Method and device for correcting blood pressure measurements

Abstract

The present application provides a kind of blood pressure measurement value correction method and device, it is related to biomedical signal processing technical field.The present application synchronously acquires sleeve pressure signal, photoelectric volume pulse wave signal, electrocardiogram signal, biological impedance signal and three-axis acceleration signal and other multi-modal physiological signals;Time stamp alignment and preprocessing are carried out to the signal, and original pulse wave conduction time is extracted;Motion interference, sleeve fit interference, heart rate variation interference and vascular tension interference and other interference indexes are extracted from multi-modal signal, and interference degree vector is formed and measurement quality grade is evaluated;Pulse wave conduction time is dynamically corrected based on interference degree vector;Individualized pulse wave conduction time and blood pressure corresponding relationship model is established, and first blood pressure measurement value and its confidence are calculated;Oscillography is used to obtain second blood pressure measurement value and estimate its error range;Adaptive weight coefficient is determined according to confidence and error range, and blood pressure value is corrected in combination with measurement quality grade.

Description

Technical Field

[0001] This invention relates to the field of biomedical signal processing technology, specifically to a method and apparatus for correcting blood pressure measurements. Background Technology

[0002] Blood pressure measurement plays a crucial role in clinical medicine and daily health management as an important means of diagnosing and monitoring cardiovascular diseases. With the development of medical technology and the increasing demand for health monitoring, blood pressure measurement technology is constantly evolving and improving.

[0003] Traditional oscillometric blood pressure measurement determines blood pressure values ​​by detecting pressure oscillation signals during cuff inflation and deflation. While offering good measurement stability, it suffers from limitations such as single-measurement limitations and inability to perform continuous monitoring. In recent years, cuffless blood pressure measurement technology based on pulse wave transit time has received widespread attention and research. Prior art (CN107854123A) discloses a cuffless continuous blood pressure monitoring method and device. This method collects heartbeat signals or electrocardiogram signals and photoplethysmography (PPG) pulse wave signals, calculates the pulse wave arrival time, and establishes a personal blood pressure calculation model to achieve continuous blood pressure monitoring. Prior art (CN118830818A and CN114224304A) discloses dynamic cuffless continuous blood pressure measurement methods. These methods utilize neural networks to establish a model relating pulse wave transit time information, pulse wave waveform information, and personal characteristic information to blood pressure, comprehensively considering cardiovascular parameters related to blood pressure.

[0004] To improve the accuracy and reliability of blood pressure measurement, researchers have begun exploring blood pressure measurement methods based on multimodal signal fusion. Prior art (CN119837507A) discloses a cuffless, continuously calibrable blood pressure estimation method, establishing a blood pressure estimation model and a blood pressure calibration model, and performing blood pressure calibration based on actively measured ECG and PPG pulse waves and multimodal information. Prior art (CN113647921A) discloses a blood pressure measurement method that improves measurement accuracy by acquiring two real-time photoplethysmography (PPG) pulse wave signals and combining them with the calibrated blood pressure value corresponding to the pulse wave conduction time, dynamically adjusting weighting parameters.

[0005] However, existing technologies still have the following technical problems: First, although the oscillometric method has good stability, a single measurement is affected by factors such as cuff position and inflation / deflation speed, and it cannot provide continuous blood pressure monitoring; second, the accuracy of the pulse wave transit time method is easily affected by physiological interferences such as human movement, heart rate variability, and changes in vascular tension, resulting in large fluctuations and low reliability of the measurement results; finally, existing schemes that fuse multiple signals usually lack dynamic evaluation of signal quality and fine correction for interference, and the fusion strategy is relatively fixed, resulting in low reliability of the final blood pressure output value in complex measurement environments. These problems limit the accuracy and reliability of blood pressure measurement technology in practical applications, especially in complex measurement environments with multiple interference factors. The information disclosed in the background section is only used to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0006] The purpose of this invention is to provide a method and apparatus for correcting blood pressure measurements in order to solve the problems mentioned in the background art.

[0007] To achieve the above objectives, the present invention provides the following technical solution: S1: Within the measurement time window, the original multimodal physiological signal set of the measurement object is acquired synchronously, including cuff pressure signal, photoplethysmography pulse wave signal, electrocardiogram signal, bioimpedance signal of tissue under the cuff and triaxial acceleration signal of arm movement; S2: Establish a unified timestamp and align the signals in all the original multimodal physiological signal groups, and obtain the original pulse wave conduction time based on the time interval between the photoplethysmography pulse wave signal and the pulse wave initiation feature point in the electrocardiogram signal. S3: Based on the preprocessed original multimodal physiological signal set, extract interference indicators to characterize the degree of measurement interference. After normalizing each interference indicator, combine them to form an interference degree vector. Compare the interference degree vector with the preset interference threshold and output the corresponding measurement quality level. S4: Perform dynamic correction processing on the original pulse wave transit time based on the interference vector, and establish an individualized correspondence model between the corrected pulse wave transit time and blood pressure to obtain the first blood pressure measurement value. Based on the fitting residual of the correspondence model and the interference vector, generate the confidence level corresponding to the first blood pressure measurement value. Based on the cuff pressure signal, S5 uses the oscillometric method to extract the feature points of the cuff pressure oscillation amplitude as a function of pressure, and determines the second blood pressure measurement value by the oscillometric method based on the feature points. Based on the interference vector, a blood pressure value error estimation model is established to quantify the error range corresponding to the second blood pressure measurement value. S6: Determine the adaptive weighting coefficients of the first and second blood pressure measurements based on the confidence level of the first blood pressure measurement and the error range of the second blood pressure measurement. Set the overall adjustment factor according to the measurement quality level, and perform weighted fusion processing on the first and second blood pressure measurements to obtain the corrected blood pressure value.

[0008] Furthermore, a preset measurement time window is established, with a single blood pressure measurement cycle as the measurement time window. The single blood pressure measurement cycle is defined as the fixed duration from before the cuff is inflated to the fixed duration after the cuff is deflated. A hardware-level synchronous clock is established, and a unified timestamp is created for the signals in all the original multimodal physiological signal groups. All sensors collect data at a fixed frequency sampling rate, and each sampling point is assigned a precise timestamp, forming a synchronized time series signal matrix.

[0009] Furthermore, preprocessing operations are performed on the raw multimodal physiological signals. For the photoplethysmography (PPG) signal, a bandpass filter is used, defining filter coefficient parameters and passband to preprocess the raw PPG signal and retain its main frequency components. For the electrocardiogram (ECG) signal, an adaptive algorithm is used to define power frequency interference parameters in real time, including amplitude and phase. Based on these parameters, noise reduction of the raw ECG signal is completed. Finally, the triaxial acceleration data of arm movements are subjected to a fixed-frequency low-pass filter, smoothed using a ten-point moving average, retaining the intended motion components and eliminating high-frequency vibration noise.

[0010] Furthermore, based on the time interval between the preprocessed photoplethysmography pulse wave signal and the pulse wave initiation feature point in the electrocardiogram (ECG) signal, the original pulse wave conduction time is obtained. The specific steps are as follows: detect the R-wave peak value from the preprocessed ECG signal, that is, the highest peak value in the QRS complex of the ECG signal. Specifically, use the first derivative maximum value detection algorithm to obtain a series of R-wave time points, each point corresponding to the start of a heart contraction, thereby defining and determining the R-wave time series data. The pulse wave start point is detected from the preprocessed photoplethysmography (PPG) signal. The starting point of the PPG signal rise is found within a fixed time window. The point where the first derivative of the PPG signal exceeds 10% of the maximum derivative is found as the pulse wave arrival time. The pulse wave time series data is defined and determined in this way. Based on the detected R-wave time series data and pulse wave time series data, the original pulse wave conduction time corresponding to each heartbeat is calculated. The value of each original pulse wave conduction time is the pulse wave start time of the corresponding heartbeat minus the corresponding pulse wave start time. The peak wave time is then subtracted from the fixed physical distance delay.

[0011] Furthermore, the interference indicators include: motion interference, cuff fit interference, heart rate variability interference, and vascular tension interference; the specific steps for obtaining the motion interference value are as follows: first, using the preprocessed arm motion triaxial acceleration data, the root mean square acceleration of the triaxial acceleration at each sampling time point is calculated; then, according to the cuff inflation / deflation state at each sampling time point, an acceleration influence weighting coefficient is assigned; finally, the motion interference value at each sampling time point is the weighted average root mean square acceleration. The specific steps to obtain cuff fit interference are as follows: First, calculate the average value of the bioimpedance signal of the tissue under the cuff for a fixed time before the cuff is inflated as the baseline impedance value; find the maximum and minimum impedance values ​​during the entire measurement process within the monitoring time window, divide the difference between the maximum and minimum impedance values ​​by the baseline impedance value and then perform percentage calculation to obtain the cuff fit interference value. The specific steps for obtaining heart rate variability interference are as follows: Based on R-wave time series data, the time interval between two adjacent R waves in the electrocardiogram signal is obtained to construct RR interval sequence data. The standard deviation and mean of the RR interval are calculated, and the exponential decay factor is determined based on the mean of the RR interval sequence data. The heart rate variability interference value is obtained by dividing the standard deviation of the RR interval by the mean of the RR interval and multiplying it by the exponential decay factor. The specific steps for obtaining vascular tension interference are as follows: First, obtain the pulse amplitude component based on the photoplethysmography (PPG) signal, i.e., the peak-to-valley difference of the PPG signal within a single heartbeat cycle; then, obtain the average blood volume within a single heartbeat cycle: the average blood volume within a single heartbeat cycle is the integral of the PPG signal within the heartbeat cycle divided by the time length; the ratio of the pulse amplitude component corresponding to a single heartbeat cycle to the average blood volume is used as the pulsation efficiency ratio of a single heartbeat cycle, and the ratio of the median pulsation efficiency ratio to the median heartbeat cycle length is used as the vascular tension interference value. Furthermore, after normalizing each interference index, an interference degree vector is formed. This vector is then compared with a preset interference threshold to output the corresponding measurement quality level. Specifically, the measurement quality level is divided into three levels: low, medium, and high. For motion interference, cuff fit interference, heart rate variability interference, and vascular tension interference, interference thresholds are set for each interference index. If the number of interference indices exceeding their corresponding interference thresholds is greater than or equal to three, the measurement quality level is high. If the number of interference indices exceeding their corresponding interference thresholds is greater than one and less than three, the measurement quality level is medium. If the number of interference indices exceeding their corresponding interference thresholds is greater than zero and less than or equal to one, the measurement quality level is low. The measurement quality level is divided into three levels: low, medium, and high, and an overall adjustment factor is set based on the measurement quality level. Furthermore, the original pulse wave conduction time is compensated and adjusted according to the magnitude of each component in the interference vector. First, the maximum correction amplitude, reference vascular tension value, and correction sensitivity are defined, and the vascular tension correction factor is calculated. Then, the original pulse wave conduction time is multiplied by the vascular tension correction factor to obtain the preliminarily corrected pulse wave conduction time series. The cuff pressure effect is corrected on the preliminarily corrected pulse wave conduction time series. First, the cuff pressure value corresponding to each pulse wave conduction time is obtained. Then, the systolic and diastolic blood pressure at that moment are estimated by simplifying the relationship. The pressure correction factor is obtained based on the preliminarily estimated diastolic and systolic blood pressure. The pulse wave conduction time is corrected by the cuff pressure value corresponding to each pulse wave conduction time and the pressure correction factor. Finally, median filtering is used to remove abnormal pulse wave conduction times to obtain the corrected pulse wave conduction time.

[0012] Furthermore, based on the corrected pulse wave transit time, an individualized correspondence model between pulse wave transit time and blood pressure is established. The specific steps for establishing the model are as follows: extract parameters describing individualized vascular characteristics from the corrected pulse wave transit time. These parameters include: the average pulse wave transit time, the standard deviation of pulse wave transit time, the correlation coefficient between pulse wave transit time and blood pressure, and the rate of change of pulse wave transit time with respect to pressure. Substitute each corrected pulse wave transit time into the individualized correspondence model between pulse wave transit time and blood pressure to obtain the first blood pressure measurement value, which includes the systolic blood pressure and diastolic blood pressure in the correspondence model. The model confidence is obtained based on the interquartile range of the estimated systolic blood pressure and the interquartile range of the estimated diastolic blood pressure.

[0013] Furthermore, the oscillometric method is used to extract feature points of the cuff pressure oscillation amplitude as a function of pressure, and the second blood pressure measurement value is determined based on these feature points. This second blood pressure measurement value includes both oscillometric systolic and diastolic blood pressure measurements. The error ranges for the oscillometric blood pressure measurements also correspond to the error ranges for systolic and diastolic blood pressure measurements. Adaptive weighting coefficients include adaptive fusion weighting coefficients for systolic and diastolic blood pressure. The systolic blood pressure fusion weighting coefficient is determined by the model confidence level and the error range of the oscillometric systolic blood pressure measurement. The error range is determined; the diastolic blood pressure fusion weight coefficient is determined by the model confidence level and the error range of the oscillometric diastolic blood pressure measurement; based on the measurement quality level, there are three levels: low, medium, and high, each corresponding to three overall adjustment factors. The adaptive weight coefficient is adjusted by matching the overall adjustment factors with the measurement quality level to obtain the final systolic blood pressure fusion weight coefficient and diastolic blood pressure fusion weight coefficient; based on the final systolic blood pressure fusion weight coefficient and diastolic blood pressure fusion weight coefficient, the blood pressure measurement value based on pulse wave transit time and the blood pressure oscillometric measurement value are weighted and fused to obtain the final corrected blood pressure value.

[0014] The present invention also provides an apparatus for correcting blood pressure measurements, the apparatus being used to perform the above-described method for correcting blood pressure measurements, comprising: Multimodal signal synchronous acquisition module: used to synchronously acquire the original multimodal physiological signal set of the measurement object within the measurement time window, including cuff pressure signal, photoplethysmography pulse wave signal, electrocardiogram signal, bioimpedance signal of tissue under cuff and triaxial acceleration signal of arm movement; Pulse wave conduction time extraction module: used to establish a unified timestamp and align the signals in all raw multimodal physiological signal groups, and obtain the raw pulse wave conduction time based on the time interval between the photoplethysmography pulse wave signal and the pulse wave initiation feature point in the electrocardiogram signal; Interference Quantification and Quality Grading Module: This module is used to extract interference indicators to characterize the degree of measurement interference based on the preprocessed raw multimodal physiological signal set. After normalizing each interference indicator, it is combined to form an interference degree vector. The interference degree vector is compared with a preset interference threshold, and the corresponding measurement quality level is output. The individualized blood pressure model calculation module is used to perform dynamic correction processing on the original pulse wave transit time based on the interference vector, and establish an individualized correspondence model between the corrected pulse wave transit time and blood pressure to obtain the first blood pressure measurement value. Based on the fitting residual of the correspondence model and the interference vector, the confidence level corresponding to the first blood pressure measurement value is generated. Oscillometric Blood Pressure and Error Estimation Module: Based on the cuff pressure signal, the module uses the oscillometric method to extract feature points of the cuff pressure oscillation amplitude as a function of pressure, and determines the second blood pressure measurement value using the oscillometric method based on the feature points. It also establishes a blood pressure value error estimation model based on the interference vector to quantify the error range corresponding to the second blood pressure measurement value. Adaptive weighted fusion correction module: It is used to determine the adaptive weight coefficients of the first blood pressure measurement and the second blood pressure measurement based on the confidence level of the first blood pressure measurement and the error range of the second blood pressure measurement, set the overall adjustment factor according to the measurement quality level, and perform weighted fusion processing on the first blood pressure measurement and the second blood pressure measurement to obtain the corrected blood pressure value.

[0015] Compared with existing technologies, the beneficial effects of this invention are as follows: By introducing multi-dimensional interference indicators, various disturbances in the measurement process can be comprehensively quantified, providing a basis for subsequent correction and fusion, and significantly improving the measurement robustness in complex scenarios; it not only performs fine dynamic correction of the easily disturbed pulse wave transit time, but also combines the continuous measurement advantages of the pulse wave transit time method with the stability advantages of the oscilloscope method, and through an adaptive weighted fusion algorithm, ensures that the final output blood pressure value maintains high accuracy under various conditions; while outputting the final blood pressure value, it also provides the measurement quality level, the confidence level of the first blood pressure value, and the error range of the second blood pressure value, allowing users or doctors to intuitively understand the reliability of this measurement; based on the corrected pulse wave transit time, an individualized blood pressure model can be established, which can better adapt to the vascular characteristics of different users and avoid the systematic errors brought about by the general model. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of the overall method flow of the present invention; Figure 2 This is a graph showing the first blood pressure measurement data of an embodiment of the present invention; Figure 3 This is a graph showing the final blood pressure data in an embodiment of the present invention; Figure 4 This is a schematic block diagram of the overall device of the present invention. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0018] It should be noted that, unless otherwise defined, the technical or scientific terms used in this invention should have the ordinary meaning understood by one of ordinary skill in the art to which this invention pertains. The terms "first," "second," and similar terms used in this invention do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" mean that the element or object preceding the word encompasses the elements or objects listed following the word and their equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are used only to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0019] Example: Please see Figures 1-3 The present invention provides a technical solution: The method for correcting blood pressure measurements includes the following steps: S1: Within the measurement time window, the original multimodal physiological signal set of the measurement object is acquired synchronously, including cuff pressure signal, photoplethysmography pulse wave signal, electrocardiogram signal, bioimpedance signal of tissue under the cuff and triaxial acceleration signal of arm movement; In this embodiment, a preset measurement time window is used, with a single blood pressure measurement cycle as the measurement time window. The single blood pressure measurement cycle is defined as the fixed duration from before the cuff is inflated to the fixed duration after the cuff is deflated. A hardware-level synchronous clock is established, and a unified timestamp is created for the signals in all the original multimodal physiological signal groups. All sensors collect data at a fixed frequency sampling rate, and each sampling point is assigned a precise timestamp to form a synchronized time series signal matrix.

[0020] The high-precision synchronous acquisition of multimodal signals in this step provides a unified time reference for all subsequent calibration steps. Compared with single-mode or asynchronous acquisition in existing technologies, this step ensures the precise temporal relationship between different physiological signals through 1000Hz synchronous sampling and unified timestamp alignment. This is also the basis for accurate calculation of pulse wave propagation time and interference quantification analysis. It eliminates timing errors caused by sensor clock differences, making the input signals of subsequent steps inherently consistent. The high sampling rate preserves signal details, providing information for feature extraction and interference analysis. The established original signal vector serves as the sole data source for all processing, ensuring the traceability and repeatability of the entire method.

[0021] S2: Establish a unified timestamp and align the signals in all the original multimodal physiological signal groups, and obtain the original pulse wave conduction time based on the time interval between the photoplethysmography pulse wave signal and the pulse wave initiation feature point in the electrocardiogram signal. In this embodiment, the original multimodal physiological signals are preprocessed. Specifically, the preprocessing of the photoplethysmography (PPG) signal uses a bandpass filter. The filter coefficient parameters and passband are defined. The specific formula for preprocessing the original PPG signal is as follows: in, for Preprocessed photoplethysmography (PPG) signal at the sampling time point These are the filter coefficient parameters. Let the filter order be . for The original photoplethysmography (PPG) signal at the sampling time point; the main frequency components of the PPG signal are retained; among which The impulse response of the Butterworth bandpass filter is 0.5~10 Hz, which is used to extract the dominant frequency component of the pulse wave.

[0022] For the preprocessing of ECG signals, an adaptive algorithm is used to define power frequency interference parameters in real time. These parameters include power frequency interference amplitude and phase. Based on these parameters, the original ECG signal is preprocessed and denoised. The specific formulas for ECG signal preprocessing are as follows: in, for Preprocessed ECG signals at sampling time points for The original photoelectrocardiogram signal at the sampling time point, For the power frequency interference amplitude, for The corresponding power frequency interference phase; the purpose of this formula is to eliminate 50 Hz fundamental frequency and second harmonic interference through real-time estimation using an adaptive algorithm. A sine wave superposition method is used for adaptive estimation. and It can track power frequency changes in real time. The 50 Hz fundamental frequency and its second harmonic (100 Hz) are the main sources of interference. The main energy sources have been covered; Indicates the harmonic order of power frequency interference. This corresponds to a 50 Hz base frequency, which is the main frequency of power grid interference. This corresponds to the 100 Hz second harmonic, which is an overtone of the fundamental frequency. If harmonics are ignored, a 50 Hz overtone interference may still remain in the ECG, leading to false R-wave detections. Parameters The estimation error directly affects the elimination effect; if it is too large or too small, it will introduce residual interference.

[0023] The triaxial acceleration data of the arm movement is subjected to a fixed-frequency low-pass filter and smoothed using a ten-point moving average to retain the intentional motion components and eliminate high-frequency vibration noise.

[0024] In this embodiment, the original pulse wave conduction time is obtained based on the time interval between the preprocessed photoplethysmography pulse wave signal and the pulse wave initiation feature point in the electrocardiogram signal. The specific steps are as follows: detect the R wave peak value from the preprocessed electrocardiogram signal, that is, the highest peak value in the QRS complex in the electrocardiogram signal. Specifically, use the first derivative maximum value detection algorithm to obtain a series of R wave time points, each point corresponding to the start of a heart contraction, thereby defining and determining the R wave time series data. In this embodiment, specifically: defining the detected R-wave time series data. The specific formula is as follows: in, For the total number of heartbeats, For the first Each heartbeat corresponds to Peak wave time point, for Preprocessed ECG signals at sampling time points; The pulse wave onset point is detected from the preprocessed photoplethysmography (PPG) signal. Within a fixed time window, the starting point of the PPG signal rise is located, and the point where the first derivative of the PPG signal exceeds 10% of the maximum derivative is identified as the pulse wave arrival time. This is used to define and determine the pulse wave time series data. In this embodiment, the pulse wave time series is defined. The specific formula is as follows: in, For the first The pulse wave start time point corresponding to each heartbeat To determine the length of the time window for finding the starting point of the pulse wave rise, Preprocessed photoplethysmography signal at the sampling time point; Based on the detected R-wave time series data and pulse wave time series data, the original pulse wave conduction time corresponding to each heartbeat is calculated. The value of each original pulse wave conduction time is the pulse wave start time of the corresponding heartbeat minus the corresponding pulse wave start time. The peak pulse time point, minus a fixed physical distance delay, is defined as follows: Define the original pulse wave propagation time series. The specific formula is as follows: in, For the first The original pulse wave conduction time corresponding to each heartbeat Delay for fixed physical distance.

[0025] It is obtained directly from the time difference, and its accuracy depends on the accurate detection of the starting point of the R wave and photoplethysmography pulse wave signal; a 1 ms time difference corresponds to a blood pressure change of about 1 mmHg, so a sampling rate of 1000Hz ensures the precision of the original pulse wave transmission time.

[0026] This step first converts the raw signal into a clean physiological signal and extracts the most basic parameters, providing direct input for subsequent interference quantization and correction. Unlike general filtering in existing technologies, this step employs targeted bandpass filtering and power frequency cancellation, and the filtered signal is directly used for feature detection, avoiding intermediate storage and repeated processing, demonstrating the continuity of the chain design.

[0027] In this step, bandpass filtering effectively removes low-frequency drift and high-frequency noise, highlighting pulse wave characteristics; adaptive power frequency cancellation targets 50 Hz and its harmonics, improving the accuracy of ECG R-wave detection; the extraction of basic parameters lays the foundation for all subsequent steps, for example... The R-wave sequence will be used for the correction in step 4, and for the calculation of heart rate variability in step 3. This step facilitates signal plotting, the first-level physiological information feature extractor, and ensures that subsequent interference quantification and model correction are based on clean and accurate physiological indicators.

[0028] S3: Based on the preprocessed original multimodal physiological signal set, extract interference indicators to characterize the degree of measurement interference. After normalizing each interference indicator, combine them to form an interference degree vector. Compare the interference degree vector with the preset interference threshold and output the corresponding measurement quality level. In this embodiment, the interference indicators include: motion interference, cuff fit interference, heart rate variability interference, and vascular tension interference. The specific steps for obtaining the motion interference value are as follows: First, using preprocessed triaxial acceleration data of the arm, the root mean square acceleration (RMSE) of the triaxial acceleration at each sampling time point is calculated. Then, an acceleration influence weighting coefficient is assigned according to the cuff inflation / deflation state at each sampling time point. Finally, the motion interference value at each sampling time point is the weighted average RMS acceleration. In this embodiment, the calculation formula is as follows: in, This represents the motion interference value. This represents the total number of sampling time points. for The acceleration corresponding to the sampling time point affects the weighting coefficient. for The root mean square acceleration of the three axes at the sampling time points; where The value is 2 during the inflation and deflation phases and 1 during other phases to emphasize the impact of motion during critical periods; in this embodiment of the simulation experiment... The overall motion intensity was reflected by the RMS composite acceleration. The pressure is doubled during the inflation / deflation phase because the cuff pressure is sensitive to changes at this time, and slight movements can easily introduce artifacts. The thresholds of 0.02 and 0.05 g are derived from experimental statistics: RMS < 0.02 g at normal rest, and can reach 0.05 g with slight movements.

[0029] The specific steps for obtaining cuff fit interference are as follows: First, calculate the average value of the bioimpedance signal of the tissue under the cuff for a fixed period before the cuff is inflated as the baseline impedance value; within the monitoring time window, find the maximum and minimum impedance values ​​during the entire measurement process, divide the difference between the maximum and minimum impedance values ​​by the baseline impedance value, and then perform percentage calculation to obtain the cuff fit interference value; in this embodiment, the formula for calculating the cuff fit interference value is as follows: in, This represents the interference value for cuff fit. Define the maximum impedance parameter within the monitoring time window. Define the minimum impedance parameter within the monitoring time window. Baseline impedance parameters; This embodiment simulates the stability of the cuff in contact with the arm. The bioimpedance change is utilized because cuff loosening causes fluctuations in electrode contact impedance. The ratio of the peak value to the baseline can be normalized to represent the relative change. A value greater than 15% indicates a good fit, while a value greater than 15% suggests a possible loosening.

[0030] The specific steps for obtaining heart rate variability interference are as follows: Based on R-wave time series data, obtain the time interval between two adjacent R waves in the electrocardiogram signal to construct RR interval sequence data. Calculate the standard deviation and mean of the RR intervals, and determine the exponential decay factor based on the mean of the RR interval sequence data. The heart rate variability interference value is obtained by dividing the standard deviation of the RR intervals by the mean of the RR intervals and multiplying by the exponential decay factor. The specific formula is as follows: in, This represents the heart rate variability interference value. The standard deviation of the RR interval series data is given. This represents the average value of the RR interval series data. Standard deviation The mean is used, and the exponential term penalizes variability dilution when the heart rate is too slow or the RR is too long. In this embodiment, the simulation experiment... , It's the coefficient of variation, but even if the absolute variation is large when the RR is too long, the relative variation may not be large. However, a slow heart rate itself affects the accuracy of blood pressure measurement, so it's multiplied by... ,when The index begins to decay when the heart rate is <75 bpm, suppressing variability and preventing misdiagnosis. The thresholds of 0.15 and 0.25 are adjusted based on clinical diagnostic criteria for arrhythmia.

[0031] The specific steps for obtaining vascular tension interference are as follows: First, obtain the pulse amplitude component based on the photoplethysmography (PPG) signal, i.e., the peak-to-valley difference of the PPG signal within a single cardiac cycle. Then, obtain the average blood volume within a single cardiac cycle: the average blood volume within a single cardiac cycle is the integral of the PPG signal within the cardiac cycle divided by the time length. The ratio of the pulse amplitude component to the average blood volume corresponding to a single cardiac cycle is used as the pulsatility efficiency ratio for that single cardiac cycle. The ratio of the median pulsatility efficiency ratio to the median cardiac cycle length is used as the vascular tension interference value. The specific formula is as follows: in, For the first The pulse amplitude component corresponding to each heartbeat For the first Average blood volume components per cardiac cycle This is the vascular tension interference value; it reflects the ratio of pulse wave amplitude to baseline and is related to vascular compliance. Dividing by the pulse cycle is to eliminate the influence of heart rate, so that the indicator reflects the relative tension change of the pulse. Taking the median in the denominator can resist abnormal pulsation interference. A larger value indicates higher vascular tension and a stiffer blood vessel; in this embodiment, a simulation experiment... Its value will be used for PTT correction in step 4.

[0032] After normalizing each interference index, an interference degree vector is formed. This vector is then compared with a preset interference threshold to output the corresponding measurement quality level. Specifically, the measurement quality level is divided into three levels: low, medium, and high. For motion interference, cuff fit interference, heart rate variability interference, and vascular tension interference, interference thresholds are set for each interference index. If the number of interference indices exceeding their corresponding interference thresholds is greater than or equal to 3, the measurement quality level is high; if the number of interference indices exceeding their corresponding interference thresholds is greater than 1 and less than 3, the measurement quality level is medium; and if the number of interference indices exceeding their corresponding interference thresholds is greater than 0 and less than or equal to 1, the measurement quality level is low. The measurement quality level is divided into three levels, and an overall adjustment factor is set based on the measurement quality level. S4: Perform dynamic correction processing on the original pulse wave transit time based on the interference vector, and establish an individualized correspondence model between the corrected pulse wave transit time and blood pressure to obtain the first blood pressure measurement value. Based on the fitting residual of the correspondence model and the interference vector, generate the confidence level corresponding to the first blood pressure measurement value. In this embodiment, the original pulse wave conduction time is compensated and adjusted according to the magnitude of each component in the interference vector. First, the maximum correction amplitude, the reference vascular tension value, and the correction sensitivity are defined, and the vascular tension correction factor is calculated. Then, the original pulse wave conduction time is multiplied by the vascular tension correction factor to obtain the preliminarily corrected pulse wave conduction time sequence. The specific formula is as follows: in, This is the corrected pulse wave conduction time. For the maximum correction range, For reference vascular tension values, To correct the sensitivity parameters. The function was chosen because the effect of vascular tension on PTT has a saturation effect: when When the deviation from the normal range occurs, the correction amount tends to be constant to avoid over-correction. In this embodiment, 0.25 is Typical mean The width of the transition zone is controlled at 0.1. A value of 0.18 indicates a maximum correction of 18%, consistent with the range of PTT changes caused by vascular stiffness in the literature. As the pulse wave transit time increases, the pulse wave transit time should decrease; therefore, the correction factor is less than 1. The initially corrected pulse wave transit time series is then corrected for cuff pressure effects. First, the cuff pressure value corresponding to each pulse wave transit time is obtained. Then, the systolic and diastolic blood pressure at that moment are estimated using a simplified relationship. Based on the initially estimated diastolic and systolic blood pressures, a pressure correction factor is obtained. The pulse wave transit time is then corrected using the cuff pressure value corresponding to each pulse wave transit time and the pressure correction factor. Finally, median filtering is used to remove abnormal pulse wave transit times, resulting in the corrected pulse wave transit time. The formula is as follows: in, For the first Cuff pressure at the onset time of the pulse wave corresponding to each heartbeat. For maximum cuff pressure, As the pressure dependence correction coefficient, in this embodiment , This is the corrected pulse wave conduction time.

[0033] This step is for A dual physical correction was performed: vascular compliance correction and pressure-dependent correction, resulting in a more realistic representation of vascular conditions. Unlike existing technologies that directly use the raw pulse wave transit time, this step fully considers the effects of changes in vascular tension and cuff pressure on pulse wave velocity, making pulse wave transit time a more stable indicator for blood pressure estimation. Function introduction Nonlinear corrections were applied to simulate the non-monotonic effect of vascular stiffness on pulse wave transit time. Exponential decay was used to correct for pressure dependence, consistent with the physiological principle that increased pressure leads to vascular wall tension and a decrease in pulse wave transit time. The generated vascular model parameters include the mean, variability, tension, and pressure sensitivity of pulse wave transit time, providing rich, personalized information for subsequent pulse wave transit time-based blood pressure calculations. For the overall scheme, this step serves as a bridge between interference quantification and blood pressure calculation, ensuring that subsequent pulse wave transit time-based blood pressure estimation is based on the corrected, reliable data.

[0034] Based on the corrected pulse wave transit time, an individualized correspondence model between pulse wave transit time and blood pressure is established; the first blood pressure measurement value based on the pulse wave transit time is calculated according to the correspondence model; and the confidence level corresponding to the first blood pressure measurement value is generated based on the fitting residual of the correspondence model and the interference vector. In this embodiment, an individualized correspondence model between pulse wave transit time and blood pressure is established based on the corrected pulse wave transit time. The specific steps for establishing the model are as follows: extracting parameters describing individualized vascular characteristics from the corrected pulse wave transit time, including: the average pulse wave transit time, the standard deviation of pulse wave transit time, the correlation coefficient between pulse wave transit time and blood pressure, and the rate of change of pulse wave transit time with respect to pressure; substituting each corrected pulse wave transit time into the individualized correspondence model between pulse wave transit time and blood pressure to obtain the first blood pressure measurement value, including: the systolic blood pressure and the diastolic blood pressure of the correspondence model, and obtaining the model confidence based on the interquartile range of the systolic blood pressure estimate and the interquartile range of the diastolic blood pressure estimate.

[0035] In this embodiment, a simulated measurement was performed on a healthy adult. Multimodal physiological signals were simultaneously acquired within a single blood pressure measurement cycle. After preprocessing, interference quantification, and correction, the raw pulse wave conduction time for 30 cardiac cycles and the corresponding cuff pressure were obtained. After vascular tension correction After being modified for stress dependence And beat-by-beat systolic blood pressure based on the corrected PTT estimate. and diastolic blood pressure The original experimental data are shown in Table 1 below. Table 1 details the complete processing chain from raw signal to corrected parameters for 30 cardiac cycles. (See also...) Figure 3 , The heart rate fluctuates slightly with the heart cycle, reflecting physiological variability; The blood pressure decreased linearly from 160 mmHg to 60 mmHg, simulating the deflation process of the cuff. After correction for vascular tension, Comparison A slight increase of approximately 8% reflects the effect of vascular compliance on pulse wave propagation time. Further pressure-dependent corrections are introduced. back, The cuff gradually increases in pressure as the pressure decreases, consistent with the physiological principle that reduced pressure leads to decreased vascular wall tension and slower pulse wave transmission velocity. Based on beat-by-beat blood pressure estimates (117.61~118.89 mmHg) and (78.41~79.26 mmHg) and The negative correlation, with fluctuations within normal physiological limits, demonstrates the PTT method's ability to capture continuous changes in blood pressure. This table provides reliable, physically corrected input for subsequent fusion steps.

[0036] This step maps the corrected pulse wave transit time series to blood pressure values ​​and introduces an individualized model and confidence assessment, overcoming the drawback of the traditional PTT method requiring frequent calibration. It employs a method consistent with the logarithmic relationship between pulse wave velocity and blood pressure, which has a physiological basis. For new users, calibration is performed quickly through measurement, while for experienced users, historical data is used for recursive updates, achieving adaptive learning. The confidence score is calculated based on the consistency of the pulse wave transit time series, intuitively reflecting the reliability of the measurement results.

[0037] S5: Based on the cuff pressure signal, the oscillometric method is used to extract the feature points of the cuff pressure oscillation amplitude as a function of pressure, and the second blood pressure measurement value is determined by the oscillometric method based on the feature points. A blood pressure value error estimation model is established based on the interference vector to quantify the error range corresponding to the second blood pressure measurement value. In this embodiment, the oscillometric method is used to extract feature points of the cuff pressure oscillation amplitude as a function of pressure, and the second blood pressure measurement value is determined based on these feature points. The second blood pressure measurement value includes both oscillometric systolic blood pressure and oscillometric diastolic blood pressure measurements. The error ranges for the oscillometric blood pressure measurements also correspond to the error ranges for systolic and diastolic blood pressure measurements. The adaptive weighting coefficients include adaptive fusion weighting coefficients for systolic and diastolic blood pressure. The systolic blood pressure fusion weighting coefficient is determined by the model confidence level and the error range of the oscillometric systolic blood pressure measurement. The diastolic blood pressure fusion weighting coefficient is determined by the model confidence level and the error range of the oscillometric diastolic blood pressure measurement. In this embodiment, the oscillometric systolic blood pressure measurement value is 120 mmHg and the oscillometric diastolic blood pressure measurement value is 80 mmHg.

[0038] This step extracts the oscillometric blood pressure value from the cuff pressure signal and provides an error estimate based on the interference vector, which is not available in the traditional oscillometric method. Gaussian fitting is used to extract the oscillatory wave envelope, which is more robust to noise than the traditional ratio method and can obtain a continuous envelope. The error estimate quantifies the interference of S3 into possible deviations in the blood pressure value, so that the oscillometric result is no longer an isolated value, but an estimate with a confidence interval.

[0039] S6: Determine the adaptive weighting coefficients of the first and second blood pressure measurements based on the confidence level of the first blood pressure measurement and the error range of the second blood pressure measurement. Set the overall adjustment factor according to the measurement quality level, and perform weighted fusion processing on the first and second blood pressure measurements to obtain the corrected blood pressure value.

[0040] Based on the measurement quality level being divided into three levels—low, medium, and high—corresponding to three overall adjustment factors, in this embodiment, after setting the evaluation thresholds for each of the above-mentioned interference indicators to complete the comprehensive evaluation of the measurement quality level, the measurement quality level of the simulation experiment is medium. The adaptive weight coefficient is adjusted by matching the measurement quality level with the overall adjustment factor to obtain the final systolic blood pressure fusion weight coefficient and diastolic blood pressure fusion weight coefficient. Specifically, For shrinkage pressure fusion weight, The diastolic blood pressure fusion weight is determined by the confidence level of the first blood pressure measurement obtained from the error range of oscillometric blood pressure and the correspondence model between individualized pulse wave transit time and blood pressure, as shown in the following formula: in, For shrinkage pressure fusion weight, For diastolic blood pressure fusion weights, This represents the estimated error range for systolic blood pressure in the second blood pressure measurement using the oscillometric method. This represents the estimation error range of diastolic blood pressure in the second blood pressure measurement using the oscillometric method; in this embodiment, The confidence level of the first blood pressure measurement obtained from the individualized model of the relationship between pulse wave conduction time and blood pressure. The baseline error of the first blood pressure measurement value is determined by an individualized model of the correlation between pulse wave transit time and blood pressure; in the simulation experiment of this embodiment... , , , The measurement quality level is divided into three levels: low, medium, and high, each corresponding to three overall adjustment factors, as shown in the following formula: in, For quality grade, This is the quality level adjustment factor; in this embodiment... Based on adjustments to the systolic blood pressure fusion weight and the diastolic blood pressure fusion weight, the specific formula is as follows: in, To adjust the systolic blood pressure fusion weight after adjusting the factor, The adjusted factor is the diastolic blood pressure fusion weight; in the simulation experiment of this embodiment... , Based on the final systolic blood pressure fusion weighting coefficient and diastolic blood pressure fusion weighting coefficient, the blood pressure measurement value based on pulse wave transit time and the blood pressure oscillometric measurement value are weighted and fused to obtain the final corrected blood pressure value. The specific formula is as follows: in, To correct the systolic blood pressure, To correct diastolic blood pressure, The systolic blood pressure is the second blood pressure measurement. This refers to the diastolic pressure in the second blood pressure measurement using the oscillometric method. The systolic blood pressure is the first blood pressure measurement. This refers to the diastolic blood pressure in the first blood pressure measurement; the corrected blood pressure value can be found in the reference section. Figure 3 , Figure 3 The final blood pressure correction value output by this technical solution is presented, with one fused blood pressure result corresponding to each cardiac cycle. The range is between 119.43 and 119.74 mmHg. The range was between 79.64 and 79.84 mmHg, with an overall fluctuation of less than ±0.3 mmHg, which is consistent with the original estimate obtained by the PTT method. The fluctuation (approximately 1.3 mmHg) is more stable and converges towards the oscillometric reference value (120 / 80 mmHg). This result demonstrates the advantage of the fusion strategy: while preserving the details of beat-by-beat physiological fluctuations, it utilizes the stability of the oscillometric method to suppress random errors and residual interference in the PTT method. (Systolic blood pressure fusion weight) =0.7598, diastolic blood pressure fusion weight =0.7766, indicating that the oscillometric method dominates the final result, but the continuous information from the PTT method still contributes approximately 24% of the weight through weighting. This table visually demonstrates the practical effect of multimodal fusion technology in improving the accuracy and robustness of blood pressure measurement, providing a more reliable beat-by-beat blood pressure reference for clinical decision-making.

[0041] This step combines the results of the first and second blood pressure measurements and calculates the overall reliability, achieving a complementary advantage of the two methods. The weights are dynamically adjusted based on the oscillometric error and the confidence level of the first blood pressure measurement. The fusion formula uses the inverse square of the error for weighting, which conforms to the optimal estimation theory. The overall reliability not only integrates the reliability of the two methods but also introduces a quality indicator for penalty, giving the final result a degree of reliability. The beneficial effect is that the final blood pressure value utilizes the stability of the oscillometric method while incorporating the continuous information of the PTT method, making it more accurate than either method. At the same time, the reliability index provides decision support for users.

[0042] Please see Figure 4 The present invention also provides an apparatus for correcting blood pressure measurements, the apparatus being used to perform the above-described method for correcting blood pressure measurements, comprising: Multimodal signal synchronous acquisition module: used to synchronously acquire the original multimodal physiological signal set of the measurement object within the measurement time window, including cuff pressure signal, photoplethysmography pulse wave signal, electrocardiogram signal, bioimpedance signal of tissue under cuff and triaxial acceleration signal of arm movement; Pulse wave conduction time extraction module: used to establish a unified timestamp and align the signals in all raw multimodal physiological signal groups, and obtain the raw pulse wave conduction time based on the time interval between the photoplethysmography pulse wave signal and the pulse wave initiation feature point in the electrocardiogram signal; Interference Quantification and Quality Grading Module: This module is used to extract interference indicators to characterize the degree of measurement interference based on the preprocessed raw multimodal physiological signal set. After normalizing each interference indicator, it is combined to form an interference degree vector. The interference degree vector is compared with a preset interference threshold, and the corresponding measurement quality level is output. The individualized blood pressure model calculation module is used to perform dynamic correction processing on the original pulse wave transit time based on the interference vector, and establish an individualized correspondence model between the corrected pulse wave transit time and blood pressure to obtain the first blood pressure measurement value. Based on the fitting residual of the correspondence model and the interference vector, the confidence level corresponding to the first blood pressure measurement value is generated. Oscillometric Blood Pressure and Error Estimation Module: Based on the cuff pressure signal, the module uses the oscillometric method to extract feature points of the cuff pressure oscillation amplitude as a function of pressure, and determines the second blood pressure measurement value using the oscillometric method based on the feature points. It also establishes a blood pressure value error estimation model based on the interference vector to quantify the error range corresponding to the second blood pressure measurement value. Adaptive weighted fusion correction module: It is used to determine the adaptive weight coefficients of the first blood pressure measurement and the second blood pressure measurement based on the confidence level of the first blood pressure measurement and the error range of the second blood pressure measurement, set the overall adjustment factor according to the measurement quality level, and perform weighted fusion processing on the first blood pressure measurement and the second blood pressure measurement to obtain the corrected blood pressure value.

[0043] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0044] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented in software, the above embodiments can be implemented, in whole or in part, as a computer program product. Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution.

[0045] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0046] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A method for calibrating blood pressure measurements, characterized in that, The specific steps include: S1: Within the measurement time window, the original multimodal physiological signal set of the measurement object is acquired synchronously, including cuff pressure signal, photoplethysmography pulse wave signal, electrocardiogram signal, bioimpedance signal of tissue under the cuff and triaxial acceleration signal of arm movement; S2: Establish a unified timestamp and align the signals in all the original multimodal physiological signal groups, and obtain the original pulse wave conduction time based on the time interval between the photoplethysmography pulse wave signal and the pulse wave initiation feature point in the electrocardiogram signal. S3: Based on the preprocessed original multimodal physiological signal set, extract interference indicators to characterize the degree of measurement interference. After normalizing each interference indicator, combine them to form an interference degree vector. Compare the interference degree vector with the preset interference threshold and output the corresponding measurement quality level. S4: Perform dynamic correction processing on the original pulse wave transit time based on the interference vector, and establish an individualized correspondence model between the corrected pulse wave transit time and blood pressure to obtain the first blood pressure measurement value. Based on the fitting residual of the correspondence model and the interference vector, generate the confidence level corresponding to the first blood pressure measurement value. S5: Based on the cuff pressure signal, the oscillometric method is used to extract the feature points of the cuff pressure oscillation amplitude as a function of pressure, and the second blood pressure measurement value is determined by the oscillometric method based on the feature points. A blood pressure value error estimation model is established based on the interference vector to quantify the error range corresponding to the second blood pressure measurement value. S6: Determine the adaptive weighting coefficients of the first and second blood pressure measurements based on the confidence level of the first blood pressure measurement and the error range of the second blood pressure measurement. Set the overall adjustment factor according to the measurement quality level, and perform weighted fusion processing on the first and second blood pressure measurements to obtain the corrected blood pressure value.

2. The method for correcting blood pressure measurements according to claim 1, characterized in that: A preset measurement time window is used, with a single blood pressure measurement cycle as the measurement time window. The single blood pressure measurement cycle is defined as the fixed duration from before the cuff is inflated to the fixed duration after the cuff is deflated. A hardware-level synchronous clock is established, and a unified timestamp is created for all signals in the original multimodal physiological signal group. All sensors collect data at a fixed frequency sampling rate, and each sampling point is assigned a precise timestamp to form a synchronized time series signal matrix.

3. The method for correcting blood pressure measurements according to claim 1, characterized in that: Preprocessing operations are performed on the raw multimodal physiological signals. For the photoplethysmography (PPG) signal, a bandpass filter is used, defining filter coefficients and passband to preserve the main frequency components. For the electrocardiogram (ECG) signal, an adaptive algorithm is used to define power frequency interference parameters in real time, including amplitude and phase. Noise reduction is achieved based on these parameters. Finally, the triaxial acceleration data of arm movements are subjected to a fixed-frequency low-pass filter with ten-point moving average smoothing to retain the intended motion components and eliminate high-frequency vibration noise.

4. The method for correcting blood pressure measurements according to claim 3, characterized in that: Based on the time interval between the preprocessed photoplethysmography pulse wave signal and the pulse wave initiation feature point in the electrocardiogram (ECG) signal, the original pulse wave conduction time is obtained. The specific steps are as follows: detect the R-wave peak value from the preprocessed ECG signal, that is, the highest peak value in the QRS complex of the ECG signal. Specifically, use the first derivative maximum value detection algorithm to obtain a series of R-wave time points, each point corresponding to the start of a heart contraction, thereby defining and determining the R-wave time series data. The pulse wave start point is detected from the preprocessed photoplethysmography (PPG) signal. The starting point of the PPG signal rise is found within a fixed time window. The point where the first derivative of the PPG signal exceeds 10% of the maximum derivative is found as the pulse wave arrival time. The pulse wave time series data is defined and determined in this way. Based on the detected R-wave time series data and pulse wave time series data, the original pulse wave conduction time corresponding to each heartbeat is calculated. The value of each original pulse wave conduction time is the pulse wave start time of the corresponding heartbeat minus the corresponding pulse wave start time. The peak wave time is then subtracted from the fixed physical distance delay.

5. The method for correcting blood pressure measurements according to claim 4, characterized in that: Interference indicators include: motion interference, cuff fit interference, heart rate variability interference, and vascular tension interference. The specific steps for obtaining motion interference values ​​are as follows: First, using the preprocessed triaxial acceleration data of the arm, the root mean square acceleration of the triaxial acceleration at each sampling time point is calculated. Then, according to the cuff inflation / deflation state at each sampling time point, an acceleration influence weighting coefficient is assigned. Finally, the motion interference value at each sampling time point is the weighted average root mean square acceleration. The specific steps to obtain cuff fit interference are as follows: First, calculate the average value of the bioimpedance signal of the tissue under the cuff for a fixed time before the cuff is inflated as the baseline impedance value; find the maximum and minimum impedance values ​​during the entire measurement process within the monitoring time window, divide the difference between the maximum and minimum impedance values ​​by the baseline impedance value and then perform percentage calculation to obtain the cuff fit interference value. The specific steps for obtaining heart rate variability interference are as follows: Based on R-wave time series data, the time interval between two adjacent R waves in the electrocardiogram signal is obtained to construct RR interval sequence data. The standard deviation and mean of the RR interval are calculated, and the exponential decay factor is determined based on the mean of the RR interval sequence data. The heart rate variability interference value is obtained by dividing the standard deviation of the RR interval by the mean of the RR interval and multiplying it by the exponential decay factor. The specific steps for obtaining vascular tension interference are as follows: First, obtain the pulse amplitude component based on the photoplethysmography (PPG) signal, i.e., the peak-to-valley difference of the PPG signal within a single heartbeat cycle. Then, obtain the average blood volume within a single heartbeat cycle: the average blood volume within a single heartbeat cycle is the integral of the PPG signal within the heartbeat cycle divided by the time length. The ratio of the pulse amplitude component corresponding to a single heartbeat cycle to the average blood volume is used as the pulsation efficiency ratio for that single heartbeat cycle. The ratio of the median pulsation efficiency ratio to the median heartbeat cycle length is used as the vascular tension interference value.

6. The method for correcting blood pressure measurements according to claim 5, characterized in that: After normalizing each interference index, an interference degree vector is formed. The interference degree vector is compared with a preset interference threshold, and the corresponding measurement quality level is output. Specifically, the measurement quality level is divided into three levels: low, medium, and high. For motion interference, cuff fit interference, heart rate variability interference, and vascular tension interference, an interference threshold is set for each interference index. When the number of interference indices that exceed their corresponding interference threshold is greater than or equal to 3, the measurement quality level is high; when the number of interference indices that exceed their corresponding interference threshold is greater than 1 and less than 3, the measurement quality level is medium. When the number of interference indicators that exceed their corresponding interference thresholds is greater than 0 and less than or equal to 1, the measurement quality level is low. The measurement quality level is divided into three levels: low, medium, and high. The overall adjustment factor is set based on the measurement quality level.

7. The method for correcting blood pressure measurements according to claim 1, characterized in that: The original pulse wave transit time is compensated and adjusted based on the magnitude of each component in the interference vector. First, the maximum correction amplitude, reference vascular tension value, and correction sensitivity are defined, and the vascular tension correction factor is calculated. Then, the original pulse wave transit time is multiplied by the vascular tension correction factor to obtain a preliminary corrected pulse wave transit time series. The preliminary corrected pulse wave transit time series is then corrected for cuff pressure effects. First, the cuff pressure value corresponding to each pulse wave transit time is obtained. Then, the systolic and diastolic blood pressure at the pulse wave initiation time corresponding to each pulse wave transit time are estimated by simplifying the relationship. The pressure correction factor is obtained based on the preliminary estimated diastolic and systolic blood pressure. The pulse wave transit time is corrected by the cuff pressure value corresponding to each pulse wave transit time and the pressure correction factor. Finally, median filtering is used to remove abnormal pulse wave transit times to obtain the corrected pulse wave transit time.

8. The method for correcting blood pressure measurements according to claim 7, characterized in that: Based on the corrected pulse wave transit time, an individualized correspondence model between pulse wave transit time and blood pressure was established. The specific steps for establishing the model are as follows: extract parameters describing individualized vascular characteristics from the corrected pulse wave transit time. These parameters include: the average pulse wave transit time, the standard deviation of pulse wave transit time, the correlation coefficient between pulse wave transit time and blood pressure, and the rate of change of pulse wave transit time with respect to pressure. Substitute each corrected pulse wave transit time into the individualized correspondence model between pulse wave transit time and blood pressure to obtain the first blood pressure measurement value, which includes the systolic blood pressure and diastolic blood pressure of the correspondence model. The model confidence is obtained based on the interquartile range of the estimated systolic blood pressure and the interquartile range of the estimated diastolic blood pressure.

9. The method for correcting blood pressure measurements according to claim 8, characterized in that: The oscillometric method was used to extract the characteristic points of the cuff pressure oscillation amplitude as a function of pressure, and the second blood pressure measurement value was determined based on the characteristic points. The second blood pressure measurement using the oscillometric method includes the oscillometric systolic blood pressure measurement and the oscillometric diastolic blood pressure measurement. The error ranges for oscillometric blood pressure measurements also correspond to the error ranges for systolic blood pressure and diastolic blood pressure measurements. The adaptive weighting coefficients include the adaptive fusion weighting coefficient for systolic blood pressure and the adaptive fusion weighting coefficient for diastolic blood pressure; The systolic blood pressure fusion weight coefficient is determined by the model confidence level and the error range of the oscillometric systolic blood pressure measurement; the diastolic blood pressure fusion weight coefficient is determined by the model confidence level and the error range of the oscillometric diastolic blood pressure measurement; based on the measurement quality level, there are three levels: low, medium, and high, each corresponding to three overall adjustment factors. The adaptive weight coefficients are adjusted by matching the overall adjustment factors with the measurement quality level to obtain the final systolic blood pressure fusion weight coefficient and diastolic blood pressure fusion weight coefficient; based on the final systolic blood pressure fusion weight coefficient and diastolic blood pressure fusion weight coefficient, the blood pressure measurement value based on pulse wave transit time and the blood pressure measurement value by oscillometric method are weighted and fused to obtain the final corrected blood pressure value.

10. A device for correcting blood pressure measurements, characterized in that, The device for correcting blood pressure measurements is used to perform the method for correcting blood pressure measurements according to any one of claims 1-8, comprising: Multimodal signal synchronous acquisition module: used to synchronously acquire the original multimodal physiological signal set of the measurement object within the measurement time window, including cuff pressure signal, photoplethysmography pulse wave signal, electrocardiogram signal, bioimpedance signal of tissue under cuff and triaxial acceleration signal of arm movement; Pulse wave conduction time extraction module: used to establish a unified timestamp and align the signals in all raw multimodal physiological signal groups, and obtain the raw pulse wave conduction time based on the time interval between the photoplethysmography pulse wave signal and the pulse wave initiation feature point in the electrocardiogram signal; Interference Quantification and Quality Grading Module: This module is used to extract interference indicators to characterize the degree of measurement interference based on the preprocessed raw multimodal physiological signal set. After normalizing each interference indicator, it is combined to form an interference degree vector. The interference degree vector is compared with a preset interference threshold, and the corresponding measurement quality level is output. The individualized blood pressure model calculation module is used to perform dynamic correction processing on the original pulse wave transit time based on the interference vector, and establish an individualized correspondence model between the corrected pulse wave transit time and blood pressure to obtain the first blood pressure measurement value. Based on the fitting residual of the correspondence model and the interference vector, the confidence level corresponding to the first blood pressure measurement value is generated. Oscillometric Blood Pressure and Error Estimation Module: Based on the cuff pressure signal, the module uses the oscillometric method to extract feature points of the cuff pressure oscillation amplitude as a function of pressure, and determines the second blood pressure measurement value using the oscillometric method based on the feature points. It also establishes a blood pressure value error estimation model based on the interference vector to quantify the error range corresponding to the second blood pressure measurement value. Adaptive weighted fusion correction module: It is used to determine the adaptive weight coefficients of the first blood pressure measurement and the second blood pressure measurement based on the confidence level of the first blood pressure measurement and the error range of the second blood pressure measurement, set the overall adjustment factor according to the measurement quality level, and perform weighted fusion processing on the first blood pressure measurement and the second blood pressure measurement to obtain the corrected blood pressure value.

Images