Noninvasive dynamic blood pressure monitoring method and device, storage medium and computer

Abstract

The invention discloses a non-invasive dynamic blood pressure monitoring method and device, a storage medium and a computer. The method comprises the following steps: acquiring a first waveform signal and a second waveform signal respectively acquired by a first photoelectric sensor and a second photoelectric sensor; monitoring whether the first waveform signal is at a first amplitude extreme point, and determining a first time point corresponding to the first amplitude extreme point when the first waveform signal is at the first amplitude extreme point; when monitoring that the first waveform signal is at the first amplitude extreme point, monitoring whether the second waveform signal is at the second amplitude extreme point, and when the second waveform signal is at the second amplitude extreme point, determining a second time point corresponding to the second amplitude extreme point; and determining the time length between the first time point and the second time point as pulse transmission time, and inputting the pulse transmission time into the blood pressure prediction model to obtain a blood pressure monitoring value of the blood pressure monitoring object. According to the technical scheme, the accuracy of monitoring the blood pressure of the user can be improved.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a non-invasive dynamic blood pressure monitoring method, device, storage medium and computer. Background Art

[0002] With advances in medical technology and growing interest in cardiovascular health, people are increasingly concerned about cardiovascular diseases. Hypertension, as one of the most common cardiovascular diseases, has received widespread attention. The ability to continuously and accurately monitor the blood pressure of hypertensive patients is crucial for the prevention, diagnosis, and treatment of these diseases. In the field of blood pressure monitoring, photoplethysmography (PPG), an emerging non-contact biosignal monitoring technology, has gained widespread application. Currently, devices using PPG technology typically consist of a light source (such as an LED) and a light detector (such as a photodiode). Light emitted by the light source enters the skin tissue. As it penetrates the tissue, it is absorbed or scattered by blood and other tissue components, while some of the light is reflected back to the light detector, which receives the received light signal as a PPG signal. Because blood volume changes periodically as the heart pumps, the PPG signal collected by the light detector also exhibits periodic fluctuations, causing the PPG signal waveform to synchronize with the pulse wave waveform.

[0003] PPG technology, with its advantages of being non-invasive, convenient, and low-cost, has shown great potential in the field of blood pressure monitoring. Currently, the main blood pressure monitoring method using PPG technology is the electrocardiogram (ECG)-photoplethysmography (PPG) blood pressure monitoring method. The ECG-PPG blood pressure monitoring method collects the patient's ECG and PPG signals and determines the patient's blood pressure value based on the ECG and PPG signals, enabling non-invasive estimation of the patient's arterial blood pressure.

[0004] However, ECG signals are relatively weak and susceptible to interference. In actual use, factors such as electromagnetic interference in the hospital environment and changes in the patient's skin condition will affect the ECG signal, causing interference during reception and transmission of the ECG signal, which may affect the stability and reliability of the ECG signal, making the acquired ECG signal unable to accurately represent the patient's electrocardiogram condition, thereby making the ECG-PPG blood pressure monitoring method less accurate in blood pressure monitoring. Summary of the Invention

[0005] In view of this, the present application provides a non-invasive dynamic blood pressure monitoring method, device, storage medium and computer, the main purpose of which is to solve the technical problem of low monitoring accuracy when monitoring the user's blood pressure.

[0006] According to a first aspect of the present invention, a non-invasive dynamic blood pressure monitoring method is provided, the method comprising:

[0007] Acquiring in real time a first waveform signal acquired by a first photoelectric sensor and a second waveform signal acquired by a second photoelectric sensor, wherein the first photoelectric sensor is disposed at a proximal end of a blood pressure monitoring subject and the second photoelectric sensor is disposed at a distal end of the blood pressure monitoring subject;

[0008] monitoring whether the first waveform signal is at a first amplitude extreme value point, and when the first waveform signal is at the first amplitude extreme value point, determining a first time point corresponding to the first amplitude extreme value point and monitoring whether the second waveform signal is at a second amplitude extreme value point, and when the second waveform signal is at the second amplitude extreme value point, determining a second time point corresponding to the second amplitude extreme value point;

[0009] The time length between the first time point and the second time point is determined as the pulse transmission time, and the pulse transmission time is input into a pre-trained blood pressure prediction model so that the blood pressure prediction model obtains the blood pressure monitoring value of the blood pressure monitoring object based on the pulse transmission time.

[0010] In an optional embodiment, the first photoelectric sensor includes a first light detector and a first light emitter; the first light emitter is used to transmit a first light signal to the skin tissue of the blood pressure monitoring subject, and the first light detector is used to collect the first light signal reflected by the skin tissue to obtain the first waveform signal; the second photoelectric sensor includes a second light detector and a second light emitter; the second light emitter is used to transmit a second light signal to the skin tissue of the blood pressure monitoring subject, and the second light detector is used to collect the second light signal reflected by the skin tissue to obtain the second waveform signal.

[0011] In an optional embodiment, the blood pressure prediction model is a regression model; the training method of the blood pressure prediction model includes: obtaining multiple intercept training data groups and multiple regression coefficient training data groups, wherein the intercept training data group includes a preset DC component, a preset pulse signal extreme value, a first preset pulse transmission time value and a first preset blood pressure value, and the regression coefficient training data group includes a second preset pulse transmission time value and a second preset blood pressure value; based on the preset DC component, the preset pulse signal extreme value, the first preset pulse transmission time value and the first preset blood pressure value in the multiple intercept training data groups, determining the intercept value of the regression model, and bringing the intercept value into the regression model to obtain a process blood pressure prediction model; based on the second preset pulse transmission time value and the second preset blood pressure value in the multiple regression coefficient training data groups, determining the regression coefficient value of the process blood pressure prediction model to obtain the blood pressure prediction model.

[0012] In an optional embodiment, the real-time acquisition of the first waveform signal collected by the first photoelectric sensor and the second waveform signal collected by the second photoelectric sensor includes: acquiring the first waveform signal collected by the first photoelectric sensor and the second waveform signal collected by the second photoelectric sensor; performing denoising processing on the first waveform signal to obtain the first waveform signal after denoising, and performing denoising processing on the second waveform signal to obtain the second waveform signal after denoising; respectively determining whether there are signal loss points in the first waveform signal and the second waveform signal; when there are signal loss points in the first waveform signal, performing interpolation processing on the signal loss points to obtain the first waveform signal after interpolation processing, and when there are signal loss points in the second waveform signal, performing interpolation processing on the signal loss points to obtain the second waveform signal after interpolation processing.

[0013] In an optional embodiment, when there is a signal loss point in the first waveform signal, the signal loss point is interpolated to obtain a first waveform signal after interpolation, and when there is a signal loss point in the second waveform signal, the signal loss point is interpolated to obtain a second waveform signal after interpolation, the method also includes: normalizing the first waveform signal and the second waveform signal respectively to obtain the normalized first waveform signal and the second waveform signal.

[0014] In an optional embodiment, the blood pressure prediction model includes a systolic pressure prediction sub-model and a diastolic pressure prediction sub-model; inputting the pulse transmission time into the pre-trained blood pressure prediction model so that the blood pressure prediction model obtains the blood pressure monitoring value of the blood pressure monitoring object based on the pulse transmission time, including: inputting the pulse transmission time into the systolic pressure prediction sub-model so that the systolic pressure prediction sub-model obtains the systolic pressure value of the blood pressure monitoring object based on the pulse transmission time; inputting the pulse transmission time into the diastolic pressure prediction sub-model so that the diastolic pressure prediction sub-model obtains the diastolic pressure value of the blood pressure monitoring object based on the pulse transmission time; and determining the mean arterial blood pressure value of the blood pressure monitoring object based on the systolic pressure value and the diastolic pressure value.

[0015] In an optional embodiment, after inputting the pulse transmission time into a pre-trained blood pressure prediction model so that the blood pressure prediction model obtains the blood pressure monitoring value of the blood pressure monitoring object based on the pulse transmission time, the method further includes: comparing the blood pressure monitoring value with a preset normal blood pressure range to determine whether the blood pressure monitoring value is within the normal blood pressure range; when the blood pressure monitoring value is not within the normal blood pressure range, sending a blood pressure abnormality alarm message to a remote host computer.

[0016] According to a second aspect of the present invention, a non-invasive dynamic blood pressure monitoring device is provided, the device comprising:

[0017] a waveform acquisition module, configured to acquire, in real time, a first waveform signal acquired by a first photoelectric sensor and a second waveform signal acquired by a second photoelectric sensor, wherein the first photoelectric sensor is disposed at the proximal end of the blood pressure monitoring subject and the second photoelectric sensor is disposed at the distal end of the blood pressure monitoring subject;

[0018] a waveform monitoring module, configured to monitor whether the first waveform signal is at a first amplitude extreme value point, and when the first waveform signal is at the first amplitude extreme value point, determine a first time point corresponding to the first amplitude extreme value point and monitor whether the second waveform signal is at a second amplitude extreme value point, and when the second waveform signal is at the second amplitude extreme value point, determine a second time point corresponding to the second amplitude extreme value point;

[0019] The result output module is used to determine the time length between the first time point and the second time point as the pulse transmission time, and input the pulse transmission time into a pre-trained blood pressure prediction model so that the blood pressure prediction model obtains the blood pressure monitoring value of the blood pressure monitoring object based on the pulse transmission time.

[0020] According to a third aspect of the present invention, a storage medium is provided, on which a computer program is stored. When the program is executed by a processor, the above-mentioned non-invasive dynamic blood pressure monitoring method is implemented.

[0021] According to a fourth aspect of the present invention, a computer is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the above-mentioned non-invasive dynamic blood pressure monitoring method when executing the program.

[0022] The present invention provides a non-invasive dynamic blood pressure monitoring method, device, storage medium and computer. First, a first photoelectric sensor and a second photoelectric sensor are determined to be arranged at different positions on the patient's body to collect light signals reflected by the patient's skin; then, a first waveform signal collected by the first photoelectric sensor is monitored in real time to determine whether the first waveform signal is at a first amplitude extreme point, and when the first waveform signal is at the first amplitude extreme point, a first time point when the first photoelectric sensor is at the first amplitude extreme point is determined; then, when the first waveform signal is monitored to be at the first amplitude extreme point, a second waveform signal collected by the second photoelectric sensor is monitored to determine whether the second waveform signal is at a second amplitude extreme point, and when the second waveform signal is at the second amplitude extreme point, a second time point when the second photoelectric sensor is at the second amplitude extreme point is determined; finally, the pulse transmission time is input into a pre-trained blood pressure prediction model, so that the blood pressure prediction model obtains the patient's blood pressure monitoring value based on the pulse transmission time. The technical solution provided by this application uses two photoelectric sensors at different locations to respectively collect light signals reflected by the patient's skin as PPG signals, and calculates the time difference between the PPG signals collected by the two photoelectric sensors. This time difference is used to determine the patient's pulse transmission time, and the pulse transmission time is input into a blood pressure prediction model to output the patient's blood pressure value through the model. In this way, without the need to collect the patient's electrocardiogram signal, the patient's blood pressure value is determined based on the dual PPG signals and a deep learning model, avoiding the problem of inaccurate blood pressure calculation caused by interference with the electrocardiogram signal, thereby improving the monitoring accuracy during blood pressure monitoring.

[0023] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of this application. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0025] Figure 1 A schematic diagram showing a flow chart of a non-invasive dynamic blood pressure monitoring method provided by an embodiment of the present invention;

[0026] Figure 2 A schematic diagram of a configuration of a first photoelectric sensor and a second photoelectric sensor provided by an embodiment of the present invention is shown;

[0027] Figure 3 A schematic diagram of a first waveform signal and a second waveform signal provided by an embodiment of the present invention is shown;

[0028] Figure 4 A schematic diagram showing a first waveform signal before and after denoising, and a second waveform signal before and after denoising, provided by an embodiment of the present invention;

[0029] Figure 5 A schematic structural diagram of a non-invasive dynamic blood pressure monitoring device provided by an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0030] The present invention will be described in detail below with reference to the accompanying drawings and in combination with embodiments. It should be noted that, unless there is a conflict, the embodiments and features in the embodiments of the present application can be combined with each other.

[0031] Currently, the blood pressure monitoring method using PPG technology is mainly the ECG-PPG blood pressure monitoring method. The ECG-PPG blood pressure monitoring method collects the patient's ECG signal and PPG signal, and determines the patient's blood pressure value based on the ECG signal and PPG signal, thereby achieving non-invasive estimation of the patient's arterial blood pressure. However, the ECG signal is relatively weak and susceptible to interference. In actual use, factors such as electromagnetic interference in the hospital environment and changes in the patient's skin condition will affect the ECG signal, causing the ECG signal to be interfered with during reception and transmission, which may affect the stability and reliability of the ECG signal, making the acquired ECG signal unable to accurately represent the patient's electrocardiogram condition, thereby making the ECG-PPG blood pressure monitoring method less accurate in blood pressure monitoring.

[0032] In one embodiment, Figure 1 As shown, a non-invasive dynamic blood pressure monitoring method is provided. The method is described by taking the application of the method to a non-invasive dynamic blood pressure monitoring system as an example. Here, the non-invasive dynamic blood pressure monitoring system includes a controller, a first photoelectric sensor, and a second photoelectric sensor. The controller can be a computer device such as a microcontroller, and the non-invasive dynamic blood pressure monitoring method is executed by the controller. Furthermore, the non-invasive dynamic blood pressure monitoring method includes the following steps:

[0033] 101. Acquire in real time a first waveform signal collected by a first photoelectric sensor and a second waveform signal collected by a second photoelectric sensor.

[0034] The first and second photosensors are respectively disposed at different locations on the body of a subject undergoing blood pressure monitoring; the subject may be a patient or user undergoing blood pressure monitoring. During actual blood pressure monitoring, the first photosensor may be disposed at the proximal end of the subject, and the second photosensor may be disposed at the distal end of the subject. The proximal end refers to the end or location on the subject's body that is closer to the heart, and the distal end refers to the end or location on the subject's body that is further away from the heart than the proximal end. The proximal and distal ends may be located on the same artery.

[0035] Further, such as Figure 2 As shown, the first photoelectric sensor includes a first light detector 211 and a first light emitter 212; the second photoelectric sensor includes a second light detector 221 and a second light emitter 222. The first light detector 211, the first light emitter 212, the second light detector 221 and the second light emitter 222 can be arranged on the signal collector 200.

[0036] In which, when performing blood pressure monitoring, the signal collector 200 can be set at the finger, wrist or arm of the blood pressure monitoring object, the first light emitter 212 is used to transmit a first light signal to the skin tissue of the blood pressure monitoring object, and the first light detector 211 is used to collect the first light signal reflected by the skin tissue to obtain the first waveform signal; specifically, when the heart beats, the blood volume in the peripheral blood vessels undergoes pulsating changes, and the first light emitter 212 transmits a light signal to the patient's skin. The light signal emitted by the first light emitter 212 is reflected by the skin tissue and received by the first light detector 211, and the first light detector 211 determines the waveform of the received light signal, and determines the waveform of the light signal as the first waveform signal.

[0037] Furthermore, the second light emitter 222 is used to emit a second light signal toward the skin tissue of the blood pressure monitoring subject, and the second light detector 221 is used to collect the second light signal reflected by the skin tissue to obtain the second waveform signal. Here, when the heart beats, the second light emitter 222 emits a light signal toward the patient's skin. The light signal emitted by the second light emitter 222 is reflected by the skin tissue and received by the second light detector 221. The second light detector 221 determines the waveform of the received light signal and determines the waveform of the light signal as the second waveform signal. Here, the first waveform signal and the second waveform signal can serve as PPG waveforms.

[0038] Furthermore, the first light detector 211 can send the collected first waveform signal to the controller in real time, and the second light detector 221 can send the collected second waveform signal to the controller in real time, so that the controller can execute subsequent steps.

[0039] Furthermore, after receiving the first waveform signal and the second waveform signal, the controller can perform waveform morphological feature analysis on the first waveform signal and the second waveform signal to determine the waveform features of the first waveform signal and the second waveform signal. Specifically, the form of the first waveform signal and the second waveform signal can be as follows: Figure 3 As shown, here, it can be determined that there are certain correlation characteristic parameters between the main wave height, dicrotic wave height, descending isthmus relative height, etc. of the first waveform signal PPG1 and the second waveform signal PPG2 and the blood pressure, so as to determine the waveform characteristics of the first waveform signal PPG1 and the second waveform signal PPG2.

[0040] Specifically, you can create a PeakFinder class to configure and initialize parameters and data structures related to the "AMPD" peak detection algorithm, including setting the zoom range, cache size, and warm-up parameters. The AMPD (Automatic Multiscale Peak Detection) algorithm is an automatic multiscale peak detection method specifically designed to accurately identify peaks from periodic or quasi-periodic signals, while PeakFinder is a signal processing tool primarily used to detect peaks in signals.

[0041] Here, we can create two double-ended queues to store the original and downsampled data points in the first waveform signal PPG1 and the second waveform signal PPG2, respectively, and set properties for the peak detection state. Furthermore, we can create a reset method to reset the internal state of the PeakFinder object, ensuring that the PeakFinder object is in a known and consistent state each time a new peak detection task is started, thereby ensuring the accuracy of the algorithm.

[0042] Furthermore, the "AMPD" peak detection algorithm is used to extract the PTT pulse transmission time (PTT) from the first waveform signal (PPG1) and the second waveform signal (PPG2). The algorithm first stores and pre-processes the downsampled data. The downsampled data is then compared at different scales to detect possible peaks and valleys, and the results are stored in pdata. The peak index is then refined by searching a range around a given peak (calculated using self.downsample) to find a more accurate peak index for the waveform signal.

[0043] 102. Monitor whether the first waveform signal is at a first amplitude extreme point. When the first waveform signal is at the first amplitude extreme point, determine a first time point corresponding to the first amplitude extreme point and monitor whether the second waveform signal is at a second amplitude extreme point. When the second waveform signal is at the second amplitude extreme point, determine a second time point corresponding to the second amplitude extreme point.

[0044] The first amplitude extreme point of the first waveform signal can be a peak or a trough of the first waveform signal, and the second amplitude extreme point of the second waveform signal can also be a peak or a trough of the second waveform signal. When the first amplitude extreme point is a peak, the second amplitude extreme point must also be a peak. Similarly, when the first amplitude extreme point is a trough, the second amplitude extreme point must also be a trough. For ease of description, the first and second amplitude extreme points in this embodiment are both described as peaks of the signal. The same applies to the case where both the first and second amplitude extreme points are troughs of the signal.

[0045] Here, because the first photoelectric sensor is arranged at the proximal end of a large artery of the blood pressure monitoring object, and the second photoelectric sensor is arranged at the distal end of the artery, the received second waveform signal will lag behind the first waveform signal. After the received first waveform signal is at the peak, the received second waveform signal will be at the corresponding peak. The time that the second waveform signal lags behind the first waveform signal is the pulse transmission time PTT (Pulse Transit Time).

[0046] Here, the received first waveform signal and the second waveform signal may be in the form of Figure 3As shown, the first waveform signal PPG1 and the second waveform signal PPG2 are time-domain signals. Here, PPG waveform morphology analysis can be performed on the collected first and second waveform signals PPG1 and PPG2 to determine characteristic parameters associated with blood pressure, such as the main wave height, dicrotic wave height, and relative height of the descending mid-isthmus. This can be used to determine whether the first and second waveform signals PPG1 and PPG2 are at peaks or valleys. Specifically, a PeakFinder class can be created to configure and initialize parameters and data structures related to the "AMPD" peak detection algorithm, including setting the scaling range, buffer size, and warmup parameters. Then, two double-ended queues can be created to store the original and downsampled data points of the first and second waveform signals PPG1 and PPG2, respectively, and properties for peak detection status can be set. Furthermore, the data points of the first and second waveform signals PPG1 and PPG2 are preprocessed to obtain the downsampled first and second waveform signals PPG1 and PPG2. Here, a reset method can be created to reset the internal state of the PeakFinder object, ensuring that the PeakFinder object is in a known and consistent state each time a new peak detection task begins, thereby ensuring the accuracy of the algorithm. Next, the downsampled first waveform signal PPG1 is compared at different amplitude scales to detect the peak value of the first waveform signal PPG1, and the peak value detection result is stored in pdata. Here, a range around the peak value can be calculated based on self.downsample to search, and the peak value index can be refined to find a more accurate peak index. When the peak value within the range around the peak value is the largest, it is determined that the first waveform signal PPG1 is at its peak.

[0047] Further, when it is determined that the first waveform signal PPG1 is at the peak at this time, it can be determined that the first waveform signal PPG1 is at the first amplitude extreme point. At this time, the time point when the first waveform signal PPG1 is at the first amplitude extreme point is determined as the first time point.

[0048] Furthermore, when the first waveform signal PPG1 is detected to be at a peak, the second waveform signal PPG2 may be detected to be at a peak. The method for detecting whether the second waveform signal PPG2 is at a peak is the same as the method for detecting whether the first waveform signal PPG1 is at a peak, and will not be further described here.

[0049] Furthermore, when the second waveform signal PPG2 is monitored to be at a peak, it can be determined that the second waveform signal PPG2 is at a second amplitude extreme point. At this time, the time point when the second waveform signal PPG2 is at the second amplitude extreme point is determined as the second time point.

[0050] 103. Determine the time length between the first time point and the second time point as the pulse transmission time, and input the pulse transmission time into a pre-trained blood pressure prediction model, so that the blood pressure prediction model obtains the blood pressure monitoring value of the blood pressure monitoring object based on the pulse transmission time.

[0051] Specifically, the length of time the second time point lags behind the first time point, i.e., the time difference between the first and second time points, can be determined and used as the pulse transit time. Furthermore, the pulse transit time is input into a blood pressure prediction model to obtain the blood pressure monitoring value of the blood pressure monitoring subject.

[0052] Here, the blood pressure prediction model includes a pre-trained systolic pressure prediction sub-model and a pre-trained diastolic pressure prediction sub-model; wherein, the systolic pressure prediction sub-model is used to receive the pulse transmission time and output the systolic pressure value based on the pulse transmission time; the diastolic pressure prediction sub-model is used to receive the pulse transmission time and output the diastolic pressure value based on the pulse transmission time.

[0053] Furthermore, the pulse transmission time can be input into the systolic pressure prediction sub-model so that the systolic pressure prediction sub-model can obtain the systolic pressure value of the blood pressure monitoring object based on the pulse transmission time; and the pulse transmission time can be input into the diastolic pressure prediction sub-model so that the diastolic pressure prediction sub-model can obtain the diastolic pressure value of the blood pressure monitoring object based on the pulse transmission time.

[0054] Furthermore, based on the systolic blood pressure value and the diastolic blood pressure value, the mean arterial blood pressure value of the blood pressure monitoring subject is determined. Specifically, the mean arterial blood pressure value can be calculated according to Formula 1:

[0055]

[0056] Among them, MAP is the mean arterial blood pressure, DBP is the diastolic blood pressure, and SBP is the systolic blood pressure.

[0057] The non-invasive dynamic blood pressure monitoring method provided in this embodiment can use two photoelectric sensors at different locations to respectively collect light signals reflected by the patient's skin as PPG signals, calculate the time difference between the PPG signals collected by the two photoelectric sensors, determine the patient's pulse transmission time based on this time difference, and input the pulse transmission time into a blood pressure prediction model to output the patient's blood pressure value through the model. In this way, without the need to collect the patient's electrocardiogram signal, the patient's blood pressure value is determined based on dual PPG signals and a deep learning model, avoiding the problem of inaccurate calculation of the patient's blood pressure value due to interference with the electrocardiogram signal, thereby improving the monitoring accuracy during blood pressure monitoring.

[0058] In an optional embodiment, the method for obtaining the first waveform signal collected by the first photoelectric sensor and the second waveform signal collected by the second photoelectric sensor in real time in step 101 includes:

[0059] First, the first waveform signal collected by the first photosensor and the second waveform signal collected by the second photosensor are obtained. Then, the first waveform signal is subjected to denoising to obtain a denoised first waveform signal, and the second waveform signal is subjected to denoising to obtain a denoised second waveform signal.

[0060] Specifically, a filtering threshold can be set, and a digital filtering device such as a bandpass filter, a low-pass finite impulse response filter, a moving average filter, etc. can be used to perform denoising processing on the first waveform signal and the second waveform signal originally collected as the PPG signal to eliminate interference factors such as environmental noise and motion artifacts. Here, if Figure 4 As shown, the signal quality of the first waveform signal and the second waveform signal after the denoising process is significantly improved compared with the first waveform signal and the second waveform signal before the denoising process.

[0061] Then, it is determined whether there is a signal loss point in the first waveform signal and the second waveform signal, respectively. The signal loss point may be an unknown data point in the waveform signal.

[0062] Here, the first waveform signal and the second waveform signal can be subjected to Fourier transform or wavelet transform respectively to observe their spectral characteristics. If certain frequency components of the first waveform signal are abnormally abrupt or missing, it indicates that there are unknown data points in the first waveform signal. If certain frequency components of the second waveform signal are abnormally abrupt or missing, it indicates that there are unknown data points in the second waveform signal. In addition, the time series characteristics of the first waveform signal and the second waveform signal can also be checked, such as the periodicity, trend, and jump point of the signal. When the signal of the first waveform signal suddenly jumps or fluctuates irregularly, it indicates that there are unknown data points in the first waveform signal. Similarly, when the signal of the second waveform signal suddenly jumps or fluctuates irregularly, it indicates that there are unknown data points in the second waveform signal.

[0063] Then, when there is a signal loss point in the first waveform signal, the signal loss point is interpolated to obtain a first waveform signal after interpolation processing, and when there is a signal loss point in the second waveform signal, the signal loss point is interpolated to obtain a second waveform signal after interpolation processing.

[0064] Specifically, linear interpolation, polynomial interpolation, and wavelet coefficient-based interpolation can be used to interpolate the signal loss points in the first waveform signal and the second waveform signal to obtain complete first waveform signals and second waveform signals.

[0065] The technical solution provided in this application can perform denoising and interpolation processing on the first waveform signal and the second waveform signal collected by the first photoelectric sensor and the second photoelectric sensor respectively, eliminate the noise and interference signals in the waveform signals, ensure the continuity and integrity of the waveform signals, and thereby improve the accuracy of blood pressure monitoring.

[0066] In an optional embodiment, when there is a signal loss point in the first waveform signal, the signal loss point is interpolated to obtain a first waveform signal after interpolation, and when there is a signal loss point in the second waveform signal, the signal loss point is interpolated to obtain a second waveform signal after interpolation, the method further includes: normalizing the first waveform signal and the second waveform signal respectively to obtain the normalized first waveform signal and the second waveform signal, and performing the subsequent processing in steps 102 and 103 based on the normalized first waveform signal and the second waveform signal.

[0067] Here, normalizing the signal is an important step in signal processing, especially when it is necessary to compare signals from different individuals or under different acquisition conditions. Normalization can eliminate signal amplitude differences, baseline drift, and other influencing factors, thereby improving the comparability between signals.

[0068] Specifically, the collected first waveform signal and the second waveform signal can be subjected to minimum-maximum normalization (Min-Max Normalization), and the amplitudes of the waveform signals from different individuals and under different collection conditions can be scaled to a fixed range, such as [0, 1] or [-1, 1], so that the waveform signals from different individuals and under different collection conditions are comparable. Here, normalization algorithms such as Z-Score standardization and amplitude normalization can also be used to normalize the first waveform signal and the second waveform signal. This application does not limit this and is also applicable to this embodiment.

[0069] The technical solution provided in this application can normalize the first waveform signal and the second waveform signal collected in real time before feature extraction is performed on the first waveform signal and the second waveform signal, so that the first waveform signal and the second waveform signal collected in different environments and scenarios are comparable, making this method applicable to different blood pressure monitoring environments.

[0070] In an optional embodiment, the blood pressure prediction model may be a regression model, and the regression model may be in the form of a linear regression model or a nonlinear regression model. When the regression model is a linear regression model, the form of the regression model is shown in Formula 2:

[0071] y=kx+b (2)

[0072] Among them, y represents the blood pressure monitoring value, x represents the pulse transmission time, k is the regression coefficient value, and b is the intercept value.

[0073] Furthermore, when the regression model is a nonlinear regression model, the form of the regression model is shown in Formula 3:

[0074]

[0075] Among them, y represents the blood pressure monitoring value, x represents the pulse transmission time, k is the regression coefficient value, and b is the intercept value.

[0076] Furthermore, the training method of the blood pressure prediction model includes:

[0077] First, a plurality of intercept training data sets and a plurality of regression coefficient training data sets are obtained. The intercept training data set includes a preset direct current component, a preset pulse signal extreme value, a first preset pulse transmission time value, and a first preset blood pressure value, and the regression coefficient training data set includes a second preset pulse transmission time value and a second preset blood pressure value. Here, the preset direct current component can be a preset blood pressure direct current component (Direct Current Component, I DC ), in blood pressure monitoring, I DC It can be used to describe non-pulsatile components in the PPG signal, which are related to basic blood volume, respiration, the sympathetic nervous system, and body temperature regulation. Furthermore, the preset pulse signal extreme values ​​include the minimum blood pressure intensity (Minimum Intensity, Inin) and the maximum blood pressure intensity (Maximum Intensity, Imax). Furthermore, the first preset blood pressure value can include a first preset systolic pressure value and a first preset diastolic pressure value.

[0078] Here, the preset DC component, preset pulse signal extreme value, first preset pulse transmission time value, first preset systolic pressure value and first preset diastolic pressure value in a set of intercept training data groups can be the blood pressure DC component, blood pressure minimum intensity, blood pressure maximum intensity, pulse transmission time value, systolic pressure value and diastolic pressure value collected respectively during the real blood pressure monitoring process based on PPG technology in history.

[0079] Furthermore, the second preset blood pressure value may include a second preset systolic pressure value and a second preset diastolic pressure value. Here, the second preset pulse transit time value, the second preset systolic pressure value, and the second preset diastolic pressure value in the set of regression coefficient training data may be pulse transit time values ​​collected during actual blood pressure monitoring in history, and the systolic pressure value and diastolic pressure value of the patient determined based on the pulse transit time values.

[0080] Here, when obtaining the above parameters through blood pressure monitoring, a FeatureExtractor class object can be created. By calling the PeakFinder class and a double-ended queue, the peak / valley features of the PPG signal can be extracted from the data. Each time the peak / valley features are extracted, the internal state of the FeatureExtractor object is reset, two PPG signals are received, and the PeakFinder object is called to detect the peaks and valleys. Based on these peak and valley features, the blood pressure DC component, minimum blood pressure intensity, maximum blood pressure intensity, and pulse transmission time values ​​are updated or calculated as characteristic values. Based on these characteristic values, the patient's blood pressure value is determined.

[0081] Then, based on the preset DC component, the preset pulse signal extreme value, the first preset pulse transmission time value, and the first preset blood pressure value in the plurality of intercept training data sets, the intercept value of the regression model is determined, and the intercept value is substituted into the regression model to obtain a process blood pressure prediction model. Here, because the blood pressure prediction model includes a systolic pressure prediction sub-model and a diastolic pressure prediction sub-model, the systolic pressure prediction sub-model and the diastolic pressure prediction sub-model are regression models, respectively, and the systolic pressure prediction sub-model and the diastolic pressure prediction sub-model can be trained separately. Furthermore, the process blood pressure prediction model here includes a process systolic pressure prediction sub-model and a process diastolic pressure prediction sub-model.

[0082] Specifically, when the regression model is a linear regression model, the intercept values ​​corresponding to the systolic pressure prediction sub-model and the diastolic pressure prediction sub-model can be calculated respectively.

[0083] Formula 5 calculates the intercept value of the systolic blood pressure prediction submodel:

[0084]

[0085] b=A1+B1ln(I DC )(5)

[0086] Where C is a constant, A1, B1, C1 and D1 are the first, second, third and fourth undetermined coefficients calibrated by experiments, b is the intercept value, P SBP is the first preset systolic blood pressure value, I DC is the preset DC component, PTT is the first preset pulse transmission time value, and Imax is the maximum blood pressure intensity.

[0087] Then, "A1+B1ln(I DC )” as the unknown number, and substitute multiple intercept training data sets into Formula 4 respectively to solve “A1+B1ln(I DC )” and set “A1+B1ln(I DC )” is determined as the intercept value of the process shrinkage prediction sub-model.

[0088] Here, the intercept value of the diastolic pressure prediction sub-model can be calculated based on Formula 6 and Formula 7, where Formula 6 is:

[0089]

[0090] b=A2+B2ln(I DC )(7)

[0091] Where C is a constant, A2, B2, C2 and D2 are the fifth, sixth, seventh and eighth undetermined coefficients calibrated by experiments, and P DBP is the first preset diastolic pressure value, I DC is the preset DC component, PTT is the first preset pulse transmission time value, and Imin is the minimum blood pressure intensity.

[0092] Then, "A2+B2ln(I DC )” as the unknown number, and substitute multiple intercept training data sets into Formula 6 respectively to solve “A2+B2ln(I DC )” and set “A2+B2ln(I DC The value of )" was determined as the intercept value of the process diastolic pressure prediction sub-model.

[0093] Furthermore, based on the second preset pulse transmission time values ​​and the second preset blood pressure values ​​in the plurality of regression coefficient training data sets, the regression coefficient values ​​of the process blood pressure prediction model are determined to obtain the blood pressure prediction model.

[0094] Here, when the systolic pressure prediction sub-model is a linear regression model, the calculated intercept value of the process systolic pressure prediction sub-model can be substituted into Formula 2, and the regression coefficient value k can be used as an unknown number. Each group of regression coefficient training data groups can be substituted into Formula 2 respectively, and the second preset systolic pressure value in the regression coefficient training data group can be used as the blood pressure monitoring value y in Formula 2, and the second preset pulse transmission time value in the regression coefficient training data group can be used as the pulse transmission time x in Formula 2 to solve the value of the regression coefficient value k, so as to obtain the regression coefficient value in the process systolic pressure prediction sub-model, and to obtain a complete systolic pressure prediction sub-model.

[0095] Conversely, when the systolic pressure prediction sub-model is a nonlinear regression model, the calculated intercept value of the process systolic pressure prediction sub-model can be substituted into Formula 3, and the regression coefficient value k can be used as an unknown number. Each group of regression coefficient training data groups can be substituted into Formula 3 respectively, and the second preset systolic pressure value in the regression coefficient training data group can be used as the blood pressure monitoring value y in Formula 3, and the second preset pulse transmission time value in the regression coefficient training data group can be used as the pulse transmission time x in Formula 3 to solve the value of the regression coefficient value k and obtain the regression coefficient value in the process systolic pressure prediction sub-model.

[0096] Furthermore, after obtaining the regression coefficient value and the intercept value in the process systolic blood pressure prediction sub-model, the trained systolic blood pressure prediction sub-model can be obtained.

[0097] Furthermore, when the diastolic pressure prediction sub-model is a linear regression model, the calculated intercept value of the process diastolic pressure prediction sub-model can be substituted into Formula 2, and the regression coefficient value k can be used as an unknown number. Each group of regression coefficient training data groups can be substituted into Formula 2 respectively, and the second preset diastolic pressure value in the regression coefficient training data group can be used as the blood pressure monitoring value y in Formula 2, and the second preset pulse transmission time value in the regression coefficient training data group can be used as the pulse transmission time x in Formula 2 to solve the value of the regression coefficient value k and obtain the regression coefficient value in the process diastolic pressure prediction sub-model.

[0098] Furthermore, after obtaining the regression coefficient value and the intercept value in the process diastolic pressure prediction sub-model, the trained diastolic pressure prediction sub-model can be obtained.

[0099] Conversely, when the diastolic pressure prediction sub-model is a nonlinear regression model, the calculated intercept value of the process diastolic pressure prediction sub-model can be substituted into Formula 3, and the regression coefficient value k can be used as an unknown number. Each group of regression coefficient training data groups can be substituted into Formula 3 respectively, and the second preset diastolic pressure value in the regression coefficient training data group can be used as the blood pressure monitoring value y in Formula 3, and the second preset pulse transmission time value in the regression coefficient training data group can be used as the pulse transmission time x in Formula 2 to solve the value of the regression coefficient value k and obtain the regression coefficient value in the process diastolic pressure prediction sub-model.

[0100] Furthermore, after obtaining the regression coefficient value and the intercept value in the process diastolic pressure prediction sub-model, the trained diastolic pressure prediction sub-model can be obtained.

[0101] After obtaining the systolic pressure prediction sub-model and the diastolic pressure prediction sub-model, the trained model can be verified using an independent data set to evaluate the model's prediction accuracy and generalization ability, so as to optimize the model performance by continuously adjusting the model parameters. The technical solution provided in this application can train the systolic pressure prediction sub-model and the diastolic pressure prediction sub-model in the blood pressure prediction model separately, so that the trained systolic pressure prediction sub-model and the diastolic pressure prediction sub-model can automatically extract the characteristic parameters in the PPG signal and perform accurate blood pressure prediction, thereby giving the model strong learning and generalization capabilities, and being able to adapt to blood pressure changes in different individuals and under different monitoring conditions.

[0102] Furthermore, after inputting the pulse transit time into a pre-trained blood pressure prediction model in step 103 so that the blood pressure prediction model obtains a blood pressure monitoring value of the blood pressure monitoring subject based on the pulse transit time, the method further includes: first, comparing the blood pressure monitoring value with a preset normal blood pressure range to determine whether the blood pressure monitoring value is within the normal blood pressure range; wherein the blood pressure monitoring value may include a systolic pressure value and a diastolic pressure value output by the model, and a calculated mean arterial blood pressure value; the normal blood pressure range may include a normal systolic pressure range, a normal diastolic pressure range, and a normal mean arterial blood pressure range; wherein the normal systolic pressure range, the normal diastolic pressure range, and the normal mean arterial blood pressure range may be numerical ranges of systolic pressure, diastolic pressure, and mean arterial blood pressure values ​​of healthy persons, respectively, as determined through experiments or tests; if the patient's systolic pressure value is not within the normal systolic pressure range, or the diastolic pressure value is not within the normal diastolic pressure range, or the mean arterial blood pressure value is not within the normal mean arterial blood pressure range, it is determined that the blood pressure monitoring value is not within the normal blood pressure range, and further, it can be determined that the patient's blood pressure is abnormal.

[0103] Then, if the blood pressure monitoring value is not within the normal blood pressure range, a blood pressure abnormality alarm message is sent to a remote host computer. The host computer can be a computer terminal held by a doctor. If the patient's blood pressure monitoring value is not within the normal blood pressure range, the abnormal blood pressure condition can be promptly detected and handled, and the patient's blood pressure value can be sent to the host computer to provide the user with timely medical intervention and health management advice.

[0104] The technical solution provided in this application can perform real-time analysis of dual-channel PPG signals, and input the pulse transmission time obtained from the analysis into the blood pressure prediction model in real time to obtain continuous blood pressure monitoring results, and store the monitored blood pressure data locally or on a cloud server. At the same time, the blood pressure monitoring results can be displayed to users and medical staff in real time through mobile devices, computers and other terminals, thereby improving the monitoring effect of patients' blood pressure.

[0105] The noninvasive ambulatory blood pressure monitoring method provided in this embodiment can collect first and second waveform signals from different parts of the patient's body as PPG signals using a first photoelectric sensor and a second photoelectric sensor, respectively, to achieve dual-channel photoplethysmography (PPG) signal acquisition. The collected original first and second waveform signals are preprocessed using digital filtering techniques, interpolation algorithms, and normalization processing to eliminate noise and interference signals in the first and second waveform signals, ensuring the continuity and integrity of the first and second waveform signals. Furthermore, the pulse transit time is determined based on the time difference between the first and second waveform signals. This pulse transit time is used as input to a blood pressure prediction model to obtain the patient's real-time systolic and diastolic blood pressure values ​​and calculate the patient's mean arterial pressure. Furthermore, the method can also determine whether the patient's systolic, diastolic, and mean arterial pressure values ​​are abnormal. If any of the patient's blood pressure parameters are abnormal, an abnormality alarm is issued, enabling timely detection of abnormal blood pressure conditions and providing timely medical intervention and health management recommendations to the user.

[0106] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, storage, and display, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties. In addition, the numbers corresponding to the various steps in the above embodiments serve only as identifiers and do not limit the order in which the steps are executed. The order in which the steps are executed in each embodiment can be set according to actual circumstances.

[0107] Further, as Figure 1 The present embodiment provides a non-invasive dynamic blood pressure monitoring device, such as Figure 5 As shown, the device includes: a waveform acquisition module 51, a waveform monitoring module 52 and a result output module 53.

[0108] The waveform acquisition module 51 can be used to acquire in real time a first waveform signal acquired by a first photoelectric sensor and a second waveform signal acquired by a second photoelectric sensor, wherein the first photoelectric sensor is disposed at the proximal end of the blood pressure monitoring subject and the second photoelectric sensor is disposed at the distal end of the blood pressure monitoring subject;

[0109] The waveform monitoring module 52 may be configured to monitor whether the first waveform signal is at a first amplitude extreme value point, and when the first waveform signal is at the first amplitude extreme value point, determine a first time point corresponding to the first amplitude extreme value point and monitor whether the second waveform signal is at a second amplitude extreme value point, and when the second waveform signal is at the second amplitude extreme value point, determine a second time point corresponding to the second amplitude extreme value point;

[0110] The result output module 53 can be used to determine the time length between the first time point and the second time point as the pulse transmission time, and input the pulse transmission time into the pre-trained blood pressure prediction model so that the blood pressure prediction model obtains the blood pressure monitoring value of the blood pressure monitoring object based on the pulse transmission time.

[0111] In a specific application scenario, the waveform acquisition module 51 can be specifically used to obtain the first waveform signal collected by the first photoelectric sensor and the second waveform signal collected by the second photoelectric sensor; perform denoising on the first waveform signal to obtain the first waveform signal after denoising, and perform denoising on the second waveform signal to obtain the second waveform signal after denoising; determine whether there are signal loss points in the first waveform signal and the second waveform signal respectively; when there are signal loss points in the first waveform signal, perform interpolation on the signal loss points to obtain the first waveform signal after interpolation, and when there are signal loss points in the second waveform signal, perform interpolation on the signal loss points to obtain the second waveform signal after interpolation.

[0112] In a specific application scenario, the waveform acquisition module 51 can be specifically used to perform normalization processing on the first waveform signal and the second waveform signal respectively to obtain the normalized first waveform signal and the second waveform signal.

[0113] In a specific application scenario, the result output module 53 can be specifically used to input the pulse transmission time into the systolic pressure prediction sub-model, so that the systolic pressure prediction sub-model obtains the systolic pressure value of the blood pressure monitoring object based on the pulse transmission time; input the pulse transmission time into the diastolic pressure prediction sub-model, so that the diastolic pressure prediction sub-model obtains the diastolic pressure value of the blood pressure monitoring object based on the pulse transmission time; and determine the mean arterial blood pressure value of the blood pressure monitoring object based on the systolic pressure value and the diastolic pressure value.

[0114] In a specific application scenario, the result output module 53 can be used to compare the blood pressure monitoring value with the preset normal blood pressure range to determine whether the blood pressure monitoring value is within the normal blood pressure range; when the blood pressure monitoring value is not within the normal blood pressure range, a blood pressure abnormality alarm message is sent to a remote host computer.

[0115] Based on the above Figure 1 The method shown in FIG. 1 is a method for performing the above-mentioned operation. Accordingly, this embodiment further provides a storage medium on which a computer program is stored. When the program is executed by a processor, the above-mentioned Figure 1 The non-invasive ambulatory blood pressure monitoring method shown.

[0116] Based on this understanding, the technical solution of the present application can be embodied in the form of a software product. The software product to be identified can be stored in a non-volatile storage medium (which can be a CD-ROM, USB flash drive, mobile hard disk, etc.), including a number of instructions for enabling a computer (which can be a personal computer, server, or network device, etc.) to execute the methods described in each implementation scenario of the present application.

[0117] Based on the above Figure 1 The method shown, and Figure 5 In order to achieve the above-mentioned purpose, the embodiment of the non-invasive dynamic blood pressure monitoring device shown in the figure further provides a computer, which can be a personal computer, a server, a smart phone, a tablet computer, a smart watch, or other network devices, etc. The computer includes a storage medium and a processor; the storage medium is used to store computer programs and an operating system; the processor is used to execute the computer program to achieve the above-mentioned Figure 1 The method shown.

[0118] Optionally, the computer may further include an internal memory, a communication interface, a network interface, a camera, a radio frequency (RF) circuit, a sensor, an audio circuit, a Wi-Fi module, a display, an input device such as a keyboard, etc. Optionally, the communication interface may further include a USB interface, a card reader interface, etc. The network interface may optionally include a standard wired interface, a wireless interface (such as a Wi-Fi interface), etc.

[0119] Those skilled in the art will understand that the computer structure for identifying an operation action provided in this embodiment does not constitute a limitation on the computer, and may include more or fewer components, or a combination of certain components, or different component arrangements.

[0120] The storage medium may also include an operating system and a network communication module. The operating system is a program that manages the computer hardware and the software resources to be identified, supporting the operation of the information processing program and other software and / or programs to be identified. The network communication module is used to enable communication between components within the storage medium and with other hardware and software in the information processing computer.

[0121] Through the description of the above embodiments, those skilled in the art can clearly understand that the present application can be implemented by means of software plus the necessary general hardware platform, or by hardware. By applying the technical solution of the present application, first, a first waveform signal collected by a first photoelectric sensor and a second waveform signal collected by a second photoelectric sensor are acquired in real time, wherein the first photoelectric sensor is arranged at the proximal end of the blood pressure monitoring subject, and the second photoelectric sensor is arranged at the distal end of the blood pressure monitoring subject; then, the first waveform signal is monitored to see if it is at a first amplitude extreme point. When the first waveform signal is at the first amplitude extreme point, a first time point corresponding to the first amplitude extreme point is determined, and the second waveform signal is monitored to see if it is at a second amplitude extreme point. When the second waveform signal is at the second amplitude extreme point, a second time point corresponding to the second amplitude extreme point is determined; finally, the time length between the first time point and the second time point is determined as the pulse transmission time, and the pulse transmission time is input into a pre-trained blood pressure prediction model so that the blood pressure prediction model obtains the blood pressure monitoring value of the blood pressure monitoring subject based on the pulse transmission time. Compared with the existing technology, the accuracy of blood pressure monitoring of the user can be improved.

[0122] Those skilled in the art will understand that the accompanying drawings are only schematic diagrams of a preferred implementation scenario, and the modules or processes in the accompanying drawings are not necessarily required to implement the present application. Those skilled in the art will understand that the modules in the devices in the implementation scenario can be distributed in the devices of the implementation scenario according to the implementation scenario description, or can be changed accordingly and located in one or more devices different from the implementation scenario. The modules of the above-mentioned implementation scenario can be combined into one module, or can be further split into multiple sub-modules.

[0123] The serial numbers of the above application are for descriptive purposes only and do not represent the advantages or disadvantages of the implementation scenarios. The above disclosure only discloses several specific implementation scenarios of the present application, but the present application is not limited thereto. Any changes that can be conceived by those skilled in the art should fall within the scope of protection of the present application.

Claims

1. A non-invasive dynamic blood pressure monitoring method, characterized in that: The method comprises: Acquiring in real time a first waveform signal acquired by a first photoelectric sensor and a second waveform signal acquired by a second photoelectric sensor, wherein the first photoelectric sensor is disposed at a proximal end of a blood pressure monitoring subject and the second photoelectric sensor is disposed at a distal end of the blood pressure monitoring subject; monitoring whether the first waveform signal is at a first amplitude extreme value point, and when the first waveform signal is at the first amplitude extreme value point, determining a first time point corresponding to the first amplitude extreme value point and monitoring whether the second waveform signal is at a second amplitude extreme value point, and when the second waveform signal is at the second amplitude extreme value point, determining a second time point corresponding to the second amplitude extreme value point; The time length between the first time point and the second time point is determined as the pulse transmission time, and the pulse transmission time is input into a pre-trained blood pressure prediction model so that the blood pressure prediction model obtains the blood pressure monitoring value of the blood pressure monitoring object based on the pulse transmission time.

2. The method according to claim 1, characterized in that The first photosensor includes a first light detector and a first light emitter; The first light emitter is used to emit a first light signal toward the skin tissue of the blood pressure monitoring subject, and the first light detector is used to collect the first light signal reflected by the skin tissue to obtain the first waveform signal; The second photosensor includes a second light detector and a second light emitter; The second light emitter is used to emit a second light signal to the skin tissue of the blood pressure monitoring subject, and the second light detector is used to collect the second light signal reflected by the skin tissue to obtain the second waveform signal.

3. The method according to claim 1, characterized in that The real-time acquisition of the first waveform signal collected by the first photoelectric sensor and the second waveform signal collected by the second photoelectric sensor includes: Acquire the first waveform signal collected by the first photoelectric sensor and the second waveform signal collected by the second photoelectric sensor; Performing denoising on the first waveform signal to obtain a denoised first waveform signal, and performing denoising on the second waveform signal to obtain a denoised second waveform signal; Determining whether there is a signal loss point in the first waveform signal and the second waveform signal respectively; When there is a signal loss point in the first waveform signal, interpolation processing is performed on the signal loss point to obtain a first waveform signal after interpolation processing; and when there is a signal loss point in the second waveform signal, interpolation processing is performed on the signal loss point to obtain a second waveform signal after interpolation processing.

4. The method according to claim 3, characterized in that When a signal loss point exists in the first waveform signal, interpolation processing is performed on the signal loss point to obtain a first waveform signal after interpolation processing; and when a signal loss point exists in the second waveform signal, interpolation processing is performed on the signal loss point to obtain a second waveform signal after interpolation processing, the method further includes: Normalization processing is performed on the first waveform signal and the second waveform signal respectively to obtain the normalized first waveform signal and the second waveform signal.

5. The method according to claim 1, wherein The blood pressure prediction model includes a systolic pressure prediction sub-model and a diastolic pressure prediction sub-model; Inputting the pulse transmission time into a pre-trained blood pressure prediction model so that the blood pressure prediction model obtains a blood pressure monitoring value of the blood pressure monitoring subject based on the pulse transmission time includes: inputting the pulse transit time into the systolic blood pressure prediction sub-model so that the systolic blood pressure prediction sub-model obtains the systolic blood pressure value of the blood pressure monitoring subject based on the pulse transit time; inputting the pulse transit time into the diastolic pressure prediction sub-model so that the diastolic pressure prediction sub-model obtains the diastolic pressure value of the blood pressure monitoring subject based on the pulse transit time; The mean arterial blood pressure value of the blood pressure monitoring subject is determined based on the systolic blood pressure value and the diastolic blood pressure value.

6. The method according to claim 1, characterized in that The blood pressure prediction model is a regression model; the training method of the blood pressure prediction model includes: Acquire multiple intercept training data sets and multiple regression coefficient training data sets, wherein the intercept training data sets include a preset DC component, a preset pulse signal extreme value, a first preset pulse transit time value, and a first preset blood pressure value, and the regression coefficient training data sets include a second preset pulse transit time value and a second preset blood pressure value; Determining an intercept value of the regression model based on the preset DC component, the preset pulse signal extreme value, the first preset pulse transmission time value, and the first preset blood pressure value in the plurality of intercept training data sets, and substituting the intercept value into the regression model to obtain a process blood pressure prediction model; Based on the second preset pulse transmission time values ​​and the second preset blood pressure values ​​in the plurality of regression coefficient training data sets, the regression coefficient values ​​of the process blood pressure prediction model are determined to obtain the blood pressure prediction model.

7. The method according to claim 5, characterized in that After inputting the pulse transit time into a pre-trained blood pressure prediction model so that the blood pressure prediction model obtains a blood pressure monitoring value of the blood pressure monitoring subject based on the pulse transit time, the method further includes: Comparing the blood pressure monitoring value with a preset normal blood pressure range to determine whether the blood pressure monitoring value is within the normal blood pressure range; When the blood pressure monitoring value is not within the normal blood pressure range, a blood pressure abnormality alarm message is sent to a remote host computer.

8. A non-invasive dynamic blood pressure monitoring device, characterized in that: The device comprises: a waveform acquisition module, configured to acquire, in real time, a first waveform signal acquired by a first photoelectric sensor and a second waveform signal acquired by a second photoelectric sensor, wherein the first photoelectric sensor is disposed at the proximal end of the blood pressure monitoring subject and the second photoelectric sensor is disposed at the distal end of the blood pressure monitoring subject; a waveform monitoring module, configured to monitor whether the first waveform signal is at a first amplitude extreme value point, and when the first waveform signal is at the first amplitude extreme value point, determine a first time point corresponding to the first amplitude extreme value point and monitor whether the second waveform signal is at a second amplitude extreme value point, and when the second waveform signal is at the second amplitude extreme value point, determine a second time point corresponding to the second amplitude extreme value point; The result output module is used to determine the time length between the first time point and the second time point as the pulse transmission time, and input the pulse transmission time into a pre-trained blood pressure prediction model so that the blood pressure prediction model obtains the blood pressure monitoring value of the blood pressure monitoring object based on the pulse transmission time.

9. A storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images