A blood pressure measurement method, electronic device, computer storage medium and computer program product

Abstract

Embodiments of the present application provide a blood pressure measurement method, an electronic device, a computer storage medium and a computer program product. The method comprises: determining a blood pressure measurement result of a user according to acquired first PPG data by using a pre-trained blood pressure prediction model; the first PPG data is obtained based on a photoplethysmogram (PPG) signal collected from the user; model parameters of the pre-trained blood pressure prediction model comprise a first parameter and a second parameter; the first parameter is obtained by training a first device; and the second parameter is obtained by continuing to perform transfer learning on the model by using a second device while the first parameter is frozen. The first PPG data can be collected by using a device with a PPG signal collection function, so that the cost of blood pressure measurement is reduced. Meanwhile, the second parameter in the pre-trained blood pressure prediction model is obtained by using the second device, so that the time and resource consumption required for retraining the blood pressure prediction model are reduced, and the accuracy of blood pressure measurement can be improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of blood pressure detection, and in particular to a blood pressure measurement method, an electronic device, a computer storage medium and a computer program product. BACKGROUND

[0002] Hypertension refers to long-term and persistent increase in blood pressure, and is a common chronic disease. Hypertension can cause cardiovascular diseases and affect health. For patients with hypertension, blood pressure needs to be monitored regularly, and the corresponding treatment plan needs to be adjusted according to the blood pressure to avoid the occurrence of cardiovascular diseases.

[0003] In related technologies, blood pressure can be measured by professional equipment such as cuffs and stethoscopes, but it needs to be operated and interpreted by medical personnel, and blood pressure measurement is complex and inefficient.

[0004] In order to reduce the complexity of blood pressure measurement and improve the efficiency of blood pressure measurement, a wearable device with blood pressure measurement function is provided in related technologies, such as a bracelet, a watch, etc. The wearable device can collect photoplethysmogram (PPG) signals and electrocardiogram (ECG) signals, determine pulse transit time according to the PPG signals and ECG signals, and then estimate blood pressure using the pulse transit time.

[0005] However, the wearable device with PPG signal and ECG signal acquisition function has a high price, which increases the cost of blood pressure measurement. At the same time, the wearable device may be affected by specific factors of the user (such as the elastic modulus of the blood vessel, the thickness of the blood vessel wall, the motion state, etc.) during signal acquisition, resulting in inaccurate signals and further low accuracy of blood pressure measurement. SUMMARY

[0006] The blood pressure measurement method, the electronic device, the computer storage medium and the computer program product provided by the present application solve the problems of high cost and low accuracy of blood pressure measurement.

[0007] To achieve the above-mentioned purpose, the technical scheme adopted by the present application is as follows:

[0008] In a first aspect, the present application provides a blood pressure measurement method, which is applied to a first device and can include: determining a blood pressure measurement result of a user according to acquired first PPG data by using a pre-trained blood pressure prediction model; the first PPG data is obtained based on photoplethysmogram (PPG) signals collected from the user, and the model parameters of the pre-trained blood pressure prediction model include a first parameter and a second parameter, the first parameter is obtained by training the first device, and the second parameter is obtained by continuing to perform transfer learning on the model under the condition that the first parameter is frozen.

[0009] It should be understood that, on the one hand, compared with a wearable device with functions of collecting PPG signals and ECG signals, the application can collect first PPG data through a device with a PPG signal collection function, thereby reducing the cost of blood pressure measurement; on the other hand, the second device is used to perform transfer learning on the model under the condition that the first parameters are frozen to obtain the second parameters, without the need to repeatedly train all parameters, thereby reducing the time and resource consumption required for retraining the model, and at the same time, by taking the first parameters and the second parameters obtained through transfer learning as model parameters of a pre-trained blood pressure prediction model, the performance of the model is improved, thereby improving the accuracy of blood pressure measurement.

[0010] In combination with the first aspect, in a possible implementation, the second parameters can be obtained in the following manner: obtaining second PPG data and gold standard data corresponding to the second PPG data, then sending the second PPG data and the gold standard data to the second device, and further calibrating the second initial parameters in the pre-trained parameters according to the second PPG data and the gold standard data through the second device to obtain the second parameters; wherein the second PPG data is obtained based on PPG signals collected from a user, the gold standard data is obtained based on actual blood pressure collected from the user, the pre-trained parameters are parameters of a trained blood pressure prediction model, the trained blood pressure model is obtained based on PPG training data and standard blood pressure training data, and the pre-trained parameters include the first initial parameters and the second initial parameters, and the first parameters are the first initial parameters.

[0011] It should be understood that, since part of the layers in the trained blood pressure prediction model have learned general feature representations through the training data set, there is no need to retrain the trained blood pressure prediction model. In this way, by calibrating the second initial parameters, the efficiency of fine-tuning can be improved and the computing resources can be saved on the basis of improving the accuracy of blood pressure measurement.

[0012] In combination with the first aspect, in a possible implementation, the second PPG data includes first data and second data, the first data is obtained based on PPG signals collected from a user in a first state, and the second data is obtained based on PPG signals collected from the user in a second state; the gold standard data includes first standard data and second standard data, the first standard data is a standard blood pressure of the user in the first state, and the second standard data is a standard blood pressure of the user in the second state.

[0013] It should be understood that, by collecting PPG calibration data and gold standard data of a user in different states, the data dimension can be enriched, so that the result output by the fine-tuned blood pressure prediction model is more accurate, thereby improving the accuracy of blood pressure measurement.

[0014] With reference to the first aspect, in a possible implementation, before the first data and the second data are taken as the second PPG data and the first standard data and the second standard data are taken as the gold standard data, the method further includes: determining whether an absolute value of a difference between the first standard data and the second standard data is greater than or equal to a preset difference value; if the absolute value of the difference between the first standard data and the second standard data is greater than or equal to the preset difference value, taking the first data and the second data as the second PPG data and taking the first standard data and the second standard data as the gold standard data; and if the absolute value of the difference between the first standard data and the second standard data is less than the preset difference value, reacquiring new second data and new second standard data until the difference between the new second standard data and the first standard data is greater than or equal to the preset difference value, taking the first data and the new second data as the second PPG data, and taking the first standard data and the new second standard data as the gold standard data.

[0015] It should be understood that, in order to ensure that the first data and the second data have differences and to enrich the data dimensions, the difference between the first standard data and the second standard data needs to be greater than or equal to the preset difference value before the second PPG data and the gold standard data are determined.

[0016] With reference to the first aspect, in a possible implementation, determining the blood pressure measurement result of the user according to the first PPG data by using the pre-trained blood pressure prediction model includes: determining a first test result according to the first PPG data by using the pre-trained blood pressure prediction model; and determining the blood pressure measurement result of the user according to an initial test result, a confidence decay algorithm, and the first test result, where the initial test result is an initial test result output by the pre-trained blood pressure prediction model.

[0017] It should be understood that, since the effectiveness of the calibration of the pre-trained blood pressure prediction model gradually decreases over time, the accuracy of blood pressure measurement decreases. In order to ensure the accuracy of blood pressure measurement, the confidence of each blood pressure measurement result can be adjusted by using the confidence decay algorithm in this embodiment, so as to ensure the accuracy of blood pressure measurement.

[0018] With reference to the first aspect, in a possible implementation, determining the blood pressure measurement result of the user according to the initial test result, the confidence decay algorithm, and the first test result includes: multiplying the first test result and the confidence decay algorithm to obtain a first product; multiplying the initial test result and a complement of the confidence decay algorithm to obtain a second product; and taking a sum of the first product and the second product as the blood pressure measurement result of the user. The complement of the confidence decay algorithm means 1 minus a set (or value) obtained by the confidence decay algorithm.

[0019] It should be understood that, by adjusting the confidence of each blood pressure measurement result by using the confidence decay algorithm, the accuracy of blood pressure measurement can be ensured.

[0020] With reference to the first aspect, in a possible implementation, the pre-trained blood pressure prediction model comprises a feature extraction layer and a linear layer, the first parameters are parameters of the feature extraction layer, and the second parameters are parameters of the linear layer.

[0021] With reference to the second aspect, the application provides a blood pressure measurement method applied to the second device, which can comprise: determining a test result of a user according to acquired first PPG data by using a pre-trained blood pressure prediction model; wherein the first PPG data is obtained by collecting a photoplethysmogram (PPG) signal of the user by using a first device; and model parameters of the pre-trained blood pressure prediction model comprise first parameters and second parameters, the first parameters are obtained by training the first device, and the second parameters are obtained by continuing to perform transfer learning on the model by using the second device under the condition that the first parameters are frozen.

[0022] It should be understood that, on the one hand, compared with a wearable device that is capable of collecting PPG signals and ECG signals, the application can collect first PPG data by using a device that is capable of collecting PPG signals, thereby reducing the cost of blood pressure measurement; on the other hand, based on transfer learning, the second parameters in the pre-trained blood pressure prediction model are obtained by using the second device, without the need to repeatedly train all parameters, thereby reducing the time and resource consumption required for retraining the blood pressure prediction model, and meanwhile, the accuracy of blood pressure measurement can be improved.

[0023] It should be noted that, in the embodiments of the application, blood pressure measurement can be performed on a user by using the first device, or blood pressure measurement can be performed on a user by using the second device, which is not limited herein.

[0024] With reference to the second aspect, in a possible implementation, the second parameters are obtained in the following manner: receiving second PPG data and gold standard data corresponding to the second PPG data sent by the first device, and then calibrating second initial parameters in pre-trained parameters according to the second PPG data and the gold standard data to obtain the second parameters; wherein the second PPG data is obtained based on a PPG signal collected by the first device from the user, the gold standard data is obtained based on an actual blood pressure collected from the user, the pre-trained parameters are parameters of a trained blood pressure prediction model, the trained blood pressure model is obtained based on PPG training data and standard blood pressure training data, the pre-trained parameters comprise first initial parameters and second initial parameters, and the first parameters are the first initial parameters.

[0025] It should be understood that, since part of layers in the trained blood pressure prediction model have learned general feature representations through a training data set, the trained blood pressure prediction model does not need to be retrained. In this way, by calibrating the second initial parameters, the efficiency of fine-tuning can be improved, and the computing resources can be saved on the basis of improving the accuracy of blood pressure measurement.

[0026] In a third aspect, the present application provides an electronic device, comprising: a processor and a memory;

[0027] Wherein, one or more computer programs are stored in the memory, the one or more computer programs comprise instructions; when the instructions are executed by the processor, the electronic device executes the blood pressure measurement method as described in the first aspect, or executes the blood pressure measurement method as described in the second aspect.

[0028] In a fourth aspect, the present application provides a computer storage medium, comprising computer instructions, when the computer instructions run on an electronic device, the electronic device executes the blood pressure measurement method as described in the first aspect, or executes the blood pressure measurement method as described in the second aspect.

[0029] In a fifth aspect, the present application provides a computer program product, when the computer program product runs on a computer, the computer executes the blood pressure measurement method as described in the first aspect, or executes the blood pressure measurement method as described in the second aspect. BRIEF DESCRIPTION OF DRAWINGS

[0030] Figure 1 A schematic diagram of a blood pressure measurement method provided by an embodiment of the present application;

[0031] Figure 2a A schematic diagram of a model training process of blood pressure measurement provided by an embodiment of the present application;

[0032] Figure 2b A schematic diagram of a PPG standard data and gold standard data acquisition process provided by an embodiment of the present application;

[0033] Figure 3a A schematic diagram of a linear attenuation curve provided by an embodiment of the present application;

[0034] Figure 3b A schematic diagram of a nonlinear attenuation curve provided by an embodiment of the present application;

[0035] Figure 3c A schematic diagram of another nonlinear attenuation curve provided by an embodiment of the present application;

[0036] Figure 4 A schematic diagram of a test result comparison provided by an embodiment of the present application;

[0037] Figure 5 A scene schematic diagram of a blood pressure measurement method provided by an embodiment of the present application. DETAILED DESCRIPTION

[0038] In order to make the following embodiments clear and simple, first give a brief introduction of related technologies:

[0039] Blood pressure means the lateral pressure exerted by blood on the walls of blood vessels per unit area when blood flows within the blood vessels. When the heart contracts, blood is pumped into the aorta, forming a pulse wave. The pulse wave propagates outward along the arterial system, and the pressure of the pulse wave on the blood vessels during propagation is called blood pressure. Blood pressure can include systolic pressure and diastolic pressure.

[0040] Systolic blood pressure (SBP) means the maximum pressure exerted by blood on the walls of blood vessels when the heart contracts. Systolic pressure can be measured as the first number in a blood pressure reading, for example, "120" in 120 / 80 mmHg is the systolic pressure.

[0041] Diastolic blood pressure (DBP) means the minimum pressure exerted by blood on the walls of blood vessels when the heart relaxes. Diastolic pressure can be measured as the second number in a blood pressure reading, for example, "80" in 120 / 80 mmHg is the diastolic pressure.

[0042] Normal blood pressure means that the systolic pressure is less than 120 mmHg and the diastolic pressure is less than 80 mmHg without the use of antihypertensive drugs. Wherein, mmHg (millimeter of mercury) is the unit of blood pressure.

[0043] Photoplethysmography (PPG) is a physiological signal, and the PPG signal is used to represent the change in blood volume in the blood vessels inside the measured tissue site. When the heart contracts, blood is pumped into the arteries, causing the arteries to dilate, and the light intensity received by the photoelectric sensor decreases, producing a PPG signal peak; when the heart relaxes, the arteries contract, and the light intensity received by the photoelectric sensor increases, and the PPG signal decreases. PPG signals can be applied to, but not limited to, heart rate monitoring, blood oxygen saturation detection (referred to as blood oxygen detection), blood pressure monitoring, cardiovascular health assessment (such as arterial compliance, arterial aging, etc.), and the like.

[0044] Electrocardiogram (ECG) is a graph that records the electrical activity of the heart. Through an electrocardiograph, the weak electrical signals generated by the heart during contraction and relaxation can be captured and converted into waveforms. These waveforms mainly include P wave, QRS complex and T wave, etc., which represent the electrical activity of the atrium and ventricle, respectively. The P wave represents the potential change of atrial muscle depolarization, the QRS complex represents the potential change of ventricular muscle depolarization, and the T wave represents the potential change of rapid repolarization of the ventricle.

[0045] Moens-Korteweg equation is used to describe the propagation speed of pressure waves in a liquid (such as water in a water pipe).

[0046] VGG (Visual Geometry Group) network is a kind of deep convolutional neural network architecture.

[0047] ResNet (Residual Network) is a kind of convolutional neural network (CNN) architecture in the field of deep learning. The core idea of ResNet is to introduce residual blocks (Residual Block) and connect the input directly to the output through skip connections (SkipConnection), thereby solving the problem of gradient vanishing and gradient explosion in deep neural networks.

[0048] EfficientNet is a kind of convolutional neural network. EfficientNet optimizes depth, width and resolution through compound scaling strategy to balance performance and computational efficiency.

[0049] Resting means that the user is in a static or calm state, such as the user is sitting, lying, standing, and there is no motion that affects the heart rate to change significantly.

[0050] Transfer learning is a method of machine learning that uses a model trained on one task (source task) to solve another related but different task (target task). By transferring existing knowledge and experience, it can reduce the time and resource consumption of retraining the model on the new task, and may improve the performance of the model.

[0051] The related technology provides a wearable device with the functions of collecting PPG signals and ECG signals, uses the wearable device to collect PPG signals and ECG signals, and then estimates blood pressure according to PPG signals and ECG signals, thereby realizing the measurement of blood pressure.

[0052] However, the first device with the functions of collecting PPG signals and ECG signals is relatively expensive, increasing the cost of blood pressure measurement. Moreover, the collected PPG signals and ECG signals may be affected by specific factors of the human body (such as the elastic modulus of blood vessels, the thickness of blood vessel walls, motion conditions, etc.), resulting in poor signal quality.

[0053] As an example, a physiological model of blood pressure is established according to the Moens-Korteweg equation, as shown in the following formula (1):

[0054] Formula (1)

[0055] wherein, is the elastic modulus of the blood vessel, is the thickness of the blood vessel wall, , is a constant, is the blood density, is the radius of the blood vessel.

[0056] As can be seen from formula (1), the specific factors affecting blood pressure include, but are not limited to, the elastic modulus of the blood vessel of the user, the thickness of the blood vessel wall, and the radius of the blood vessel, and the like. The manner of measuring blood pressure by using the PPG signal and the ECG signal in the related art can cause low accuracy of blood pressure measurement.

[0057] Based on the above problems, in combination with Figure 1 , the embodiment of the present application provides a blood pressure measurement method, which, through a training process, a fine-tuning process, and a test process, aims to reduce the cost of blood pressure measurement and improve the accuracy of blood pressure measurement. For ease of illustration, the training process, the fine-tuning process, and the test process are described below in combination with Figure 1 .

[0058] In combination with 10 in Figure 1 , the training process of blood pressure measurement provided by the embodiment of the present application can be:

[0059] Step 1: Obtain PPG data and standard blood pressure data corresponding to the PPG data.

[0060] PPG data means a data set obtained after sampling, recording, and processing a PPG signal. The PPG signal in the PPG data is in the format of a time sequence, representing the beating of the heart and the change in blood volume. In one possible implementation manner, the PPG data can be obtained based on a first device with PPG signal acquisition function.

[0061] The first device means an electronic device with PPG signal acquisition capability that can be worn by a user, such as a watch, a bracelet, a portable detector, a patch-type detection device, and the like, which is not specifically limited here. In the following embodiments, the first device can be exemplarily described by taking a watch as an example.

[0062] Standard blood pressure data means data including a plurality of standard blood pressures. The standard blood pressure (which can also be referred to as actual blood pressure) means the actual blood pressure value of the user, and the standard blood pressure includes systolic pressure and diastolic pressure. In one possible implementation manner, the standard blood pressure of the user can be obtained by using a sphygmomanometer, a blood pressure meter, or other blood pressure measurement devices.

[0063] As an example, assuming that the systolic pressure / diastolic pressure is 140 / 90 mmHg, the standard blood pressure is 140 / 90 mmHg; and assuming that the systolic pressure / diastolic pressure is 110 / 70 mmHg, the standard blood pressure is 110 / 70 mmHg.

[0064] Step 2: Determine a training data set according to the PPG data and standard blood pressure data.

[0065] The training data set means a data set used for training the blood pressure prediction model, and the training data set includes PPG training data and standard blood pressure training data corresponding to the PPG training data. As an example, assuming that the PPG training data includes PPG1 and PPG2, the standard blood pressure data includes standard blood pressure 1 corresponding to PPG1 and standard blood pressure 2 corresponding to PPG2. The training data set can be obtained by data alignment, data slicing, etc. of the PPG data and the standard blood pressure data.

[0066] Data alignment means adjusting data of different sources, different formats or different time points to a unified format or time reference for comparison, analysis or fusion. In a possible implementation, the PPG data and the standard blood pressure data can be one-to-one corresponding in time, that is, each PPG signal corresponds to a standard blood pressure, so as to ensure the accuracy and reliability of the subsequent training process.

[0067] As an example, assuming that the PPG data includes PPG1, PPG2, PPG3, PPG4 and PPG5, and the standard blood pressure data includes standard blood pressure 1, standard blood pressure 2, standard blood pressure 3, standard blood pressure 4 and standard blood pressure 5, by aligning the PPG data with the standard blood pressure data, PPG1 corresponds to standard blood pressure 1, PPG2 corresponds to standard blood pressure 2, PPG3 corresponds to standard blood pressure 3, PPG4 corresponds to standard blood pressure 4, and PPG5 corresponds to standard blood pressure 5.

[0068] Data slicing means that slicing refers to extracting a part of data from an initial data set for a specific analysis or processing task. In a possible implementation, the data set after aligning the PPG data and the standard blood pressure data can be subjected to data slicing processing to divide the training data set, the verification data set and the test data set according to a proportion, so as to train and evaluate.

[0069] As an example, assuming that the PPG data includes PPG1, PPG2, PPG3, PPG4 and PPG5, and the standard blood pressure data includes standard blood pressure 1, standard blood pressure 2, standard blood pressure 3, standard blood pressure 4 and standard blood pressure 5, the PPG data and the standard blood pressure data can be subjected to data slicing processing to obtain a training data set: the processed PPG data includes PPG2, PPG3 and PPG4, and the processed standard blood pressure data includes standard blood pressure 2, standard blood pressure 3 and standard blood pressure 4.

[0070] Step 3: Train the blood pressure prediction model by using the training data set.

[0071] The blood pressure prediction model (may also be referred to as an initial blood pressure prediction model) means an initial network model for predicting blood pressure, such as Figure 2a As shown in the figure, the blood pressure prediction model can include a feature extraction layer and a linear layer.

[0072] The feature extraction layer means a layer for feature extraction of PPG training data. The network structure of the feature extraction layer can be, but is not limited to, VGG, ResNet, EfficientNet, etc., which is not limited here.

[0073] The linear layer (Linear Layer) is also referred to as a fully connected layer (Fully Connected Layer, FCL) or a dense layer (Dense Layer). The linear layer is used for linear transformation of input data. The linear layer is located after the feature extraction layer, used to receive these high-level abstract features and integrate them into the final output. In this embodiment, the output of the feature extraction layer is taken as the input data of the linear layer, and the output of the feature extraction layer is integrated to obtain the training result. The training result includes systolic pressure and diastolic pressure.

[0074] In a possible implementation, each PPG signal in the PPG training data in the training data set can be input into the blood pressure prediction model to obtain a training result corresponding to each PPG signal, and then the blood pressure prediction model is optimized according to the plurality of training results and the standard blood pressure training data to obtain the trained blood pressure prediction model.

[0075] It should be understood that the training of the blood pressure prediction model in the embodiments of the present application can adopt supervised training. Supervised training (also referred to as supervised learning) means a training process of adjusting the parameters of the blood pressure prediction model by using a set of known categories of samples (i.e. training data set) to achieve the required performance. Through supervised learning, prediction and classification can be more accurately performed to minimize prediction error. Since in supervised learning, each training result provides a corresponding standard blood pressure, the blood pressure prediction model can directly learn from errors and adjust its performance, which is beneficial to the rapid convergence and performance improvement of the blood pressure prediction model.

[0076] It should be noted that the training process of the blood pressure prediction model in the embodiments of the present application can be implemented based on the first device. Training the blood pressure prediction model based on the first device can make full use of existing resources, improve resource utilization, reduce training delay, facilitate rapid adjustment and optimization of the blood pressure prediction model, and avoid time and bandwidth consumption in the data transmission process, thereby improving training efficiency.

[0077] Step 4: Obtain and store the pre-training parameters of the trained blood pressure prediction model.

[0078] The trained blood pressure prediction model means a network trained by a training data set. The pre-training parameter means a parameter corresponding to the trained blood pressure prediction model, and the pre-training parameter can include feature extraction layer parameters (also referred to as first initial parameters) and linear layer parameters (also referred to as second initial parameters). The feature extraction layer parameters mean parameters corresponding to the feature extraction layer, and the linear layer parameters mean parameters corresponding to the linear layer.

[0079] It should be noted that, due to the influence of specific factors in the process of PPG signal acquisition, the quality of the acquired signal may be poor, and thus the accuracy of blood pressure measurement is low. In order to improve the accuracy of blood pressure measurement, after training the blood pressure prediction model, the trained blood pressure prediction model can also be fine-tuned.

[0080] In combination with 20 shown in Figure 1 , the fine-tuning process (also referred to as calibration process) of blood pressure measurement provided by the embodiments of the present application can be:

[0081] Step one: acquiring PPG calibration data and gold standard data corresponding to the PPG calibration data.

[0082] The PPG calibration data (also referred to as second PPG data) means data used for fine-tuning the trained blood pressure prediction model. The PPG calibration data can include first calibration data and second calibration data. The first calibration data means PPG data acquired by the user in a resting condition, and the second calibration data means PPG data acquired by the user in a motion condition.

[0083] The gold standard data (i.e. standard blood pressure data) means data including multiple standard blood pressures, and the gold standard data is used to represent the actual blood pressure condition of the user. In one possible implementation, in the process of acquiring PPG calibration data, the systolic pressure and diastolic pressure of the user can be synchronously acquired by using a sphygmomanometer, a blood pressure meter or other devices, and then the actual systolic pressure and diastolic pressure are taken as the standard blood pressure, and the obtained multiple standard blood pressures are taken as the gold standard data.

[0084] As an example, in combination with 20 shown in Figure 2b , the acquisition process of PPG calibration data and gold standard data can be:

[0085] A1: prompting the user to calibrate.

[0086] It should be understood that in the embodiments of the present application, in order to ensure the accuracy of blood pressure measurement and improve the accuracy of blood pressure measurement, the user can be prompted by the blood pressure measurement application in the first device to calibrate when using the blood pressure measurement function for the first time or again after a preset time, so as to fine-tune the trained blood pressure prediction model. The prompting methods include but are not limited to button display, full-screen display, vibration display, voice prompt, etc., which are not limited here.

[0087] A2: The user is at rest.

[0088] It should be understood that the blood pressure of the user in different states may also be different, for example, the blood pressure of the user in the case of exercise is higher than that in the case of sleep. In order to improve the accuracy of blood pressure measurement, the calibration data of the user in the resting state (which can also be referred to as the first state) and the exercise state (which can also be referred to as the second state) can be obtained. The user can rest for a period of time to collect data, so that the blood pressure is stable within a certain range, and avoid poor quality of collected data caused by large fluctuations, for example, the user can rest for three minutes before collecting data.

[0089] A3: Collecting first calibration data and first standard data corresponding to the first calibration data.

[0090] The first calibration data (which can also be referred to as the first data) means the PPG data collected by the first device when the user is in the resting state, and the first standard data means the gold standard data corresponding to the first calibration data.

[0091] In the embodiments of the present application, the first calibration data can be collected by the first device, and the first standard data can be collected by a sphygmomanometer, a blood pressure meter or other devices. The first calibration data and the first standard data can be collected at the same time, the first calibration data can be collected first and then the first standard data can be collected, or the first standard data can be collected first and then the first calibration data can be collected, which are not limited here.

[0092] A4: The user exercises.

[0093] It should be understood that in order to improve the accuracy of blood pressure measurement, the user can be prompted to exercise after collecting the first calibration data and the first standard data, so that the blood pressure during exercise is different from the blood pressure during rest, for example, the user can exercise for one minute before collecting data.

[0094] A5: Collecting second calibration data and second standard data corresponding to the second calibration data.

[0095] The second calibration data (may also be referred to as second data) means PPG data collected by the first device when the user is in a motion state (or after the motion ends), and the second standard data means gold standard data corresponding to the second calibration data.

[0096] In the embodiments of the present application, the second calibration data can be collected by the first device, the second standard data can be collected by a sphygmomanometer, a tonometer or other devices, and the second calibration data and the second standard data can be collected simultaneously. Alternatively, the second calibration data can be collected first, and then the second standard data can be collected. Alternatively, the second standard data can be collected first, and then the second calibration data can be collected, which is not limited specifically herein.

[0097] A6: determining whether an absolute value of a difference between the systolic pressure in the first standard data and the systolic pressure in the second standard data is greater than or equal to a preset difference value.

[0098] If the absolute value of the difference between the systolic pressure in the first standard data and the systolic pressure in the second standard data is greater than or equal to the preset difference value, step A7 is performed; otherwise, step A4 is performed.

[0099] The preset difference value means an absolute value of a difference between the systolic pressure of the first standard data and the systolic pressure of the second standard data that meets the output condition, for example, the preset difference value is 5 mmHg.

[0100] It should be understood that, in order to ensure that there is a difference between the data collected at rest and the data collected during motion, the output is required only when the absolute value of the difference between the systolic pressure in the first standard data and the systolic pressure in the second standard data is greater than or equal to the preset difference value, otherwise the user needs to perform motion again, and collect second calibration data (may also be referred to as new second calibration data) and second standard data (may also be referred to as new second standard data) again.

[0101] A7: taking the first calibration data and the second calibration data as PPG calibration data, and taking the first standard data and the second standard data as gold standard data.

[0102] It should be understood that, in the embodiments of the present application, by collecting PPG calibration data and gold standard data in different states of the user, the data dimension can be enriched, so that the result output by the fine-tuned blood pressure prediction model is more accurate, and the accuracy of blood pressure measurement is improved.

[0103] Step two: uploading the PPG calibration data and the gold standard data to the second device.

[0104] The second device means a device with more computing resources in communication connection with the first device, such as a mobile phone, a computer, a tablet, and the like, which are not limited herein. The second device and the first device can be connected through Bluetooth, a local area network, WLAN, and the like, which are not limited herein.

[0105] It should be understood that, due to the limited computing resources of the first device, the fine-tuning process of the blood pressure prediction model can be performed by the second device to save resources. Therefore, in the embodiments of the present application, the collected PPG calibration data and gold standard data can be sent to the second device, and the second device can fine-tune the trained blood pressure prediction model using the PPG calibration data and the gold standard data.

[0106] Step three: fine-tuning the linear layer in the trained blood pressure prediction model by the second device using the PPG calibration data and the gold standard data.

[0107] It should be understood that, in the embodiments of the present application, the feature extraction layer can be frozen first, that is, the feature extraction layer is not fine-tuned, and the feature extraction parameters remain unchanged.

[0108] It should be understood that, since the feature extraction layer in the trained blood pressure prediction model has learned the general feature representation through the training data set, the general feature representation already contains rich information and prior knowledge for the new task (i.e., the fine-tuning process), so there is no need to fine-tune the whole trained blood pressure prediction model.

[0109] At the same time, since the linear layer is responsible for feature combination and classification according to the characteristics of the specific task, fine-tuning the linear layer can make it better adapt to the data distribution and label space of the new task, and there is no need to change the overall structure or most of the parameters of the trained blood pressure prediction model. The fine-tuning of the linear layer requires less computing resources and time than the retraining of the whole model, which can improve the efficiency of fine-tuning and save computing resources on the basis of improving the accuracy of blood pressure measurement.

[0110] In addition, retraining the whole model may cause overfitting, so by keeping the feature extraction parameters of the feature extraction layer in the trained blood pressure prediction model unchanged and using its generalization ability, the risk of overfitting can be reduced.

[0111] Step four: taking the feature extraction parameters (which can also be referred to as first parameters or first initial parameters) of the feature extraction layer and the linear layer parameters (which can also be referred to as second parameters) of the fine-tuned linear layer as model parameters of the trained blood pressure prediction model.

[0112] The model parameters can be saved in the second device, or transmitted by the second device to the first device and saved by the first device, which are not limited herein.

[0113] It should be understood that, on the one hand, by collecting PPG calibration data and gold mark data of the user in different states, the data dimension can be enriched, so that the result output by the fine-tuned blood pressure prediction model is more accurate, and the accuracy of blood pressure measurement is improved. On the other hand, based on transfer learning, the linear layer in the blood pressure prediction model is fine-tuned by the second device, without the need for repeated training, reducing the time and resource consumption required for retraining the blood pressure prediction model, while improving the accuracy of blood pressure measurement.

[0114] After obtaining the fine-tuned blood pressure prediction model (referred to as pre-trained blood pressure prediction model), the test process of blood pressure measurement provided by the embodiments of the present application can be: Figure 1

[0115] Step A: obtaining test data by the first device.

[0116] The test data (also referred to as first PPG data) means the PPG data collected by the user when measuring blood pressure.

[0117] Step B: inputting the test data into the pre-trained blood pressure prediction model to obtain a test result.

[0118] Among them, the pre-trained blood pressure prediction model in the embodiment means the trained blood pressure prediction model after fine-tuning. The test result means the systolic pressure and diastolic pressure predicted by the pre-trained blood pressure prediction model.

[0119] In a possible implementation, the pre-trained blood pressure prediction model saved on the first device can be used to predict the blood pressure of the user according to the test data, and obtain the test result (also referred to as blood pressure measurement result). The linear layer parameters of the fine-tuned linear layer can be sent to the first device by the second device for subsequent blood pressure measurement.

[0120] It should be noted that since the second device only fine-tunes the linear layer in the blood pressure prediction model, the parameters of the feature extraction on the first device and the second device are the same, and in order to improve the transmission efficiency, the second device can only transmit the linear layer parameters to the first device. In other possible implementations, the second device can also transmit the model parameters to the first device, which is not limited here.

[0121] In another possible implementation, the first device can send the test data to the second device, and the pre-trained blood pressure prediction model on the second device can measure the blood pressure of the user according to the test data to obtain the test result. It should be understood that since the second device has higher computing power, blood pressure measurement by the pre-trained blood pressure prediction model on the second device can improve the speed of blood pressure measurement.​

[0122] Based on the above test process, it is found in further research that the effectiveness of the calibration for the pre-trained blood pressure prediction model gradually decreases over time, resulting in a decrease in the accuracy of blood pressure measurement. In order to ensure the accuracy of blood pressure measurement, the embodiments of the present application also provide a confidence reduction algorithm for adjusting the confidence of each test result, as shown in the following formula (2):

[0123] Formula (2)

[0124] wherein, is the final test result (also referred to as the blood pressure measurement result) displayed to the user, is the current test result (also referred to as the first test result), is the initial test result (also referred to as the initial test result), is the confidence decay algorithm.

[0125] In the embodiments of the present application, the decay mode of the confidence decay algorithm can include linear decay (as shown in Figure 3a ) and non-linear decay (as shown in Figure 3b and Figure 3c ), which includes but is not limited to exponential decay, power-law decay, etc., without specific limitation here.

[0126] Linear decay means that the decay amount of the actual value (the current test result) is linearly related to the initial value (the initial test result of the blood pressure prediction model) or a certain reference value, i.e., the decay amount increases at a constant ratio as the time or distance increases.

[0127] Non-linear decay means that the decay amount of the actual value is not linearly related to the initial value or a certain reference value, i.e., the decay rate is not constant, but changes as the time or distance increases.

[0128] Exponential decay is a special non-linear decay mode, in which the decay rate of the actual value is directly proportional to its current value, i.e., the decay rate decreases as the actual value decreases.

[0129] Power-law decay is another special non-linear decay mode, in which the decay rate of a physical quantity or signal is directly proportional to the power of the current value.

[0130] Based on the blood pressure measurement method provided in the above embodiments, in combination with Figure 4 , as shown in Figure 4 , the first test result (L1 and L2) obtained by the blood pressure measurement method of the related art is shown in a, L1 represents the fitting line of multiple systolic pressures in the first test result, and L2 represents the fitting line of multiple diastolic pressures in the first test result.Figure 4 b shows the second test result (L3 and L4) obtained by the blood pressure measurement method of the embodiment of the present application, L3 represents the fitting line of the plurality of systolic pressures in the second test result, and L4 represents the fitting line of the plurality of diastolic pressures in the second test result. K1 represents the reference line of the plurality of actual systolic pressures, and K2 represents the reference line of the plurality of actual diastolic pressures, which are obtained based on the blood pressure measurement on the user.

[0131] The correlation coefficients between the fitting lines of the first test result obtained by the blood pressure measurement method of the related art and the reference lines are r1=0.395 and r2=0.321, respectively; the correlation coefficients between the fitting lines of the second test result obtained by the blood pressure measurement method provided by the embodiment of the present application and the reference lines are r3=0.755 and r4=0.807, respectively. r1 represents the correlation coefficient between L1 and K1, r2 represents the correlation coefficient between L2 and K2, r3 represents the correlation coefficient between L3 and K1, and r4 represents the correlation coefficient between L4 and K2.

[0132] The correlation coefficient (r) is used to represent the degree of linear correlation between two variables. The value range of the correlation coefficient is [-1, 1], wherein 1 represents complete positive correlation, -1 represents complete negative correlation, and 0 represents no correlation. The correlation coefficient is an index for measuring the strength and direction of the linear relationship between two variables. When r is close to 1 or -1, it indicates that there is a strong linear relationship between the two variables; when r is close to 0, it indicates that there is almost no linear relationship between the two variables.

[0133] From Figure 4 It can be seen that the blood pressure measurement method provided by the embodiment of the present application has a higher correlation degree between the test result and the actual test result, that is, compared with the related art, the blood pressure measurement method provided by the embodiment of the present application improves the accuracy of blood pressure measurement.

[0134] Based on the blood pressure measurement method provided in the above embodiment, in combination with Figure 5 As shown in the figure, the embodiment of the present application also provides a scene diagram of blood pressure measurement, the first device is taken as an example of a watch, and the second device is taken as an example of a mobile phone. The process of blood pressure measurement can be:

[0135] Step a: The user clicks the blood pressure test through the mobile phone.

[0136] The mobile phone and the watch establish a communication connection, the mobile phone can respond to the click operation of the user and send a test instruction to the watch through Bluetooth or the like, and the watch responds to the test instruction sent by the mobile phone.

[0137] Step b: The watch prompts the user to calibrate.

[0138] If the user performs blood pressure measurement for the first time or performs blood pressure measurement at a preset time interval, calibration can be performed to improve the accuracy of blood pressure measurement. It should be understood that, in a possible implementation, if the user performs blood pressure measurement again within the preset time, calibration can not be performed; the user can also be set to perform calibration each time, and the number of calibrations or calibration conditions are not limited herein.

[0139] Step c: The watch collects PPG calibration data of the user and sends the PPG data to the mobile phone.

[0140] It should be noted that the watch sends the collected data to the mobile phone for calibration. After calibration is completed, the parameters of the blood pressure prediction model can be saved or sent to the watch for saving, and the number of calibrations is not limited herein. The calibration process can refer to the fine-tuning process of blood pressure measurement in the above embodiment, and is not repeated here.

[0141] If calibration fails, the watch can prompt the user to perform calibration again; if calibration succeeds, the watch can prompt the user to perform blood pressure measurement.

[0142] Step d: After the mobile phone completes calibration, the watch prompts the user to perform blood pressure measurement.

[0143] It should be noted that if blood pressure measurement is performed by the watch, the watch can output the test result according to the collected PPG data; if blood pressure measurement is performed by the mobile phone, the mobile phone can send the collected PPG data to the mobile phone, and the mobile phone can output the test result according to the PPG data.

[0144] Step e: The watch or the mobile phone displays the test result of the user.

[0145] It should be understood that, by performing blood pressure measurement after calibration, the accuracy of blood pressure measurement can be improved.

[0146] The terms "first", "second", and "third" and the like in the specification and claims of the present application and the description of the drawings are used to distinguish different objects, and are not used to limit a specific order.

[0147] In the embodiments of the present application, the words "exemplary" or "for example" are used to mean serving as an example, instance, or illustration, and not necessarily as preferred or advantageous over other embodiments or designs. In fact, any embodiment or design that is described as "exemplary" or "for example" in the embodiments of the present application is not necessarily to be construed as preferred or advantageous over other embodiments or designs. Rather, use of the words "exemplary" or "for example" is intended to present concepts in a particular manner.

[0148] The technical solutions of the embodiments can be embodied in the form of a software product, and the computer software product is stored in a storage medium, and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) or a processor to execute all or part of the steps of the methods described in the embodiments. The aforementioned storage medium includes: a flash memory, a mobile hard disk, a read-only memory, a random access memory, a magnetic disk, or an optical disk, and various media that can store program codes.

[0149] The above merely provides a specific implementation of the present application, but the protection scope of the present application is not limited thereto, and any change or replacement within the technical scope disclosed in the present application should be covered in the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A method for measuring blood pressure, characterized in that, The method is applied to a first device and includes: Acquire first PPG data; the first PPG data is obtained based on the photoplethysmography (PPG) signal collected from the user. The user's blood pressure measurement result is determined based on the first PPG data using a pre-trained blood pressure prediction model on the first device; or, the first PPG data is sent to a second device, and the user's blood pressure measurement result is determined based on the first PPG data using a pre-trained blood pressure prediction model on the second device. The model parameters of the pre-trained blood pressure prediction model include a first parameter and a second parameter. The first parameter is obtained by training the blood pressure prediction model using the first device, and the second parameter is obtained by the second device through transfer learning when the first parameter is frozen. The second parameter is obtained in the following way: Acquire second PPG data and corresponding gold standard data; the second PPG data includes first data and second data, the first data is the PPG signal collected from the user by the first device in a first state, and the second data is the PPG signal collected from the user in a second state; the gold standard data includes first standard data and second standard data, the first standard data is the user's standard blood pressure in the first state, and the second standard data is the user's standard blood pressure in the second state; the first state and the second state are different. Send the second PPG data and the gold label data to the second device; Using the second device, the second initial parameter in the pre-training parameters is calibrated based on the second PPG data and the gold standard data to obtain the second parameter; the pre-training parameter is the parameter obtained by training the blood pressure prediction model based on PPG training data and standard blood pressure training data using the first device; the pre-training parameter includes a first initial parameter and a second initial parameter, wherein the first parameter is the first initial parameter.

2. The blood pressure measurement method according to claim 1, characterized in that, Before using the first data and the second data as the second PPG data, and before using the first standard data and the second standard data as the gold standard data, the method further includes: Determine whether the absolute value of the difference between the first standard data and the second standard data is greater than or equal to a preset difference; If the absolute value of the difference between the first standard data and the second standard data is greater than or equal to the preset difference, then the first data and the second data are used as the second PPG data, and the first standard data and the second standard data are used as gold standard data. If the absolute value of the difference between the first standard data and the second standard data is less than the preset difference, then new second data and new second standard data are acquired again until the difference between the new second standard data and the first standard data is greater than or equal to the preset difference. The first data and the new second data are used as the second PPG data, and the first standard data and the new second standard data are used as the gold standard data.

3. The blood pressure measurement method according to claim 1, characterized in that, The process of determining the user's blood pressure measurement result based on the first PPG data using a pre-trained blood pressure prediction model includes: The first test result is determined based on the first PPG data using the pre-trained blood pressure prediction model. Based on the initial test results, the confidence decay algorithm, and the first test results, the user's blood pressure measurement results are determined; the initial test results are the initial test results output by the pre-trained blood pressure prediction model for the user.

4. The blood pressure measurement method according to claim 3, characterized in that, The step of determining the user's blood pressure measurement result based on the initial test result, the confidence decay algorithm, and the first test result includes: Multiply the first test result by the confidence value of the confidence decay algorithm to obtain the first product; The initial test result is multiplied by the uncertainty value of the confidence decay algorithm to obtain the second product; The sum of the first product and the second product is taken as the user's blood pressure measurement result.

5. The blood pressure measurement method according to any one of claims 1-4, characterized in that, The pre-trained blood pressure prediction model includes a feature extraction layer and a linear layer, wherein the first parameter is the parameter of the feature extraction layer and the second parameter is the parameter of the linear layer.

6. A method for measuring blood pressure, characterized in that, Applied to a second device, including: Acquire first PPG data; the first PPG data is obtained by the first device from the photoplethysmography (PPG) signal collected from the user. The blood pressure measurement result of the user is determined based on the first PPG data using a pre-trained blood pressure prediction model. The model parameters of the pre-trained blood pressure prediction model include a first parameter and a second parameter. The first parameter is obtained by training the blood pressure prediction model using the first device, and the second parameter is obtained by the second device through transfer learning when the first parameter is frozen. The second parameter is obtained in the following way: The system receives second PPG data and corresponding gold standard data sent by the first device. The second PPG data includes first data and second data, where the first data is the PPG signal collected from the user by the first device in a first state, and the second data is the PPG signal collected from the user in a second state. The gold standard data includes first standard data and second standard data, where the first standard data is the user's standard blood pressure in the first state, and the second standard data is the user's standard blood pressure in the second state. The first state and the second state are different. Based on the second PPG data and the gold standard data, the second initial parameter in the pre-training parameters is calibrated to obtain the second parameter; the pre-training parameter is the parameter obtained by training the blood pressure prediction model based on PPG training data and standard blood pressure training data through the first device; the pre-training parameter includes a first initial parameter and a second initial parameter, wherein the first parameter is the first initial parameter.

7. An electronic device, characterized in that, include: Processor and memory; The memory stores one or more computer programs, each including instructions; when the instructions are executed by the processor, the electronic device performs the blood pressure measurement method as described in any one of claims 1-5, or performs the blood pressure measurement method as described in claim 6.

8. A computer storage medium, characterized in that, The device includes computer instructions that, when executed on an electronic device, enable the electronic device to perform the blood pressure measurement method as described in any one of claims 1-5, or to perform the blood pressure measurement method as described in claim 6.

9. A computer program product, characterized in that, When the computer program product is run on a computer, the computer performs the blood pressure measurement method as described in any one of claims 1-5, or performs the blood pressure measurement method as described in claim 6.

Images