Sleeveless continuous calibratable blood pressure estimation method and system, equipment and medium

Abstract

The invention discloses a cuff-free continuous calibratable blood pressure estimation method and system, equipment and a medium. The method comprises the following steps: establishing a blood pressure estimation model and a blood pressure calibration model, continuously estimating blood pressure based on PPG signals, calibrating blood pressure based on actively measured ECG and PPG pulse waves, and calibrating blood pressure based on actively measured ECG and PPG pulse waves and multi-modal information. The system comprises an acquisition module, a calculation module and a transmission module. According to the scheme, the cuff-free continuous calibratable blood pressure estimation process is more scientific and reasonable, the blood pressure measurement process of a user is more comfortable, the final blood pressure measurement result is accurate and is not affected by interference, meanwhile, the purpose of continuous blood pressure monitoring without frequent manual intervention is achieved, and the blood pressure measurement efficiency is improved. And the influence of the accuracy of blood pressure reading caused by individual differences is reduced.

Description

Technical Field

[0001] The present invention relates to the field of blood pressure measurement technology, and in particular to a cuffless, continuously calibrated blood pressure estimation method, system, device, and medium with a full emergency procedure accompanied by artificial intelligence-assisted guidance. Background Art

[0002] The statements in this section merely provide background technical information related to the present application and do not necessarily constitute prior art.

[0003] Blood pressure refers to the lateral pressure exerted by blood on the blood vessel wall per unit area. As a key physiological parameter of the human body, blood pressure is not only an important indicator of cardiovascular health, but also an important basis for assisting clinical diagnosis and treatment. Therefore, rapid and accurate blood pressure measurement is crucial. At present, blood pressure measurement methods are mainly divided into two categories: invasive and non-invasive.

[0004] Invasive blood pressure measurement: Invasive blood pressure measurement is considered the "gold standard" for blood pressure monitoring. This method uses an arterial catheter inserted directly into the artery to obtain continuous blood pressure readings. Although this method can provide the most accurate blood pressure data, it is complicated to operate and is accompanied by a series of potential risks, including hematoma, infection, arteriovenous fistula, and nerve damage. Therefore, invasive blood pressure measurement is mainly used for critically ill patients with unstable hemodynamics who need close monitoring.

[0005] Non-invasive blood pressure measurement method: Non-invasive blood pressure measurement method has been widely used in daily medical practice due to its simplicity and safety. Among them, the Korotkoff sounds method and the oscillometric method are the two most common methods. The Korotkoff sounds method relies on a stethoscope to capture Korotkoff sounds, but this method has significant individual differences and cannot provide continuous blood pressure data. The oscillometric method estimates blood pressure values ​​through the oscillation waves generated during the inflation and deflation of the cuff. However, this method is affected by the tightness of the cuff and the measurement time interval, resulting in limited measurement accuracy. In addition, neither method can provide continuous systolic and diastolic blood pressure data within the same cardiac cycle, which may interfere with the doctor's diagnosis.

[0006] Unlike traditional methods, continuous blood pressure monitoring devices can continuously monitor blood pressure and provide systolic and diastolic blood pressure data within the same cardiac cycle. At present, the mainstream non-invasive continuous non-invasive blood pressure monitoring methods include constant volume method, arterial tension method, pulse wave transit time (PTT) method, pulse wave characteristic parameter measurement method, etc. However, the arterial tension method and constant volume method require a continuously pressurized "cuff" device for interval measurement, which has limitations such as continuous pressurization, high risk of local ischemia, complex measurement device and high technical requirements, so clinical promotion is limited. In contrast, the PTT method and pulse wave characteristic parameter measurement method mainly use the continuous sampling of biological signals of PPG (Photoplethysmography, PPG) and ECG (Electrocardiogram, ECG) photoelectric sensors to conveniently and continuously measure PTT, and do not require equipment such as inflatable cuffs. They have the advantages of lower sensor accuracy requirements and simplified measurement devices, are suitable for long-term blood pressure monitoring, and eliminate the discomfort caused by the cuff, and have broad application prospects.

[0007] In the process of implementing the present invention, the applicant found that the currently used PTT method and pulse wave characteristic parameter determination method have at least the following problems:

[0008] 1. In the process of estimating blood pressure using photoplethysmography (PPG) signals, existing methods mainly rely on pulse wave characteristic parameters. Although this method can effectively track the changing trend of individual blood pressure, it is insufficient in reflecting the absolute value of blood pressure, especially when dealing with significant physiological differences between different individuals. It cannot provide accurate measurement data.

[0009] 2. In addition, although calculating the pulse wave transit time through the electrocardiogram (ECG) signal provides a potential alternative for blood pressure monitoring and theoretically supports continuous monitoring over a long period of time, the portability and comfort limitations of ECG equipment, as well as the measurement process that requires the user to actively participate, restrict the continuity and convenience of this method in practical applications.

[0010] 3. The pulse wave characteristic parameter determination method will encounter various noise interferences in the actual measurement process. These interferences will cause the quality of PPG data to deteriorate, thereby affecting the clarity of the extracted features, and further affecting the accuracy of blood pressure measurement.

[0011] Therefore, how to solve the problems existing in the existing cuffless continuous blood pressure monitoring technology, such as the inability to handle individual physiological differences, the limitations on continuity and convenience, and the poor measurement accuracy, has become the subject to be studied and solved by the present invention. Summary of the invention

[0012] The object of the present invention is to provide a cuff-free, continuously calibrable blood pressure estimation method, system, device and medium.

[0013] To achieve the above object, a first aspect of the present invention provides a cuffless continuous calibrated blood pressure estimation method, the cuffless continuous calibrated blood pressure estimation method comprising:

[0014] Establish a blood pressure estimation model and a blood pressure calibration model, establish a blood pressure estimation model based on multivariate independent variable PPG features, and establish a blood pressure calibration model based on ECG features and / or multimodal information;

[0015] Continuously estimate blood pressure based on PPG signals, continuously collect PPG signals, intercept the continuously collected PPG signals in a fixed time window, process the PPG signals within the window, extract PPG features after PPG signal processing, and estimate systolic and diastolic pressures based on the PPG features and the blood pressure estimation model;

[0016] Blood pressure calibration based on actively measured ECG and PPG pulse waves, collecting ECG signals and PPG signals, intercepting the collected ECG signals and PPG signals in a fixed time window, processing the ECG signals and PPG signals in the window, aligning the processed ECG signals and PPG signals, extracting ECG signal features and PPG features based on the aligned ECG signals and PPG signals, combining the PPG features and ECG signal features with a mapping relationship and adjusting the absolute baseline of blood pressure based on the pulse wave transmission time, and performing blood pressure calibration on the blood pressure estimation based on the blood pressure calibration model and the absolute baseline of blood pressure;

[0017] Blood pressure calibration based on actively measured ECG and PPG pulse waves and multimodal information, collects ECG signals and PPG signals, intercepts the collected ECG signals and PPG signals in a fixed time window, processes the ECG signals and PPG signals in the window, aligns the processed ECG signals and PPG signals, extracts ECG signal features based on the aligned ECG signals and PPG signals, combines PPG features and ECG signal features with a mapping relationship and the user's personal multimodal information, adjusts the individualized level baseline based on the pulse wave conduction time, and performs blood pressure calibration on the blood pressure estimation based on the blood pressure calibration model and the individualized level baseline.

[0018] The second aspect of the present invention provides a cuffless continuous calibrable blood pressure estimation system, the system is applied to the cuffless continuous calibrable blood pressure estimation method as described in the first aspect of the present invention, the system includes a collection module, a calculation module and a transmission module;

[0019] The acquisition module is used for continuous signal acquisition and active signal acquisition, wherein PPG signals are continuously acquired and ECG signals are actively acquired;

[0020] The calculation module has a built-in blood pressure estimation model and a blood pressure calibration model. The calculation module is used for continuous blood pressure estimation based on PPG signals, blood pressure calibration based on actively measured ECG and PPG pulse waves, and blood pressure calibration based on actively measured ECG and PPG pulse waves and multimodal information;

[0021] The transmission module is responsible for uploading the calculated blood pressure parameters and their timestamps to the receiving terminal.

[0022] The third aspect of the present invention proposes a smart wearable device for measuring blood pressure, which includes the system as described in the second aspect of the present invention, the smart wearable device for measuring blood pressure is worn on the finger or wrist of the user, the acquisition module includes a PPG sensor and an ECG sensor, the PPG sensor is facing the human body part, one end of the electrode in the ECG sensor is in contact with the surface of human skin, and the other end is selectively contacted by the other finger of the user to collect ECG signals, and the smart wearable device for measuring blood pressure is selectively connected to the receiving terminal signal.

[0023] A fourth aspect of the present invention provides a readable storage medium having a control program stored thereon. When the control program is executed by a computing module, the computing module executes the steps of the method described in the first aspect of the present invention.

[0024] The relevant contents of the present invention are explained as follows:

[0025] 1. The implementation of the above technical solution of the present invention is to solve the problems existing in the existing cuffless continuous blood pressure monitoring technology, such as the inability to handle individual physiological differences, limited continuity and convenience, and poor measurement accuracy. A cuffless continuous calibrated blood pressure estimation method, a cuffless continuous calibrated blood pressure estimation system, a blood pressure measurement intelligent wearable device and a readable storage medium are innovatively designed to improve the quality of PPG data and enhance the reliability of the extracted features, thereby overcoming the influence of noise interference on the accuracy of blood pressure measurement. In addition, by combining the PTT calibration algorithm of the ECG signal, personal static feature data, dynamic physiological signal data and blood glucose level data are added as features for modeling, so as to achieve the purpose of continuous blood pressure monitoring without frequent manual intervention, and reduce the influence of individual differences on the accuracy of blood pressure readings. In the cuffless continuous calibrated blood pressure estimation method, a blood pressure estimation model based on multivariate independent variable PPG features is first established, and a blood pressure calibration model based on ECG features and / or multimodal information is established to provide a scientific, reasonable, accurate, and quickly usable model basis for subsequent continuous blood pressure estimation based on PPG signals, blood pressure calibration based on actively measured ECG and PPG pulse waves, and blood pressure calibration based on actively measured ECG and PPG pulse waves and multimodal information; in the step of continuous blood pressure estimation based on PPG signals, the continuously collected PPG signals are intercepted with a fixed time window for subsequent processing, the PPG signals in the window are processed to remove interference to obtain a smoother PPG signal, the PPG features are extracted after the PPG signal is processed, and the systolic and diastolic pressures are estimated based on the PPG features and the blood pressure estimation model, so as to obtain the blood pressure parameters of the user during daily long-term wear, and the systolic pressure can be directly or indirectly estimated from the PPG features, which makes it simpler to obtain more accurate blood pressure parameters for daily reference and saves computing power, so that the smart wearable device for blood pressure measurement consumes less power and has a longer wearing time;In the step of blood pressure calibration based on actively measured ECG and PPG pulse waves, there is a process of indirectly inferring the blood pressure baseline level by using electrocardiogram (ECG) and photoplethysmogram (PPG) signals, and estimating the absolute blood pressure by the pulse wave conduction time PTT. In this step, the collected ECG signal and PPG signal are intercepted in a fixed time window, the ECG signal and PPG signal in the window are processed, and the processed ECG signal and PPG signal are aligned, so that the subsequent blood pressure processing and calibration can be calibrated with synchronized signals, making the signal processing and calibration process more unified, and making the PPG features and ECG signal features have a mapping relationship, and then combining the PPG features and ECG signal features with a mapping relationship and adjusting the blood pressure absolute level baseline based on the pulse wave conduction time, the process of blood pressure calibration for blood pressure estimation based on the blood pressure calibration model and the blood pressure absolute level baseline is more accurate, so that a more accurate absolute blood pressure can be output, thereby providing users with more accurate absolute blood pressure. Get more accurate daily blood pressure measurements; in the blood pressure calibration step based on actively measured ECG and PPG pulse waves and multimodal information, multimodal information is added to synchronously participate in the adjustment, and the individualized horizontal baseline is adjusted by the combination of the user's personal multimodal information and the pulse wave transmission time PTT to avoid inaccurate blood pressure measurement due to individual physiological differences of the user. The blood pressure estimation is calibrated by the individualized horizontal baseline and blood pressure calibration model, so that the blood pressure data measured by different individual users when using their own exclusive blood pressure measurement smart wearable devices are more in line with their own conditions, providing users with more stable, accurate and convenient blood pressure monitoring for evaluating daily blood pressure, and users do not need to use ECG measurement for a long time. They only need to actively measure the ECG signal within a certain period to obtain the absolute horizontal baseline or individualized horizontal baseline of blood pressure, and combine the blood pressure calibration model to calibrate the blood pressure estimation based on the continuous PPG signal, which is more convenient and burden-free to use. ;

[0026] 2. In the first aspect of the above technical solution, in the step of establishing a blood pressure calibration model based on ECG features and / or multimodal information, the multimodal information includes at least two of the following: personal static feature data, dynamic physiological signal data, and blood glucose level data. The personal static feature data includes age, gender, weight, height, health status, medication status, etc., the dynamic physiological signal data includes heart rate data, pulse wave data, blood oxygen data, etc., and the blood glucose level data includes fasting blood glucose, postprandial blood glucose and daily blood glucose data, etc., so as to establish a blood pressure calibration model related to different individual users, so that the blood pressure measured by different individual users is more accurate.

[0027] 3. In the first aspect of the above technical solution, in the step of continuously estimating blood pressure based on PPG signals, the user wears a blood pressure measuring smart wearable device that continuously measures blood pressure, wherein the blood pressure measuring smart wearable device is worn on the fingertips or wrists of the human body and continuously collects PPG signals to reflect changes in blood volume caused by heart beats;

[0028] In the step of blood pressure calibration based on actively measured ECG and PPG pulse waves or blood pressure calibration based on actively measured ECG and PPG pulse waves and multimodal information, the user actively measures, one end of the electrode in the blood pressure measurement smart wearable device is attached to the fingertip or wrist skin surface, and the other finger of the user touches the other end of the electrode to complete ECG signal data collection for at least a set time.

[0029] Through the implementation of the above method, users do not need to use ECG measurement for a long time. They only need to actively collect ECG signal data at intervals of half a day, a day or several days, and the time for each ECG signal data collection can be at least one and a half minutes, two minutes or other set time, which greatly improves the user experience and comfort, and will not limit the user's daily use.

[0030] 4. In the first aspect of the above technical solution, the step of processing the PPG signal in the window includes PPG signal preprocessing, PPG signal period division, and PPG signal quality verification; wherein:

[0031] PPG signal preprocessing: for the PPG signal in the window, a bandpass filter is applied to remove high-frequency noise and low-frequency drift in the PPG signal, and then a moving average or filter is used to reduce short-term fluctuations in the PPG signal, thereby obtaining a smoother PPG signal;

[0032] PPG signal cycle division: Based on the preprocessed PPG signal, the signal is divided into multiple cardiac cycles through peak and trough detection, providing a basis for subsequent feature calculation;

[0033] PPG signal quality verification: perform quality verification on multiple segmented cardiac cycles.

[0034] By implementing the above method, the processed PPG signal has a higher quality, so as to more accurately estimate the user's blood pressure level. The preprocessing step may include two parts: denoising and smoothing. Denoising removes high-frequency noise (such as muscle activity) and low-frequency drift (such as slow movement or ambient temperature changes) in the signal by applying a bandpass filter (such as a Butterworth filter). Smoothing reduces short-term fluctuations in the signal through a moving average or a Savitzky-Golay filter, thereby obtaining a smoother PPG signal.

[0035] 5. In the first aspect of the above technical solution, in the step of dividing the PPG signal cycle, a peak detection algorithm is used to detect the peak value of the PPG signal, and the PPG signal is divided into several cycles according to the peak and trough marks of two adjacent heart cycle signals, and the first and last cycles are removed to ensure that the cycle data is clean, and the low peak position is adjusted to be relative to the new starting point; the peak and trough detection can be achieved by a threshold method, derivative analysis or other adaptive algorithms. The detected peaks and troughs are used to divide the signal into multiple cardiac cycles, providing a basis for subsequent feature calculations.

[0036] In the step of PPG signal quality verification, the peaks and troughs of the waveform after segmentation are detected by curvature analysis. The positions of the peaks and troughs correspond to the points where the curvature changes significantly. For the peaks, the curvature reaches the local maximum, and the curve changes from convex to concave. The troughs correspond to the local minimum of the curvature, and the curve changes from concave to convex. The changes in curvature are analyzed to accurately find the positions of the peaks and troughs.

[0037] To ensure signal quality, multiple segmented cardiac cycles are quality checked. The quality check step may include checking the consistency of cycle length, signal integrity, and other factors that may affect feature extraction. Only cardiac cycle signals that pass the quality check will be used for subsequent feature calculations.

[0038] 6. In the first aspect of the above technical solution, in the step of processing the ECG signal and the PPG signal in the window, the ECG signal in the window is subjected to filtering technology to remove noise and interference signals in the ECG signal, and the PPG signal in the window is subjected to a bandpass filter to remove high-frequency noise and low-frequency drift in the PPG signal, and then a moving average or a filter is used to reduce short-term fluctuations in the PPG signal, thereby obtaining a smoother PPG signal;

[0039] In the step of extracting ECG signal features based on the aligned ECG signal and PPG signal, the position of the R wave is detected in the ECG signal, and the position of the peak is detected in the PPG signal, wherein the R wave is the waveform with the largest amplitude and the most stable in the ECG waveform, representing the beginning of ventricular contraction, and the peak represents the time point when the arterial pulse wave reaches the peripheral tissue.

[0040] This design is used to align the processed ECG signal and PPG signal so that the subsequent blood pressure processing and calibration can be calibrated with synchronized signals, making the signal processing and calibration process more unified, and the process of blood pressure calibration based on the blood pressure calibration model and the absolute blood pressure level baseline and individualized level baseline for blood pressure estimation more accurate.

[0041] 7. In the first aspect of the above technical solution, the low cutoff frequency of the PPG signal bandpass filter is set to 0.4Hz, and the low cutoff frequency of the ECG signal bandpass filter is set to 2.5Hz; the high cutoff frequency of the PPG signal is set to 4Hz, and the high cutoff frequency of the ECG signal is set to 50Hz. This is used to suppress the high-frequency noise and 50Hz power frequency interference generated by the light source and electronic devices, to cope with the influence of light and electrical noise, so that the final calibrated blood pressure value is more accurate.

[0042] 8. In the first aspect of the above technical solution, in the step of extracting PPG features after PPG signal processing and the step of extracting ECG signal features and PPG features according to the aligned ECG signal and PPG signal, the following signal features are used to describe the shape and dynamic characteristics of each signal waveform:

[0043] Enhancement Index AI: It indicates the ratio of the main peak to the diastolic peak amplitude in a single cycle waveform to reflect the enhancement characteristics of the PPG waveform. The calculation method is AI=amp dia / amp sys , where amp sys Indicates the main peak amplitude, amp dia It is expressed as the diastolic peak amplitude;

[0044] Inflection point area ratio IPA: It represents the ratio of the area before and after the trough point of the dicrotic wave in a single cycle waveform. The calculation method is IPA=area dia ⁄area sys , where area sys is the area of ​​the left part of the waveform, representing the part before the trough of the dicrotic wave, while area dia It is the part after the trough point of the heavy beat;

[0045] Systolic amplitude SA: represents the amplitude of the main peak in a single cycle waveform, recorded as amp sys ;

[0046] The time difference between the main wave and the diastolic peak TD: represents the time difference between the main wave peak and the diastolic peak in a single cycle waveform, calculated as TD=t dia -t sys , where t dia is the secondary peak arrival time, t sys is the main wave peak arrival time;

[0047] Main wave duration HRDN: It indicates the time difference between the main wave peak and the dicrotic wave trough in a single cycle waveform. It is calculated as HRDN=t dic -t sys , where t dic is the arrival time of the re-beat trough, t sysis the main wave peak arrival time;

[0048] Main peak interval PPI: represents the time difference between two adjacent period waveform pulses; the calculation method is , where t Sys is the main wave peak arrival time, is the arrival time of the main peak of the previous cycle;

[0049] Main wave peak width PW: represents the width of the main wave peak pulse waveform of a single cycle waveform, calculated as follows: ,in represents the half-amplitude arrival time on the left side of the main peak pulse, where It indicates the arrival time of the half amplitude on the right side of the main peak pulse;

[0050] Intra-cycle inflection wave and harmonic wave area ratio IHAR: It represents the energy ratio of the inflection wave and the periodic waveform in the Fourier domain of a single periodic waveform, and is calculated as follows: , where pow dia Indicates the energy spectrum of the turn wave in the Fourier domain, pow sum Represents the energy spectrum of the entire periodic waveform in the Fourier domain.

[0051] Through the above specific settings of the shape and dynamic characteristics of the relevant signal waveforms in the PPG features and ECG signal features, the parameters describe the amplitude, area, time interval and other characteristics of the waveform, which can help analyze the morphological characteristics of the waveform, such as the enhancement effect, pulse propagation characteristics and the relative relationship between different parts, so as to establish a data foundation for the application of blood pressure estimation model and blood pressure calibration model.

[0052] 9. In the first aspect of the above technical solution, in the step of establishing a blood pressure estimation model based on multivariate independent variable PPG features, a multivariate linear regression model is used to establish a linear relationship between the multivariate independent variable PPG features AI, IPA, SA, TD, HRTD, PPI, PW, IHAR and the blood pressure difference PP and the average blood pressure MAP, respectively, where PP=SBP-DBP, and the least squares method is used to estimate unknown parameters;

[0053] ;

[0054] ;

[0055] in is the error term, whose mean is zero;

[0056] After calculating MAP and PP, the systolic blood pressure (SBP) and diastolic blood pressure (DBP) are reconstructed as follows:

[0057] DBP=SBP-PP;

[0058] ;

[0059] in represents the fitting coefficient of the linear fitting model of PP; Represents the fitting coefficients of the linear fit.

[0060] The specific steps of establishing the blood pressure estimation model in the above manner make it possible to quickly and accurately estimate the systolic and diastolic blood pressure of the user during continuous measurement. Since the data has been pre-processed, the blood pressure estimation result is already relatively accurate.

[0061] 10. In the first aspect of the above technical solution, in the step of establishing a blood pressure calibration model based on ECG features and / or multimodal information, a large number of synchronized ECG and PPG signals and corresponding blood pressure measurements are collected, personal static feature data including age, gender, weight, height, health status, medication status, and personal blood sugar level data are collected, and a blood pressure calibration model between blood pressure and PTT, personal static features, and dynamic physiological signals is established using statistical or machine learning methods. This makes the application of the blood pressure calibration model more accurate.

[0062] 11. In the first aspect of the above technical solution, the step of establishing a blood pressure calibration model based on ECG features includes the following steps:

[0063] Calculate the pulse wave transit time PTT: PTT is defined as the pulse wave transit time from the heart to the distal artery, measured from the time interval between the R wave of the ECG and the peak of the PPG wave: , where t P is the time when the photoplethysmogram (PPG) reaches its peak, t R It is the time when the R wave appears in the electrocardiogram (ECG) signal;

[0064] The calibrated mean blood pressure MAP is obtained based on the relationship between vascular elasticity and blood flow * :

[0065] MAP * =C × PTT;

[0066] Where C represents the compliance parameter of the blood vessel, which is defined as the volume change caused by a unit pressure change. There is a linear relationship between PTT and blood pressure;

[0067] According to the calibrated MAP * , calculate the calibrated SBP * and DBP * .

[0068] The above are the specific steps of the blood pressure calibration model that uses only ECG for calibration. Through the implementation of this method, in a uniform and elastic vascular system, the increase in blood pressure will lead to further expansion of the vascular wall, which will speed up the propagation of the pressure wave and shorten the PTT. On the contrary, if the blood pressure decreases, the expansion of the vascular wall will decrease, the propagation speed of the pressure wave will slow down, and the PTT will be prolonged. There is a linear relationship between PTT and blood pressure. According to this linear relationship, the absolute baseline of blood pressure is adjusted to obtain the calibrated average blood pressure MAP. * , calculate the calibrated SBP * and DBP * , making the blood pressure data measured by the blood pressure calibration based on the actively measured ECG and PPG pulse waves more accurate.

[0069] 12. In the first aspect of the above technical solution, the step of establishing a blood pressure calibration model based on ECG features and multimodal information includes the following steps:

[0070] Confirm the type of multimodal information, which includes personal static features X sta , and dynamic physiological sign parameters X dyn And the measured blood pressure BP tr ; Personal static characteristics mainly include gender, age, weight, height, and medical history, and dynamic physiological sign parameters include heart rate, blood oxygen, and blood sugar measurement;

[0071] Calculate the compliance of the blood vessels. The compliance of the blood vessels can be expressed as the following function:

[0072] ;

[0073] Among them, α is a constant coefficient, E represents Young's modulus, h represents the thickness of the arterial wall, ρ represents the blood density, and r represents the arterial radius;

[0074] Calculate the calibrated individual vascular fitness parameter C * By establishing E, h, ρ, r and X sta and X dyn The multivariate linear relationship function f is used to calculate the calibrated individual vascular adaptability parameter C * :

[0075] ;

[0076] ;

[0077] Calibrated blood pressure via multimodality:

[0078] ;

[0079] ;

[0080] .

[0081] The above are the specific steps of the blood pressure calibration model calibrated using ECG features and multimodal information, which is more in line with the various characteristics of each individual user to provide users with more accurate blood pressure measurement and estimation, and provide data with more medical reference value.

[0082] Due to the application of the above scheme, the present invention has the following advantages and effects compared with the prior art:

[0083] 1. Through the implementation of the technical solution of the present invention, in view of the problems existing in the existing cuffless continuous blood pressure monitoring technology, such as the inability to handle individual physiological differences, limited continuity and convenience, and poor measurement accuracy, a cuffless continuous calibrated blood pressure estimation method, a cuffless continuous calibrated blood pressure estimation system, a blood pressure measurement intelligent wearable device and a readable storage medium are innovatively designed to improve the quality of PPG data and enhance the reliability of the extracted features, thereby overcoming the influence of noise interference on the accuracy of blood pressure measurement. In addition, by combining the PTT calibration algorithm of the ECG signal, personal static feature data, dynamic physiological signal data and blood glucose level data are added as features for modeling, so as to achieve the purpose of continuous blood pressure monitoring without frequent manual intervention, and reduce the influence of individual differences on the accuracy of blood pressure readings.

[0084] 2. Through the implementation of the technical solution of the present invention, in the cuffless continuous calibrated blood pressure estimation method, a blood pressure estimation model based on multivariate independent variable PPG features is first established, and a blood pressure calibration model based on ECG features and / or multimodal information is established, so as to provide a scientific, reasonable, accurate and quickly usable model basis for subsequent continuous blood pressure estimation based on PPG signals, blood pressure calibration based on actively measured ECG and PPG pulse waves, and blood pressure calibration based on actively measured ECG and PPG pulse waves and multimodal information; in the step of continuous blood pressure estimation based on PPG signals, by continuously sampling the blood pressure, the blood pressure is estimated based on the blood pressure of the patient, and the blood pressure is calibrated based on the blood pressure of the patient. The collected PPG signals are intercepted in a fixed time window for subsequent processing, and the PPG signals in the window are processed to remove interference to obtain a smoother PPG signal. After the PPG signal is processed, the PPG features are extracted, and the systolic and diastolic pressures are estimated based on the PPG features and the blood pressure estimation model, so as to obtain the blood pressure parameters of the user during daily long-term wear, and the systolic pressure can be directly or indirectly estimated from the PPG features, which makes it simpler to obtain more accurate blood pressure parameters for daily reference and saves computing power, so that the smart wearable device for blood pressure measurement consumes less power and can be worn longer.

[0085] 3. Through the implementation of the technical solution of the present invention, in the step of blood pressure calibration based on actively measured ECG and PPG pulse waves, it is a process of indirectly inferring the blood pressure baseline level by using electrocardiogram (ECG) and photoelectric volumetric pulse wave (PPG) signals, and estimating the absolute blood pressure by the pulse wave transmission time PTT. In this step, the collected ECG signal and PPG signal are intercepted with a fixed time window, the ECG signal and PPG signal in the window are processed, and the processed ECG signal and PPG signal are aligned, so that the subsequent blood pressure processing and calibration can be calibrated with synchronized signals, so that the signal processing and calibration process are more unified, and the PPG feature and the ECG signal feature have a mapping relationship. In the subsequent combination of the PPG feature and the ECG signal feature with a mapping relationship and the blood pressure absolute level baseline is adjusted based on the pulse wave transmission time, the process of blood pressure calibration for blood pressure estimation based on the blood pressure calibration model and the blood pressure absolute level baseline is more accurate, so that a more accurate absolute blood pressure can be output, thereby obtaining a more accurate daily blood pressure measurement value for the user.

[0086] 4. Through the implementation of the technical solution of the present invention, in the blood pressure calibration step based on actively measured ECG and PPG pulse waves and multimodal information, multimodal information is added to synchronously participate in the adjustment. The individualized horizontal baseline is adjusted by the combination of the user's personal multimodal information and the pulse wave transmission time PTT to avoid inaccurate blood pressure measurement due to individual physiological differences of the user. The blood pressure estimation is calibrated by the individualized horizontal baseline and the blood pressure calibration model, so that the blood pressure data measured by different individual users when using their own exclusive blood pressure measurement smart wearable devices are more in line with their own conditions, providing users with more stable, accurate and convenient blood pressure monitoring for evaluating daily blood pressure, and users do not need to use ECG measurement for a long time. Users only need to actively measure ECG signals within a certain period to obtain an absolute blood pressure horizontal baseline or an individualized horizontal baseline, and combine the blood pressure calibration model to calibrate the blood pressure estimation based on the continuous PPG signal, which is more convenient and burden-free to use.

[0087] 5. In summary, through the establishment of the blood pressure estimation model and the blood pressure calibration model, and the implementation of continuous blood pressure estimation based on PPG signals, blood pressure calibration based on actively measured ECG and PPG pulse waves, and blood pressure calibration based on actively measured ECG and PPG pulse waves and multimodal information, the multi-dimensional and cross-judgment use of the parameter characteristics of the continuously measured PPG signal, the actively measured ECG signal, and the collected multimodal information makes the process of cuffless continuous calibrated blood pressure estimation more scientific and reasonable, making the user's blood pressure measurement process more comfortable, and making the final blood pressure measurement results accurate and unaffected by interference, while achieving the purpose of continuous blood pressure monitoring without frequent manual intervention, reducing the impact of individual differences on the accuracy of blood pressure readings. BRIEF DESCRIPTION OF THE DRAWINGS

[0088] Figure 1 It is a flow chart of a method for estimating blood pressure without a cuff and with continuous calibration according to an embodiment of the present invention;

[0089] Figure 2 It is a schematic diagram of the overall process of the cuffless continuous calibrable blood pressure estimation method according to an embodiment of the present invention;

[0090] Figure 3 It is a block diagram of the composition of the cuffless continuous calibrable blood pressure estimation system according to an embodiment of the present invention;

[0091] Figure 4 This is a schematic diagram of comparing multiple groups of data before and after signal preprocessing according to an embodiment of the present invention;

[0092] Figure 5 It is a schematic diagram of the effect of waveform segmentation after filtering according to an embodiment of the present invention;

[0093] Figure 6 It is a schematic diagram of the effect of feature extraction after waveform segmentation according to an embodiment of the present invention. DETAILED DESCRIPTION

[0094] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are described in detail below in conjunction with the accompanying drawings. In the following description, many specific details are set forth to facilitate a full understanding of the present application. However, the present application can be implemented in many other ways different from those described herein, and those skilled in the art can make similar improvements without violating the connotation of the present application, so the present application is not limited by the specific embodiments disclosed below.

[0095] The terms used herein are only for describing specific embodiments and are not intended to be limiting of the present invention. Singular forms such as "a", "this", "here", "this" and "the" as used herein also include plural forms.

[0096] The terms “first”, “second”, etc. used in this document do not specifically refer to an order or sequence, nor are they used to limit the present case. They are only used to distinguish components or operations described with the same technical terms.

[0097] As used herein, “connected” or “positioned” may refer to two or more components or devices being in direct physical contact with each other, or being in indirect physical contact with each other, or may refer to two or more components or devices operating or moving with each other.

[0098] The words “include,” “including,” and “have” used in this document are open-ended terms, meaning including but not limited to.

[0099] The terms used in this document generally have the ordinary meaning of each term used in this field, in the context of the case and in the specific context, unless otherwise noted. Certain terms used to describe the present invention will be discussed below or elsewhere in this specification to provide additional guidance to those skilled in the art in describing the present invention.

[0100] The present invention aims to improve the quality of PPG data and enhance the reliability of extracted features, so as to overcome the influence of noise interference on the accuracy of blood pressure measurement, and by combining the PTT calibration algorithm of ECG signals, add personal static feature data, dynamic physiological signal data and blood glucose level data as features for modeling, so as to achieve the purpose of continuous blood pressure monitoring without frequent manual intervention, and reduce the influence of individual differences on the accuracy of blood pressure readings. Therefore, a cuffless continuously calibrated blood pressure estimation method, a cuffless continuously calibrated blood pressure estimation system, a blood pressure measurement intelligent wearable device and a readable storage medium are innovatively designed.

[0101] Embodiment 1: Embodiment 1 of the present invention provides a cuffless continuous calibrated blood pressure estimation method, the cuffless continuous calibrated blood pressure estimation method comprising:

[0102] Establish a blood pressure estimation model and a blood pressure calibration model, establish a blood pressure estimation model based on multivariate independent variable PPG features, and establish a blood pressure calibration model based on ECG features and / or multimodal information;

[0103] Continuously estimate blood pressure based on PPG signals, continuously collect PPG signals, intercept the continuously collected PPG signals in a fixed time window, process the PPG signals within the window, extract PPG features after PPG signal processing, and estimate systolic and diastolic pressures based on the PPG features and the blood pressure estimation model;

[0104] Blood pressure calibration based on actively measured ECG and PPG pulse waves, collecting ECG signals and PPG signals, intercepting the collected ECG signals and PPG signals in a fixed time window, processing the ECG signals and PPG signals in the window, aligning the processed ECG signals and PPG signals, extracting ECG signal features and PPG features based on the aligned ECG signals and PPG signals, combining the PPG features and ECG signal features with a mapping relationship and adjusting the absolute baseline of blood pressure based on the pulse wave transmission time, and performing blood pressure calibration on the blood pressure estimation based on the blood pressure calibration model and the absolute baseline of blood pressure;

[0105] Blood pressure calibration based on actively measured ECG and PPG pulse waves and multimodal information, collects ECG signals and PPG signals, intercepts the collected ECG signals and PPG signals in a fixed time window, processes the ECG signals and PPG signals in the window, aligns the processed ECG signals and PPG signals, extracts ECG signal features based on the aligned ECG signals and PPG signals, combines PPG features and ECG signal features with a mapping relationship and the user's personal multimodal information, adjusts the individualized level baseline based on the pulse wave conduction time, and performs blood pressure calibration on the blood pressure estimation based on the blood pressure calibration model and the individualized level baseline.

[0106] In the above steps of establishing a blood pressure estimation model and a blood pressure calibration model, a blood pressure estimation model based on multivariate independent variable PPG features is first established, and a blood pressure calibration model based on ECG features and / or multimodal information is established, so as to provide a scientific, reasonable, accurate and quickly usable model foundation for subsequent continuous blood pressure estimation based on PPG signals, blood pressure calibration based on actively measured ECG and PPG pulse waves, and blood pressure calibration based on actively measured ECG and PPG pulse waves and multimodal information.

[0107] In the above-mentioned step of continuous blood pressure estimation based on PPG signals, the continuously collected PPG signals are intercepted in a fixed time window for subsequent processing, and the PPG signals in the window are processed to remove interference to obtain a smoother PPG signal. After the PPG signal is processed, the PPG features are extracted, and the systolic and diastolic blood pressures are estimated based on the PPG features and the blood pressure estimation model, so as to obtain the blood pressure parameters of the user during daily long-term wear, and the systolic blood pressure can be directly or indirectly estimated from the PPG features, which makes it simpler to obtain more accurate blood pressure parameters for daily reference and saves computing power, so that the smart wearable device for blood pressure measurement consumes less power and can be worn longer.

[0108] In the above-mentioned blood pressure calibration step based on actively measured ECG and PPG pulse waves, it is a process of indirectly inferring the blood pressure baseline level by using electrocardiogram (ECG) and photoplethysmogram (PPG) signals, and estimating the absolute blood pressure by the pulse wave conduction time PTT. In this step, the collected ECG signal and PPG signal are intercepted with a fixed time window, the ECG signal and PPG signal in the window are processed, and the processed ECG signal and PPG signal are aligned, so that the subsequent blood pressure processing and calibration can be calibrated with synchronized signals, so that the signal processing and calibration process are more unified, and the PPG feature and the ECG signal feature have a mapping relationship, and in the subsequent combination of the PPG feature and the ECG signal feature with a mapping relationship and the blood pressure absolute level baseline is adjusted based on the pulse wave conduction time, the process of blood pressure calibration for blood pressure estimation based on the blood pressure calibration model and the blood pressure absolute level baseline is more accurate, so that a more accurate absolute blood pressure can be output, so as to obtain more accurate daily blood pressure measurement values ​​for users.

[0109] In the above-mentioned blood pressure calibration steps based on active measurement of ECG and PPG pulse waves and multimodal information, multimodal information is added to synchronously participate in the adjustment. The individualized horizontal baseline is adjusted by the combination of the user's personal multimodal information and the pulse wave conduction time PTT to avoid inaccurate blood pressure measurement due to individual physiological differences of the user. The blood pressure estimation is calibrated by the individualized horizontal baseline and the blood pressure calibration model, so that the blood pressure data measured by different individual users when using their own exclusive blood pressure measurement smart wearable devices are more in line with their own conditions, providing users with more stable, accurate and convenient blood pressure monitoring for evaluating daily blood pressure, and users do not need to use ECG measurement for a long time. Users only need to actively measure the ECG signal within a certain period to obtain the absolute blood pressure horizontal baseline or the individualized horizontal baseline, and use the blood pressure calibration model to calibrate the blood pressure estimation based on the continuous PPG signal, which is more convenient and burden-free to use.

[0110] In one embodiment of the present invention, in the step of establishing a blood pressure calibration model based on ECG features and / or multimodal information, the multimodal information includes at least two of the following: personal static feature data, dynamic physiological signal data, and blood sugar level data. The personal static feature data includes age, gender, weight, height, health status, medication status, etc., the dynamic physiological signal data includes heart rate data, pulse wave data, blood oxygen data, etc., and the blood sugar level data includes fasting blood sugar, postprandial blood sugar, and daily blood sugar data, etc., so as to establish a blood pressure calibration model related to different individual users, so that the blood pressure measured by different individual users is more accurate.

[0111] In another embodiment of the present invention, in the step of continuously estimating blood pressure based on PPG signals, a user wears a blood pressure measuring smart wearable device that continuously measures blood pressure, wherein the blood pressure measuring smart wearable device is worn on a fingertip or wrist of a human body, and continuously collects PPG signals to reflect changes in blood volume caused by heart beats;

[0112] In the step of blood pressure calibration based on actively measured ECG and PPG pulse waves or blood pressure calibration based on actively measured ECG and PPG pulse waves and multimodal information, the user actively measures, one end of the electrode in the blood pressure measurement smart wearable device is attached to the fingertip or wrist skin surface, and the other finger of the user touches the other end of the electrode to complete ECG signal data collection for at least a set time.

[0113] Through the implementation of the above method, users do not need to use ECG measurement for a long time. They only need to actively collect ECG signal data at intervals of half a day, a day or several days, and the time for each ECG signal data collection can be at least one and a half minutes, two minutes or other set time, which greatly improves the user experience and comfort, and will not limit the user's daily use.

[0114] In another embodiment of the embodiment of the present invention, the step of processing the PPG signal in the window includes PPG signal preprocessing, PPG signal period division, and PPG signal quality verification; wherein:

[0115] PPG signal preprocessing: for the PPG signal in the window, a bandpass filter is applied to remove high-frequency noise and low-frequency drift in the PPG signal, and then a moving average or filter is used to reduce short-term fluctuations in the PPG signal, thereby obtaining a smoother PPG signal;

[0116] PPG signal cycle division: Based on the preprocessed PPG signal, the signal is divided into multiple cardiac cycles through peak and trough detection, providing a basis for subsequent feature calculation;

[0117] PPG signal quality verification: perform quality verification on multiple segmented cardiac cycles.

[0118] By implementing the above method, the processed PPG signal has a higher quality, so as to more accurately estimate the user's blood pressure level. The preprocessing step may include two parts: denoising and smoothing. Denoising removes high-frequency noise (such as muscle activity) and low-frequency drift (such as slow movement or ambient temperature changes) in the signal by applying a bandpass filter (such as a Butterworth filter). Smoothing reduces short-term fluctuations in the signal through a moving average or a Savitzky-Golay filter, thereby obtaining a smoother PPG signal.

[0119] In one embodiment of the present invention, in the step of dividing the PPG signal cycle, a peak detection algorithm is used to detect the peak value of the PPG signal, and the PPG signal is divided into several cycles according to the peak and trough marks of two adjacent heart cycle signals, and the first and last cycles are removed to ensure that the cycle data is clean, and the low peak position is adjusted to be relative to the new starting point; the peak and trough detection can be achieved by a threshold method, derivative analysis or other adaptive algorithms. The detected peaks and troughs are used to divide the signal into multiple cardiac cycles, providing a basis for subsequent feature calculations.

[0120] In the step of PPG signal quality verification, the peaks and troughs of the waveform after segmentation are detected by curvature analysis. The positions of the peaks and troughs correspond to the points where the curvature changes significantly. For the peaks, the curvature reaches the local maximum, and the curve changes from convex to concave. The troughs correspond to the local minimum of the curvature, and the curve changes from concave to convex. The changes in curvature are analyzed to accurately find the positions of the peaks and troughs.

[0121] To ensure signal quality, multiple segmented cardiac cycles are quality checked. The quality check step may include checking the consistency of cycle length, signal integrity, and other factors that may affect feature extraction. Only cardiac cycle signals that pass the quality check will be used for subsequent feature calculations.

[0122] In another embodiment of the present invention, in the step of processing the ECG signal and the PPG signal in the window, a filtering technique is used to remove noise and interference signals in the ECG signal in the window, and a bandpass filter is used to remove high-frequency noise and low-frequency drift in the PPG signal in the window, and then a moving average or a filter is used to reduce short-term fluctuations in the PPG signal, so as to obtain a smoother PPG signal;

[0123] In the step of extracting ECG signal features based on the aligned ECG signal and PPG signal, the position of the R wave is detected in the ECG signal, and the position of the peak is detected in the PPG signal, wherein the R wave is the waveform with the largest amplitude and the most stable in the ECG waveform, representing the beginning of ventricular contraction, and the peak represents the time point when the arterial pulse wave reaches the peripheral tissue.

[0124] This design is used to align the processed ECG signal and PPG signal so that the subsequent blood pressure processing and calibration can be calibrated with synchronized signals, making the signal processing and calibration process more unified, and the process of blood pressure calibration based on the blood pressure calibration model and the absolute blood pressure level baseline and individualized level baseline for blood pressure estimation more accurate.

[0125] In another embodiment of the present invention, the low cutoff frequency of the PPG signal bandpass filter is set to 0.4Hz, and the low cutoff frequency of the ECG signal bandpass filter is set to 2.5Hz; the high cutoff frequency of the PPG signal is set to 4Hz, and the high cutoff frequency of the ECG signal is set to 50Hz. This is used to suppress the high-frequency noise and 50Hz power frequency interference generated by the light source and electronic devices to cope with the influence of light and electrical noise, so that the final calibrated blood pressure value is more accurate.

[0126] In one embodiment of the present invention, in the step of extracting PPG features after PPG signal processing and the step of extracting ECG signal features and PPG features according to the aligned ECG signal and PPG signal, the following signal features are used to describe the shape and dynamic characteristics of each signal waveform:

[0127] Enhancement Index AI: It indicates the ratio of the main peak to the diastolic peak amplitude in a single cycle waveform to reflect the enhancement characteristics of the PPG waveform. The calculation method is AI=amp dia / amp sys , where amp sys Indicates the main peak amplitude, amp dia It is expressed as the diastolic peak amplitude;

[0128] Inflection point area ratio IPA: It represents the ratio of the area before and after the trough point of the dicrotic wave in a single cycle waveform. The calculation method is IPA=area dia ⁄area sys , where area sys is the area of ​​the left part of the waveform, representing the part before the trough of the dicrotic wave, while area dia It is the part after the trough point of the heavy beat;

[0129] Systolic amplitude SA: represents the amplitude of the main peak in a single cycle waveform, recorded as amp sys ;

[0130] The time difference between the main wave and the diastolic peak TD: represents the time difference between the main wave peak and the diastolic peak in a single cycle waveform, calculated as TD=t dia -t sys , where t dia is the secondary peak arrival time, t sys is the main wave peak arrival time;

[0131] Main wave duration HRDN: It indicates the time difference between the main wave peak and the dicrotic wave trough in a single cycle waveform. It is calculated as HRDN=t dic -t sys , where t dic is the arrival time of the re-beat trough, t sysis the main wave peak arrival time;

[0132] Main peak interval PPI: represents the time difference between two adjacent period waveform pulses; the calculation method is , where t Sys is the main wave peak arrival time, is the arrival time of the main peak of the previous cycle;

[0133] Main wave peak width PW: represents the width of the main wave peak pulse waveform of a single cycle waveform, calculated as follows: ,in represents the half-amplitude arrival time on the left side of the main peak pulse, where It indicates the arrival time of the half amplitude on the right side of the main peak pulse;

[0134] Intra-cycle inflection wave and harmonic wave area ratio IHAR: It represents the energy ratio of the inflection wave and the periodic waveform in the Fourier domain of a single periodic waveform, and is calculated as follows: , where pow dia Indicates the energy spectrum of the turn wave in the Fourier domain, pow sum Represents the energy spectrum of the entire periodic waveform in the Fourier domain.

[0135] Through the above specific settings of the shape and dynamic characteristics of the relevant signal waveforms in the PPG features and ECG signal features, the parameters describe the amplitude, area, time interval and other characteristics of the waveform, which can help analyze the morphological characteristics of the waveform, such as the enhancement effect, pulse propagation characteristics and the relative relationship between different parts, so as to establish a data foundation for the application of blood pressure estimation model and blood pressure calibration model.

[0136] In another embodiment of the present invention, in the step of establishing a blood pressure estimation model based on multivariate independent variable PPG features, a multivariate linear regression model is used to establish a linear relationship between the multivariate independent variable PPG features AI, IPA, SA, TD, HRTD, PPI, PW, and IHAR and the blood pressure difference PP and the average blood pressure MAP, respectively, where PP=SBP-DBP, and the least squares method is used to estimate unknown parameters;

[0137] ;

[0138] ;

[0139] in is the error term, whose mean is zero;

[0140] After calculating MAP and PP, the systolic blood pressure (SBP) and diastolic blood pressure (DBP) are reconstructed as follows:

[0141] DBP=SBP-PP;

[0142] ;

[0143] in represents the fitting coefficient of the linear fitting model of PP; Represents the fitting coefficients of the linear fit.

[0144] The specific steps of establishing the blood pressure estimation model in the above manner make it possible to quickly and accurately estimate the systolic and diastolic blood pressure of the user during continuous measurement. Since the data has been pre-processed, the blood pressure estimation result is already relatively accurate.

[0145] In another embodiment of the present invention, the step of establishing a blood pressure calibration model based on ECG features includes the following steps:

[0146] Calculate the pulse wave transit time PTT: PTT is defined as the pulse wave transit time from the heart to the distal artery, measured from the time interval between the R wave of the ECG and the peak of the PPG wave: ;

[0147] The calibrated mean blood pressure MAP is obtained based on the relationship between vascular elasticity and blood flow * :

[0148] MAP * =C × PTT;

[0149] Where C represents the compliance parameter of the blood vessel, which is defined as the volume change caused by a unit pressure change. There is a linear relationship between PTT and blood pressure;

[0150] According to the calibrated MAP * , calculate the calibrated SBP * and DBP * .

[0151] The above are the specific steps of the blood pressure calibration model that uses only ECG for calibration. Through the implementation of this method, in a uniform and elastic vascular system, the increase in blood pressure will lead to further expansion of the vascular wall, which will speed up the propagation of the pressure wave and shorten the PTT. On the contrary, if the blood pressure decreases, the expansion of the vascular wall will decrease, the propagation speed of the pressure wave will slow down, and the PTT will be prolonged. There is a linear relationship between PTT and blood pressure. According to this linear relationship, the absolute baseline of blood pressure is adjusted to obtain the calibrated average blood pressure MAP. * , calculate the calibrated SBP * and DBP * , making the blood pressure data measured by the blood pressure calibration based on the actively measured ECG and PPG pulse waves more accurate.

[0152] In one embodiment of the present invention, the step of establishing a blood pressure calibration model based on ECG features and multimodal information includes the following steps:

[0153] Confirm the type of multimodal information, which includes personal static features X sta , and dynamic physiological sign parameters X dyn And the measured blood pressure BP tr ; Personal static characteristics mainly include gender, age, weight, height, and medical history, and dynamic physiological sign parameters include heart rate, blood oxygen, and blood sugar measurement;

[0154] Calculate the compliance of the blood vessels. The compliance of the blood vessels can be expressed as the following function:

[0155] ;

[0156] Among them, α is a constant coefficient, E represents Young's modulus, h represents the thickness of the arterial wall, ρ represents the blood density, and r represents the arterial radius;

[0157] Calculate the calibrated individual vascular fitness parameter C * By establishing E, h, ρ, r and X sta and X dyn The multivariate linear relationship function f is used to calculate the calibrated individual vascular adaptability parameter C * :

[0158] ;

[0159] ;

[0160] Calibrated blood pressure via multimodality:

[0161] ;

[0162] ;

[0163] .

[0164] The above are the specific steps of the blood pressure calibration model calibrated using ECG features and multimodal information, which is more in line with the various characteristics of each individual user to provide users with more accurate blood pressure measurement and estimation, and provide data with more medical reference value.

[0165] Embodiment 2: Embodiment 2 of the present invention proposes a cuffless continuous calibrated blood pressure estimation system, the system is applied to the cuffless continuous calibrated blood pressure estimation method as described in Embodiment 1 of the present invention, the system includes a collection module, a calculation module and a transmission module;

[0166] The acquisition module is used for continuous signal acquisition and active signal acquisition, wherein PPG signals are continuously acquired and ECG signals are actively acquired;

[0167] The calculation module has a built-in blood pressure estimation model and a blood pressure calibration model. The calculation module is used for continuous blood pressure estimation based on PPG signals, blood pressure calibration based on actively measured ECG and PPG pulse waves, and blood pressure calibration based on actively measured ECG and PPG pulse waves and multimodal information;

[0168] The transmission module is responsible for uploading the calculated blood pressure parameters and their timestamps to the receiving terminal.

[0169] Embodiment 3. Embodiment 3 of the present invention proposes a smart wearable device for measuring blood pressure, which includes the system as described in Embodiment 2 of the present invention. The smart wearable device for measuring blood pressure is worn on the finger or wrist of the user, and the acquisition module includes a PPG sensor and an ECG sensor. The PPG sensor is facing the human body part, one end of the electrode in the ECG sensor is in contact with the surface of human skin, and the other end is selectively contacted by the other finger of the user to collect ECG signals, and the smart wearable device for measuring blood pressure is selectively connected to the receiving terminal signal.

[0170] The smart wearable device for blood pressure measurement can be designed as a ring or a watch style, which not only improves wearing comfort but also provides users with more accurate blood pressure estimation parameters.

[0171] Embodiment 4: Embodiment 4 of the present invention proposes a readable storage medium, on which a control program is stored. When the control program is executed by a computing module, the computing module executes the steps of the method described in Embodiment 1 of the present invention.

[0172] The composition and working principle of the present invention are described below in conjunction with the hardware modules, the measurement implementation process and the main flow of the method in the embodiments of the present invention.

[0173] Figure 3 As shown, the hardware module structure of the cuffless continuous calibrable blood pressure estimation system of the present invention is shown, which mainly includes a collection module, a calculation module and a transmission module.

[0174] The acquisition module includes a PPG sensor and an ECG module for active measurement:

[0175] PPG sensor: The optical sensor is worn on the fingertips or wrists of the human body and continuously collects PPG signals to reflect the changes in blood volume caused by heart beats;

[0176] ECG sensor: When the user actively measures, one end of the electrode is attached to the fingertip or wrist skin surface, and the other finger touches the other end of the electrode to complete ECG signal data collection for at least one and a half minutes.

[0177] The computing module is responsible for processing the collected PPG and ECG signals and estimating blood pressure.

[0178] When only PPG signals are collected: The calculation module continuously estimates systolic and diastolic blood pressure through an algorithm based on PPG characteristic parameters.

[0179] When the user actively measures the ECG signal: the calculation module first aligns the PPG and ECG signals, and then adjusts the absolute blood pressure baseline and the individualized blood pressure baseline through an algorithm centered on the pulse wave transmission time (PTT) to calibrate the blood pressure baseline.

[0180] The transmission module is responsible for uploading the calculated blood pressure parameters and their timestamps to the receiving terminal, which can be a smartphone, tablet or other smart device, so that users can view and manage their blood pressure data at any time.

[0181] The overall algorithm flow of the cuffless continuous calibrable blood pressure estimation method based on PPG and ECG signals is mainly divided into (1) continuous blood pressure estimation process based on PPG signals, (2) blood pressure calibration process based on active measurement of ECG combined with PPG, (3) blood pressure calibration process based on active measurement of ECG combined with PPG and multi-modal information, such as Figure 2 shown.

[0182] The cuffless blood pressure estimation process based on continuous PPG signals can be referred to as shown in Table 1 below:

[0183]

[0184] Table 1

[0185] 1. Signal acquisition: Continuous signal acquisition is performed using a photoplethysmography (PPG) sensor. PPG sensors are usually worn on the fingertips or wrists and can continuously record changes in blood volume caused by heart beats.

[0186] 2. Signal windowing: The continuously collected PPG signal is intercepted in a fixed time window for subsequent processing.

[0187] 3. Signal preprocessing: To improve signal quality, the signal within the window is preprocessed. The preprocessing step includes two parts: denoising and smoothing. Denoising removes high-frequency noise (such as muscle activity) and low-frequency drift (such as slow movement or ambient temperature changes) in the signal by applying a bandpass filter (such as a Butterworth filter). Smoothing reduces short-term fluctuations in the signal through a moving average or Savitzky-Golay filter, resulting in a smoother PPG signal.

[0188] 4. Cycle division: Based on the preprocessed PPG signal, the signal is divided into multiple cardiac cycles by peak and trough detection. Peak and trough detection can be achieved by threshold method, derivative analysis or other adaptive algorithms. The detected peaks and troughs are used to divide the signal into multiple cardiac cycles, providing a basis for subsequent feature calculations.

[0189] 5. Signal quality verification: To ensure signal quality, perform quality verification on multiple cardiac cycles after segmentation. The quality verification step may include checking the consistency of cycle length, signal integrity, and other factors that may affect feature extraction. Only cardiac cycle signals that pass the quality verification will be used for subsequent feature calculations.

[0190] 6. PPG feature calculation: Extract pulse wave feature parameters from the PPG signal that has passed the quality check, usually the waveform morphology features of PPG, including but not limited to pulse time, rise time, fall time, waveform width and height, etc. These feature parameters will be used for subsequent blood pressure estimation.

[0191] 7. Blood pressure estimation based on PPG features: Collect a large number of synchronized ECG and PPG signals and corresponding blood pressure measurements (such as standard blood pressure values ​​obtained by a cuff-type blood pressure monitor). Use statistical or machine learning methods (such as linear regression, support vector machine, neural network, etc.) to establish a mathematical model between blood pressure and PPG features. The goal of the model is to estimate systolic blood pressure (SBP) and diastolic blood pressure (DBP) directly or indirectly from PPG features.

[0192] The blood pressure calibration method based on actively measured ECG and PPG signals is a process that uses electrocardiogram (ECG) and photoplethysmography (PPG) signals to indirectly estimate the baseline blood pressure level. This method mainly estimates absolute blood pressure based on pulse transit time (PTT). PTT refers to the time required for blood to be transported to a certain distal site in the arterial system after the heart ejects blood, which is closely related to the blood pressure level.

[0193] The blood pressure calibration process based on actively measured ECG and PPG signals can be referred to as shown in Table 2 below:

[0194]

[0195] Table 2

[0196] 1. Signal acquisition: Use an electrocardiogram device to record the ECG signal; use a photoplethysmogram device to record the PPG signal, which reflects the changes in tissue blood volume caused by heart beating.

[0197] 2. Signal windowing alignment: The collected PPG and ECG signals are intercepted with a fixed time window and aligned along the time axis for subsequent processing.

[0198] 3. Signal preprocessing: Use appropriate filtering techniques (such as bandpass filters) to remove noise and interference signals from the ECG signal to ensure accurate identification of the R wave.

[0199] 4. ECG feature point detection: Detect the position of the R wave in the ECG signal. The R wave is the largest and most stable waveform in the ECG waveform, representing the beginning of ventricular contraction; detect the position of the peak in the PPG signal. The peak usually indicates the time point when the arterial pulse wave reaches the peripheral tissue.

[0200] 5. PTT calculation: PTT is defined as the time interval from the R wave of the ECG to the peak of the PPG.

[0201] 6. Blood pressure calibration: Collect a large number of synchronized ECG and PPG signals and corresponding blood pressure measurements (such as standard blood pressure values ​​obtained by a cuff-type sphygmomanometer), establish a blood pressure calibration model based on ECG features, combine the PPG features and ECG signal features with a mapping relationship, adjust the absolute blood pressure baseline based on the pulse wave transmission time, and calibrate the blood pressure estimation based on the blood pressure calibration model and the absolute blood pressure baseline.

[0202] The other steps of the blood pressure calibration process based on active measurement of ECG combined with PPG and multimodal information are the same as the blood pressure calibration process based on active measurement of ECG and PPG signals. The difference is that in the blood pressure calibration in step 6, a large number of synchronized ECG and PPG signals and corresponding blood pressure measurements (such as standard blood pressure values ​​obtained by cuff-type blood pressure monitors) are collected, and personal static characteristic data, including age, gender, weight, height, health status, medication status, etc., are collected, and personal blood glucose level data are collected. Use statistical or machine learning methods (such as linear regression, support vector machine, neural network, etc.) to establish a mathematical model between blood pressure and PTT, personal static characteristics and dynamic physiological signals. The goal of the model is to predict an individualized level baseline from PTT, personal static characteristics and dynamic physiological signals.

[0203] Next, the specific key steps of signal preprocessing, period division, signal quality verification, and feature extraction in the above three processes: "(1) continuous blood pressure estimation process based on PPG signals, (2) blood pressure calibration process based on active measurement of ECG combined with PPG, and (3) blood pressure calibration process based on active measurement of ECG combined with PPG and multimodal information" are explained.

[0204] For signal preprocessing:

[0205] The pulse signal originates from the beating activity of the heart. When the photoplethysmography (PPG) technology is used for detection, the collected signal is often weak and easily interfered by various noises, which in turn affects the purity of the signal. The main sources of interference include electromagnetic interference generated by the equipment, power frequency interference (such as 50 / 60Hz power frequency interference), and noise caused by factors such as human breathing movement, body displacement, and relative offset of the sensor. In practical applications, these noises will not only reduce the integrity of the signal, but may also cause important physiological information to be masked, thereby affecting the accuracy of subsequent analysis and diagnosis results.

[0206] The present invention uses a bandpass filter to filter PPG and ECG signals. The low cutoff frequency of the PPG signal bandpass filter is set to 0.4 Hz, and the low cutoff frequency of the ECG signal bandpass filter is set to 2.5 Hz, in order to effectively deal with the signal deviation caused by human body movement; the high cutoff frequency of the PPG signal is set to 4 Hz, and the high cutoff frequency of the PPG signal is set to 50 Hz, in order to suppress the high-frequency noise and 50 Hz power frequency interference generated by the light source and electronic devices. It mainly deals with the influence of light and electrical noise.

[0207] Multiple sets of data are pre-processed, and the comparison before and after can be referenced Figure 4 shown.

[0208] For period division:

[0209] PPG waveform period division. Although the shape of the PPG signal will be significantly different for people of different ages and physical conditions, its periodicity remains unchanged. Therefore, as long as the starting point and the highest point of each pulse cycle can be reasonably found, the pulse wave can be divided and the corresponding blood pressure changes can be further observed.

[0210] The present invention uses a peak detection algorithm (such as peak detection, threshold detection, differential detection, etc.) to detect the peak of the PPG signal, and divides the PPG signal into several cycles according to the peak and trough marks of two adjacent heartbeat cycle signals. The first and last cycles are removed to ensure that the cycle data is clean. The low peak position is adjusted to be relative to the new starting point.

[0211] The filtered waveform is divided into single cycles to prepare for feature extraction. Figure 5 Demonstrates the effect of segmenting the waveform after filtering.

[0212] For signal quality verification:

[0213] High-quality signals are the premise for ensuring the accuracy and reliability of data analysis. Therefore, developing effective signal quality assessment mechanisms and screening algorithms is crucial to improving the quality of signal processing results. Since signal quality varies, this step removes data without sub-peaks and troughs (removed at a certain ratio). The specific steps are as follows:

[0214] By analyzing the time domain characteristics (such as peak interval), frequency domain characteristics (such as power spectral density), and waveform characteristics (such as peak shape, rising and falling edge slopes) of the PPG waveform, key information that helps to assess cardiovascular function and hemodynamic status can be extracted.

[0215] Through the implementation of the above embodiments, through the multi-dimensional and cross-judgment use of parameter characteristics such as the target pressure sequence, the pulse amplitude signal sequence, etc. during the pressurization process and the decompression process of the blood pressure measurement, the process of the cuffless continuous calibrated blood pressure estimation is made more scientific and reasonable, which not only makes the user's blood pressure measurement process more comfortable, but also makes the final blood pressure measurement result accurate and not affected by interference.

[0216] The peaks and troughs of the segmented waveform can be detected by curvature analysis. By analyzing the change in the curvature of the waveform, the extreme points of the waveform can be accurately located, so as to better understand the overall structure and change trend of the waveform. Curvature is a geometric concept that describes the degree of curvature of a curve at a certain point; in mathematics, curvature can be used to quantify the degree of curvature of a curve at a certain point. For a one-dimensional signal y=f(x), the curvature k at the point (x, y) is defined as follows:

[0217] (1);

[0218] in:

[0219] It is the first derivative of the function f(x), representing the slope of the tangent line.

[0220] It is the second derivative of the function f(x), which represents the rate of change of the slope of the tangent line.

[0221] Its physical meaning is:

[0222] First Derivative It represents the slope of the tangent line to the curve at point x.

[0223] Second Derivative It represents the reciprocal of the distance from the center of curvature of the curve at point x to the curve, that is, the curvature of the curve.

[0224] Typically, the locations of crests and troughs correspond to points where the curvature changes significantly. For crests, the curvature reaches a local maximum and the curve changes from convex to concave. The troughs correspond to local minima of curvature and the curve changes from concave to convex. Therefore, analyzing the changes in curvature can help us accurately find the locations of crests and troughs.

[0225] Figure 6 Shows how Figure 5 The feature extraction effect after waveform segmentation is shown in Figure 1. The waveform signal quality is verified based on the above feature points. The signal quality verification process is as follows:

[0226] 1. The main peak height within the cycle is greater than 0.3 times the average main peak height (single patient data) and less than 1.7 times the average main peak height.

[0227] 2. The minimum value within a cycle is greater than 1.7 times the average minimum value (single patient data) and less than 0.3 times the average minimum value (the minimum value within a cycle after normalization is less than 0).

[0228] 3. The cycle length is less than 4 / 3 times the I value (the I value is calculated by the sampling frequency and the pulse wave signal data of a single patient) and greater than 2 / 3 times the I value.

[0229] 4. The main peak height within a cycle is the maximum value within the cycle.

[0230] About feature point extraction:

[0231] The feature point extraction refers to the definitions of the above multivariate independent variables PPG features AI, IPA, SA, TD, HRTD, PPI, PW, and IHAR, which will not be repeated here. The features are used to describe the shape and dynamic characteristics of each signal waveform. These parameters describe the amplitude, area, time interval and other characteristics of the waveform, which can help analyze the morphological characteristics of the waveform, such as the enhancement effect, pulse propagation characteristics and the relative relationship between different parts.

[0232] About the blood pressure estimation model:

[0233] The embodiment of the present invention uses a multivariate linear regression model to establish the linear relationship between the multivariate independent variables PPG features AI, SA, IPA, PW, PPI, TD, HRTD, and IHAR and the blood pressure difference (PP, SBP-DBP), and the mean blood pressure (MAP), and the least squares method is used to estimate the unknown parameters.

[0234] (2);

[0235] (3);

[0236] After calculating MAP and PP, the systolic blood pressure (SBP) and diastolic blood pressure (DBP) are reconstructed as follows:

[0237] DBP=SBP-PP (4);

[0238] .

[0239] About the Blood Pressure Calibration Model:

[0240] To calibrate using ECG only:

[0241] 1. Calculate PTT (Pulse Transit Time): PTT is defined as the pulse wave transmission time from the heart to the distal artery, which can be measured by the time interval from the R wave of the ECG to the PPG peak. The details are as follows:

[0242] (6);

[0243] 2. The Compliance Chamber Model is used clinically to describe the relationship between vascular elasticity and blood flow. In this model, blood vessels are viewed as one or a series of elastic chambers that can expand and contract with changes in internal pressure. When the heart pumps blood, the pressure increases and the blood vessel walls expand outward; when the heart relaxes, the blood vessel walls help push the blood forward due to their elastic recoil force. This elastic property is measured by "compliance", which is the amount of volume change caused by a unit pressure change.

[0244] (7);

[0245] Where C represents the compliance of the blood vessel, which is a priori model parameter and is defined as the volume change caused by a unit pressure change. According to the model assumption, there is a linear relationship between PTT and blood pressure. This shows that in a uniform and elastic vascular system, an increase in blood pressure will lead to further expansion of the vascular wall, which will accelerate the propagation speed of the pressure wave and shorten the PTT. On the contrary, if the blood pressure decreases, the expansion of the vascular wall decreases, the propagation speed of the pressure wave slows down, and the PTT is prolonged.

[0246] 3. Obtaining calibrated MAP using the elastic chamber model * , substituting into formula (5) can calculate the calibrated SBP * and DBP * .

[0247] Calibration using ECG and multimodal information:

[0248] 1. Multimodal information comes from the information filled in the APP and the information collected by other device sources, mainly including personal static features X sta , and dynamic physiological sign parameters X dyn And the measured blood pressure BP tr The personal static characteristics mainly include gender, age, weight, height, whether there is a history of disease, etc., which are actively filled in by the user in the APP on the computer, mobile phone or tablet. The dynamic physiological sign parameters mainly include measuring heart rate, blood oxygen and blood sugar, which can be filled in by the user or obtained through the device networking.

[0249] 2. The compliance of blood vessels is related to the elasticity of the arterial wall, the Young's modulus of the arterial wall, the thickness of the arterial wall, the blood density, and the arterial radius. That is, the compliance of blood vessels in formula (7) can be expressed as the following function:

[0250] (8);

[0251] 3. The present invention establishes a regression model and uses the measured blood pressure as BP tr Regression target, using least squares to build a regression model:

[0252] (9);

[0253] (10);

[0254] 4. The final multi-modal calibrated blood pressure can be expressed as:

[0255] (11);

[0256] ;

[0257] (13).

[0258] Through the implementation of the above embodiments, the problems existing in the existing cuffless continuous blood pressure monitoring technology, such as the inability to handle individual physiological differences, limitations on continuity and convenience, and poor measurement accuracy, are solved, thereby overcoming the influence of noise interference on the accuracy of blood pressure measurement. In addition, by combining the PTT calibration algorithm of the ECG signal, personal static feature data, dynamic physiological signal data and blood glucose level data are added as features for modeling, thereby achieving the purpose of continuous blood pressure monitoring without frequent manual intervention, reducing the influence of individual differences on the accuracy of blood pressure readings, and achieving the purpose of the present invention.

[0259] The above embodiments are only for illustrating the technical concept and features of the present invention, and their purpose is to enable people familiar with the technology to understand the content of the present invention and implement it accordingly, and they cannot be used to limit the protection scope of the present invention. Any equivalent changes or modifications made according to the spirit of the present invention should be included in the protection scope of the present invention.

Claims

1. A cuffless continuous calibrable blood pressure estimation method, characterized in that: The cuff-free continuous calibratable blood pressure estimation method comprises: Establish a blood pressure estimation model and a blood pressure calibration model, establish a blood pressure estimation model based on multivariate independent variable PPG features, and establish a blood pressure calibration model based on ECG features and / or multimodal information; Continuously estimate blood pressure based on PPG signals, continuously collect PPG signals, intercept the continuously collected PPG signals in a fixed time window, process the PPG signals within the window, extract PPG features after PPG signal processing, and estimate systolic and diastolic pressures based on the PPG features and the blood pressure estimation model; Blood pressure calibration based on actively measured ECG and PPG pulse waves, collecting ECG signals and PPG signals, intercepting, aligning and processing the collected ECG signals and PPG signals in a fixed time window, extracting ECG signal features and PPG features based on the aligned ECG signals and PPG signals, combining the PPG features and ECG signal features with a mapping relationship and adjusting the blood pressure absolute level baseline based on the pulse wave transmission time, and performing blood pressure calibration on the blood pressure estimation based on the blood pressure calibration model and the blood pressure absolute level baseline; Blood pressure calibration based on actively measured ECG and PPG pulse waves and multimodal information, collect ECG signals and PPG signals, intercept, align and process the collected ECG signals and PPG signals in a fixed time window, extract ECG signal features according to the aligned ECG signals and PPG signals, combine the PPG features and ECG signal features with a mapping relationship and the user's personal multimodal information, adjust the individualized level baseline based on the pulse wave transmission time, and calibrate the blood pressure estimation based on the blood pressure calibration model and the individualized level baseline.

2. The cuffless continuous calibratable blood pressure estimation method according to claim 1, characterized in that: In the step of establishing a blood pressure calibration model based on ECG characteristics and / or multimodal information, the multimodal information includes at least two of the following: personal static characteristic data, dynamic physiological signal data, and blood glucose level data.

3. The cuffless continuous calibratable blood pressure estimation method according to claim 2, characterized in that: In the step of continuously estimating blood pressure based on PPG signals, the user wears a blood pressure measuring smart wearable device that continuously measures blood pressure, wherein the blood pressure measuring smart wearable device is worn on the fingertips or wrists of the human body and continuously collects PPG signals to reflect changes in blood volume caused by heart beats; In the step of blood pressure calibration based on actively measured ECG and PPG pulse waves or blood pressure calibration based on actively measured ECG and PPG pulse waves and multimodal information, the user actively measures, one end of the electrode in the blood pressure measurement smart wearable device is attached to the fingertip or wrist skin surface, and the other finger of the user touches the other end of the electrode to complete ECG signal data collection for at least a set time.

4. The cuffless continuous calibratable blood pressure estimation method according to claim 1, characterized in that: The step of processing the PPG signal within the window includes PPG signal preprocessing, PPG signal period division, and PPG signal quality verification; wherein: PPG signal preprocessing: for the PPG signal in the window, a bandpass filter is applied to remove high-frequency noise and low-frequency drift in the PPG signal, and then a moving average or filter is used to reduce short-term fluctuations in the PPG signal, thereby obtaining a smoother PPG signal; PPG signal cycle division: Based on the preprocessed PPG signal, the signal is divided into multiple cardiac cycles through peak and trough detection, providing a basis for subsequent feature calculation; PPG signal quality verification: perform quality verification on multiple segmented cardiac cycles.

5. The cuffless continuous calibratable blood pressure estimation method according to claim 4, characterized in that: In the step of dividing the PPG signal cycle, a peak detection algorithm is used to detect the peak value of the PPG signal, and the PPG signal is divided into several cycles according to the peak and trough marks of two adjacent heartbeat cycle signals, and the first and last cycles are removed to ensure that the cycle data is clean, and the low peak position is adjusted to be relative to the new starting point; In the step of PPG signal quality verification, the peaks and troughs of the waveform after segmentation are detected by curvature analysis. The positions of the peaks and troughs correspond to the points where the curvature changes significantly. For the peaks, the curvature reaches the local maximum, and the curve changes from convex to concave. The troughs correspond to the local minimum of the curvature, and the curve changes from concave to convex. The changes in curvature are analyzed to accurately find the positions of the peaks and troughs.

6. The cuffless continuous calibratable blood pressure estimation method according to claim 4, characterized in that: In the step of processing the ECG signal and the PPG signal in the window, the ECG signal in the window is subjected to filtering technology to remove noise and interference signals in the ECG signal, and the PPG signal in the window is subjected to a bandpass filter to remove high-frequency noise and low-frequency drift in the PPG signal, and then a moving average or a filter is used to reduce short-term fluctuations in the PPG signal, thereby obtaining a smoother PPG signal; In the step of extracting ECG signal features based on the aligned ECG signal and PPG signal, the position of the R wave is detected in the ECG signal, and the position of the peak is detected in the PPG signal, wherein the R wave is the waveform with the largest amplitude and the most stable in the ECG waveform, representing the beginning of ventricular contraction, and the peak represents the time point when the arterial pulse wave reaches the peripheral tissue.

7. The cuffless continuous calibratable blood pressure estimation method according to claim 6, characterized in that: The lower cutoff frequency of the bandpass filter for the PPG signal is set to 0.4 Hz, and the lower cutoff frequency of the bandpass filter for the ECG signal is set to 2.5 Hz; The high cutoff frequency for the PPG signal is set to 4 Hz, and the high cutoff frequency for the ECG signal is set to 50 Hz.

8. The cuffless continuous calibratable blood pressure estimation method according to claim 1, characterized in that: In the step of extracting PPG features after PPG signal processing and the step of extracting ECG signal features and PPG features based on the aligned ECG signal and PPG signal, the following signal features are used to describe the shape and dynamic characteristics of each signal waveform: Enhancement Index AI: It indicates the ratio of the main peak to the diastolic peak amplitude in a single cycle waveform to reflect the enhancement characteristics of the PPG waveform. The calculation method is AI=amp dia / amp sys , where amp sys Indicates the main peak amplitude, amp dia It is expressed as the diastolic peak amplitude; Inflection point area ratio IPA: It represents the ratio of the area before and after the trough point of the dicrotic wave in a single cycle waveform. The calculation method is IPA=area dia ⁄area sys , where area sys is the area of ​​the left part of the waveform, representing the part before the trough of the dicrotic wave, while area dia It is the part after the trough point of the heavy beat; Systolic amplitude SA: represents the amplitude of the main peak in a single cycle waveform, recorded as amp sys ; The time difference between the main wave and the diastolic peak TD: represents the time difference between the main wave peak and the diastolic peak in a single cycle waveform, calculated as TD=t dia -t sys , where t dia is the secondary peak arrival time, t sys is the main wave peak arrival time; Main wave duration HRDN: It indicates the time difference between the main wave peak and the dicrotic wave trough in a single cycle waveform. It is calculated as HRDN=t dic -t sys , where t dic is the arrival time of the re-beat trough, t sys is the main wave peak arrival time; Main peak interval PPI: represents the time difference between two adjacent period waveform pulses; the calculation method is , where t Sys is the main wave peak arrival time, is the arrival time of the main peak of the previous cycle; Main wave peak width PW: represents the width of the main wave peak pulse waveform of a single cycle waveform, calculated as follows: ,in represents the half-amplitude arrival time on the left side of the main peak pulse, where It indicates the arrival time of the half amplitude on the right side of the main peak pulse; Intra-cycle inflection wave and harmonic wave area ratio IHAR: It represents the energy ratio of the inflection wave and the periodic waveform in the Fourier domain of a single periodic waveform, and is calculated as follows: , where pow dia Indicates the energy spectrum of the turn wave in the Fourier domain, pow sum Represents the energy spectrum of the entire periodic waveform in the Fourier domain.

9. The cuffless continuous calibratable blood pressure estimation method according to claim 8, characterized in that: In the step of establishing a blood pressure estimation model based on multivariate independent variable PPG features, a multivariate linear regression model is used to establish a linear relationship between the multivariate independent variable PPG features AI, IPA, SA, TD, HRTD, PPI, PW, and IHAR and the blood pressure difference PP and the average blood pressure MAP, respectively, where PP=SBP-DBP, and the least squares method is used to estimate unknown parameters; ； ； in is the error term, whose mean is zero; After calculating MAP and PP, the systolic blood pressure (SBP) and diastolic blood pressure (DBP) are reconstructed as follows: DBP=SBP-PP; ； in represents the fitting coefficient of the linear fitting model of PP; Represents the fitting coefficients of the linear fit.

10. The cuffless continuous calibratable blood pressure estimation method according to claim 9, characterized in that: In the step of establishing a blood pressure calibration model based on ECG features and / or multimodal information, a large number of synchronized ECG, PPG signals and corresponding blood pressure measurements are collected, personal static feature data are collected, including age, gender, weight, height, health status, medication status, and personal blood glucose level data are collected, and statistical or machine learning methods are used to establish a blood pressure calibration model between blood pressure and PTT, personal static characteristics and dynamic physiological signals.

11. The cuffless continuous calibratable blood pressure estimation method according to claim 10, characterized in that: The steps of establishing a blood pressure calibration model based on ECG features include the following steps: Calculate the pulse wave transit time PTT: PTT is defined as the pulse wave transit time from the heart to the distal artery, measured from the time interval between the R wave of the ECG and the peak of the PPG wave: , where t P is the time when the photoplethysmogram (PPG) reaches its peak, t R It is the time when the R wave appears in the electrocardiogram (ECG) signal; The calibrated mean blood pressure MAP is obtained based on the relationship between vascular elasticity and blood flow * : MAP * =C×PTT; Where C represents the compliance parameter of the blood vessel, which is defined as the volume change caused by a unit pressure change. There is a linear relationship between PTT and blood pressure; According to the calibrated MAP * , calculate the calibrated SBP * and DBP * .

12. The cuffless continuous calibratable blood pressure estimation method according to claim 11, characterized in that: The steps of establishing a blood pressure calibration model based on ECG features and multimodal information include the following steps: Confirm the type of multimodal information, which includes personal static features X sta , and dynamic physiological sign parameters X dyn And the measured blood pressure BP tr ; Personal static characteristics mainly include gender, age, weight, height, and medical history, and dynamic physiological sign parameters include heart rate, blood oxygen, and blood sugar measurement; Calculate the compliance of the blood vessels. The compliance of the blood vessels can be expressed as the following function: ； Among them, α is a constant coefficient, E represents Young's modulus, h represents the thickness of the arterial wall, ρ represents the blood density, and r represents the arterial radius; Calculate the calibrated individual vascular fitness parameter C * By establishing E, h, ρ, r and X sta and X dyn The multivariate linear relationship function f is used to calculate the calibrated individual vascular adaptability parameter C * : ； ； Calibrated blood pressure via multimodality: ； ； 。 13. A cuffless, continuously calibrable blood pressure estimation system, characterized in that: The system is applied to the cuffless continuous calibrable blood pressure estimation method according to any one of claims 1 to 12, and the system comprises an acquisition module, a calculation module and a transmission module; The acquisition module is used for continuous signal acquisition and active signal acquisition, wherein PPG signals are continuously acquired and ECG signals are actively acquired; The calculation module has a built-in blood pressure estimation model and a blood pressure calibration model. The calculation module is used for continuous blood pressure estimation based on PPG signals, blood pressure calibration based on actively measured ECG and PPG pulse waves, and blood pressure calibration based on actively measured ECG and PPG pulse waves and multimodal information; The transmission module is responsible for uploading the calculated blood pressure parameters and their timestamps to the receiving terminal.

14. A smart wearable device for measuring blood pressure, characterized in that: The smart wearable device for measuring blood pressure includes the system as claimed in claim 13, wherein the smart wearable device for measuring blood pressure is worn on the finger or wrist of the user, the acquisition module includes a PPG sensor and an ECG sensor, the PPG sensor is facing the human body part, one end of the electrode in the ECG sensor is in contact with the surface of human skin, and the other end is selectively contacted by the other finger of the user to collect ECG signals, and the smart wearable device for measuring blood pressure is selectively connected to the receiving terminal signal.

15. A readable storage medium, characterized in that: The readable storage medium stores a control program, and when the control program is executed by a computing module, the computing module executes the steps of the method according to any one of claims 1 to 12.

Images