter.

[0030] Specifically, according to the monitoring data within a historical time period (such as within the most recent three months), a set of sample skin color information and a set of sample contact influence parameters are collected. Among them, the skin color information can be obtained through wearable devices or manual input, and the skin color information of each sample can be represented by numerical features, such as skin color RGB values; the contact influence parameter (sweating amount) is measured and recorded by sensors. Then, the predicted blood pressure based on the PPG signal under different sample skin color information and sample contact influence parameters is collected, as well as the predicted blood pressure based on the PPG signal under a preset skin color (selecting a standard skin color for comparison, such as medium skin color) and a preset contact influence parameter (standard sweating amount). Then, the error margin between the blood pressure monitoring values in the sample state and the standard state is calculated, which is the ratio of the absolute value of the difference in blood pressure monitoring to the blood pressure monitoring value in the standard state, and is labeled as the sample's second sensing error influence parameter, obtaining a set of sample second sensing error influence parameters.

[0031] Based on machine learning, a skin color sensing error influence prediction branch is constructed. The skin color sensing error influence prediction branch is used to predict the influence of skin color and sweating amount on the PPG sensor error through a machine learning model. For example, a skin color sensing error influence prediction branch is constructed based on a BP neural network, including an input layer, multiple hidden layers, and an output layer. The input data of the input layer is skin color information and contact influence parameters, and the output data of the output layer is the second sensing error influence parameter. Further, the set of sample skin color information, the set of sample contact influence parameters, and the set of sample second sensing error influence parameters are used as sample data and divided into a sample training set and a sample test set according to a certain ratio.

[0032] Further, taking the sample skin color information and the sample contact influence parameter as inputs, and using the sample second sensing error influence parameter as the supervision, the skin color sensing error influence prediction branch is supervised and trained and tested using the sample training set and the sample test set until convergence. During the training process, the neural network first performs forward propagation. The input data (skin color information and contact influence parameter) is passed into the neural network through the input layer. The input data passes through each layer (from the input layer to the hidden layer and then to the output layer), and after calculations of weighted sum and activation function, the predicted value is finally output. Then, the loss function is calculated using the result of forward propagation and the actual value (the second sensing error influence parameter). The loss function measures the gap between the predicted value and the actual value. Then, the backpropagation algorithm is used to calculate the gradient of the loss function with respect to each weight. The gradient descent algorithm is used to optimize the weights and update the weights of the neural network to minimize the loss. An optimization algorithm (such as the Adam optimizer or SGD) is used to adjust the weights and biases in the neural network so that the model output is closer to the true value in each iteration. The processes of forward propagation, loss calculation, backpropagation, and weight update are repeatedly executed until the network converges. The convergence criterion is usually that the change in the loss function is very small or the set number of iterations is reached. During the testing process, an unseen test set is used to evaluate the generalization ability of the model. The data (skin color information, contact influence parameter) in the test set is input into the trained neural network, and the predicted second sensing error influence parameter is output. The predicted results of the test set are compared with the actual results, and the performance metrics of the model are calculated. If the test results meet the expected effect, the trained skin color sensing error influence prediction branch is obtained.

[0033] Finally, the skin color information and the contact influence parameter are input into the skin color sensing error influence prediction branch for prediction, and the second sensing error influence parameter is output. The second sensing error influence parameter represents the magnitude of the error in the PPG signal caused by the skin color information and the contact influence, and can accurately identify and quantify the errors caused by skin color and contact factors.

[0034] The user blood pressure prediction module 13 is used to perform user blood pressure prediction based on the motion signal to obtain the first predicted blood pressure range, perform user blood pressure prediction based on the PPG signal to obtain the predicted blood pressure, and perform error compensation based on the first sensing error influence parameter and the second sensing error influence parameter to obtain the second predicted blood pressure range.

[0035] Furthermore, the user blood pressure prediction module 13 is further used for: According to the blood pressure monitoring data of the target user within the historical time, collect a set of sample motion signals, and collect the blood pressure ranges of the target user under different sample motion signals, and label to obtain a set of sample first predicted blood pressure ranges; based on machine learning, construct a motion blood pressure prediction branch; use the set of sample motion signals and the set of sample first predicted blood pressure ranges to perform supervised training and testing on the motion blood pressure prediction branch until convergence; input the motion signal into the motion blood pressure prediction branch, and predict and output to obtain the first predicted blood pressure range.

[0036] Specifically, first, according to the blood pressure monitoring data of the target user within the historical time (such as within the recent three months), collect a set of sample motion signals (motion type, motion intensity, etc.); then, collect the blood pressure ranges of the target user under different sample motion signals (including the low blood pressure range and the high blood pressure range), and label them as the sample first predicted blood pressure ranges, to obtain a set of sample first predicted blood pressure ranges, where each sample records the change range of the blood pressure (i.e., the blood pressure range, including the low blood pressure range and the high blood pressure range) of the target user under a specific motion state.

[0037] Next, construct a motion blood pressure prediction branch based on machine learning, that is, use a machine learning model (such as a regression model or a neural network) to construct a motion blood pressure prediction branch for predicting the blood pressure range of the user according to the motion state. For example, construct a motion blood pressure prediction branch based on a generative adversarial network. The generative adversarial network is a deep learning model composed of two neural networks (a generator and a discriminator), and generates realistic data through adversarial training. Among them, the task of the generator is to generate a predicted blood pressure range according to the input motion signal. The generator tries to generate real blood pressure range data to make it look like real data; the task of the discriminator is to judge whether the input blood pressure range data is real or generated. The discriminator makes a judgment on the output of the generator and gives feedback according to its judgment result. The generator receives the motion signal as input and generates a predicted blood pressure range. The goal of the generator is to generate data similar to the real blood pressure range according to the motion signal. The generator model can use a multi-layer perceptron or a convolutional neural network; the discriminator receives the motion signal and the generated blood pressure range (generated by the generator) as input, and outputs a binary classification result to judge whether the blood pressure range is real or generated by the generator. The task of the discriminator is to distinguish real data from generated data as accurately as possible.

[0038] Next, use the sample motion signal set and the sample first predicted blood pressure interval set as training data, and divide them into a training set and a test set according to a certain ratio; then, use the sample motion signal as the input and the sample first predicted blood pressure interval as the supervision, and use the training set and the test set to perform supervised training and testing on the motion blood pressure prediction branch until convergence; during the training process, the generator receives the sample motion signal as the input and generates a predicted blood pressure interval; then, the discriminator evaluates the difference between the generated blood pressure interval and the true blood pressure interval, and according to the feedback of the discriminator, the generator updates the parameters to optimize the generated blood pressure interval; the discriminator receives a true blood pressure interval and the blood pressure interval output by the generator as the input, and the discriminator outputs a probability indicating the likelihood that the blood pressure interval is from the true data; then the discriminator updates the parameters according to the difference between the generated blood pressure interval and the true blood pressure interval to optimize its ability to judge the generated data and the true data. The training of the generative adversarial network is an alternating process. In each training, the parameters of both the generator and the discriminator are updated simultaneously. In each iterative training stage, first the discriminator is trained, and then the generator is trained, alternating until the blood pressure interval generated by the generator is close enough to the true blood pressure interval and the discriminator cannot easily distinguish the generated data from the true data. Among them, the loss function of the generator is usually based on the feedback of the discriminator, and the generator is optimized by maximizing the probability that the discriminator judges the generated data as true data; the loss function of the discriminator is usually based on the error of its judgment result, and the goal is to minimize the error between the generated data and the true data to ensure that the discriminator can accurately distinguish the true data from the generated data. During the testing process, input the sample motion signal in the test set, and the generator will use these inputs to generate a predicted blood pressure interval; then evaluate the accuracy of the blood pressure interval predicted by the generator. Metrics such as mean square error can be used to measure the difference between the generated blood pressure interval and the true blood pressure interval. When the generative adversarial network training is completed and the performance reaches the expectation, the trained motion blood pressure prediction branch is obtained.

[0039] Finally, input the motion signal into the motion blood pressure prediction branch for prediction, and output the first predicted blood pressure interval, which provides a blood pressure fluctuation range for the user in a motion state.

[0040] Further, the user blood pressure prediction module 13 is further configured to: Input the PPG signal into the pre-trained optoelectronic blood pressure prediction branch, and predict and output the predicted blood pressure. Among them, the optoelectronic blood pressure prediction branch is constructed based on machine learning and is supervised and trained to convergence using the sample PPG signal set and the sample blood pressure set of the target user within the historical time; calculate the sensing error influence parameter according to the first sensing error influence parameter and the second sensing error influence parameter; use the sensing error influence parameter to perform error compensation on the predicted blood pressure to obtain the second predicted blood pressure interval.

[0041] Specifically, first, an optoelectronic blood pressure prediction branch is constructed based on machine learning. The optoelectronic blood pressure prediction branch is used to predict blood pressure through a machine learning model. For example, an optoelectronic blood pressure prediction branch is constructed based on a BP neural network. Among them, the optoelectronic blood pressure prediction branch includes an input layer, multiple hidden layers, and an output layer. Among them, the input data of the input layer is the PPG signal, and the output data of the output layer is the blood pressure data. Then, the sample PPG signal set and the sample blood pressure set within the historical time of the target user are used as training data. With the sample PPG signal as the input and the sample blood pressure as the supervision, the optoelectronic blood pressure prediction branch is supervised and trained. During the training, the PPG signal is input into the input layer of the neural network, and then processed through the hidden layer, and finally the predicted blood pressure value is generated. By calculating the difference between the loss function (such as the mean square error) and the actual blood pressure value, the network adjusts the weights of the output layer. Then, the backpropagation algorithm calculates the gradient according to the loss function and adjusts the weights and biases in the network to reduce the error. Usually, the gradient descent method (or its variants, such as the Adam optimizer) is used to optimize the weights to minimize the loss function. During the training process, the neural network continuously iteratively updates the parameters through forward propagation and backpropagation until convergence. Convergence usually means that the loss function no longer decreases significantly, or reaches the preset number of training epochs, and the trained optoelectronic blood pressure prediction branch is obtained. Then, the PPG signal is input into the trained optoelectronic blood pressure prediction branch for prediction, and the predicted blood pressure is output.

[0042] Then, the first sensing error influence parameter (representing the influence of motion-related errors on blood pressure prediction) and the second sensing error influence parameter (representing the error influence on blood pressure prediction due to factors such as skin color and contact) are added together to obtain the sensing error influence parameter, which represents the total error influence on blood pressure prediction under the combined action of all external factors (such as motion, skin color, contact, etc.). Further, the sensing error influence parameter is used to proportionally expand the predicted blood pressure to obtain the second predicted blood pressure interval. For example, assuming that the sensing error influence parameter is 10%, the predicted systolic blood pressure is 120, and the diastolic blood pressure is 80, then the predicted systolic blood pressure interval is from 120*(1 - 10%) to 120*(1 + 10%), that is, the predicted systolic blood pressure interval is from 108 to 132, and the predicted diastolic blood pressure interval is from 80*0.9 to 80*1.1, which is 72 to 88, and the predicted systolic blood pressure interval and the predicted diastolic blood pressure interval are set as the second predicted blood pressure interval.

[0043] The error coefficient annotation module 14 is used to calculate and obtain the blood pressure monitoring error coefficient according to the first predicted blood pressure interval and the second predicted blood pressure interval, and annotate the predicted blood pressure as the blood pressure monitoring result.

[0044] Furthermore, the error coefficient annotation module 14 is further used for: Obtain the intersection of the first predicted blood pressure range and the second predicted blood pressure range to obtain an intersection blood pressure range; calculate the ratios of the intersection blood pressure range to the first predicted blood pressure range and the second predicted blood pressure range respectively, and calculate to obtain a first blood pressure monitoring error coefficient and a second blood pressure monitoring error coefficient; calculate a blood pressure monitoring error coefficient according to the first blood pressure monitoring error coefficient and the second blood pressure monitoring error coefficient; use the blood pressure monitoring error coefficient to label the predicted blood pressure to obtain a blood pressure monitoring result.

[0045] Specifically, perform an intersection operation on the first predicted blood pressure range and the second predicted blood pressure range to obtain an intersection blood pressure range. The intersection blood pressure range represents the blood pressure range that is commonly satisfied between the two (including the high-pressure intersection and the low-pressure intersection). Then calculate the ratio of the length of the intersection blood pressure range to the length of the first predicted blood pressure range, subtract the ratio from 1, and set the difference between the two as the first blood pressure monitoring error coefficient. The first blood pressure monitoring error coefficient reflects the degree of difference between the first predicted blood pressure range and the intersection blood pressure range. The smaller the error coefficient, the more accurate the predicted blood pressure range. On the other hand, calculate the ratio of the length of the intersection blood pressure range to the length of the second predicted blood pressure range, subtract the ratio from 1, and set the difference between the two as the second blood pressure monitoring error coefficient. The second blood pressure monitoring error coefficient reflects the degree of difference between the second predicted blood pressure range and the intersection blood pressure range. A smaller error coefficient indicates a smaller difference between the two blood pressure ranges and higher reliability of the predicted blood pressure, while a larger error coefficient indicates a larger difference and possible larger errors in the predicted blood pressure.

[0046] Further perform a weighted calculation on the first blood pressure monitoring error coefficient and the second blood pressure monitoring error coefficient. Among them, since the accuracy of predicting the user's blood pressure through motion signals is relatively low, the weight of the second blood pressure monitoring error coefficient is much greater than the weight of the first blood pressure monitoring error coefficient. For example, set the first weight to 0.1 and the second weight to 0.9, and then perform a weighted calculation to obtain a blood pressure monitoring error coefficient. The weighted blood pressure monitoring error coefficient obtained through the weighted calculation can be used to evaluate the accuracy and reliability of the overall blood pressure prediction. Finally, use the blood pressure monitoring error coefficient to label the predicted blood pressure to obtain a blood pressure monitoring result, so as to provide more accurate blood pressure monitoring data for the user.

[0047] In summary, the intelligent non-invasive blood pressure real-time monitoring device based on multi-modal provided by the present invention has the following technical effects: By collecting PPG signals and motion signals from the target user, PPG signals and motion signals are obtained, and the skin color information of the target user is collected. Then, based on the motion signals, the analysis of the influence of sensing errors and the prediction of contact influence parameters are carried out to obtain the first sensing error influence parameter and the contact influence parameter. Based on the skin color information and the contact influence parameter, the analysis of the influence of sensing errors is carried out to obtain the second sensing error influence parameter. Then, based on the motion signals, the user's blood pressure is predicted to obtain the first predicted blood pressure range. On the other hand, based on the PPG signals, the user's blood pressure is predicted to obtain the predicted blood pressure. Based on the first sensing error influence parameter and the second sensing error influence parameter, error compensation is carried out to obtain the second predicted blood pressure range. Finally, based on the first predicted blood pressure range and the second predicted blood pressure range, the blood pressure monitoring error coefficient is calculated, and the predicted blood pressure is marked as the blood pressure monitoring result. That is to say, by combining multi-modal sensor data for sensing error analysis, it is possible to accurately identify and quantify the sensing errors caused by different factors, improving the accuracy and reliability of obtaining sensing errors. Then, by marking the predicted blood pressure according to the error analysis results, feedback information about the measurement reliability can be provided to the user, helping the user effectively identify the reliability of the predicted blood pressure, so as to provide more accurate, comprehensive and reliable blood pressure monitoring data for the user.

[0048] Embodiment 2. Based on the same inventive concept as the multi-modal intelligent non-invasive blood pressure real-time monitoring device in the foregoing embodiment, the present invention also provides a multi-modal intelligent non-invasive blood pressure real-time monitoring method. Please refer to the attached Figure 2 , including: By using a PPG sensor and a motion sensor, PPG signals and motion signals are collected from the target user to obtain PPG signals and motion signals, and the skin color information of the target user is collected; Based on the motion signals, the analysis of the influence of sensing errors and the prediction of contact influence parameters are carried out to obtain the first sensing error influence parameter and the contact influence parameter. Based on the skin color information and the contact influence parameter, the analysis of the influence of sensing errors is carried out to obtain the second sensing error influence parameter; Based on the motion signals, the user's blood pressure is predicted to obtain the first predicted blood pressure range. Based on the PPG signals, the user's blood pressure is predicted to obtain the predicted blood pressure. Based on the first sensing error influence parameter and the second sensing error influence parameter, error compensation is carried out to obtain the second predicted blood pressure range; Based on the first predicted blood pressure range and the second predicted blood pressure range, the blood pressure monitoring error coefficient is calculated, and the predicted blood pressure is marked as the blood pressure monitoring result.

[0049] Further, the intelligent non-invasive blood pressure real-time monitoring method based on multi-modalities further includes: collecting PPG signals and motion signals of a target user within a recent preset time window through a PPG sensor and a motion sensor to obtain PPG signals and motion signals, wherein the motion sensor is a gyroscope; and acquiring the skin color information of the target user collected and recorded within a historical time period.

[0050] Further, the intelligent non-invasive blood pressure real-time monitoring method based on multi-modalities further includes: collecting a sample motion signal set according to the monitoring data within a historical time period, and collecting the error magnitudes between different sample motion signals and the blood pressure predicted based on PPG signals in a non-moving state, which are labeled as sample first sensing error influence parameters to obtain a sample first sensing error influence parameter set; collecting the skin sweating parameters of the user under different sample motion signals, which are labeled as sample contact influence parameters to obtain a sample contact influence parameter set; constructing a motion sensing error influence prediction branch and a contact influence prediction branch by using machine learning; using the sample motion signal set as input features, and respectively using the sample first sensing error influence parameter set and the sample contact influence parameter set as output features to perform supervised training and testing on the motion sensing error influence prediction branch and the contact influence prediction branch until convergence; and inputting the motion signals into the motion sensing error influence prediction branch and the contact influence prediction branch respectively to predict and output the first sensing error influence parameter and the contact influence parameter.

[0051] Further, the intelligent non-invasive blood pressure real-time monitoring method based on multi-modalities further includes: collecting a sample skin color information set and a sample contact influence parameter set according to the monitoring data within a historical time period; collecting the error magnitudes between different sample skin color information and sample contact influence parameters and the blood pressure predicted based on PPG signals compared with a preset skin color and a preset contact influence parameter, which are labeled as sample second sensing error influence parameters to obtain a sample second sensing error influence parameter set; constructing a skin color sensing error influence prediction branch based on machine learning; using the sample skin color information set, the sample contact influence parameter set and the sample second sensing error influence parameter set to perform supervised training and testing on the skin color sensing error influence prediction branch until convergence; and inputting the skin color information and the contact influence parameter into the skin color sensing error influence prediction branch to predict and output the second sensing error influence parameter.

[0052] Further, the intelligent non-invasive blood pressure real-time monitoring method based on multi-modalities further includes: according to the blood pressure monitoring data of the target user within a historical time period, collecting a set of sample motion signals, and collecting the blood pressure intervals of the target user under different sample motion signals, and obtaining a set of sample first predicted blood pressure intervals through annotation; based on machine learning, constructing a motion blood pressure prediction branch; using the set of sample motion signals and the set of sample first predicted blood pressure intervals to perform supervised training and testing on the motion blood pressure prediction branch until convergence; inputting the motion signal into the motion blood pressure prediction branch, and predicting and outputting to obtain a first predicted blood pressure interval.

[0053] Further, the intelligent non-invasive blood pressure real-time monitoring method based on multi-modalities further includes: inputting the PPG signal into a pre-trained optoelectronic blood pressure prediction branch, and predicting and outputting to obtain a predicted blood pressure, wherein the optoelectronic blood pressure prediction branch is constructed based on machine learning and is supervised and trained to convergence using a set of sample PPG signals and a set of sample blood pressures of the target user within a historical time period; calculating a sensing error influence parameter according to the first sensing error influence parameter and the second sensing error influence parameter; using the sensing error influence parameter to perform error compensation on the predicted blood pressure to obtain a second predicted blood pressure interval.

[0054] Further, the intelligent non-invasive blood pressure real-time monitoring method based on multi-modalities further includes: obtaining the intersection of the first predicted blood pressure interval and the second predicted blood pressure interval to obtain an intersection blood pressure interval; respectively calculating the ratios of the intersection blood pressure interval to the first predicted blood pressure interval and the second predicted blood pressure interval, and calculating to obtain a first blood pressure monitoring error coefficient and a second blood pressure monitoring error coefficient; calculating a blood pressure monitoring error coefficient according to the first blood pressure monitoring error coefficient and the second blood pressure monitoring error coefficient; using the blood pressure monitoring error coefficient to annotate the predicted blood pressure to obtain a blood pressure monitoring result.

[0055] The various embodiments in this specification are described in a progressive manner, and the key point of each embodiment is the difference from other embodiments. The intelligent non-invasive blood pressure real-time monitoring device and specific examples in the foregoing Embodiment 1 are equally applicable to the intelligent non-invasive blood pressure real-time monitoring method in this embodiment. Through the foregoing detailed description of the intelligent non-invasive blood pressure real-time monitoring device, those skilled in the art can clearly know the intelligent non-invasive blood pressure real-time monitoring method in this embodiment. Therefore, for the sake of brevity of the specification, it will not be elaborated here. For the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the description of the method part.

[0056] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the broadest scope consistent with the principles and novel features disclosed herein.

[0057] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the present invention and its equivalent technologies, the present invention is also intended to include these changes and variations.

Claims

1. An intelligent non-invasive blood pressure real-time monitoring device based on multi-modalities, characterized in that, The device includes a PPG sensor and a motion sensor, and further includes: A data acquisition module, configured to collect PPG signals and motion signals of a target user through the PPG sensor and the motion sensor, obtain the PPG signals and the motion signals, and collect the skin color information of the target user; An error impact analysis module, configured to perform sensing error impact analysis and contact impact parameter prediction according to the motion signals, obtain a first sensing error impact parameter and a contact impact parameter, and perform sensing error impact analysis according to the skin color information and the contact impact parameter to obtain a second sensing error impact parameter; A user blood pressure prediction module, configured to perform user blood pressure prediction according to the motion signals, obtain a first predicted blood pressure range, perform user blood pressure prediction according to the PPG signals, obtain a predicted blood pressure, and perform error compensation based on the first sensing error impact parameter and the second sensing error impact parameter to obtain a second predicted blood pressure range; An error coefficient annotation module, configured to calculate a blood pressure monitoring error coefficient according to the first predicted blood pressure range and the second predicted blood pressure range, and annotate the predicted blood pressure as a blood pressure monitoring result.

2. The intelligent non-invasive blood pressure real-time monitoring device based on multi-modalities according to claim 1, wherein Collecting PPG signals and motion signals of a target user through the PPG sensor and the motion sensor, obtaining the PPG signals and the motion signals, and collecting the skin color information of the target user includes: Collecting PPG signals and motion signals of a target user within a most recent preset time window through the PPG sensor and the motion sensor, obtaining the PPG signals and the motion signals, where the motion sensor is a gyroscope; Obtaining the skin color information of the target user collected and recorded within a historical time.

3. The intelligent non-invasive blood pressure real-time monitoring device based on multi-modalities according to claim 1, wherein Performing sensing error impact analysis and contact impact parameter prediction according to the motion signals, obtaining a first sensing error impact parameter and a contact impact parameter includes: According to the monitoring data within a historical time, collecting a sample motion signal set, and collecting the error magnitudes between different sample motion signals and the blood pressure predicted based on the PPG signals in the non-moving state, annotating them as sample first sensing error impact parameters, and obtaining a sample first sensing error impact parameter set; Collecting the skin sweating parameters of the user under different sample motion signals, annotating them as sample contact impact parameters, and obtaining a sample contact impact parameter set; Using machine learning to construct a motion sensing error impact prediction branch and a contact impact prediction branch; Using the sample motion signal set as input features, and respectively using the sample first sensing error impact parameter set and the sample contact impact parameter set as output features, performing supervised training and testing on the motion sensing error impact prediction branch and the contact impact prediction branch until convergence; Inputting the motion signals into the motion sensing error impact prediction branch and the contact impact prediction branch respectively, and predicting and outputting to obtain a first sensing error impact parameter and a contact impact parameter.

4. The multi-modal-based intelligent non-invasive blood pressure real-time monitoring device according to claim 1, characterized in that, Performing sensing error impact analysis according to the skin color information and the contact impact parameter to obtain a second sensing error impact parameter includes: According to the monitoring data within a historical time, collecting a sample skin color information set and a sample contact impact parameter set; Collect the error range of blood pressure prediction based on the PPG signal when collecting the skin color information of different samples and the sample contact influence parameters compared with the preset skin color and the preset contact influence parameters, label it as the sample second sensing error influence parameter, and obtain the sample second sensing error influence parameter set; Based on machine learning, construct a skin color sensing error influence prediction branch; Use the sample skin color information set, the sample contact influence parameter set, and the sample second sensing error influence parameter set to perform supervised training and testing on the skin color sensing error influence prediction branch until convergence; Input the skin color information and the contact influence parameter into the skin color sensing error influence prediction branch, and predict and output to obtain the second sensing error influence parameter.

5. The intelligent non-invasive blood pressure real-time monitoring device based on multi-modal according to claim 1, wherein According to the motion signal, perform user blood pressure prediction to obtain the first predicted blood pressure range, including: According to the blood pressure monitoring data of the target user within the historical time, collect the sample motion signal set, and collect the blood pressure range of the target user under different sample motion signals, and label to obtain the sample first predicted blood pressure range set; Based on machine learning, construct a motion blood pressure prediction branch; Use the sample motion signal set and the sample first predicted blood pressure range set to perform supervised training and testing on the motion blood pressure prediction branch until convergence; Input the motion signal into the motion blood pressure prediction branch, and predict and output to obtain the first predicted blood pressure range.

6. The intelligent non-invasive blood pressure real-time monitoring device based on multi-modalities according to claim 1, wherein, According to the PPG signal, perform user blood pressure prediction to obtain the predicted blood pressure, and perform error compensation based on the first sensing error influence parameter and the second sensing error influence parameter to obtain the second predicted blood pressure range, including: Input the PPG signal into the pre-trained optoelectronic blood pressure prediction branch, and predict and output to obtain the predicted blood pressure, where the optoelectronic blood pressure prediction branch is constructed based on machine learning and is supervised and trained to convergence using the sample PPG signal set and the sample blood pressure set of the target user within the historical time; According to the first sensing error influence parameter and the second sensing error influence parameter, calculate to obtain the sensing error influence parameter; Use the sensing error influence parameter to perform error compensation on the predicted blood pressure to obtain the second predicted blood pressure range.

7. The multi-modal-based intelligent non-invasive blood pressure real-time monitoring device according to claim 1, characterized in that, According to the first predicted blood pressure range and the second predicted blood pressure range, calculate to obtain the blood pressure monitoring error coefficient, and label the predicted blood pressure, including: Obtain the intersection of the first predicted blood pressure range and the second predicted blood pressure range to obtain the intersection blood pressure range; Calculate the ratios of the intersection blood pressure range to the first predicted blood pressure range and the second predicted blood pressure range respectively, and calculate to obtain the first blood pressure monitoring error coefficient and the second blood pressure monitoring error coefficient; According to the first blood pressure monitoring error coefficient and the second blood pressure monitoring error coefficient, calculate to obtain the blood pressure monitoring error coefficient; Use the blood pressure monitoring error coefficient to label the predicted blood pressure to obtain the blood pressure monitoring result.

8. An intelligent non-invasive blood pressure real-time monitoring method based on multi-modalities, characterized in that, Executed by the multi-modal based intelligent non-invasive blood pressure real-time monitoring device according to any one of claims 1 to 7, further including: Through the PPG sensor and the motion sensor, collect the PPG signal and the motion signal of the target user to obtain the PPG signal and the motion signal, and collect the skin color information of the target user; Conduct sensing error impact analysis and contact impact parameter prediction based on the motion signal to obtain the first sensing error impact parameter and contact impact parameter, and conduct sensing error impact analysis based on the skin color information and contact impact parameter to obtain the second sensing error impact parameter; Conduct user blood pressure prediction based on the motion signal to obtain the first predicted blood pressure range, conduct user blood pressure prediction based on the PPG signal to obtain the predicted blood pressure, and perform error compensation based on the first sensing error impact parameter and the second sensing error impact parameter to obtain the second predicted blood pressure range; Calculate the blood pressure monitoring error coefficient based on the first predicted blood pressure range and the second predicted blood pressure range, and label the predicted blood pressure as the blood pressure monitoring result.

Images