CN113558590A - Blood pressure monitoring method and device based on electrocardio-piezoelectric pulse wave coupling - Google Patents

Blood pressure monitoring method and device based on electrocardio-piezoelectric pulse wave coupling Download PDF

Info

Publication number
CN113558590A
CN113558590A CN202110748592.5A CN202110748592A CN113558590A CN 113558590 A CN113558590 A CN 113558590A CN 202110748592 A CN202110748592 A CN 202110748592A CN 113558590 A CN113558590 A CN 113558590A
Authority
CN
China
Prior art keywords
pulse wave
piezoelectric
blood pressure
piezoelectric pulse
electrocardio
Prior art date
Legal status (The legal status 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 status listed.)
Withdrawn
Application number
CN202110748592.5A
Other languages
Chinese (zh)
Inventor
吴化平
魏昶
张灿
梁利华
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Zhejiang University of Technology ZJUT
Original Assignee
Zhejiang University of Technology ZJUT
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 Zhejiang University of Technology ZJUT filed Critical Zhejiang University of Technology ZJUT
Priority to CN202110748592.5A priority Critical patent/CN113558590A/en
Publication of CN113558590A publication Critical patent/CN113558590A/en
Withdrawn legal-status Critical Current

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
    • A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
    • A61B5/316—Modalities, i.e. specific diagnostic methods
    • A61B5/318—Heart-related electrical modalities, e.g. electrocardiography [ECG]
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
    • A61B5/316—Modalities, i.e. specific diagnostic methods
    • A61B5/318—Heart-related electrical modalities, e.g. electrocardiography [ECG]
    • A61B5/346—Analysis of electrocardiograms
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
    • A61B5/316—Modalities, i.e. specific diagnostic methods
    • A61B5/318—Heart-related electrical modalities, e.g. electrocardiography [ECG]
    • A61B5/346—Analysis of electrocardiograms
    • A61B5/349—Detecting specific parameters of the electrocardiograph cycle
    • A61B5/352—Detecting R peaks, e.g. for synchronising diagnostic apparatus; Estimating R-R interval
    • 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
    • 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
    • A61B5/7235—Details of waveform analysis
    • A61B5/725—Details of waveform analysis using specific filters therefor, e.g. Kalman or adaptive filters
    • 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
    • A61B5/7235—Details of waveform analysis
    • A61B5/7264—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems

Abstract

The invention discloses a blood pressure monitoring method and equipment based on electrocardio-piezoelectric pulse wave coupling. Acquiring a piezoelectric pulse wave signal and an electrocardiosignal; preprocessing the piezoelectric pulse wave signals and the electrocardiosignals to obtain a signal sequence, and extracting characteristics; calculating pulse transmission time PTT and heart rate HR according to the characteristics and segmenting the signal sequence; and establishing a convolutional neural network-long-term memory neural network (CNN-LSTM) model based on mixed characteristics, training the model by using the segmented signal sequence, and measuring the blood pressure by using the trained model.

Description

Blood pressure monitoring method and device based on electrocardio-piezoelectric pulse wave coupling
Technical Field
The application belongs to a blood pressure measuring method and equipment in the field of human health detection, and particularly relates to a machine learning blood pressure measuring method and equipment based on mixed characteristics of piezoelectric pulse wave signals and electrocardiosignals.
Background
Blood Pressure refers to the Pressure that Blood exerts on the arterial wall when the heart contracts and relaxes, pushing Blood to flow around the body in Blood vessels, and is generally represented by the Systolic (SBP) and Diastolic (DBP) pressures. Systolic pressure refers to the blood pressure at which the blood pressure reaches a maximum value in the mid-systolic period of the heart chamber, and diastolic pressure refers to the blood pressure at which the blood pressure gradually decreases during ventricular diastole and reaches a minimum value when ventricular ejection is stopped
Pulse wave measurement is currently the most popular non-invasive blood pressure measurement. The principle can be mainly divided into Pulse Wave Transit Time (PTT) or Pulse Wave Velocity (PWV) and Pulse Wave parametric Analysis (PWA).
The PTT method fits a blood pressure formula by using the transmission time between two locations in the artery, and calculates the blood pressure. The pulse wave parameter analysis method is to process pulse wave signals, extract specified characteristics from the pulse wave signals, and then establish a mapping relation between the characteristics and blood pressure. The two technologies can measure and obtain the blood pressure in a portable way, and have important significance for diagnosing the hypertension.
However, the current work on the blood pressure monitoring technology has some problems, mainly including the following points: firstly, the single PTT method has lower precision and does not meet the medical standard; secondly, pulse wave signals utilized by researchers for researching blood pressure measurement technology are basically from photoelectric pulse wave signals, the photoelectric pulse wave signals mostly collect the volume change of capillary vessels at the tail ends of human limbs, and the characteristics of weak signals, heavy pulse waves and the like are not obvious and are easily influenced by environmental noise. Thirdly, at present, a method of manually extracting features is mostly adopted when analyzing pulse wave features, the effectiveness of the pulse wave features is determined after limited sample experiments, different researchers extract features differently, and no unified standard exists. When using the electrocardiographic signal, only the ECG is used as a reference signal to calculate the PTT parameter, however, the electrocardiographic signal is an important physiological parameter reflecting the health of the heart, and some characteristics of the waveform, such as the Q-wave T-wave time interval, can reflect the change of the blood pressure.
Disclosure of Invention
The invention aims to overcome the defects in the prior art and provide the blood pressure measuring method which is simple in measurement, high in precision and more suitable for daily blood pressure monitoring.
The key point of the main technology is that the machine learning technology is utilized to automatically extract useful signals in electrocardiosignals and piezoelectric pulse waves, and the conduction time and heart rate characteristics of the pulse waves are mixed to measure the blood pressure.
In order to achieve the purpose, the technical scheme adopted by the invention is as follows:
a blood pressure monitoring method based on electrocardio-piezoelectric pulse wave coupling mainly comprises the following steps:
a) acquiring a piezoelectric pulse wave signal and an electrocardiosignal;
b) preprocessing the piezoelectric pulse wave signals and the electrocardiosignals to obtain a signal sequence, and extracting characteristics;
c) calculating pulse transmission time PTT and heart rate HR according to the characteristics, and segmenting the signal sequence;
d) and establishing a CNN-LSTM model based on mixed characteristics, training the model by using the segmented signal sequence, and measuring the blood pressure by using the trained model.
In the step a), acquiring the piezoelectric pulse wave signals and the electrocardiosignals specifically adopts a polyvinylidene fluoride (PVDF) piezoelectric sensor and a conditioning circuit to acquire the pressure pulse wave signals, and adopts the electrocardiosensor and the conditioning circuit to acquire the electrocardiosignals, wherein the acquisition frequency is 200 Hz.
In the step b), the piezoelectric pulse wave signal and the electrocardiosignal are preprocessed, and the method specifically comprises the following steps: and baseline drift and high-frequency noise interference are removed.
In the step b), filtering the electrocardiosignal by using a high-order Butterworth filter to perform band-pass filters with cut-off frequencies of 0.67Hz and 40Hz respectively, and then filtering by using a morphological filter to obtain an electrocardiosignal sequence; and then extracting characteristics by adopting a time and amplitude dual-threshold method, wherein the method comprises the steps of first-order difference, amplitude threshold screening and time threshold screening which are sequentially carried out, and the main wave peak of the piezoelectric pulse wave is obtained by processing and is used as the characteristics of the piezoelectric pulse wave.
In the step b), the piezoelectric pulse wave signals are subjected to band-pass filtering by using band-pass Butterworth filters with cut-off frequencies of 0.5 to 10Hz respectively, and then are filtered by using a morphological filter to obtain a piezoelectric pulse wave signal sequence; and then extracting by adopting a self-adaptive threshold method, wherein the self-adaptive threshold method comprises the steps of band-pass filtering, square linear amplification, window integration, low-pass filtering and R peak calibration which are sequentially carried out, and the R wave peak of the electrocardiosignal is obtained by processing and is used as the characteristic of the electrocardiosignal.
In the step c), the method specifically comprises the following steps:
c.1) calculating the time difference between the R wave crest of the electrocardiosignal and the main wave crest of the piezoelectric pulse wave signal as the pulse transmission time PTT according to the following formula:
Figure BDA0003145163970000021
in the formula, P [ i]The position serial number of the ith main wave peak of the piezoelectric pulse wave signal in the preprocessed piezoelectric pulse wave signal sequence is R [ i]The position serial number, F, of the ith R wave peak of the electrocardiosignal in the preprocessed electrocardiosignal sequenceSFor the sampling frequency, PTT [ i ]]PTT data of the ith pulse transmission time; n represents the total number of dominant/R wave peaks.
Calculating the interval time of two adjacent electrocardio R wave peaks according to the following formula to determine the heart rate HR:
Figure BDA0003145163970000031
wherein HR [ i ] is the ith heart rate HR data;
c.2) after the pulse transmission time PTT and the heart rate HR are obtained through calculation, segmenting the electrocardiosignal sequence and the piezoelectric pulse wave signal sequence, wherein the segmenting steps are mainly as follows:
selecting original data with fixed duration, wherein the original data are an electrocardiosignal sequence and a piezoelectric pulse wave signal sequence;
dividing according to the main wave crest of the piezoelectric pulse wave signal, and dividing continuous data between every three pulse wave crests as a period, so as to divide the original data into a section of data section corresponding to each period; the segmentation of the invention is based on piezoelectric pulse wave signals, and the segmentation range is two heartbeat cycles, namely data between three pulse wave peaks.
And (3) interpolating each data segment through one-dimensional second-order spline interpolation, then normalizing, connecting the interpolated and normalized data segments according to the time sequence of the heartbeat cycle and the sequence of the first electrocardio wave and the second pulse wave to obtain each data sequence, and intercepting the first six data sequences a1-a 6.
In the step d), the CNN-LSTM model based on the mixed characteristics comprises a CNN neural network and LSTM and fully-connected neural network models, the CNN neural network comprises two continuous convolution modules, and each convolution module is mainly formed by sequentially connecting a convolution layer, a discarding layer and a pooling layer; the LSTM and full-connection neural network model comprises a bidirectional long-short-time memory neural network (BiLSTM) module and a full-connection module (Dense), the bidirectional long-short-time memory neural network module is mainly formed by sequentially connecting a batch standardization layer, a bidirectional long-short-time memory network layer and a rejection layer, and the full-connection module is formed by sequentially connecting the full-connection layer, the rejection layer and the full-connection layer; a obtained by processing the piezoelectric pulse wave signal and the electrocardio signal1To a6And as the input of the CNN neural network, extracting blood pressure related characteristics by using the CNN neural network, mixing the blood pressure related characteristics and sequences of the electrocardiosignal sequence and the piezoelectric pulse wave signal sequence through a fusion layer, and inputting the mixture into the LSTM and the fully-connected neural network model for processing and outputting signals of the systolic pressure SBP and the diastolic pressure DBP.
In the step d), the electrocardiosignals and the piezoelectric pulse wave signals which are actually measured are processed according to the steps b) to c), and then input into the trained model to be output, so that the measurement result is obtained.
Secondly, a blood pressure detection device based on electrocardiosignals and piezoelectric pulse waves:
the system comprises:
the signal acquisition unit is used for acquiring electrocardiosignals and piezoelectric pulse wave signals and carrying out hardware conditioning on the signals to obtain relatively clean signals;
the data transmission unit is used for transmitting the electrocardiosignals and the piezoelectric pulse wave signals obtained by the hardware to the mobile phone terminal;
the mobile phone terminal is internally provided with a mobile phone software unit for receiving the electrocardio-piezoelectric signals obtained by the data transmission unit, processing the data according to the method of claim 7 and calculating to obtain a systolic pressure SBP and a diastolic pressure DBP.
The hardware conditioning mainly comprises low-pass filtering, high-pass filtering, an amplifying circuit and notch filtering.
And the data transmission unit transmits the original electrocardio and piezoelectric pulse wave digital signals to a mobile phone software unit in the mobile phone terminal in a BLE Bluetooth mode according to a formulated protocol.
The mobile phone software unit needs to analyze a Bluetooth packet data protocol, process signals, input an algorithm model, calculate to obtain SBP and DBP, and display signal waveforms in real time, display calculation results of the SBP and the DBP and store user data.
The invention has the beneficial effects that:
1. the accuracy of blood pressure measurement is improved.
2. The adaptive population range of blood pressure measurement is expanded.
The signals utilized by the invention are piezoelectric pulse wave signals and electrocardio signals, the piezoelectric pulse wave signals are used as the blood vessel pressure change to be more directly reflected, and compared with photoelectric pulse wave signals, the pulse wave signals can retain more useful information, contain more characteristic information and have higher signal intensity. And there are features in the cardiac electrical signal that have a high correlation with blood pressure. The invention utilizes the convolutional neural network to automatically extract the characteristics of the piezoelectric pulse wave and the electrocardio pulse wave, mixes PTT and HR parameters which are proved to have high correlation with blood pressure to be used as the input of an algorithm model, and then measures the systolic pressure and the diastolic pressure based on a deep learning model of the bidirectional long-and-short time memory neural network for better considering the dependence in the time domain direction required by the continuous monitoring of signals, thereby bringing the effect of the accuracy of blood pressure measurement. The neural network is used for training, the prior knowledge of artificial feature extraction can be effectively avoided by automatically extracting features, useful features are extracted from crowds with different health conditions, and the adaptive population range of blood pressure measurement is expanded.
Drawings
FIG. 1 is a flow chart of a blood pressure measurement method;
fig. 2 is a structure view of the smart bracelet 1;
fig. 3 is a diagram of a smart bracelet structure 2;
FIG. 4 is a flow chart of the ECG signal pre-processing;
FIG. 5 is a schematic diagram of the effect of extracting the R wave peak of the electrocardiographic signal;
FIG. 6 is a flow chart of piezoelectric pulse wave signal preprocessing;
FIG. 7 is a schematic diagram illustrating the effect of extracting the main wave peak of the piezoelectric pulse wave;
FIG. 8 is a schematic diagram of data partitioning;
FIG. 9 is a diagram of a neural network model topology;
fig. 10 is a schematic diagram of the terminal software.
In the figure: electrocardio-electrode 1, piezoresistive sensor 2, electrocardio-electrode 3, piezoelectric pulse wave sensor 4 and wrist surface 5.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is further described in detail with reference to the accompanying drawings and examples. The specific embodiments described herein are merely illustrative of the invention and are not to be construed as limiting the invention.
The blood pressure measurement designed by the invention is shown in figure 1, and the conditions of the embodiment are as follows:
a) acquiring a piezoelectric pulse wave signal and an electrocardiosignal;
in the step a), acquiring piezoelectric pulse wave signals and electrocardiosignals, specifically, acquiring pressure pulse wave signals by adopting a polyvinylidene fluoride (PVDF) piezoelectric sensor and a conditioning circuit, acquiring original signals by the PVDF piezoelectric sensor, analyzing the original signals by the conditioning circuit to obtain the pressure pulse wave signals, acquiring electrocardiosignals by adopting the electrocardiosensor and the conditioning circuit, acquiring the original signals by the electrocardiosensor, analyzing the original signals by the conditioning circuit to obtain the electrocardiosignals, wherein the acquisition frequency is 200 Hz.
Specifically implement and adopt intelligent bracelet to carry out piezoelectricity pulse wave signal and electrocardiosignal's measurement, electrocardioelectrode 1, piezoresistive sensor 2 have been arranged to the bottom surface in the wrist surface 5 of intelligent bracelet, and electrocardioelectrode 3 has been arranged to the top surface outside the wrist surface 5 of intelligent bracelet, is equipped with piezoelectricity pulse wave sensor 4 near wrist blood vessel department at the watchband.
The part is mainly obtained by a hardware part of the system, the embodiment of the invention is a wearable intelligent bracelet, signals are acquired by the intelligent bracelet, and the structural schematic diagram of the intelligent bracelet is shown in fig. 2 and 3.
The bracelet detects pulse wave signals at the radial artery of the wrist and single-lead electrocardiosignals between the wrist and fingers at the same time and is worn at the wrist.
The bracelet hardware system mainly comprises the following units.
The signal sensor unit, the signal conditioning unit, the digital-to-analog conversion unit and the main controller processing Bluetooth sending unit.
The signal sensor unit mainly comprises an electrocardio sensor and a piezoelectric pulse wave sensor.
The electrocardiosignal is measured by adopting single-lead connection electrocardio, only two electrodes are needed, the sensor adopts a dry electrode with gold deposited on the surface, the size is small, and the portability of the device is higher. The electrocardio-electrode 1 is contacted with the surface of the wrist, and the electrocardio-electrode 3 is contacted with the finger of the other hand.
Piezoelectric pulse wave sensor 4 adopts HK1205PVDF piezoelectric sensor, and the piezoelectric constant of polyvinylidene fluoride PVDF membrane is high than other materials, and consequently is high to the detection precision of stress strain than other materials, and the detection limit is lower, can be applicable to the detection of small strain stress better. The piezoelectric sensor is in contact with the radial artery of the wrist.
The wrist strap of the intelligent bracelet is utilized to give a certain pretightening force to the piezoelectric sensor, so that a piezoelectric pulse wave signal with higher quality can be obtained. The detection of the pretightening force is carried out through a piezoresistive sensor 2 which is a piezoresistive sensor, and the pressure value of the piezoresistive sensor is approximately equal to the pretightening force of the piezoelectric sensor.
The signal conditioning unit is divided into an electrocardiosignal conditioning unit and a piezoelectric pulse wave conditioning unit.
The human body ECG signal has the characteristics of weak property, low frequency, randomness and the like, and the ECG signal at the wrist and the finger is weaker. The ECG electrode of the device is contacted with dry skin to collect electrocardiosignals on the surface of a human body, and the dry skin presents high impedance characteristic under low voltage. Therefore, to accurately acquire the ECG signal, the circuit for ECG signal acquisition has severe requirements such as high gain, high common mode rejection ratio, high input impedance, and strict frequency response range.
Therefore, in order to obtain a clean and high-quality electrocardiosignal, the electrocardiosignal conditioning unit needs to condition the electrocardiosignal, and as shown in fig. 4, the electrocardiosignal conditioning unit mainly comprises a primary amplifying circuit, a low-pass filter circuit, a high-pass filter circuit, a main amplifying circuit and a voltage lifting circuit.
The electrocardio input belongs to high-impedance input, and in order to avoid the reduction of input voltage, the designed electrocardio amplifying circuit meets the condition of high input impedance, so the amplifying circuit of the system is constructed by two stages of amplifiers. The amplitude of the original cardiac signal is only a few mV, thus amplifying it at least by a few hundred times. In order to reduce the hardware space and improve the system integration, the hardware system uses an instrument amplifier as an input amplifier and simultaneously performs primary amplification on signals. The amplifier model is AD620 by TI corporation. The gain of the first-stage amplifying circuit is 10 times.
According to the recommended standard of American Heart Association (AHA) and American College of Cardiology (ACC), the signal passband of the amplifying circuit of the electrocardio monitoring device is recommended to be set to be 0.67Hz to 40Hz, but the common signal range of the electrocardio signals is mostly within 20Hz, and the invention sets the cut-off frequency of the filter passband to be 0.4Hz to 25Hz according to the requirements of blood pressure monitoring and specific experimental tests, and is realized by a low-pass filter and a high-pass filter.
The low-pass filter circuit is a fourth-order filter constructed by two-order unity-gain active low-pass filters. The operational amplifier is an OPA2188 amplifier of TI company by comprehensively considering the volume and cost factors. The cut-off frequency of the low-pass filter is 25 Hz.
The high-pass filter circuit is constructed by a unity-gain second-order active high-pass filter, an OPA2188 amplifier of TI company is selected as an operational amplifier, and the cut-off frequency of the high-pass filter is 0.4 Hz.
The main amplifying circuit mainly comprises a main amplifier, the signal is further amplified, the gain of the secondary amplification is about 20 times, and the secondary amplification and the primary amplification can achieve about 200 times of amplification effect.
Because the analog-to-digital converter ADC selected by the invention can only convert analog signals with positive voltage, and electrocardiosignals may have negative voltage, a voltage lifting circuit needs to be designed. The lifting circuit is realized by adopting a differential amplifier, the input voltage of the negative input end is obtained by dividing the voltage of a positive power supply and a negative power supply, the gain is unity, and the voltage of the negative input end can be adjusted by adjusting the size of the variable amplifier.
The piezoelectric pulse wave signal acquisition is obtained by adopting a PVDF piezoelectric film sensor. PVDF film sensor obtains that the pulse wave signal is comparatively weak, and voltage is at the millivolt level, and certain noise can be brought into to sensor itself and hardware circuit, motion interference etc. consequently piezoelectricity pulse wave signal need be through enlarging and filtering processing, and piezoelectricity pulse wave signal conditioning unit mainly includes: the device comprises a signal amplifying unit, a band-pass filtering unit and a voltage lifting unit.
The pulse wave main amplification circuit designed by the invention is constructed by a reverse amplifier, and the amplification factor is about 20 times.
The input signal and the circuit have noise such as white noise, sensor noise and the like, and the frequency of useful information of pulse waves is basically within 10Hz, so the cut-off frequency of the low-pass filter circuit designed by the invention is about 10Hz, and the low-pass filter circuit consists of an active first-order inverse low-pass filter.
Due to the characteristics of the piezoelectric sensor, negative signals may appear in the piezoelectric signal, so a voltage raising circuit is needed to raise the piezoelectric pulse wave signal to a voltage range suitable for ADC conversion.
The clean simulated electrocardiosignals and piezoelectric pulse wave signals can be obtained through the signal sensor unit.
The digital-to-analog conversion unit mainly comprises an ADS115 chip.
The electrocardiosignal and the piezoelectric pulse wave signal after signal conditioning are analog signals, and the analog signals are required to be converted into digital signals for further processing signal data. The analog-digital conversion unit designed by the invention mainly comprises ADS1115, wherein the ADS1115 is an analog-digital conversion chip of TI company and is compatible with I2The chip has four single-ended inputs or two differential inputs, the conversion rate is up to 860SPS, the voltage conversion range can reach +/-256 mV to +/-6.144V, the measurement of signals with precise magnitude can be realized, and the power consumption during continuous conversion is only 150 muA. The sampling rate of the signal of the invention is 250Hz, and the sampling and conversion requirements of the invention are met.
The clean electrocardio and piezoelectric pulse wave digital signals can be obtained through the steps.
The main controller processing Bluetooth sending unit is mainly responsible for collecting digital signals and sending the signals to the terminal software in a wireless mode.
The main controller signal acquisition frequency is 200Hz, then the electrocardio and piezoelectric pulse wave data are packaged by a certain protocol and sent to the terminal software through the low-power Bluetooth.
b) Preprocessing the piezoelectric pulse wave signals and the electrocardiosignals to obtain a signal sequence, and extracting characteristics;
in the step b), filtering the electrocardiosignals by using a high-order Butterworth filter according to the characteristic that the frequency of the electrocardio useful signals is mostly between 0.67Hz and 40Hz, and then filtering by using a morphological filter to obtain an electrocardiosignal sequence; and then extracting characteristics by adopting a time and amplitude dual-threshold method, wherein the method comprises the steps of first-order difference, amplitude threshold screening and time threshold screening which are sequentially carried out, and the main wave peak of the piezoelectric pulse wave is obtained by processing and is used as the characteristics of the piezoelectric pulse wave.
S2.1, the structure of the morphological filter constructed by the electrocardiosignal R is as follows:
Figure BDA0003145163970000081
wherein H (n) is the filtered signal; s (n) is an original signal; OC [ S (n) ] is to open and close S (n); CO [ S (n) ] is the closing and opening operation of S (n). n denotes the position of the current data in the sequence.
The shape of the morphological filter is linear with a width of the structuring element of 61.
S3.2, the specific steps of extracting the electrocardiosignal R wave are as follows:
the method comprises the following steps: the electrocardiosignals are subjected to band-pass filtering. The QRS energy in the electrocardiosignal is concentrated near 8-16Hz, the P wave energy is concentrated at 3-12Hz, and the T wave peak energy is 0-7 Hz. Through experimental tests, the QRS wave can be well highlighted by filtering the original signal by using a 40-order Butterworth band-pass filter with the frequency of 15-25Hz, meanwhile, the interference of removing P waves, T waves and other waves can be reduced, and the later-stage R wave extraction is facilitated.
Step two: the electrocardiosignal is then squared and linearly amplified. The signal after band-pass filtering is subjected to square operation, and the square operation can make all data into positive numbers and simultaneously equivalently perform nonlinear amplification on the signal.
Step three: the electrocardiosignal is window integrated. The QRS segment is the place where the energy transformation of the electrocardiosignal is most severe, and the waveform information of the QRS can be obtained and amplified through window integration. The key to the moving window integration is the selection of the window width, which should be similar to the width of the QRS segment. The QRS wave band is generally 0.06s-0.10s, not more than 0.11s, the sampling frequency of the invention is 200Hz, and the window width is set to 25 after the experiment. The integrated cardiac electric wave peak is obvious, and the rising area and the plateau period of the peak are mainly QRS sections.
Step four: the zero-phase Butterworth low-pass filtering with the cut-off frequency of 10Hz is adopted, the waveform is smoother after the low-pass filtering, the wave crest is more obvious, and the wave crest monitoring precision is favorably improved.
Step five: and judging the QRS wave band. Setting a low threshold and a high threshold, and temporarily determining that a QRS wave is detected when the detected peak signal is greater than the low threshold. And respectively carrying out self-adaptive adjustment on the low threshold and the high threshold according to the amplitude of the currently detected peak.
In order to ensure that the low and high thresholds are not too low, the lower limit of the two thresholds is set, and the specific adjustment method of the high and low thresholds is shown as the following formula:
Figure BDA0003145163970000082
Figure BDA0003145163970000091
in the formula (4), THRhFor high threshold, THRlThe initial high threshold and the low threshold are selected according to a specific signal and need to be adjusted to be a low threshold, and peak refers to the amplitude of the currently detected peak. According to the signal after the pretreatment collected by the equipment, the initial high threshold value is 0.8, and the low threshold value is 0.5. mean (peak)f) The average value of the first eight electrocardio wave peak values of the currently detected wave peak is referred to. THRhlimAnd THRllimThe lower limits of the high threshold and the low threshold change are respectively, and the two values are respectively 0.4 and 0.25.
Step six: and calibrating the R peak. The position of the QRS section signal can be detected through the double threshold value method signal detection. Because delay is caused at the filtering part and the window integral part, the R peak can be detected only by subtracting the signal delay after the wave peak is detected in the step five, but because the signal acquired by the hardware system of the invention is a discrete signal, the delay may have slight offset in the process, and the R wave detection accuracy is improved by adopting an R wave return detection mode. After the peak after the step five is detected, the position of the peak stays in the second half of the QRS wave band of the original signal, so the invention searches the peak of the R wave through the slope in the range of a window before the peak, and the size of the window is set to be 20. The R wave is the wave with the most drastic change, the absolute value of the slope product obtained by the forward difference and the backward difference is the largest, the range of the forward difference and the backward difference is set to be 4, the R wave is an extreme point, and the R wave in the window can be detected through the condition.
Step seven: in order to reduce the false detection of the R wave, screening of a refractory period is introduced, because the heart rate of normal people of a human body is 60-100bpm, the heart rate of some patients can reach 160 times/minute, and meanwhile, the phenomena of irregular heart rate and the like can occur to some patients, the sampling frequency is combined, and when two R wave peaks are detected in a window of 48 sampling points, namely 0.24s, a point with a higher peak value is selected as the R wave peak.
And storing the extracted sequence number of the R wave crest in a sequence R [ i ].
In the step b), according to the characteristic that the useful frequency of the piezoelectric pulse wave signal is mainly within 10Hz, a band-pass Butterworth filter with the cut-off frequency of 0.5-10 Hz is used for carrying out band-pass filtering to remove main baseline drift and high-frequency noise, and then a morphological filter is used for carrying out filtering to remove irregular noise signals to obtain a piezoelectric pulse wave signal sequence; then, an adaptive threshold method is adopted for extraction, the steps of band-pass filtering, square linear amplification, window integration, low-pass filtering and R peak calibration are sequentially carried out, the R wave peak of the electrocardiosignal is obtained through processing and is used as the characteristic of the electrocardiosignal, and the extraction effect of the R wave peak of the electrocardiosignal is shown in fig. 5.
The piezoelectric pulse wave signal preprocessing flow is shown in fig. 6.
S2.1, the structure of the morphological filter constructed by the piezoelectric pulse wave signals is as follows:
Figure BDA0003145163970000092
wherein H (n) is the filtered signal; s (n) is an original signal; OC [ S (n) ] is to open and close S (n); CO [ S (n) ] is the closing and opening operation of S (n). n denotes the position of the current data in the sequence.
The shape of the morphological filter is linear with a width of the structuring element of 61.
S2.2、
The detection of the main wave crest of the piezoelectric pulse wave is as follows:
the intelligent bracelet performs low-pass filtering and baseline drift removal processing on the pulse waves, the obtained pulse wave waveform is smooth and stable, and the main wave peak of the pulse waves is detected by adopting a differential double-threshold method.
The specific steps are as follows:
the method comprises the following steps: and calculating a first-order difference function of the pulse wave, wherein if the zero crossing point of the first-order difference indicates that the point is a maximum value point, the main wave crest of the pulse wave is in a maximum value set meeting the condition.
And step two, calculating the average value of the maximum values, setting the initial amplitude threshold value to be 0.4 times of the average value of the maximum values, and updating the peak threshold value to be 0.4 times of the current peak after the peak is detected.
Step three: and setting a time threshold, wherein if only one maximum value point exists in the time threshold, the point is a pulse wave crest, and if two maximum value points exist in the time threshold, a point with a larger amplitude is a pulse wave main wave crest.
Step four: in order to reduce false detection of the main wave peak of the pulse wave, screening of a 'refractory period' is introduced, because the heart rate of normal people of a human body is 60-100 times/minute, the heart rate of some patients can reach 160 times/minute, and meanwhile, the phenomena of irregular heart rate and the like can occur to some patients, the sampling frequency is combined, when two main wave peaks of the pulse wave are detected in a window with 48 sampling points, namely 0.24s, a point with a higher peak value is selected as the main wave peak of the pulse wave, and the extraction effect of the main wave peak of the piezoelectric pulse wave is shown in fig. 7.
And storing the sequence number of the extracted pulse wave main wave crest in a sequence P [ i ].
c) Calculating pulse transmission time PTT and heart rate HR according to the characteristics and segmenting the signal sequence;
c.1) calculating the time difference between the R wave crest of the electrocardiosignal and the main wave crest of the piezoelectric pulse wave signal as the pulse transmission time PTT according to the following formula:
Figure BDA0003145163970000101
in the formula, P [ i]The position serial number of the ith main wave peak of the piezoelectric pulse wave signal in the preprocessed piezoelectric pulse wave signal sequence is R [ i]The position serial number, F, of the ith R wave peak of the electrocardiosignal in the preprocessed electrocardiosignal sequenceSAt a sampling frequency of 250Hz, PTT [ i ]]PTT data of the ith pulse transmission time; n represents the total number of dominant wave peak/R wave peak'
Calculating the interval time of two adjacent electrocardio R wave peaks according to the following formula to determine the heart rate HR:
Figure BDA0003145163970000102
wherein HR [ i ] is the ith heart rate HR data;
c.2) after the pulse transmission time PTT and the heart rate HR are obtained through calculation, segmenting the electrocardiosignal sequence and the piezoelectric pulse wave signal sequence, wherein the segmenting steps are mainly as follows:
raw data of a fixed duration, for example 10s, i.e. electrocardiograph and pulse wave data of 2000 sampling points, are selected. The original data are an electrocardiosignal sequence and a piezoelectric pulse wave signal sequence;
dividing according to the main wave peak of the piezoelectric pulse wave signal, as shown in fig. 8, dividing the continuous data between every three pulse wave peaks as a period, thereby dividing the original data into a data segment corresponding to each period; interpolating each data segment by one-dimensional second-order spline interpolation, normalizing after interpolation, connecting the interpolated and normalized data segments according to the time sequence of heartbeat cycle and the sequence of first electrocardio and second pulse wave to obtain each data sequence, and intercepting the first six data sequences a1-a6. For example, the first heart-beat period interpolated and normalized electrocardiosignal data segment and the first heart-beat period interpolated and normalized pulse wave signal data segment are connected to form a data sequence a1After interpolation and normalization of the second heartbeat cycleThe electrocardiosignal data segment is connected with the pulse wave signal data segment after the interpolation and the standardization of the second heartbeat period to form a data sequence a2. The dimension of the original data feature vector is (6, 1600), and the dimension of the auxiliary input feature vector is (6, 2).
d) And establishing a CNN-LSTM model based on mixed characteristics, training the model by using the segmented signal sequence, and measuring the blood pressure by using the trained model.
In the step d), the CNN-LSTM model based on the mixed characteristics comprises a CNN neural network and LSTM and fully-connected neural network models, the CNN neural network comprises two continuous convolution modules, and each convolution module is mainly formed by sequentially connecting a convolution layer, a discarding layer and a pooling layer; the LSTM and fully-connected neural network model comprises a bidirectional long-short time memory neural network (BiLSTM) module and a fully-connected module (Dense). The bidirectional long-short time memory neural network module is mainly formed by sequentially connecting a batch standardization layer, a bidirectional long-short time memory network layer and a abandonment layer. The full-connection module is formed by connecting a full-connection layer 1, a discarding layer and a full-connection layer 2 at one time.
The method comprises the steps of taking a1-a6 obtained by processing piezoelectric pulse wave signals and electrocardio signals as input of a CNN neural network, extracting blood pressure related features by using the CNN neural network, mixing the blood pressure related features and sequences of electrocardio signal sequences and piezoelectric pulse wave signal sequences through a fusion layer, inputting the mixture to an LSTM and a fully-connected neural network model, and processing and outputting signals of systolic pressure SBP and diastolic pressure DBP. Specifically, the model is trained, and after the iteration is performed for 300 times, the model is stored as the trained model.
The designed algorithm model structure is shown in fig. 9.
The raw input 1 and the input features 2 are used as input of the algorithm model, and the following is a detailed description.
The convolution operation is first performed on input 1. The method comprises two layers of one-dimensional convolutional layers, wherein the first layer of convolutional layer learns the basic characteristics of input data, and the second layer of convolutional layer learns higher-level characteristics. The number of filters of each convolution layer is 128, the length of a convolution window is 3, and after two one-dimensional convolutions, the two one-dimensional convolutions are connected with a dropout layer with the size of 0.4 and a maximum pooling layer with the size of 2, so that overfitting can be effectively prevented by the dropout layer and the pooling layer, and the robustness of the model is improved. And (3) connecting the features extracted by the CNN with auxiliary input to obtain input features 1 with feature dimensionality (6, 128) after convolution operation, and adding a Batch normalization layer (BN) to solve the problems of gradient extinction and gradient explosion in training.
The resulting input feature 1 is connected with the input feature 2 to obtain a hybrid feature with dimension (6, 66), and the batch normalization layer is connected.
The bidirectional long-short time memory network layer (BilSTM) is connected, the inside of the BilSTM comprises two LSTM layers, the two LSTM layers learn the dependence of characteristics on time through two directions and finally summarize the dependence into a characteristic variable, the number of units of the BilSTM selected by the invention is 64, and therefore, the vector length is 128. From the feature dimensions input to BilSTM, the time step is 6, and the input is the mixture feature.
After learning the characteristics through the BilSTM layer, adding two full-connected layers, wherein the first full-connected layer adopts a ReLU activation function and carries out batch standardization, accelerating model convergence and improving model precision. Since the predicted blood pressure is a continuous value, the second layer fully-connected layer does not need an activation function, and the output of the second layer fully-connected function is the SBP and DBP, the last of which is the final predicted systolic and diastolic blood pressures.
The Mean Square Error (MSE) is selected as a loss function for training the algorithm model, the process of minimizing the loss function is the process of continuously training the neural network, and along with the reduction of the loss function, the difference between the predicted value and the true value of the neural network is smaller, the Error is smaller, and the model effect is better.
The mean square error of the algorithm model is calculated by the formula:
Figure BDA0003145163970000121
in the formula, N is the total number of samples,
Figure BDA0003145163970000122
as a result of the calibration of the blood pressure, yiAnd calculating the obtained blood pressure by using an algorithm model.
The algorithm model of the invention adopts Adam (adaptive motion estimation) optimization, the Adam optimization can effectively improve the training speed and stability of the neural network model, and the initial learning rate is set to be 0.01.
And storing the training to obtain an algorithm model.
Then the collected 10s data is preprocessed, parameters are calculated, and data are segmented and input into a model, and then the blood pressure measurement result is obtained.
In the step d), the electrocardiosignals and the piezoelectric pulse wave signals which are actually measured are processed according to the steps b) to c), and then input into the trained model to be output to obtain the measurement result.
The invention also comprises a terminal software which comprises several modules of wireless communication, data processing and interface display.
The wireless communication is mainly responsible for BLE Bluetooth communication with equipment, and after data are obtained through Bluetooth, the data are analyzed according to an agreed protocol.
The data processing part comprises two parts of data preprocessing and algorithm model prediction, and the steps and the method are designed in chapter four, mainly used for preprocessing the data and then calculating the systolic pressure and the diastolic pressure by using the designed algorithm model.
The interface display part mainly comprises two parts, namely, electrocardio and pulse wave data analyzed by the wireless communication part are converted into visual signal waveforms, and a user can visually observe an electrocardiogram and a piezoelectric pulse wave diagram. The other part is to display the calculated systolic and diastolic blood pressure data.
The terminal software is written by utilizing a Qt programming tool, the maximum advantage is that the terminal software can span multiple platforms and is beneficial to being transplanted on different platforms, and the programming language of Qt is mainly based on C + + language and can also directly run functions and files written by Python language. The Qt contains many libraries, including a Bluetooth library, a chart library, etc., which can be called conveniently, and the interface is shown in FIG. 10.
The method utilizes a convolutional neural network to automatically extract the characteristics of piezoelectric pulse waves and electrocardio pulse waves, mixes PTT and HR parameters which are proved to have high correlation with blood pressure and serves as the input of an algorithm model, then measures systolic pressure and diastolic pressure based on a deep learning model of a bidirectional long-and-short time memory neural network in order to better consider the dependence of continuous monitoring of signals in the time domain direction, can improve the accuracy of blood pressure measurement, and is suitable for measuring crowds in a wider range. Meanwhile, the intelligent bracelet comprises the lower computer equipment and the terminal software, the lower computer equipment can acquire the needed electrocardiosignals and piezoelectric pulse wave signals and wirelessly transmit the electrocardiosignals and the piezoelectric pulse wave signals to the terminal software by utilizing Bluetooth, the terminal software acquires the acquired signal data, the blood pressure is measured based on the method to acquire and display the blood pressure measurement data, and meanwhile, the shape of the signal waveform can be displayed in real time.

Claims (10)

1. A blood pressure monitoring method based on electrocardio-piezoelectric pulse wave coupling is characterized by mainly comprising the following steps:
a) acquiring a piezoelectric pulse wave signal and an electrocardiosignal;
b) preprocessing the piezoelectric pulse wave signals and the electrocardiosignals to obtain a signal sequence, and extracting characteristics;
c) calculating pulse transmission time PTT and heart rate HR according to the characteristics, and segmenting the signal sequence;
d) and establishing a CNN-LSTM model based on mixed characteristics, training the model by using the segmented signal sequence, and measuring the blood pressure by using the trained model.
2. The method for monitoring blood pressure based on electrocardio-piezoelectric pulse wave coupling as claimed in claim 1, wherein: in the step a), acquiring the piezoelectric pulse wave signals and the electrocardiosignals specifically adopts a polyvinylidene fluoride (PVDF) piezoelectric sensor and a conditioning circuit to acquire the pressure pulse wave signals, and adopts the electrocardiosensor and the conditioning circuit to acquire the electrocardiosignals, wherein the acquisition frequency is 200 Hz.
3. The method for monitoring blood pressure based on electrocardio-piezoelectric pulse wave coupling as claimed in claim 1, wherein: in the step b), filtering the electrocardiosignal by using a high-order Butterworth filter to perform band-pass filters with cut-off frequencies of 0.67Hz and 40Hz respectively, and then filtering by using a morphological filter to obtain an electrocardiosignal sequence; and then extracting characteristics by adopting a time and amplitude dual-threshold method, wherein the method comprises the steps of first-order difference, amplitude threshold screening and time threshold screening which are sequentially carried out, and the main wave peak of the piezoelectric pulse wave is obtained by processing and is used as the characteristics of the piezoelectric pulse wave.
4. The method for monitoring blood pressure based on electrocardio-piezoelectric pulse wave coupling as claimed in claim 1, wherein: in the step b), the piezoelectric pulse wave signals are subjected to band-pass filtering by using band-pass Butterworth filters with cut-off frequencies of 0.5 to 10Hz respectively, and then are filtered by using a morphological filter to obtain a piezoelectric pulse wave signal sequence; and then extracting by adopting a self-adaptive threshold method, wherein the self-adaptive threshold method comprises the steps of band-pass filtering, square linear amplification, window integration, low-pass filtering and R peak calibration which are sequentially carried out, and the R wave peak of the electrocardiosignal is obtained by processing and is used as the characteristic of the electrocardiosignal.
5. The method for monitoring blood pressure based on electrocardio-piezoelectric pulse wave coupling as claimed in claim 1, wherein: in the step c), the method specifically comprises the following steps:
c.1) calculating the time difference between the R wave crest of the electrocardiosignal and the main wave crest of the piezoelectric pulse wave signal as the pulse transmission time PTT according to the following formula:
Figure FDA0003145163960000011
in the formula, P [ i]Is the piezoelectric pulse of the ith main wave peak of the piezoelectric pulse wave signal after pretreatmentPosition number occupied in wave signal sequence, R [ i ]]The position serial number, F, of the ith R wave peak of the electrocardiosignal in the preprocessed electrocardiosignal sequenceSFor the sampling frequency, PTT [ i ]]PTT data of the ith pulse transmission time; n represents the total number of dominant/R wave peaks.
Calculating the interval time of two adjacent electrocardio R wave peaks according to the following formula to determine the heart rate HR:
Figure FDA0003145163960000021
wherein HR [ i ] is the ith heart rate HR data;
c.2) after the pulse transmission time PTT and the heart rate HR are obtained through calculation, segmenting the electrocardiosignal sequence and the piezoelectric pulse wave signal sequence, wherein the segmenting steps are mainly as follows:
selecting original data with fixed duration, wherein the original data are an electrocardiosignal sequence and a piezoelectric pulse wave signal sequence; dividing according to the main wave crest of the piezoelectric pulse wave signal, and dividing continuous data between every three pulse wave crests as a period, so as to divide the original data into a section of data section corresponding to each period; interpolating each data segment by one-dimensional second-order spline interpolation, normalizing after interpolation, connecting the interpolated and normalized data segments according to the time sequence of heartbeat cycle and the sequence of first electrocardio and second pulse wave to obtain each data sequence, and intercepting the first six data sequences a1-a6。
6. The method for monitoring blood pressure based on electrocardio-piezoelectric pulse wave coupling as claimed in claim 1, wherein: in the step d), the CNN-LSTM model based on the mixed characteristics comprises a CNN neural network and LSTM and fully-connected neural network models, the CNN neural network comprises two continuous convolution modules, and each convolution module is mainly formed by sequentially connecting a convolution layer, a discarding layer and a pooling layer; the LSTM and full-connection neural network model comprises a bidirectional long-short-time memory neural network (BiLSTM) module and a full-connection module (Dense), the bidirectional long-short-time memory neural network module is mainly formed by sequentially connecting a batch standardization layer, a bidirectional long-short-time memory network layer and a rejection layer, and the full-connection module is formed by sequentially connecting the full-connection layer, the rejection layer and the full-connection layer;
a obtained by processing the piezoelectric pulse wave signal and the electrocardio signal1To a6The CNN neural network is used as the input of the CNN neural network to extract blood pressure related characteristics, the blood pressure related characteristics and sequences of the electrocardiosignal sequence and the piezoelectric pulse wave signal sequence are mixed through a fusion layer and then input to the LSTM and full-connection neural network model to process and output signals of the systolic pressure SBP and the diastolic pressure DBP.
7. The method for monitoring blood pressure based on electrocardio-piezoelectric pulse wave coupling as claimed in claim 1, wherein: in the step d), the electrocardiosignals and the piezoelectric pulse wave signals which are actually measured are processed according to the steps b) to c), and then input into the trained model to be output, so that the measurement result is obtained.
8. The utility model provides a blood pressure check out test set based on electrocardiosignal and piezoelectricity pulse wave which characterized in that:
the system comprises:
the signal acquisition unit is used for acquiring electrocardiosignals and piezoelectric pulse wave signals and carrying out hardware conditioning on the signals to obtain relatively clean signals;
the data transmission unit is used for transmitting the electrocardiosignals and the piezoelectric pulse wave signals obtained by the hardware to the mobile phone terminal;
the mobile phone terminal is internally provided with a mobile phone software unit for receiving the electrocardio-piezoelectric signals obtained by the data transmission unit, processing the data according to the method of claim 7 and calculating to obtain a systolic pressure SBP and a diastolic pressure DBP.
9. The blood pressure detecting apparatus according to claim 8, wherein: the hardware conditioning mainly comprises low-pass filtering, high-pass filtering, an amplifying circuit and notch filtering.
10. The blood pressure detecting apparatus according to claim 8, wherein: and the data transmission unit transmits the original electrocardio and piezoelectric pulse wave digital signals to a mobile phone software unit in the mobile phone terminal in a BLE Bluetooth mode according to a formulated protocol.
CN202110748592.5A 2021-07-02 2021-07-02 Blood pressure monitoring method and device based on electrocardio-piezoelectric pulse wave coupling Withdrawn CN113558590A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110748592.5A CN113558590A (en) 2021-07-02 2021-07-02 Blood pressure monitoring method and device based on electrocardio-piezoelectric pulse wave coupling

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110748592.5A CN113558590A (en) 2021-07-02 2021-07-02 Blood pressure monitoring method and device based on electrocardio-piezoelectric pulse wave coupling

Publications (1)

Publication Number Publication Date
CN113558590A true CN113558590A (en) 2021-10-29

Family

ID=78163460

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110748592.5A Withdrawn CN113558590A (en) 2021-07-02 2021-07-02 Blood pressure monitoring method and device based on electrocardio-piezoelectric pulse wave coupling

Country Status (1)

Country Link
CN (1) CN113558590A (en)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114081462A (en) * 2021-11-19 2022-02-25 齐齐哈尔大学 Heart health monitoring system based on multi-dimensional physiological information
CN115568853A (en) * 2022-09-26 2023-01-06 山东大学 Psychological stress state assessment method and system based on picoelectric signals
CN116369882A (en) * 2023-01-13 2023-07-04 汉王科技股份有限公司 Blood pressure measurement method and device, blood pressure measurement equipment and electronic equipment
CN117045216A (en) * 2023-10-11 2023-11-14 泰州市榕兴医疗用品股份有限公司 Blood index non-invasive medical analysis method based on fuzzy control

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114081462A (en) * 2021-11-19 2022-02-25 齐齐哈尔大学 Heart health monitoring system based on multi-dimensional physiological information
CN115568853A (en) * 2022-09-26 2023-01-06 山东大学 Psychological stress state assessment method and system based on picoelectric signals
CN116369882A (en) * 2023-01-13 2023-07-04 汉王科技股份有限公司 Blood pressure measurement method and device, blood pressure measurement equipment and electronic equipment
CN117045216A (en) * 2023-10-11 2023-11-14 泰州市榕兴医疗用品股份有限公司 Blood index non-invasive medical analysis method based on fuzzy control
CN117045216B (en) * 2023-10-11 2024-04-02 泰州市榕兴医疗用品股份有限公司 Blood index non-invasive medical analysis system based on fuzzy control

Similar Documents

Publication Publication Date Title
CN113558590A (en) Blood pressure monitoring method and device based on electrocardio-piezoelectric pulse wave coupling
CN106725428B (en) Electrocardiosignal classification method and device
CN104382571B (en) A kind of measurement blood pressure method and device based on radial artery pulse wave conduction time
CN106821356B (en) Cloud continuous BP measurement method and system based on Elman neural network
CN111493850A (en) Blood pressure measuring method and device based on deep neural network
CN104363824A (en) Real time QRS duration measurement in electrocardiogram
CN201157353Y (en) Detection apparatus for blood pressure and bloodstream parameter
CN111528814A (en) Method for monitoring blood pressure through machine learning based on LSTM neural network
CN109965861A (en) Continuous monitoring device when a kind of wearable non-invasive blood pressure of no cuff is long
CN109793507A (en) It is a kind of based on finger pressure oscillographic method without oversleeve blood pressure measuring device and measurement method
CN111973165B (en) Linear and nonlinear mixed non-invasive continuous blood pressure measuring system based on PPG
CN106073783B (en) Method for extracting respiration rate from photoplethysmography wave
CN111839488B (en) Non-invasive continuous blood pressure measuring device and method based on pulse wave
CN110881967A (en) Non-invasive multi-segment peripheral arterial vessel elastic function detection method and instrument thereof
Argha et al. A novel automated blood pressure estimation algorithm using sequences of Korotkoff sounds
Nidigattu et al. Non-invasive modeling of heart rate and blood pressure from a photoplethysmography by using machine learning techniques
CN107995981B (en) Data processing method for blood pressure measuring device
CN110353704B (en) Emotion evaluation method and device based on wearable electrocardiogram monitoring
CN111345801A (en) Human body beat-by-beat heart rate measuring device and method based on particle filtering
Hamad et al. Remote ECG signal monitoring and classification based on Arduino with AD8232 sensor
CN114145725A (en) PPG sampling rate estimation method based on noninvasive continuous blood pressure measurement
WO2022094317A1 (en) Continuous, and cuff-less measurement of blood pressure by pulse wave acquisition and analysis using non-invasive sensors
Jokić et al. An efficient ECG modeling for heartbeat classification
Bhoi et al. Real time acquisition and analysis of PCG and PPG signals
CN111904403A (en) Blood pressure measuring system, blood pressure measuring method, computer device and storage medium

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
WW01 Invention patent application withdrawn after publication

Application publication date: 20211029

WW01 Invention patent application withdrawn after publication