WO2023103977A1 - 基于单路脉搏波的血压监测装置、存储介质及电子设备 - Google Patents

基于单路脉搏波的血压监测装置、存储介质及电子设备 Download PDF

Info

Publication number
WO2023103977A1
WO2023103977A1 PCT/CN2022/136638 CN2022136638W WO2023103977A1 WO 2023103977 A1 WO2023103977 A1 WO 2023103977A1 CN 2022136638 W CN2022136638 W CN 2022136638W WO 2023103977 A1 WO2023103977 A1 WO 2023103977A1
Authority
WO
WIPO (PCT)
Prior art keywords
blood pressure
pulse wave
signal
calculation model
filtering
Prior art date
Application number
PCT/CN2022/136638
Other languages
English (en)
French (fr)
Inventor
刘剑
鲁子鹏
孙凤云
Original Assignee
山东大学
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 山东大学 filed Critical 山东大学
Publication of WO2023103977A1 publication Critical patent/WO2023103977A1/zh

Links

Images

Classifications

    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/02—Detecting, measuring or recording pulse, heart rate, blood pressure or blood flow; Combined pulse/heart-rate/blood pressure determination; Evaluating a cardiovascular condition not otherwise provided for, e.g. using combinations of techniques provided for in this group with electrocardiography or electroauscultation; Heart catheters for measuring blood pressure
    • A61B5/021—Measuring pressure in heart or blood vessels
    • A61B5/02108—Measuring pressure in heart or blood vessels from analysis of pulse wave characteristics
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
    • A61B5/7203—Signal processing specially adapted for physiological signals or for diagnostic purposes for noise prevention, reduction or removal
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
    • A61B5/7225—Details of analog processing, e.g. isolation amplifier, gain or sensitivity adjustment, filtering, baseline or drift compensation

Definitions

  • the invention belongs to the technical field of blood pressure detection devices, and in particular relates to a blood pressure monitoring device based on a single pulse wave, a storage medium and electronic equipment.
  • the blood pressure of the human body is constantly changing. Continuous monitoring of blood pressure is one of the measures to efficiently check the health status of hypertensive patients. Continuous blood pressure monitoring can capture abnormal fluctuations in blood pressure in time, and warn users of sudden heart attacks such as cerebral infarction and myocardial infarction in advance. Possible threat of vascular disease.
  • the types of signals that the detection equipment needs to collect are also different.
  • there are more methods used to calculate blood pressure based on oscillometric method calculate blood pressure based on ECG signal and pulse wave signal, and calculate blood pressure based on two-way pulse wave signal. Blood pressure is measured based on a single pulse wave.
  • the method of calculating blood pressure based on the oscillometric method requires periodic inflation and deflation, which has a poor user experience, and the measurement interval is long, which makes it impossible to achieve true continuous blood pressure detection.
  • the method of obtaining blood pressure based on the cooperation of ECG signal and pulse wave signal and the method of obtaining blood pressure based on two-way pulse wave has strict requirements on the cooperation of the two signals, and the operation process is complicated and difficult to implement. Moreover, the two-way signal is easily disturbed by noise, resulting in poor accuracy of calculation results.
  • the operation process of blood pressure monitoring based on single-channel pulse wave is simple, and the measuring equipment is smaller and easy to carry. It is a blood pressure continuous detection solution with better user experience.
  • the existing blood pressure calculated based on single-channel pulse waveform analysis has large errors, which cannot meet clinical needs.
  • the errors mainly come from three aspects: First, the pulse wave signal is easily affected by individual differences during the acquisition process, resulting in uneven signal quality, and the signal that contains individual differences will cover up the pulse wave waveform characteristics and the inner blood pressure. connect. Second, the continuous measurement environment of the pulse wave signal is changeable, and the degree of noise interference is different. It is difficult for the existing filtering methods to dynamically adapt to the signal filtering under different noise interference levels.
  • the third is that although the blood pressure calculation model can reflect the user's blood pressure change trend, the blood pressure benchmarks of different users are not the same, so if you want to obtain a higher precision blood pressure, you need to perform blood pressure calibration.
  • the inventors found that when detecting blood pressure based on a single pulse wave, there are still errors in the three aspects of signal acquisition process, signal processing process and calculation model, resulting in poor accuracy of blood pressure detection.
  • the present invention provides a blood pressure monitoring device, storage medium and electronic equipment based on a single-channel pulse wave, which can be analyzed from three aspects: signal acquisition process, signal processing process and calculation model. Reduce blood pressure detection errors, thereby improving the accuracy of blood pressure detection results.
  • the first aspect of the present invention provides a blood pressure monitoring device based on a single pulse wave, which includes:
  • a signal acquisition module which is used to acquire a single pulse wave signal after eliminating individual differences
  • a signal filtering module which is used to filter the single-channel pulse wave signal
  • a blood pressure calculation and calibration module which is used to extract the characteristic parameters of the filtered single-channel pulse wave signal, and obtain the blood pressure detection value based on the extracted characteristic parameters and the calibrated blood pressure calculation model;
  • the calibration process of the blood pressure calculation model is:
  • a set of calibration coefficients is calculated based on the reference blood pressure value and the extracted characteristic parameters; wherein, the calculation model of the calibration coefficient is a set polynomial function of the reference blood pressure value and the extracted characteristic parameters;
  • the calibration coefficient is used as the coefficient and index of the setting items of the known initial blood pressure calculation model to calibrate the blood pressure calculation model to obtain a calibrated blood pressure calculation model.
  • a sliding filtering algorithm with variable window length is used to filter the single pulse wave signal.
  • the preset distribution of photoelectric sensors is used to identify the wrist motion mode, and then the wrist motion mode matching
  • the photoelectric compensation strategy performs photoelectric compensation on the single-channel pulse wave signal after eliminating individual differences, and finally filters the photoelectrically compensated single-channel pulse wave signal through the decomposition and reconstruction of the empirical mode and the sliding filter algorithm with variable window length.
  • a second aspect of the present invention provides a computer-readable storage medium on which a computer program is stored, and when the program is executed by a processor, the following steps are implemented:
  • the calibration process of the blood pressure calculation model is:
  • a set of calibration coefficients is calculated based on the reference blood pressure value and the extracted characteristic parameters; wherein, the calculation model of the calibration coefficient is a set polynomial function of the reference blood pressure value and the extracted characteristic parameters;
  • the calibration coefficient is used as the coefficient and index of the setting items of the known initial blood pressure calculation model to calibrate the blood pressure calculation model to obtain the calibrated blood pressure calculation model.
  • a sliding filter algorithm with variable window length is used to filter the single-channel pulse wave signal.
  • the photoelectric sensors with preset distribution are used to identify the wrist motion mode, and then the photoelectric compensation strategy matching the wrist motion mode is used to eliminate individual differences Afterwards, the single-channel pulse wave signal is subjected to photoelectric compensation, and finally the single-channel pulse wave signal after photoelectric compensation is filtered through the decomposition and reconstruction of the empirical mode and the sliding filter algorithm with variable window length.
  • a third aspect of the present invention provides a computer device, including a memory, a processor, and a computer program stored on the memory and operable on the processor, and the processor implements the following steps when executing the program:
  • the calibration process of the blood pressure calculation model is:
  • a set of calibration coefficients is calculated based on the reference blood pressure value and the extracted characteristic parameters; wherein, the calculation model of the calibration coefficient is a set polynomial function of the reference blood pressure value and the extracted characteristic parameters;
  • the calibration coefficient is used as the coefficient of the setting item of the known initial blood pressure calculation model to calibrate the blood pressure calculation model to obtain a calibrated blood pressure calculation model.
  • a sliding filter algorithm with variable window length is used to filter the single-channel pulse wave signal.
  • the photoelectric sensors with preset distribution are used to identify the wrist motion mode, and then the photoelectric compensation strategy matching the wrist motion mode is used to eliminate individual differences Afterwards, the single-channel pulse wave signal is subjected to photoelectric compensation, and finally the single-channel pulse wave signal after photoelectric compensation is filtered through the decomposition and reconstruction of the empirical mode and the sliding filter algorithm with variable window length.
  • the present invention provides a blood pressure monitoring device based on a single-channel pulse wave, which collects a single-channel pulse wave signal after eliminating individual differences.
  • the sliding filter algorithm filters the single-channel pulse wave signal.
  • the preset distribution of photoelectric sensors is used to identify the wrist movement mode.
  • the matching photoelectric compensation strategy is used for photoelectric compensation.
  • the blood pressure detection value is calculated, and the detection based on a single pulse wave is solved.
  • the problem of poor accuracy of blood pressure detection during blood pressure reduces the error of blood pressure detection and improves the accuracy of blood pressure detection results from the three aspects of signal acquisition process, signal processing process and calculation model.
  • Fig. 1 is a schematic structural diagram of a blood pressure monitoring device based on a single pulse wave according to an embodiment of the present invention
  • Fig. 2 is a schematic diagram of a blood pressure monitoring device based on a single pulse wave according to an embodiment of the present invention
  • Figure 3 is the diagonal distribution of pressure sensors
  • Figure 4 shows the distribution of pressure sensors to the four corners
  • Figure 5 shows the diagonal distribution of pressure sensors
  • Figure 6 shows the distribution of the two sides of the pressure sensor
  • Figure 7 is a comparison chart of pulse waves measured by two users with large differences in skin color under the same light intensity
  • Figure 8 is a comparison diagram of pulse waves measured by two users after adjusting the light intensity of the light source
  • Figure 9(a) is the effect of sliding filtering with a large window and a fixed window length
  • Figure 9(b) is the effect of sliding filtering with a small window and fixed window length
  • Fig. 9(c) is a filtering effect diagram of the sliding filtering algorithm with variable window length
  • FIG. 10 is a process of filtering a signal using a sliding filtering algorithm with a variable window length according to an embodiment of the present invention
  • Fig. 11 is the central symmetrical distribution mode of the photoelectric sensor
  • Figure 12 is the diagonal distribution of photoelectric sensors
  • Figure 13 is a symmetrical distribution of photoelectric sensors along the horizontal axis of symmetry of the watch
  • Fig. 14 is the four-corner distribution mode of the photoelectric sensor
  • Fig. 15(a) is the signal under the outward twisting motion state of the wrist
  • Fig. 15(b) is the signal under the state of flexing the wrist
  • Figure 15(c) is the signal in the state of the wrist twisting inwards
  • Figure 15(d) is the signal under the state of raising the wrist
  • Figure 16(a) is the wrist twisting outward
  • Figure 16 (b) is the wrist flexion movement
  • Figure 16(c) is the inward twisting of the wrist
  • Figure 16(d) is raising the wrist
  • Figure 17 is the original signal in the state of wrist twisting inwards
  • Figure 18 is the signal after photoelectric compensation
  • Figure 19 is the signal after decomposing and recombining
  • Fig. 20 is the signal of removing motion noise obtained by sliding filtering with variable window length
  • Fig. 21 is a partial pulse wave feature extracted during the calibration process of the embodiment of the present invention.
  • this embodiment provides a blood pressure monitoring device based on a single pulse wave, which includes a signal acquisition module, a signal filtering module, and a blood pressure calculation and calibration module.
  • the signal acquisition module is used to acquire the single pulse wave signal after eliminating the individual differences.
  • the first problem to be solved in order to obtain high-quality signals is the impact of individual differences on signal acquisition. Individual differences in the signal acquisition process have an important impact on pulse wave signal shape and signal noise interference, which seriously affects subsequent signal analysis and characteristics. Extraction, the blood pressure value obtained by using this pulse wave with individual differences lacks universality, and it only performs well in monitoring the blood pressure of a specific user in a specific environment, but cannot be applied to other users. Individual differences in the signal acquisition process mainly include differences in wearing styles and individual differences in skin characteristics.
  • the individualized differences include differences in wearing styles and individualized differences in skin characteristics.
  • the differences in wearing methods include differences in tightness, different wearing positions, and different wearing flatness caused by different wearing habits of users' watches.
  • the difference in wearing methods will also affect the distance between the watch sensor and the wrist skin and the stability of the gap, that is, the length and stability of the optical path of the optical signal, and eventually lead to pulse wave Signal amplitude and signal quality vary widely.
  • the uniformity of the wearing mode is judged by detecting the set point pressure of the contact surface between the single-channel pulse wave-based blood pressure monitoring device and the skin and the pressure difference of each set point.
  • this embodiment uses a group of pressure sensors with a special distribution in a specific area on the bottom of the watch to quantitatively evaluate the wearing tightness and smoothness of wearing, and through the pressure value and pressure
  • the difference can precisely control the wearing tightness and wearing position, providing an ideal and stable measurement environment for the watch work, and at the same time eliminating the difference in signal quality caused by the user's wearing tightness and wearing flatness.
  • the possible distribution methods of pressure sensors are: distribution at the four corners of the watch, distribution on the four sides of the watch, distribution at the diagonal of the watch, distribution on the opposite side of the watch, etc.
  • Figure 3- Figure 6 shows several Distribution of possible pressure sensors.
  • the tightness of the watch worn by the user will affect the distance between the watch sensor and the wrist skin and the interference of external ambient light, resulting in errors in the measurement process.
  • the tightness of existing watches is estimated and adjusted by users, which is affected by personal factors and has uncertainty.
  • the pressure sensor is used to quantify the tightness of the wrist watch, and the measured value of the pressure sensor is extracted as an important parameter in the blood pressure calculation model, thereby ensuring accurate calculation of blood pressure.
  • adjusting the tightness of the wrist watch can keep the movement consistency of the watch and the wrist skin, avoiding the error interference caused by the deviation of the measurement position.
  • the flatness of the watch By calculating the difference between the pressure sensors distributed in a specific area, the flatness of the watch can be evaluated. By adjusting the flatness of the watch, the contact area between the watch and the wrist can be evenly stressed, so that the contact between the watch and the wrist The gap remains stable to ensure a stable measurement signal.
  • the system when the user wears the watch, the system first obtains the measured value of each thin-film pressure sensor and performs a threshold judgment. If the measured value p i of the pressure sensor is within the pressure threshold range [p imin , p imax ] , it is considered that the pressure value of the pressure sensor meets the requirements, and the settings of pimin and pimax are obtained from the experimental data. If the measured value of a pressure sensor does not meet the threshold requirements, the system will analyze the distribution position of the pressure sensor and its pressure value, and provide a prompt on how to adjust the tightness of the wristband. It should be noted that because the wrist muscles are unevenly distributed and the wrist has certain fluctuations and radians, the corresponding p imin and p imax of the thin film pressure sensors distributed in different positions are not the same.
  • the pressure difference ⁇ p m,n between the pressure sensors is calculated.
  • ⁇ p m,n p m ⁇ p n , (1 ⁇ m, n ⁇ N, m ⁇ n) N is the number of pressure sensors. If all ⁇ p m, n meet the threshold requirements, it means that the current watch is worn smoothly. If there is a pressure difference that does not meet the threshold requirements, it means that the watch is distorted or shifted. The system can obtain the way the watch is shifted according to the pressure difference and the positions of the two pressure sensors corresponding to the pressure difference. Therefore, the corresponding wrist-worn adjustment scheme is given.
  • individual differences in skin characteristics include differences in skin chroma, skin thickness, skin hair density, and skin pore size and density.
  • the light source is a light source with constant light intensity
  • these skin differences will cause different degrees of loss in the incident and outgoing process of the light measurement signal, and finally cause the obtained signal to include individual differences in the skin. If the blood pressure is calculated based on this signal, there will be large errors in the calculation results due to individual differences in the skin. In other words, measuring with a constant light source cannot eliminate measurement errors due to individual differences in skin characteristics.
  • the skin characteristic coefficient is used to quantify the photoelectric loss degree of the current user's skin characteristic, and then the light intensity of the light source is adjusted according to the skin characteristic coefficient to compensate for the light loss caused by the current user's skin characteristic, thereby eliminating the impact of individual differences in the skin of different people.
  • the influence of the measurement signal makes the result of blood pressure measurement more accurate.
  • the light generator of the watch emits a standard light with a fixed light intensity of Is to enter the skin, and the light intensity of the outgoing light is obtained after the light is collected by the photoelectric sensor .
  • the formula for calculating the skin characteristic coefficient is
  • Segmented function type divide the skin characteristic coefficient into different threshold intervals, and use different constant light intensity of the light source for the skin characteristic coefficient in different threshold intervals.
  • Fig. 7 is a comparison chart of pulse waves measured by two users with large skin color differences under the same light intensity, in which the skin color of user 1 is darker, and the skin color of user 2 is lighter. It can be clearly seen from Fig. 7 that the pulse wave amplitude measured by user 2 is relatively large and the signal quality is relatively good. This is because user 2 has fairer skin and less hair, which causes less loss to the optical signal during the incident and outgoing process, and has a higher signal-to-noise ratio, so the signal quality is higher.
  • Figure 8 is a comparison chart of pulse waves measured by two users after adjusting the light intensity of the light source. From the comparison of Figure 7 and Figure 8, it can be seen that the amplitude and quality of the pulse wave of user two have increased significantly. This is because the light intensity of the light source in the measurement process of user 2 is increased, which compensates for the loss of light in the incident and outgoing processes, and improves the signal-to-noise ratio.
  • a sliding filtering algorithm with variable window length is used to filter the single-channel pulse wave signal.
  • the preset distribution of photoelectric sensors is used to identify the wrist movement mode, and then the photoelectric compensation strategy matching the wrist movement mode is used to photoelectrically control the single-channel pulse wave signal after eliminating individual differences. Finally, through the decomposition and reconstruction of the empirical mode and the sliding filter algorithm with variable window length, the single pulse wave signal after photoelectric compensation is filtered.
  • the pulse wave signal will also be interfered by noise such as signal drift, motion noise, and random noise during the acquisition process. .
  • the existing filtering algorithms that can run on wearable devices and do not require high computing power cannot dynamically adapt to different levels of noise pollution. Under signal filtering, signal details are often filtered out or noise filtering is not thorough, and it is impossible to filter out signal noise while retaining signal details to the greatest extent.
  • the sliding filtering algorithm has a good performance in eliminating random noise interference and filtering signal drift. Moreover, the sliding filtering algorithm has a simple principle and takes up less memory, which is very suitable for wearable devices.
  • the filtering effect of the sliding filtering algorithm is greatly affected by the sliding window length. The larger the filtering window length of the sliding filtering algorithm, the more thorough the noise filtering, and the less signal details can be retained, as shown in Figure 9(a); The smaller the filtering window length of the sliding filtering algorithm, the more signal details can be preserved, but it is prone to incomplete noise filtering, as shown in Figure 9(b).
  • Fig. 9(c) is a filtering effect diagram of the sliding filtering algorithm with variable window length.
  • the signal filtering module is used for filtering the single-channel pulse wave signal by adopting a sliding filtering algorithm with a variable window length.
  • the filtering window length and filtering times of the sliding filtering algorithm are determined by the quality of the single pulse wave signal.
  • the sliding filter algorithm with variable window length refers to the quantitative evaluation of the degree of signal pollution by noise using the noise figure. Decide. Then the filter window length is determined by the relationship model between the noise coefficient and the filter window length. After the filter window length is determined, various sliding filtering methods are performed on the signal, including sliding mean filtering, sliding median filtering, and sliding weighted filtering. Select the result with the smallest noise figure among the multiple filtering results as the result of primary filtering. If the noise figure of the primary filtering result does not meet the set threshold requirements, recalculate the filter window length and repeat the filtering process until the noise figure of the filtering result meets the requirements.
  • the threshold of the noise figure is derived from experimental data and is related to the type of signal and the required signal quality.
  • the sliding filtering algorithm is improved, and the sliding filtering algorithm with variable window length is used to filter the signal, that is, the filtering window length and filtering times of the sliding filtering algorithm are dynamically determined according to the degree of noise pollution of the signal.
  • the purpose is to adapt to signal filtering under different noise environments, and to retain the effective information of the signal to the greatest extent while filtering out the noise.
  • the filtering implementation process is as follows:
  • First-order difference is performed on the original signal, and the number k 1 of zero points in the first-order difference signal is calculated.
  • the original signal is subjected to a second-order difference, and the number k 2 of zero points in the second-order difference signal is calculated.
  • the filter window length is obtained according to the model.
  • the calculation process of the empirical window length L 0 the pulse rate per minute of the human body is usually between [s1, s2].
  • the cycle length of the pulse wave sampling rate*60/pulse rate per minute, according to the model
  • the pulse wave cycle length interval can be calculated as [s3, s4].
  • N is the sliding filter window length value
  • the experimental data show that the sliding filter algorithm with variable window length has a better filtering effect on the signal acquired in a calm state or a slight motion state, but the filtering effect on the signal acquired in a violent motion state is poor.
  • a group of photoelectric sensors with a specific distribution are used in the watch.
  • the signals collected by the photoelectric sensors at different distribution positions have different performances.
  • the signal performance of multiple photoelectric signals with a specific distribution can be analyzed to obtain the motion mode and motion intensity of the wrist. For details, see the example section of motion noise removal. Distinguishable wrist movements include wrist lift, wrist flexion, wrist twist inward, and wrist twist outward.
  • Example of motion noise removal Taking the photoelectric distribution method in Figure 11 as an example, it shows the signal performance of two photoelectric sensors in the four wrist motion states of raising the wrist, bending the wrist, twisting the wrist inward, and twisting the wrist outward, as shown in Figure 15(a)- Figure 15(d), and the corresponding four wrist movement modes are shown in Figure 16(a)- Figure 16(d).
  • the signals collected by multiple photoelectric sensors are used to perform preliminary photoelectric compensation for the movement and eliminate part of the movement noise, as shown in Figure 18. Then the empirical mode decomposition and reconstruction of the photoelectric compensation signal are carried out. After decomposing and recombining the empirical modes, the motion noise can be further removed. Finally, the above-mentioned sliding filtering algorithm is used to further filter the signal to obtain a signal with most of the motion noise removed.
  • the movement mode of the wrist can be obtained.
  • signal features need to be extracted, including: time interval of signal abnormal area, time interval of rising edge, rising edge slope, peak amplitude, valley amplitude, time difference between peak points of two abnormal areas of photoelectric signals, etc., if If there are multiple signal abnormal areas, multiple extractions are required.
  • the signal abnormal area refers to the signal amplitude greatly increases or decreases within a very short period of time.
  • ⁇ is an index for evaluating the intensity of exercise
  • a 1 , ... a n are the values of extracted features.
  • ⁇ is less than the system preset threshold, it is considered that the degree of motion is not severe, and only the sliding filter algorithm with variable window length is used for signal processing. If ⁇ is smaller than the system preset threshold, the movement is considered severe.
  • the motion is compensated with the results of multiple photoelectric sensors. Taking the signal collected when the wrist is twisted inward as an example, the original signal is shown in Figure 17, and the signals collected by the two photoelectric sensors are added together as photoelectric compensation, and the effect after photoelectric compensation is shown in Figure 18.
  • the empirical mode decomposition and reconstruction of the photoelectrically compensated signal is carried out, that is, the signal is decomposed into several intrinsic mode components (IMF), and after the decomposition is completed, IMF2, IMF3, and IMF4 are added to obtain the processed signal, as shown in Figure 19 shown. Finally, the sliding filter with variable window length is performed on the signal to obtain the signal with motion noise removed, as shown in Figure 20.
  • IMF intrinsic mode components
  • a blood pressure calculation and calibration module which is used to extract the characteristic parameters of the single pulse wave signal after sliding filtering, and obtain the blood pressure detection value based on the extracted characteristic parameters and the calibrated blood pressure calculation model;
  • a pulse wave signal with better waveform quality can be obtained. Then the feature extraction of the waveform can be carried out, and the blood pressure value can be calculated by substituting the extracted features into the blood pressure calculation model built in advance.
  • the characteristics required by the blood pressure calculation model peak value of dicrotic wave P d , trough value of dicrotic wave P dv , peak value of main wave P p , maximum slope S 1 of ascending limb, area of ascending section A u , area of descending section A d , ascending section and descending section Area ratio A u /A d , time interval between dicrotic wave and main wave T dp , time interval between main wave and main wave T pp , systolic time ST, diastolic time DT, ascending branch maximum slope point and dicrotic wave
  • the time interval T md of the peak the time interval between the point of the maximum slope of the ascending branch and the trough of the dicrotic wave , etc.
  • Vessel characteristic quantity K ascending and descending branch time ratio T vdv /T dvv , systolic time ratio ST/T pp , dicrotic wave and main wave interval time ratio T dp /T pp .
  • the characteristic parameters of the single pulse wave signal after sliding filtering are extracted: the characteristic parameters whose Pearson correlation coefficient with blood pressure exceeds a set threshold (for example: 0.6).
  • X represents the reference blood pressure BP
  • Y represents a characteristic parameter that requires correlation analysis
  • Cov(X,Y) represents the covariance
  • Var[X] represents the variance of X
  • Var[Y] represents the variance of Y.
  • the blood pressure calculation model can reflect the change trend of blood pressure according to the change of pulse wave characteristics, there are large differences in the blood pressure reference value (blood pressure in a calm state) between different users, and when the pulse wave characteristics change in the same range, There are also certain differences in the magnitude of changes in blood pressure, and these differences are the third impact of blood pressure calculation errors.
  • the device needs to perform blood pressure calibration before the official measurement.
  • the purpose is to correct the coefficients of the blood pressure calculation model, so that the blood pressure calculation model is more suitable for the pulse wave characteristics of the current user, and provide higher-precision blood pressure calculation. result.
  • the calibration process of the blood pressure calculation model is:
  • a set of calibration coefficients is calculated based on the reference blood pressure value and the extracted characteristic parameters; wherein, the calculation model of the calibration coefficient is a set polynomial function of the reference blood pressure value and the extracted characteristic parameters;
  • the calibration coefficient is used as the coefficient of the setting item of the known initial blood pressure calculation model to calibrate the blood pressure calculation model to obtain a calibrated blood pressure calculation model.
  • the pulse wave features extracted during the calibration process include: dicrotic wave peak value P d , dicrotic wave trough value P dv , main wave peak value P p , maximum slope S 1 of the ascending limb, ascending section area A u , descending section area A d , rising Area ratio of descending segment A u /A d , time interval between dicrotic wave and main wave T dp , time interval between main wave and main wave T pp , systolic time ST, diastolic time DT, ascending branch maximum slope point and The time interval T md of the dicrotic wave peak, the time interval between the maximum slope point of the ascending branch and the dicrotic wave trough and other characteristics T mdv , the time T vdv from the starting point trough to the dicrotic wave trough, and the time T from the dicrotic wave trough to the terminal trough dvv , cardiovascular feature K, time ratio of ascending and descending branches
  • a1 is the peak value of the dicrotic wave
  • t1 is the systolic time
  • t2 is the time interval between the main wave and the dicrotic wave
  • t3 is the cycle time
  • t4 is the diastolic time
  • t5 is the time between the maximum slope point and the dicrotic wave interval.
  • the initial blood pressure calculation model is a preset polynomial function of the characteristic parameters.
  • b 1 f(SBP ref ,T pp ,a 1 )
  • b 2 f(SBP ref ,DT,a 2 )
  • b 3 f(SBP ref ,ST,DBP ref )
  • T dp is the time interval between the dicrotic wave and the main wave
  • SBP ref is the systolic blood pressure reference value
  • T pp is the cycle time
  • P d is the dicrotic wave DT is the diastolic time
  • ST is the systolic time.
  • f(T dp ) is a polynomial about the interval between dicrotic wave and main wave; f(SBP ref , T pp , a 1 ) is a polynomial about systolic blood pressure reference value, cycle time, and calibration parameter a 1 ; f(P d ) is a polynomial about the peak value of dicrotic wave; f(SBP ref , DT, a 2 ) is a polynomial about systolic blood pressure reference value, diastolic time, calibration parameter a 2 ; f(SBP ref , ST, DBP ref ) is about systolic blood pressure Reference value, diastolic reference value, polynomial of systolic time.
  • the reference SBP refers to the systolic blood pressure value measured by the Omron sphygmomanometer
  • the calculated SBP refers to the systolic blood pressure value obtained by the calculation model. According to the calculation, the error between the reference SBP and the calculated SBP before calibration is -3.125 ⁇ 10.334mmHg, and the error between the reference SBP and the calculated SBP after calibration is -0.875 ⁇ 3.588mmHg.
  • the reference DBP refers to the diastolic blood pressure value measured by the Omron sphygmomanometer
  • the calculated DBP refers to the diastolic blood pressure value obtained by the calculation model.
  • the calculation can get the error between the reference DBP before calibration and the calculated DBP is -1.75 ⁇ 7.450mmHg,
  • the error between the reference DBP and the calculated DBP after calibration is 1.75 ⁇ 3.122mmHg.
  • This embodiment provides a computer-readable storage medium, on which a computer program is stored, and when the program is executed by a processor, the following steps are implemented:
  • the calibration process of the blood pressure calculation model is:
  • a set of calibration coefficients is calculated based on the reference blood pressure value and the extracted characteristic parameters; wherein, the calculation model of the calibration coefficient is a set polynomial function of the reference blood pressure value and the extracted characteristic parameters;
  • the calibration coefficient is used as the coefficient of the setting item of the known initial blood pressure calculation model to calibrate the blood pressure calculation model to obtain a calibrated blood pressure calculation model.
  • a sliding filter algorithm with variable window length is used to filter the single-channel pulse wave signal.
  • the preset distribution of photoelectric sensors is used to identify the wrist movement mode, and then the photoelectric compensation strategy matching the wrist movement mode is used to photoelectrically control the single-channel pulse wave signal after eliminating individual differences. Finally, through the decomposition and reconstruction of the empirical mode and the sliding filter algorithm with variable window length, the single pulse wave signal after photoelectric compensation is filtered.
  • This embodiment provides a computer device, including a memory, a processor, and a computer program stored on the memory and operable on the processor.
  • the processor executes the program, the following steps are implemented:
  • the calibration process of the blood pressure calculation model is:
  • a set of calibration coefficients is calculated based on the reference blood pressure value and the extracted characteristic parameters; wherein, the calculation model of the calibration coefficient is a set polynomial function of the reference blood pressure value and the extracted characteristic parameters;
  • the calibration coefficient is used as the coefficient of the setting item of the known initial blood pressure calculation model to calibrate the blood pressure calculation model to obtain a calibrated blood pressure calculation model.
  • a sliding filter algorithm with variable window length is used to filter the single-channel pulse wave signal.
  • the preset distribution of photoelectric sensors is used to identify the wrist movement mode, and then the photoelectric compensation strategy matching the wrist movement mode is used to photoelectrically control the single-channel pulse wave signal after eliminating individual differences. Finally, through the decomposition and reconstruction of the empirical mode and the sliding filter algorithm with variable window length, the single pulse wave signal after photoelectric compensation is filtered.

Landscapes

  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Engineering & Computer Science (AREA)
  • Signal Processing (AREA)
  • Molecular Biology (AREA)
  • Surgery (AREA)
  • Biophysics (AREA)
  • Pathology (AREA)
  • Physiology (AREA)
  • Biomedical Technology (AREA)
  • Heart & Thoracic Surgery (AREA)
  • Medical Informatics (AREA)
  • Veterinary Medicine (AREA)
  • Physics & Mathematics (AREA)
  • Animal Behavior & Ethology (AREA)
  • General Health & Medical Sciences (AREA)
  • Public Health (AREA)
  • Cardiology (AREA)
  • Artificial Intelligence (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Psychiatry (AREA)
  • Vascular Medicine (AREA)
  • Power Engineering (AREA)
  • Measuring Pulse, Heart Rate, Blood Pressure Or Blood Flow (AREA)

Abstract

一种基于单路脉搏波的血压监测装置、存储介质及电子设备。其中该装置包括：信号采集模块，其用于采集消除个体化差异之后的单路脉搏波信号；信号滤波模块，其用于在静息状态或运动幅度小于设定阈值的情境下，采用变窗长的滑动滤波算法对单路脉搏波信号滤波；在运动幅度大于设定阈值的情境下，采用预设分布的光电传感器识别手腕运动方式，再采用相匹配的光电补偿方法进行光电补偿，再经过经验模态的分解与重构及变窗长的滑动滤波消除运动噪声；血压计算及校准模块，其用于提取滑动滤波后的单路脉搏波信号的特征参数，并基于提取的特征参数及校准后的血压计算模型，得到血压检测值。

Description

基于单路脉搏波的血压监测装置、存储介质及电子设备
本发明要求于2021年12月7日提交中国专利局、申请号为202111485941.5、发明名称为“基于单路脉搏波的血压监测装置、存储介质及电子设备”的中国专利申请的优先权，其全部内容通过引用结合在本发明中。
技术领域
本发明属于血压检测装置技术领域，尤其涉及一种基于单路脉搏波的血压监测装置、存储介质及电子设备。
背景技术
本部分的陈述仅仅是提供了与本发明相关的背景技术信息，不必然构成在先技术。
人体血压时刻处于动态变化中，血压的连续监测是高效查看高血压患者健康状态的措施之一，连续血压监测可以及时捕捉血压的异常波动变化，提前预警用户脑梗、心梗等突发性心血管疾病的可能威胁。
根据血压计算方式的不同，检测设备需要采集的信号种类也不同，目前运用较多的有基于示波法计算血压，基于心电信号和脉搏波信号计算血压，基于双路脉搏波信号计算血压，基于单路脉搏波测量血压。其中基于示波法计算血压的方式需要周期性的充气和放气，用户体验较差，而且测量间隔较长，无法实现真正的血压连续检测。基于心电信号和脉搏波信号配合获取血压以及基于双路脉搏波获取血压的方法对于两种信号的配合要求严格，操作过程复杂、实现困难。而且双路更信号容易受噪声干扰，造成计算结果精度差。
基于单路脉搏波监测血压操作过程简单，测量设备体积更小而且便于携带，是一种用户体验较好的血压连续检测方案。但是现有的基于单路脉搏波波形形态分析来计算得到的血压误差较大，无法满足临床需求。误差主要来自于三个方面：一是脉搏波信号在采集过程中容易受个体化差异的影响导致信号质量参差不齐，而且包含了个体化差异的信号会掩盖了脉搏波波形特征与血压的内在联系。二是脉搏波信号的连续测量环境多变，受噪声干扰的程度不一，现有的滤波方法很难动态适应不同噪声干扰程度下的信号滤波，经常出现信号细节被滤除或者噪声滤除不彻底的情况，为后续脉搏波的形态分析和特征提取带来极大困难。三是血压计算模型虽然能够反应用户的血压变化趋势，但是不同用户的血压基准并不相同，所以想要获得更高精度的血压，需要进行血压校准。
综上所述，发明人发现，基于单路脉搏波进行检测血压时仍然存在信号采集过程、信号处理过程和计算模型这三个方面的误差，从而导致血压检测的精度差。
发明内容
为了解决上述背景技术中存在的技术问题，本发明提供一种基于单路脉搏波的血压监测装置、存储介质及电子设备，其能够从信号采集过程、信号处理过程和计算模型这三个方面来降低血压检测误差，从而提高血压检测结果的准确性。
为了实现上述目的，本发明采用如下技术方案：
本发明的第一个方面提供了一种基于单路脉搏波的血压监测装置，其包括：
信号采集模块，其用于采集消除个体化差异之后的单路脉搏波信号；
信号滤波模块，其用于对单路脉搏波信号进行滤波；
血压计算及校准模块，其用于提取滤波后的单路脉搏波信号的特征参数，并基于提取的特征参数及校准后的血压计算模型，得到血压检测值；
其中在所述血压计算及校准模块，血压计算模型的校准过程为：
基于参考血压值及提取的特征参数，计算得到一组校准系数；其中，校准系数的计算模型为参考血压值及提取的特征参数的设定多项式函数；
将校准系数作为已知初始血压计算模型的设定项的系数和指数，来校准血压计算模型，得到校准后的血压计算模型。
在一个或多个实施例中，在所述信号滤波模块中，在静息状态或者运动幅度小于设定阈值的情境下，采用变窗长的滑动滤波算法对单路脉搏波信号滤波。
在一个或多个实施例中，在所述信号滤波模块中，在运动幅度大于设定阈值的情境下，采用预设分布的光电传感器来识别手腕运动方式，再采用与手腕运动方式相匹配的光电补偿策略对消除个体化差异之后的单路脉搏波信号进行光电补偿，最后依次通过经验模态的分解与重构及变窗长的滑动滤波算法对光电补偿后的单路脉搏波信号进行滤波。
本发明的第二个方面提供了一种计算机可读存储介质，其上存储有计算机程序，该程序被处理器执行时实现如下步骤：
采集消除个体化差异之后的单路脉搏波信号；
对单路脉搏波信号进行滤波；
提取滤波后的单路脉搏波信号的特征参数，并基于提取的特征参数及校准后的血压计算模型，得到血压检测值；
其中血压计算模型的校准过程为：
基于参考血压值及提取的特征参数，计算得到一组校准系数；其中，校准系数的计算模型为参考血压值及提取的特征参数的设定多项式函数；
将校准系数作为已知初始血压计算模型的设定项的系数和指数，来校准血压计算模型，得到 校准后的血压计算模型。
在一个或多个实施例中，在静息状态或者运动幅度小于设定阈值的情境下，采用变窗长的滑动滤波算法对单路脉搏波信号滤波。
在一个或多个实施例中，在运动幅度大于设定阈值的情境下，采用预设分布的光电传感器来识别手腕运动方式，再采用与手腕运动方式相匹配的光电补偿策略对消除个体化差异之后的单路脉搏波信号进行光电补偿，最后依次通过经验模态的分解与重构及变窗长的滑动滤波算法对光电补偿后的单路脉搏波信号进行滤波。
本发明的第三个方面提供了一种计算机设备，包括存储器、处理器及存储在存储器上并可在处理器上运行的计算机程序，所述处理器执行所述程序时实现如下步骤：
采集消除个体化差异之后的单路脉搏波信号；
对单路脉搏波信号进行滤波；
提取滤波后的单路脉搏波信号的特征参数，并基于提取的特征参数及校准后的血压计算模型，得到血压检测值；
其中血压计算模型的校准过程为：
基于参考血压值及提取的特征参数，计算得到一组校准系数；其中，校准系数的计算模型为参考血压值及提取的特征参数的设定多项式函数；
将校准系数作为已知初始血压计算模型的设定项的系数，来校准血压计算模型，得到校准后的血压计算模型。
在一个或多个实施例中，在静息状态或者运动幅度小于设定阈值的情境下，采用变窗长的滑动滤波算法对单路脉搏波信号滤波。
在一个或多个实施例中，在运动幅度大于设定阈值的情境下，采用预设分布的光电传感器来识别手腕运动方式，再采用与手腕运动方式相匹配的光电补偿策略对消除个体化差异之后的单路脉搏波信号进行光电补偿，最后依次通过经验模态的分解与重构及变窗长的滑动滤波算法对光电补偿后的单路脉搏波信号进行滤波。
与现有技术相比，本发明的有益效果是：
本发明提供了一种基于单路脉搏波的血压监测装置，其采集消除个体化差异之后的单路脉搏波信号，在静息状态或者运动幅度小于设定阈值的情境下，采用变窗长的滑动滤波算法对单路脉搏波信号滤波，在运动幅度大于设定阈值的情境下，采用预设分布的光电传感器识别手腕运动方式，针对不同的运动方式，采用相匹配的光电补偿策略进行光电补偿，然后通过经验模态的分解与重构、变窗长的滑动滤波算法消除运动噪声，基于提取的特征参数及校准后的血压计算模型，计算血压检测值，解决了基于单路脉搏波进行检测血压时血压检测的精度差的问题，从信号采集 过程、信号处理过程和计算模型这三个方面降低了血压检测误差，提高了血压检测结果的准确性。
本发明附加方面的优点将在下面的描述中部分给出，部分将从下面的描述中变得明显，或通过本发明的实践了解到。
附图说明
构成本发明的一部分的说明书附图用来提供对本发明的进一步理解，本发明的示意性实施例及其说明用于解释本发明，并不构成对本发明的不当限定。
图1是本发明实施例的基于单路脉搏波的血压监测装置结构示意图；
图2是本发明实施例的基于单路脉搏波的血压监测装置原理图；
图3是压力传感器对角分布方式；
图4是压力传感器对四角分布方式；
图5是压力传感器斜对角分布方式；
图6是压力传感器两侧边分布方式；
图7是两个肤色差异较大的用户在相同光源光强下测到的脉搏波的对比图；
图8是调节光源光强后两个用户测到的脉搏波的对比图；
图9(a)是大窗定窗长滑动滤波效果；
图9(b)是小窗定窗长滑动滤波效果；
图9(c)是采用的变变窗长的滑动滤波算法的滤波效果图；
图10是本发明实施例的使用变窗长的滑动滤波算法对信号进行滤波的流程；
图11是光电传感器中心对称分布方式；
图12是光电传感器对角分布方式；
图13是光电传感器沿沿腕表横向对称轴对称分布方式；
图14是光电传感器四角分布方式；
图15(a)是手腕向外扭动运动状态下的信号；
图15(b)是手屈腕运动状态下的信号；
图15(c)是手腕向内扭动运动状态下的信号；
图15(d)是抬腕运动状态下的信号；
图16(a)是手腕向外扭动；
图16(b)是手屈腕运动；
图16(c)是手腕向内扭动；
图16(d)是抬腕；
图17是手腕向内扭动运动状态下原始信号；
图18是光电补偿后的信号；
图19是分解重组后的信号；
图20是进行变窗长的滑动滤波得到去除运动噪声的信号；
图21是本发明实施例的校准过程中提取的部分脉搏波特征。
具体实施方式
下面结合附图与实施例对本发明作进一步说明。
应该指出，以下详细说明都是例示性的，旨在对本发明提供进一步的说明。除非另有指明，本文使用的所有技术和科学术语具有与本发明所属技术领域的普通技术人员通常理解的相同含义。
需要注意的是，这里所使用的术语仅是为了描述具体实施方式，而非意图限制根据本发明的示例性实施方式。如在这里所使用的，除非上下文另外明确指出，否则单数形式也意图包括复数形式，此外，还应当理解的是，当在本说明书中使用术语“包含”和/或“包括”时，其指明存在特征、步骤、操作、器件、组件和/或它们的组合。
实施例一
参照图1和图2，本实施例提供了一种基于单路脉搏波的血压监测装置，其包括信号采集模块、信号滤波模块和血压计算及校准模块。
(1)信号采集模块
信号采集模块，其用于采集消除个体化差异之后的单路脉搏波信号。
想要获取高精度血压，首先要采集到的高质量的脉搏波信号。想要获取高质量信号首先要解决的问题是个体化差异对信号采集的影响，信号采集过程中个体化差异对于脉搏波信号形态和信号噪声干扰都有重要影响，严重影响后续的信号分析和特征提取，使用这种具有个体化差异的脉搏波得到的血压值缺乏普适性，只在监测特定环境下的特定用户血压时表现良好，而无法适用其他用户。信号采集过程中的个体化差异主要包括佩戴方式差异化和皮肤特性个体化差异。
具体地，所述个体化差异包括佩戴方式差异化和皮肤特性个体化差异。
其中，佩戴方式差异化包括用户腕表佩戴习惯不同所导致的松紧度不同、佩戴位置不同、佩戴平整度不同等差异。这种佩戴方式差异化除了影响外界环境光干扰外，还会影响腕表传感器与手腕皮肤的间隙距离大小与间隙的平稳性，即影响光学信号的光程长短和光程稳定性，最终导致脉搏波信号的幅度和信号质量有较大差异。
在所述信号采集模块中，通过检测所述基于单路脉搏波的血压监测装置与皮肤接触面的设定点压力以及各个设定点压力差值，判断佩戴方式的统一性。
在具体实施例中，为了消除佩戴方式差异化，本实施例使用一组在腕表底面的特定区域呈现 特殊分布的压力传感器来定量的评估佩戴松紧度和佩戴平整性，并通过压力值和压力差值对佩戴松紧度、佩戴位置进行精确调控，为腕表工作提供理想且稳定的测量环境，同时消除因用户佩戴松紧度和佩戴平整度不同带来的信号质量差异。
压力传感器可能的分布方式有：在腕表的四角分布，在腕表的四边分布，在腕表的对角分布，在腕表的对边分布等等，图3-图6中展示了几种可能的压力传感器的分布方式。
具体来说，在实际测量中，用户佩戴腕表的松紧度会影响腕表传感器与手腕皮肤的距离以及外界环境光干扰，导致测量过程产生误差。现有的腕表松紧度都是由用户估量调整，受个人因素影响，具有不确定性。本实施例使用压力传感器量化腕表佩戴的松紧度，并提取压力传感器的测量值作为血压计算模型中的一个重要参数，从而保证了血压的精确计算。同时调控腕表佩戴的松紧度可以使腕表与手腕皮肤保持运动的一致性，避免了测量位置偏移带来的误差干扰。
通过计算分布于特定区域的压力传感器的差值可以评估腕表佩戴的平整性，通过调控腕表佩戴的平整性，可以使腕表与手腕接触面积受力均匀，从而使腕表与手腕的接触间隙保持稳定，保证测量信号稳定。
在具体实施过程中，当用户佩戴腕表后，系统首先获取每个薄膜压力传感器的测量值并进行阈值判断，如果压力传感器的测量值p i在压力阈值范围[p imin,p imax]内时，认为压力传感器的压力值符合要求，p imin和p imax的设置都由实验数据得到。如果有压力传感器的测量值不满足阈值要求，则系统根据该压力传感器的分布位置与其压力值进行分析，并对如何调整腕带松紧度进行提示。需要注意的是，因为手腕肌肉不均匀分布，手腕有一定起伏和弧度，所以分布于不同位置的薄膜压力传感器，其对应的p imin和p imax均不相同。
如果所有薄膜压力传感器的值均符合阈值要求，则计算压力传感器之间的压力差Δp m，n。Δp m，n＝p m-p n，(1≤m,n≤N，m≠n)N为压力传感器的数量。如果所有的Δp m，n都满足阈值要求，则说明当前腕表佩戴平整。如果有压力差值不满足阈值要求，则说明腕表的佩戴出现了扭曲或者偏移，系统根据该压力差值以及压力差值对应的两个压力传感器的位置可以得到腕表偏移的方式，从而给出相应的腕佩戴调整方案。
此外还需要说明的是，皮肤特性个体化差异包括皮肤色度不同、皮肤厚度不同、皮肤毛发密度不同、皮肤毛孔大小和密度不同等用户皮肤差异。当光源为恒光强光源时，这些皮肤差异对光测量信号的入射和出射过程造成不同程度的损耗，最终导致得到的信号包含了皮肤的个体化差异。如果根据这种信号来计算血压，则计算结果因皮肤的个体化差异存在较大误差。也就是说，采用恒定光源进行测量无法消除因皮肤特性个体化差异带来的测量误差。
本实施例利用皮肤特征系数来量化当前用户皮肤特性对光电的损耗程度，然后根据皮肤特征 系数来调节光源光强以弥补当前用户皮肤特性对光的损耗，从而消除不同人皮肤的个体化差异对测量信号影响，使得血压测量的结果更加精确。
根据压力传感器调整好用户腕表佩戴后，腕表的光发生器发出固定光强为I s的标准光入射皮肤，用光电传感器收集出射光后，得到出射光的光强I 1。计算当前出射光光强I 1与标准出射光强I 0的比值，皮肤特征系数的计算公式为
Figure PCTCN2022136638-appb-000001
其中，在所述信号采集模块中，量化皮肤特性，根据皮肤特征系数与光源光强的模型，计算光源光强调整值；调整后的光源光强为I＝I s*(a+b*ε)，其中a和b为匹配系数，ε为受试者的皮肤特征系数，
Figure PCTCN2022136638-appb-000002
I 0为标准出射光强，I 1为当前出射光强；I s为选定的标准光源光强。
皮肤特征系数与光源光强的模型有多种形式：
(a)连续函数型：光源光强调整值I＝I s*(a+b*ε)其中a，b为匹配系数，ε为受试者的皮肤特征系数,I s为选定的标准光源光强。
(b)分段函数型：给皮肤特征系数划分不同的阈值区间，对不同阈值区间内的皮肤特征系数采用不同的恒定光源光强。
例如：划分阈值k1和k2，其中k1<k2。当皮肤特征系数小于k1时，令光源光强为I a，当皮肤特征系数大于k1且小于k2时，令光源光强为I b，当皮肤特征系数大于k2时，令光源光强为I c。
调整光强的具体措施也有多种，包括：
(a)增强单个光发生器的功率。方法有多种，例如增强工作电流、增强工作电压等。
(b)增加光发生器的工作数量等。
图7为两个肤色差异较大的用户在相同光源光强下测到的脉搏波的对比图，其中用户一的皮肤色度较深，用户二的皮肤色度较浅。图7中可以清晰看出用户2测到的脉搏波幅度较大且信号质量较好。这是因为用户2的皮肤较白且毛发较少，对光信号在入射和出射过程中造成的损耗较少，信号信噪比更大，所以信号质量较高。
图8为调节光源光强后两个用户测到的脉搏波的对比图，通过图7和图8的对比可以看出用户二的脉搏波的幅度和质量都大幅增加。这是因为增大了用户2测量过程的光源光强，弥补了光在入射和出射过程中的损耗，提高了信噪比。
(2)信号滤波模块
在所述信号滤波模块中，在静息状态或者运动幅度小于设定阈值的情境下，采用变窗长的滑 动滤波算法对单路脉搏波信号滤波。
在运动幅度大于设定阈值的情境下，采用预设分布的光电传感器来识别手腕运动方式，再采用与手腕运动方式相匹配的光电补偿策略对消除个体化差异之后的单路脉搏波信号进行光电补偿，最后依次通过经验模态的分解与重构及变窗长的滑动滤波算法对光电补偿后的单路脉搏波信号进行滤波。
虽然在信号采集之前通过压力传感器调控和调节光源光强的方法消除了个体化误差对信号采集的影响，但是脉搏波信号在采集过程中还会受到信号漂移、运动噪声、随机噪声等噪声的干扰。而且因为采集信号时的环境多变，每次采集到的信号受噪声干扰的程度不同，现有的能够在可穿戴设备上运行的对算力要求不高的滤波算法无法动态适应不同噪声污染程度下的信号滤波，经常出现信号细节被滤除或者噪声滤除不彻底的情况，无法在滤除信号噪声的同时无法最大程度的保留信号细节。
滑动滤波算法在消除随机噪声干扰，滤除信号漂移现象等有较好的表现，而且滑动滤波算法原理简单，占用内存较小，非常适用于可穿戴设备。但是滑动滤波算法的滤波效果受滑动窗长的影响较大，滑动滤波算法的滤波窗长越大，噪声滤除越彻底，同时能够保留的信号细节越少，如图9(a)所示；滑动滤波算法的滤波窗长越小，能够保留的信号细节越多，但是容易出现噪声滤除不彻底的情况，如图9(b)所示。图9(c)为采用的变变窗长的滑动滤波算法的滤波效果图。
信号滤波模块，其用于采用变窗长的滑动滤波算法对单路脉搏波信号滤波。
其中，在所述信号滤波模块中，所述滑动滤波算法的滤波窗长和滤波次数由所述单路脉搏波信号的质量来决定的。
如图10所示，此处所说的变变窗长的滑动滤波算法是指用噪声系数定量评估信号受噪声污染的程度，噪声系数由原始信号的一阶差分信号和二阶差分信号的零点数决定。然后由噪声系数于滤波窗长的关系模型确定滤波窗长。确定滤波窗长后对信号进行多种滑动滤波方式，包括滑动均值滤波、滑动中值滤波、滑动加权滤波。选择多个滤波结果中噪声系数最小的结果作为一次滤波的结果。若一次滤波结果的噪声系数不满足设定的阈值要求，则重新计算滤波窗长，重复滤波过程，直至滤波结果的噪声系数满足要求。噪声系数的阈值来源于实验数据，与信号类型和需要的信号质量有关。
本实施例改进了滑动滤波算法，使用变窗长的滑动滤波算法来对信号进行滤波，即根据信号受噪声污染程度动态决定滑动滤波算法的滤波窗长和滤波次数。目的是适应不同噪声环境下的信号滤波，在滤除噪声的同时最大程度的保留信号的有效信息。
滤波实施过程如下：
将原始信号进行一阶差分，计算一阶差分信号中零点的数量k 1。将原始信号进行二阶差分，计算二阶差分信号中零点的数量k 2。根据一阶差分信号的零点数量k 1和二阶差分信号的零点数量k 2计算噪声系数，根据模型得到噪声系数：β＝f(k 1，k 2)，其中a,b为匹配系数，β为噪声系数。
零点判断标准：
(1)如果信号的相邻两个采样点数值由正变负或者由负变正，则在两点中间存在一个零点。
(2)如果信号的某一采样点数值为0，则这个点为零点。
得到噪声系数后，根据模型得到滤波窗长，滤波窗长的计算公式为L＝L 0(c+d*β)，其中c,d为匹配系数，β为噪声系数，L 0为经验窗长。
经验窗长L 0的计算过程：人体每分钟脉搏数通常在[s1,s2]之间，在采样率确定的情况下，脉搏波的周期长度＝采样率*60/每分钟脉搏数，根据模型可以计算得到脉搏波周期长度区间为[s3,s4]。经验窗长模型为：窗长L 0＝e*s3+(1-e)*s4,其中e为匹配系数。
滑动均值滤波算法：
Figure PCTCN2022136638-appb-000003
0≤k≤N，N为滑动滤波窗长值；
滑动中值滤波算法：
Figure PCTCN2022136638-appb-000004
0≤k≤N，N滑动滤波为窗长值；
得到滤波窗长后，对信号进行多种滑动滤波，包括滑动均值滤波、滑动中值滤波、滑动加权滤波等，选择噪声系数最小的结果作为滤波结果。如果该滤波结果的噪声系数符合阈值要求，则滤波过程结束，噪声系数的阈值来源于实验数据，与信号类型和需要的信号质量有关。如果噪声系数不满阈值要求，则重复上述滤波过程，直至滤波结果的噪声系数满足要求，整体流程如图10所示。滤波效果如图9(c)所示。
实验数据表明：变窗长的滑动滤波算法对于平静状态或者轻微运动状态下获取的信号有较好的滤波效果，但是对于剧烈运动下采集的信号滤波效果较差。
在本实施例中，针对剧烈运动下的信号滤波，使用的腕表中有特定分布的一组光电传感器，在不同的手腕运动状态下，不同分布位置的光电传感器采集到的信号表现不同，根据特定分布的多光电的信号表现可以分析得到手腕的运动方式和运动强度，具体见运动噪声去除例子部分。可以分辨的手腕运动方式包括抬腕、屈腕、手腕向内扭动、手腕向外扭动。
光电传感器可能的几种分布方式有：沿腕表横向对称轴对称分布，沿腕表对称中心对角对称分布，图11-图14展示了几种光电传感器可能的分布方式。
运动噪声去除例子：以图11的光电分布方式为例，展示了两个光电传感器在抬腕、屈腕、 手腕向内扭动、手腕向外扭动四种手腕运动状态下的信号表现，如图15(a)-图15(d)，进而对应的四种手腕运动方式如图16(a)-图16(d)所示。
分析针对不同的手腕运动方式，利用多个光电传感器采集的信号对运动进行初步光电补偿，消除部分运动噪声，如图18所示。然后对光电补偿信号进行经验模态分解和重构。经过经验模态的分解与重组后，运动噪声可以进一步去除。最后再利用上述的滑动滤波算法对信号进行进一步滤波处理，得到去除了大部分运动噪声的信号。
通过提取光电传感器采集的信号特征，可以得到手腕的运动方式。
以屈腕动作为例，需要提取信号特征包括：信号异常区域的时间间隔、上升沿时间间隔、上升沿斜率、峰值幅度、谷值幅度、两个光电信号的异常区域的峰值点时间差等，如果有多个信号异常区域，则需要多次提取。信号异常区域指信号幅值在极短时间内大幅增加或减少。
将提取的特征值代入模型，可以得到运动剧烈程度的评估。模型为θ＝f(a 1，a 2，...，a n)
其中θ为评价运动剧烈程度的指标，a 1，…a n为提取特征的值。
如果θ小于系统预设阈值，则认为运动程度不剧烈，则只应用变窗长的滑动滤波算法进行信号处理。如果θ小于系统预设阈值，则认为运动剧烈。针对每种手腕运动方式，用多个光电传感器的结果对运动进行补偿。以手腕向内扭动时采集的信号为例，原始信号如图17所示，将两个光电传感器采集的信号相加作为光电补偿，光电补偿后的效果如图18所示。然后对光电补偿后的信号进行经验模态分解与重构，即将信号分解成若干个内涵模态分量(IMF)，分解完成后将IMF2,IMF3,IMF4相加得到处理后的信号，如图19所示。最后对信号进行变窗长的滑动滤波得到去除运动噪声的信号如图20所示。
(3)血压计算及校准模块
血压计算及校准模块，其用于提取滑动滤波后的单路脉搏波信号的特征参数，并基于提取的特征参数及校准后的血压计算模型，得到血压检测值；
经过个体化差异消除，噪声滤除过程之后，就可以得到波形质量较好的脉搏波信号。然后就可以进行波形的特征提取，将提取的特征代入提前建好的血压计算模型中就可以计算得到血压值。
初始血压计算模型为：SBP＝f 1(x1，x2，...，xn),DBP＝f 2(x1，x2，...xn)，其中SBP为收缩压，DBP为舒张压，x1，x2，...，xn为血压计算所需要特征的值，f 1(x1，x2，...，xn)与f 2(x1，x2，...xn)代表基于相同提取特征的不同的计算模型。
血压计算模型需要的特征：重搏波峰值P d、重搏波谷值P dv、主波峰值P p、升支最大斜率S 1、 上升段面积A u、下降段面积A d、上升段下降段面积比A u/A d，重搏波与主波的时间间隔T dp、主波与主波的时间间隔T pp、收缩期时间ST、舒张期时间DT、升支最大斜率点与重搏波波峰的时间间隔T md、升支最大斜率点与重搏波波谷的时间间隔等特征T mdv，起点波谷到重搏波波谷的时间T vdv，重搏波波谷到终点波谷的时间T dvv，心血管特征量K，升支降支时间比T vdv/T dvv，收缩时间比ST/T pp，重搏波与主波的间隔时间占比T dp/T pp。
在所述血压计算及校准模块，提取滑动滤波后的单路脉搏波信号的特征参数为：与血压的皮尔逊相关系数超过设定阈值(例如：0.6)的特征参数。
皮尔逊相关系数的计算方式：
Figure PCTCN2022136638-appb-000005
其中，X表示参考血压BP,Y表示某个需要进行相关性分析的特征参数，Cov(X,Y)表示协方差，Var[X]表示X的方差，Var[Y]表示Y的方差。
虽然血压计算模型能够根据脉搏波特征的变化来反应血压的变化趋势，但是不同用户之间的血压基准值(平静状态下的血压)存在较大差别，而且脉搏波特征变化幅度相同的情况下，血压的变化幅度也存在一定的差别，这些差别是血压计算误差的第三个影响方面。
为了消除这一方面的误差影响，在正式测量之前设备需要进行血压校准，目的是对血压计算模型的系数进行校正，让血压计算模型更适应当前用户的脉搏波特征，提供更高精度的血压计算结果。
其中在所述血压计算及校准模块，血压计算模型的校准过程为：
基于参考血压值及提取的特征参数，计算得到一组校准系数；其中，校准系数的计算模型为参考血压值及提取的特征参数的设定多项式函数；
将校准系数作为已知初始血压计算模型的设定项的系数，来校准血压计算模型，得到校准后的血压计算模型。
校准过程中提取的脉搏波特征包括:重搏波峰值P d、重搏波谷值P dv、主波峰值P p、升支最大斜率S 1、上升段面积A u、下降段面积A d、上升段下降段面积比A u/A d，重搏波与主波的时间间隔T dp、主波与主波的时间间隔T pp、收缩期时间ST、舒张期时间DT、升支最大斜率点与重搏波波峰的时间间隔T md、升支最大斜率点与重搏波波谷的时间间隔等特征T mdv,起点波谷到重搏波波谷的时间T vdv,重搏波波谷到终点波谷的时间T dvv，心血管特征量K,升支降支时间比T vdv/T dvv， 收缩时间比ST/T pp，重搏波与主波的间隔时间占比T dp/T pp，部分特征展示在图21中。在图21中，a1为重搏波峰值，t1为收缩时间，t2为主波与重搏波的时间间隔，t3为周期时间，t4为舒张时间，t5为最大斜率点与重搏波的时间间隔。
在所述血压计算及校准模块中，所述初始血压计算模型为所述特征参数的预设多项式函数。
部分校准系数的计算模型为：
a 1＝f(T dp)   a 2＝f(P d)
b 1＝f(SBP ref，T pp，a 1)   b 2＝f(SBP ref，DT，a 2)   b 3＝f(SBP ref，ST，DBP ref)
其中a 1，a 2，b 1，b 2，b 3为校准系数，T dp为重搏波与主波的时间间隔，SBP ref为收缩压参考值，T pp为周期时间，P d为重搏波峰值，DT为舒张时间，ST为收缩时间。
f(T dp)为关于重搏波与主波间隔的多项式；f(SBP ref，T pp，a 1)为关于收缩压参考值、周期时间、校准参数a 1的多项式；f(P d)为关于重搏波峰值的多项式；f(SBP ref，DT，a 2)为关于收缩压参考值、舒张时间、校准参数a 2的多项式；f(SBP ref，ST，DBP ref)为关于收缩压参考值，舒张压参考值、收缩时间的多项式。
表1校正前后参考SBP与计算SBP对比
Figure PCTCN2022136638-appb-000006
其中，参考SBP指欧姆龙血压计测到的收缩压值，计算SBP是指计算模型得到的收缩压值。计算可以得到校准前参考SBP与计算SBP的误差为-3.125±10.334mmHg，而校准后参考SBP与计算SBP的误差为-0.875±3.588mmHg。
表2校正前后参考DBP与计算DBP对比
Figure PCTCN2022136638-appb-000007
其中，参考DBP指欧姆龙血压计测到的舒张压值，计算DBP是指计算模型得到的舒张压值。
计算可以得到校准前参考DBP与计算DBP的误差为-1.75±7.450mmHg，
而校准后参考DBP与计算DBP的误差为1.75±3.122mmHg。
实施例二
本实施例提供了一种计算机可读存储介质，其上存储有计算机程序，该程序被处理器执行时实现如下步骤：
采集消除个体化差异之后的单路脉搏波信号；
对单路脉搏波信号进行滤波；
提取滤波后的单路脉搏波信号的特征参数，并基于提取的特征参数及校准后的血压计算模型，得到血压检测值；
其中血压计算模型的校准过程为：
基于参考血压值及提取的特征参数，计算得到一组校准系数；其中，校准系数的计算模型为参考血压值及提取的特征参数的设定多项式函数；
将校准系数作为已知初始血压计算模型的设定项的系数，来校准血压计算模型，得到校准后的血压计算模型。
其中，在对单路脉搏波信号进行滤波的过程中：
在静息状态或者运动幅度小于设定阈值的情境下，采用变窗长的滑动滤波算法对单路脉搏波信号滤波。
在运动幅度大于设定阈值的情境下，采用预设分布的光电传感器来识别手腕运动方式，再采用与手腕运动方式相匹配的光电补偿策略对消除个体化差异之后的单路脉搏波信号进行光电补偿，最后依次通过经验模态的分解与重构及变窗长的滑动滤波算法对光电补偿后的单路脉搏波信号进行滤波。
实施例三
本实施例提供了一种计算机设备，包括存储器、处理器及存储在存储器上并可在处理器上运行的计算机程序，所述处理器执行所述程序时实现如下步骤：
采集消除个体化差异之后的单路脉搏波信号；
对单路脉搏波信号进行滤波；
提取滤波后的单路脉搏波信号的特征参数，并基于提取的特征参数及校准后的血压计算模型，得到血压检测值；
其中血压计算模型的校准过程为：
基于参考血压值及提取的特征参数，计算得到一组校准系数；其中，校准系数的计算模型为参考血压值及提取的特征参数的设定多项式函数；
将校准系数作为已知初始血压计算模型的设定项的系数，来校准血压计算模型，得到校准后的血压计算模型。
其中，在对单路脉搏波信号进行滤波的过程中：
在静息状态或者运动幅度小于设定阈值的情境下，采用变窗长的滑动滤波算法对单路脉搏波信号滤波。
在运动幅度大于设定阈值的情境下，采用预设分布的光电传感器来识别手腕运动方式，再采用与手腕运动方式相匹配的光电补偿策略对消除个体化差异之后的单路脉搏波信号进行光电补偿，最后依次通过经验模态的分解与重构及变窗长的滑动滤波算法对光电补偿后的单路脉搏波信号进行滤波。
以上所述仅为本发明的优选实施例而已，并不用于限制本发明，对于本领域的技术人员来说，本发明可以有各种更改和变化。凡在本发明的精神和原则之内，所作的任何修改、等同替换、改进等，均应包含在本发明的保护范围之内。

Claims (8)

  1. 一种基于单路脉搏波的血压监测装置，其特征在于，包括：
    信号采集模块，其用于采集消除个体化差异之后的单路脉搏波信号；
    信号滤波模块，其用于对单路脉搏波信号进行滤波；
    血压计算及校准模块，其用于提取滤波后的单路脉搏波信号的特征参数，并基于提取的特征参数及校准后的血压计算模型，得到血压检测值；
    其中在所述血压计算及校准模块，血压计算模型的校准过程为：
    基于参考血压值及提取的特征参数，计算得到一组校准系数；其中，校准系数的计算模型为参考血压值及提取的特征参数的设定多项式函数；
    将校准系数作为已知初始血压计算模型的设定项的系数，来校准血压计算模型，得到校准后的血压计算模型；
    在所述信号滤波模块中，在静息状态或者运动幅度小于设定阈值的情境下，采用变窗长的滑动滤波算法对单路脉搏波信号滤波；
    所述滑动滤波算法的滤波窗长和滤波次数由所述单路脉搏波信号的质量来决定的；
    所述变窗长的滑动滤波算法是指用噪声系数定量评估信号受噪声污染的程度，噪声系数由原始信号的一阶差分信号和二阶差分信号的零点数决定；由噪声系数与滤波窗长的关系模型确定滤波窗长。
  2. 如权利要求1所述的基于单路脉搏波的血压监测装置，其特征在于，所述个体化差异包括佩戴方式差异化和皮肤特性个体化差异。
  3. 如权利要求2所述的基于单路脉搏波的血压监测装置，其特征在于，在所述信号采集模块中，通过检测所述基于单路脉搏波的血压监测装置与皮肤接触面的设定点压力以及各个设定点压力差值，判断佩戴方式的统一性。
  4. 如权利要求2所述的基于单路脉搏波的血压监测装置，其特征在于，在所述信号采集模块中，量化皮肤特性，根据皮肤特征系数与光源光强的模型，计算光源光强调整值；调整后的光源光强为I＝I s*(a+b*ε)，其中a和b为匹配系数，ε为受试者的皮肤特征系数，
    Figure PCTCN2022136638-appb-100001
    I 0为标准出射光强，I 1为标准出射光强；I s为选定的标准光源光强。
  5. 如权利要求1所述的基于单路脉搏波的血压监测装置，其特征在于，在所述信号滤波模块中，在运动幅度大于设定阈值的情境下，采用预设分布的光电传感器来识别手腕运动方式，再采用与手腕运动方式相匹配的光电补偿策略对消除个体化差异之后的单路脉搏波信号进行光电补偿，最后依次通过经验模态的分解与重构及变窗长的滑动滤波算法对光电补偿后的单路脉搏波信号进行滤波。
  6. 如权利要求1所述的基于单路脉搏波的血压监测装置，其特征在于，在所述信号滤波模块中，所述单路脉搏波信号的质量由噪声系数决定，噪声系数定量评估信号受噪声污染的程度，噪声系数由原始信号的一阶差分信号和二阶差分信号的零点数决定。
  7. 一种计算机可读存储介质，其上存储有计算机程序，其特征在于，该程序被处理器执行时实现如下步骤：
    采集消除个体化差异之后的单路脉搏波信号；
    对单路脉搏波信号进行滤波；
    提取滤波后的单路脉搏波信号的特征参数，并基于提取的特征参数及校准后的血压计算模型，得到血压检测值；
    其中血压计算模型的校准过程为：
    基于参考血压值及提取的特征参数，计算得到一组校准系数；其中，校准系数的计算模型为参考血压值及提取的特征参数的设定多项式函数；
    将校准系数作为已知初始血压计算模型的设定项的系数，来校准血压计算模型，得到校准后的血压计算模型。
  8. 一种计算机设备，包括存储器、处理器及存储在存储器上并可在处理器上运行的计算机程序，其特征在于，所述处理器执行所述程序时实现如下步骤：
    采集消除个体化差异之后的单路脉搏波信号；
    对单路脉搏波信号进行滤波；
    提取滤波后的单路脉搏波信号的特征参数，并基于提取的特征参数及校准后的血压计算模型，得到血压检测值；
    其中血压计算模型的校准过程为：
    基于参考血压值及提取的特征参数，计算得到一组校准系数；其中，校准系数的计算模型为参考血压值及提取的特征参数的设定多项式函数；
    将校准系数作为已知初始血压计算模型的设定项的系数，来校准血压计算模型，得到校准后的血压计算模型。
PCT/CN2022/136638 2021-12-07 2022-12-05 基于单路脉搏波的血压监测装置、存储介质及电子设备 WO2023103977A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202111485941.5 2021-12-07
CN202111485941.5A CN114052688B (zh) 2021-12-07 2021-12-07 基于单路脉搏波的血压监测装置、存储介质及电子设备

Publications (1)

Publication Number Publication Date
WO2023103977A1 true WO2023103977A1 (zh) 2023-06-15

Family

ID=80228812

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2022/136638 WO2023103977A1 (zh) 2021-12-07 2022-12-05 基于单路脉搏波的血压监测装置、存储介质及电子设备

Country Status (2)

Country Link
CN (1) CN114052688B (zh)
WO (1) WO2023103977A1 (zh)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114052688B (zh) * 2021-12-07 2022-12-20 山东大学 基于单路脉搏波的血压监测装置、存储介质及电子设备

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106691406A (zh) * 2017-01-05 2017-05-24 大连理工大学 一种基于单探头光电容积脉搏波的血管弹性和血压的检测方法
EP3669773A1 (fr) * 2018-12-20 2020-06-24 Sorin CRM SAS Dispositif et procédé de détection de fibrillation ventriculaire
CN112274127A (zh) * 2020-10-28 2021-01-29 河北工业大学 一种基于单路脉搏波的无创连续血压检测方法、装置
CN114052688A (zh) * 2021-12-07 2022-02-18 山东大学 基于单路脉搏波的血压监测装置、存储介质及电子设备

Family Cites Families (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6882874B2 (en) * 2002-02-15 2005-04-19 Datex-Ohmeda, Inc. Compensation of human variability in pulse oximetry
US8948832B2 (en) * 2012-06-22 2015-02-03 Fitbit, Inc. Wearable heart rate monitor
FI126338B (en) * 2013-05-15 2016-10-14 Pulseon Oy Portable heart rate monitor
CN103607366B (zh) * 2013-11-18 2016-10-19 长安大学 一种基于滑动窗滤波的信号包络解调方法
CN104921736B (zh) * 2015-06-08 2017-08-04 浙江大学 一种包含参数估计功能滤波模块的连续血糖监测设备
CN107822608B (zh) * 2017-10-26 2020-04-17 中国民航大学 基于高斯混合模型的脉搏波特征提取方法
US11119176B2 (en) * 2017-11-27 2021-09-14 Siemens Healthcare Gmbh Automatic capture of cardiac motion by pre-scan and automated data evaluation for determination of motionless periods within the RR-interval
CN108078554A (zh) * 2018-01-05 2018-05-29 吉林大学 一种人体脉搏波信号噪声抑制方法
CN109512403B (zh) * 2018-11-05 2021-11-26 北京招通致晟科技有限公司 一种指端光电容积脉搏波检测方法、设备及系统
CN109875543B (zh) * 2019-02-01 2020-07-21 电子科技大学 用于可穿戴心率监测设备的多种健身运动状态下的心率估计方法及装置
EP4106622A4 (en) * 2020-03-20 2024-03-20 Georgia Tech Research Corporation METHODS AND SYSTEMS FOR NON-INVASIVE CUFFLESS BLOOD PRESSURE MONITORING
CN112274120B (zh) * 2020-10-28 2023-07-25 河北工业大学 一种基于单路脉搏波的无创动脉硬化检测方法、装置

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106691406A (zh) * 2017-01-05 2017-05-24 大连理工大学 一种基于单探头光电容积脉搏波的血管弹性和血压的检测方法
EP3669773A1 (fr) * 2018-12-20 2020-06-24 Sorin CRM SAS Dispositif et procédé de détection de fibrillation ventriculaire
CN112274127A (zh) * 2020-10-28 2021-01-29 河北工业大学 一种基于单路脉搏波的无创连续血压检测方法、装置
CN114052688A (zh) * 2021-12-07 2022-02-18 山东大学 基于单路脉搏波的血压监测装置、存储介质及电子设备

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
ZHU GUANG-PING, SUN HUI, ZHU FENG-QIN : "Detecting signals using adaptive mean and ridgelet transform filtering in a time-frequency domain", JOURNAL OF HARBIN ENGINEERING UNIVERSITY., vol. 31, no. 9, 15 September 2010 (2010-09-15), pages 1172 - 1178, XP093070676, DOI: 10.3969 /j.issn.1006-7043.2010.09.007 *

Also Published As

Publication number Publication date
CN114052688B (zh) 2022-12-20
CN114052688A (zh) 2022-02-18

Similar Documents

Publication Publication Date Title
WO2020119245A1 (zh) 一种基于可穿戴手环的情绪识别系统及方法
CN107569226B (zh) 基于压电传感获取hrv的方法及应用
US20190295729A1 (en) Universal non-invasive blood glucose estimation method based on time series analysis
WO2023103977A1 (zh) 基于单路脉搏波的血压监测装置、存储介质及电子设备
US10966662B2 (en) Motion-dependent averaging for physiological metric estimating systems and methods
US11350835B2 (en) Wearable device for reflecting fatigue level of human body
US20190090757A1 (en) Method and Apparatus for Estimating a Cardiovascular Characteristic Parameter, and Storage Medium for the Same
CN105054918A (zh) 一种基于脉搏反射波传输时间的血压计算方法及血压仪
CN106419879B (zh) 基于桡动脉生物传感器技术的血压动态监测系统及方法
CN111481190A (zh) 基于双路ppg信号的稳健心率测量方法和装置
CN117421539B (zh) 一种脑机接口系统数据优化处理方法
CN114145717B (zh) 基于ppg心率特征参数和运动量的睡眠状态分析方法
CN107995981B (zh) 一种用于血压测量装置的数据处理方法
CN106901732B (zh) 突变状态下肌力与肌张力的测量方法和测量装置
He et al. A new approach for daily life Blood-Pressure estimation using smart watch
CN111419219A (zh) Ppg心搏信号的预处理方法、装置及房颤检测设备
WO2023061565A1 (en) Apparatus, system and method for determining whether a person is asleep
Bali et al. Performance comparison of ANN classifiers for sleep apnea detection based on ECG signal analysis using hilbert transform
CN106175698B (zh) 睡眠状态分析中睡眠周期检测装置
CN112006669A (zh) 一种基于血压测量模型的双通道计算方法
CN113470805A (zh) 建立血压模型的方法
CN116211308A (zh) 一种高强度运动下机体疲劳评估方法
CN114052689A (zh) 一种运动状态下血压监测装置、存储介质及设备
CN117158940A (zh) 一种采用毫米波雷达的心率数据监测方法、检测系统及可穿戴装置
CN113951826B (zh) 一种用于评估睡眠状况的方法、系统及设备

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 22903399

Country of ref document: EP

Kind code of ref document: A1