CN113576438B - Non-invasive blood pressure extraction method and system - Google Patents
Non-invasive blood pressure extraction method and system Download PDFInfo
- Publication number
- CN113576438B CN113576438B CN202111033773.6A CN202111033773A CN113576438B CN 113576438 B CN113576438 B CN 113576438B CN 202111033773 A CN202111033773 A CN 202111033773A CN 113576438 B CN113576438 B CN 113576438B
- Authority
- CN
- China
- Prior art keywords
- ppg
- ppg signal
- signal
- period
- blood pressure
- 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.)
- Active
Links
- 230000036772 blood pressure Effects 0.000 title claims abstract description 80
- 238000000605 extraction Methods 0.000 title claims abstract description 31
- 230000000630 rising effect Effects 0.000 claims abstract description 15
- 230000003595 spectral effect Effects 0.000 claims abstract description 14
- 238000000034 method Methods 0.000 claims description 27
- 238000012549 training Methods 0.000 claims description 27
- 238000001228 spectrum Methods 0.000 claims description 20
- 230000035487 diastolic blood pressure Effects 0.000 claims description 19
- 230000035488 systolic blood pressure Effects 0.000 claims description 19
- 238000001514 detection method Methods 0.000 claims description 17
- 238000012545 processing Methods 0.000 claims description 17
- 238000001914 filtration Methods 0.000 claims description 16
- 238000007637 random forest analysis Methods 0.000 claims description 13
- 238000012417 linear regression Methods 0.000 claims description 12
- 238000012706 support-vector machine Methods 0.000 claims description 12
- 238000013507 mapping Methods 0.000 claims description 8
- 238000004364 calculation method Methods 0.000 claims description 7
- 238000006243 chemical reaction Methods 0.000 claims description 3
- 238000011160 research Methods 0.000 abstract description 7
- 238000000354 decomposition reaction Methods 0.000 abstract description 4
- QYSXJUFSXHHAJI-YRZJJWOYSA-N vitamin D3 Chemical compound C1(/[C@@H]2CC[C@@H]([C@]2(CCC1)C)[C@H](C)CCCC(C)C)=C\C=C1\C[C@@H](O)CCC1=C QYSXJUFSXHHAJI-YRZJJWOYSA-N 0.000 description 15
- 230000008569 process Effects 0.000 description 9
- 238000010586 diagram Methods 0.000 description 5
- 238000009530 blood pressure measurement Methods 0.000 description 4
- 238000010276 construction Methods 0.000 description 4
- 238000002790 cross-validation Methods 0.000 description 4
- 230000006870 function Effects 0.000 description 4
- 108091008698 baroreceptors Proteins 0.000 description 3
- 239000008280 blood Substances 0.000 description 3
- 210000004369 blood Anatomy 0.000 description 3
- 238000012937 correction Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 238000012544 monitoring process Methods 0.000 description 3
- 210000001774 pressoreceptor Anatomy 0.000 description 3
- 238000005070 sampling Methods 0.000 description 3
- 238000010801 machine learning Methods 0.000 description 2
- 230000004089 microcirculation Effects 0.000 description 2
- 230000000737 periodic effect Effects 0.000 description 2
- 230000001360 synchronised effect Effects 0.000 description 2
- 238000012935 Averaging Methods 0.000 description 1
- 208000024172 Cardiovascular disease Diseases 0.000 description 1
- 206010008190 Cerebrovascular accident Diseases 0.000 description 1
- 208000017667 Chronic Disease Diseases 0.000 description 1
- 206010019280 Heart failures Diseases 0.000 description 1
- 208000006011 Stroke Diseases 0.000 description 1
- 210000001367 artery Anatomy 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000017531 blood circulation Effects 0.000 description 1
- 210000004204 blood vessel Anatomy 0.000 description 1
- 230000002490 cerebral effect Effects 0.000 description 1
- 208000026106 cerebrovascular disease Diseases 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 125000004122 cyclic group Chemical group 0.000 description 1
- 238000003066 decision tree Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 201000010099 disease Diseases 0.000 description 1
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 1
- 238000006073 displacement reaction Methods 0.000 description 1
- 230000002526 effect on cardiovascular system Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 208000015181 infectious disease Diseases 0.000 description 1
- 230000010361 irregular oscillation Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 208000010125 myocardial infarction Diseases 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 230000010355 oscillation Effects 0.000 description 1
- 230000036513 peripheral conductance Effects 0.000 description 1
- 238000013186 photoplethysmography Methods 0.000 description 1
- 238000005086 pumping Methods 0.000 description 1
- 230000033764 rhythmic process Effects 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 210000003462 vein Anatomy 0.000 description 1
- 210000000707 wrist Anatomy 0.000 description 1
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
- A61B5/02125—Measuring pressure in heart or blood vessels from analysis of pulse wave characteristics of pulse wave propagation time
-
- 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
- A61B5/024—Detecting, measuring or recording pulse rate or heart rate
- A61B5/02416—Detecting, measuring or recording pulse rate or heart rate using photoplethysmograph signals, e.g. generated by infrared radiation
-
- 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
- A61B5/024—Detecting, measuring or recording pulse rate or heart rate
- A61B5/02438—Detecting, measuring or recording pulse rate or heart rate with portable devices, e.g. worn by the patient
-
- 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
- A61B5/68—Arrangements of detecting, measuring or recording means, e.g. sensors, in relation to patient
- A61B5/6801—Arrangements of detecting, measuring or recording means, e.g. sensors, in relation to patient specially adapted to be attached to or worn on the body surface
- A61B5/6802—Sensor mounted on worn items
- A61B5/681—Wristwatch-type devices
-
- 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
- A61B5/7235—Details of waveform analysis
-
- 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
- A61B5/7253—Details of waveform analysis characterised by using transforms
- A61B5/7257—Details of waveform analysis characterised by using transforms using Fourier transforms
Abstract
The invention discloses a non-invasive blood pressure extraction method and system, which take a single-period PPG signal obtained by combining a plurality of period PPG signals as a research object, carry out frequency decomposition on the single-period PPG signal by fast Fourier transform extracted by a time axis, convert the problem of researching the relation between different PPG signals and blood pressure into the relation between the spectral amplitude and the phase of a research PPG signal waveform and the blood pressure, define the time interval between the peak value of an ECG signal and the point with the maximum gradient of the rising edge of the PPG signal as pulse wave arrival time PAT, the spectral amplitude and the phase of the pulse wave arrival time PAT and the PPG signal can completely represent the physical characteristics of a tested person, train the stabling model by taking the pulse wave arrival time PAT and the spectral amplitude and the phase of the PPG signal as the input of the stabling model, and then use the trained stabling model to carry out blood pressure prediction, thereby realizing accurate judgment on the blood pressure and solving the technical problems of larger error and low accuracy of the traditional non-invasive blood pressure extraction method.
Description
Technical Field
The invention relates to the technical field of blood pressure detection, in particular to a non-invasive blood pressure extraction method and system.
Background
Blood pressure is an important physiological parameter reflecting the functions of the circulatory system of a human body, can indirectly map the pumping function, peripheral vascular resistance, heart rhythm, aortic elasticity, whole blood volume, physical state of blood and the like of the heart, is the most common chronic disease, is also a disease related to a wider age stage, is the most important risk factor of cardiovascular and cerebrovascular diseases, is a main complication such as cerebral apoplexy, myocardial infarction, heart failure and the like, and has high disability and mortality rate. Therefore, it is of great importance to monitor the blood pressure of the human body.
The existing blood pressure measuring modes are divided into a invasive blood pressure measuring mode and a noninvasive blood pressure measuring mode, wherein the invasive blood pressure measuring mode is to puncture a blood vessel, place a catheter in a vein blood vessel or an artery blood vessel of a patient, sense the flow and impact of blood, generate pressure, then externally connect a baroreceptor to the catheter, and display the blood pressure of the patient in real time through the baroreceptor; the noninvasive blood pressure measurement mode is to inflate the cuff in calm states such as no movement of a patient, and vibration change of pipeline transmission can be sensed in real time through the baroreceptors, so that the blood pressure in that period of time can be detected, and the problem of the invasive blood pressure measurement mode can be avoided, but real-time monitoring cannot be realized. In order to solve the problems of easy infection, limited movement and incapability of real-time monitoring in the existing invasive blood pressure measurement mode and the non-invasive blood pressure measurement mode, the prior art researches an intelligent bracelet for extracting blood pressure signals by utilizing pulse wave propagation time (PTT), wherein the pulse wave propagation time uses synchronous Electrocardiographic (ECG) and photo-electric volume (PPG) sensors to measure wrist microcirculation waveforms, extracts the pulse wave propagation time, finds out the relation among diastolic pressure, systolic pressure and flight time by utilizing a machine learning method, and estimates the blood pressure of a monitored person. However, the method for extracting the blood pressure has larger error and the detection accuracy needs to be improved.
Disclosure of Invention
The invention provides a non-invasive blood pressure extraction method and a system, which are used for solving the technical problems of larger error and low accuracy of the existing non-invasive blood pressure extraction method.
In view of this, a first aspect of the present invention provides a method for non-invasive blood pressure extraction, comprising:
the PPG signal acquisition device and the ECG signal acquisition device on the intelligent bracelet synchronously acquire the PPG signal and the ECG signal of the tested person;
calculating the number of periods of the PPG signal, and combining the PPG signals with the preset number of periods;
performing fast Fourier transform on the PPG signals subjected to the period combination to obtain spectrum amplitude information and phase information of the PPG signals;
from the PPG signal and the ECG signal, a pulse wave arrival time PAT is calculated, wherein pat=ecg peak -PPG D-max ,ECG peak PPG for a point in time corresponding to the peak of the ECG signal located by a sliding window D-max A horizontal axis time point corresponding to a point with the maximum gradient on the rising edge of the PPG signal obtained through first-order forward difference;
the method comprises the steps of inputting spectrum amplitude information and phase information of a PPG signal and pulse wave arrival time PAT as training data sets into a Stacking model for training to obtain a well-trained Stacking model, wherein a first layer of the Stacking model integrates a random forest prediction model, a support vector machine prediction model and a LightGBM prediction model, and a second layer is a multiple linear regression model;
and predicting the systolic pressure and the diastolic pressure of the target blood pressure detection personnel through the trained Stacking model.
Optionally, after obtaining the PPG signal and the ECG signal, before calculating the number of periods of the PPG signal, the method further comprises:
the PPG signal is input into an FIR filter for filtering, and the high-frequency component of the dicrotic peak of the PPG signal is filtered.
Optionally, calculating the number of periods of the PPG signal, combining the PPG signals of the preset number of periods, including:
calculating the period number of the PPG signal, setting a sliding window with preset length according to the period number of the PPG signal, performing baseline drift removal processing on the PPG signal, and aligning the PPG signal baseline in a linear mapping mode;
and merging PPG signals of a preset number of periods, respectively calculating the distances between the baselines and the troughs of adjacent PPG signal periods by taking the trough of the dicrotic peak of each period as a baseline when merging, calculating a distance average value, taking the distance average value as the standard distance between the baselines and the starting point and the finishing point of a single period after period merging, and unifying the period lengths of the PPG signals measured in a given time.
Optionally, the band pass frequency of the FIR filter is 2Hz, the band reject frequency is 4Hz, and the stop band attenuation is 80dB.
Optionally, when the PPG signal is input to the FIR filter for filtering, a direct current component of 3s is added to the PPG signal to cancel the delay of the FIR filter.
Optionally, the preset length of the sliding window is 0.6t, t being an average period length calculated from the number of periods of the PPG signal.
In a second aspect, the present invention provides a non-invasive blood pressure extraction system comprising:
the signal acquisition module is used for synchronously acquiring the PPG signal and the ECG signal of the tested person through the PPG signal acquisition device and the ECG signal acquisition device on the intelligent bracelet;
the signal period processing module is used for calculating the period number of the PPG signals and combining the PPG signals with the preset period number;
the signal conversion module is used for carrying out fast Fourier transform on the PPG signals after the period combination to obtain spectrum amplitude information and phase information of the PPG signals;
a PAT calculation module for calculating pulse wave arrival time PAT according to the PPG signal and the ECG signal, wherein pat=ecg peak -PPG D-max ,ECG peak PPG for a point in time corresponding to the peak of the ECG signal located by a sliding window D-max A horizontal axis time point corresponding to a point with the maximum gradient on the rising edge of the PPG signal obtained through first-order forward difference;
the model training module is used for inputting the spectrum amplitude information and the phase information of the PPG signal and the pulse wave arrival time PAT as training data sets into a Stacking model for training to obtain a trained Stacking model, wherein a random forest prediction model, a support vector machine prediction model and a LightGBM prediction model are integrated on a first layer of the Stacking model, and a multiple linear regression model is arranged on a second layer of the Stacking model;
and the blood pressure detection module is used for predicting the systolic pressure and the diastolic pressure of the target detection blood pressure personnel through the trained Stacking model.
Optionally, the method further comprises:
and the filtering module is used for inputting the PPG signal into the FIR filter for filtering and filtering out the high-frequency component of the dicrotic peak of the PPG signal.
Optionally, the signal period processing module is specifically configured to:
calculating the period number of the PPG signal, setting a sliding window with preset length according to the period number of the PPG signal, performing baseline drift removal processing on the PPG signal, and aligning the PPG signal baseline in a linear mapping mode;
and merging PPG signals of a preset number of periods, respectively calculating the distances between the baselines and the troughs of adjacent PPG signal periods by taking the trough of the dicrotic peak of each period as a baseline when merging, calculating a distance average value, taking the distance average value as the standard distance between the baselines and the starting point and the finishing point of a single period after period merging, and unifying the period lengths of the PPG signals measured in a given time.
Optionally, the band pass frequency of the FIR filter is 2Hz, the band reject frequency is 4Hz, and the stop band attenuation is 80dB.
From the above technical solutions, the embodiment of the present invention has the following advantages:
the invention provides a non-invasive blood pressure extraction method, which comprises the following steps: the PPG signal acquisition device and the ECG signal acquisition device on the intelligent bracelet synchronously acquire the PPG signal and the ECG signal of the tested person; calculating the number of periods of the PPG signal, and combining the PPG signals with the preset number of periods; performing fast Fourier transform on the PPG signals subjected to the period combination to obtain spectrum amplitude information and phase information of the PPG signals; from the PPG signal and the ECG signal, a pulse wave arrival time PAT is calculated, wherein pat=ecg peak -PPG D-max ,ECG peak PPG for a point in time corresponding to the peak of the ECG signal located by a sliding window D-max When the rising edge of the PPG signal obtained by first-order forward difference is along the transverse axis corresponding to the point with the maximum gradientA point of separation; the method comprises the steps of inputting spectrum amplitude information and phase information of a PPG signal and pulse wave arrival time PAT as training data sets into a Stacking model for training to obtain a well-trained Stacking model, wherein a first layer of the Stacking model integrates a random forest prediction model, a support vector machine prediction model and a LightGBM prediction model, and a second layer is a multiple linear regression model; and predicting the systolic pressure and the diastolic pressure of the target blood pressure detection personnel through the trained Stacking model. According to the invention, a single-period PPG signal obtained by combining a plurality of period PPG signals is taken as a research object, frequency decomposition is carried out on the single-period PPG signal through fast Fourier transform extracted by a time axis, the problem of researching the relation between different PPG signals and blood pressure is converted into the relation between the spectral amplitude and the phase of the waveform of the PPG signal and the blood pressure, the time interval between the peak value of the ECG signal and the point with the maximum gradient of the rising edge of the PPG signal is defined as pulse wave arrival time PAT, the pulse wave arrival time PAT and the spectral amplitude and the phase of the PPG signal can completely represent the physical characteristics of a tested person, the pulse wave arrival time PAT and the spectral amplitude and the phase of the PPG signal are taken as the input of a Stacking model to train the Stacking model, and the trained Stacking model is used for carrying out blood pressure prediction, so that the accurate judgment on the blood pressure can be realized, and the technical problems of large error and low accuracy of the existing non-invasive blood pressure extraction method are solved.
Drawings
For a clearer description of embodiments of the invention or of solutions according to the prior art, the figures which are used in the description of the embodiments or of the prior art will be briefly described, it being obvious that the figures in the description below are only some embodiments of the invention, from which, without the aid of inventive efforts, other relevant figures can be obtained for a person skilled in the art.
FIG. 1 is a flow chart of a non-invasive blood pressure extraction method according to an embodiment of the present invention;
FIG. 2 is a schematic flow chart of a non-invasive blood pressure extraction method according to an embodiment of the present invention;
FIG. 3 is a diagram showing a typical misalignment of the photoplethysmogram PPG signal in locating the maximum value when it is shifted from baseline;
FIG. 4 is a schematic diagram showing the baseline drift of the photoplethysmogram PPG signal;
fig. 5 is a schematic diagram of the extraction of the pulse wave arrival time PAT;
fig. 6 is an effect diagram of performing fast fourier transform on a single photoplethysmogram PPG signal to obtain its spectral amplitude and phase;
fig. 7 is a structure diagram of a Stacking fusion model provided in an embodiment of the present invention.
Detailed Description
In order to make the present invention better understood by those skilled in the art, the following description will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings, and it is apparent that the described embodiments are only some embodiments of the present invention, not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Example 1
For ease of understanding, referring to fig. 1, an embodiment of a non-invasive blood pressure extraction method according to the present invention includes:
step 101, synchronously acquiring a PPG signal and an ECG signal of a tested person through a PPG signal acquisition device and an ECG signal acquisition device on the intelligent bracelet.
In the invention, firstly, the PPG signal and the ECG signal of a tested person are synchronously acquired through an intelligent bracelet with PPG signal acquisition and ECG signal acquisition functions.
Step 102, calculating the number of periods of the PPG signal, and combining the PPG signals with the preset number of periods.
For the PPG signals, the number of periods in a preset time period (for example, 5 s) is calculated, and the periodic PPG signals in the time period are combined to be used as a study object.
Step 103, performing fast fourier transform on the PPG signals after the period combination to obtain spectrum amplitude information and phase information of the PPG signals.
The spectral amplitude and phase of the PPG signal may represent complete signal information, so in the present invention, the PPG signal after the period combination is subjected to fast fourier transform to obtain the spectral amplitude information and phase information of the PPG signal, as shown in fig. 6.
Step 104, calculating pulse wave arrival time PAT according to the PPG signal and the ECG signal, wherein pat=ecg peak -PPG D-max ,ECG peak PPG for a point in time corresponding to the peak of the ECG signal located by a sliding window D-max Is the horizontal axis time point corresponding to the point where the rising edge of the PPG signal has the maximum gradient obtained by the first-order forward difference.
The invention separates the pulse wave arrival time PAT by combining the photoplethysmography PPG signal and the synchronous electrocardiogram ECG signal, wherein the pulse wave arrival time PAT is defined as the time interval between the points with maximum gradient between the ECG peak value and the PPG rising edge, namely PAT=ECG peak -PPG D-max The point with the maximum gradient of the PPG rising edge is obtained by first order forward difference while the ECG signal peak is located by a sliding window, as shown in fig. 5. In order to further improve the detection accuracy, the subject may measure the pulse wave arrival time PAT in a plurality of periods in a short time to perform correction processing, and the correction processing may be average correction.
Step 105, inputting the spectrum amplitude information and the phase information of the PPG signal and the pulse wave arrival time PAT as training data sets into a Stacking model for training to obtain a well-trained Stacking model, wherein a first layer of the Stacking model integrates a random forest prediction model, a support vector machine prediction model and a LightGBM prediction model, and a second layer is a multiple linear regression model.
The spectrum amplitude information and the phase information of the PPG signal and the pulse wave arrival time PAT are input as a training data set of a Stacking model, the Stacking model is based on the idea of integrated learning, 3 kinds of base models of a random forest prediction model, a support vector machine prediction model and a LightGBM prediction model are fused through a Stacking technology to serve as a first layer, a multiple linear regression model is used as a second layer of the Stacking model, and a secondary regression is conducted on a prediction result of the first layer, so that the prediction of diastolic pressure and systolic pressure is achieved, the non-invasive blood pressure extraction and monitoring targets are achieved, and the prediction of the Stacking model is completed. The structure of the Stacking model is shown in fig. 7.
The construction process of the Stacking model is as follows:
(1) Obtaining an original data set s= { (PAT n ,Xk n ,Phase n ),n=1,...,N},PAT n Xk is the nth pulse wave arrival time n For the spectral amplitude of the nth PPG signal, phase n For the phase of the nth PPG signal, the original dataset is divided into 10 subsets, equal in size, denoted S' = { S, using a 10-fold cross-validation method 1 ,S 2 ,...,S k }。
(2) After the first 10-fold cross-validation training, for each subset S', each subset was taken as a tester 10 times, and the other subset was taken as the first layer 3-basis learner (i.e., random forest prediction model, support vector machine prediction model, and LightGBM prediction model) training set, and the predicted values of the 3-basis learners were respectively denoted as random forest_pred_1, svm_pred_1, lightgbm_pred_1.
Specifically, regarding the construction process of the base model LightGBM prediction model, the data set is divided through 10-fold cross validation and randomly divided into a training set and a testing machine, and the fitting of the parameters is carried out on the LightGBM prediction modelAnd->Characterizing the degree of deviation of the fit of the systolic pressure, +.>The degree of fitting deviation of the diastolic blood pressure is characterized as follows:
in the method, in the process of the invention,the predicted value and the average value of the systolic blood pressure are respectively shown, SBP is the actual value of the systolic blood pressure in the training set,/>The predicted value and the average value of the diastolic blood pressure are respectively obtained, and DBP is the actual value of the diastolic blood pressure in the training set.
Respectively makeAnd->Minimum, in the parameter adjustment process, combining variable step-length cyclic traversal to obtain optimal parameters, setting different tree depths, leaf numbers, learning rates and feature selection probability parameters, and setting the range of the initial tree depth tree_depth to be [2,4,6,8 ]]The number of leaves num_leave is set to [20,30,40,50, 60]Setting the learning rate range as [0.01,0.02,0.03,0.04,0.05 ]]Set the feature selection probability as [0.4,0.5,0.6,0.7,0.8 ]]Obtaining optimal parameters in the process of traversing the leaf number and the learning rate, dividing the range of the optimal parameters and two adjacent points into 5 intervals at equal intervals based on the optimal parameters, performing second cycle traversal, and selecting R under the condition of preferentially reducing the leaf number and the maximum tree depth in the fitting process in order to prevent the overfitting caused by deeper decision trees in the splitting growth process 2 Values (i.e. degree of fitting deviation +.>And->) Minimum model parameters.
In the construction process of the base model random forest prediction model, each group of sub-training sets respectively trains a CART decision regression tree, the CART decision regression tree is used as the dividing basis of sub-nodes according to the minimum radix coefficient, the final prediction result of each CART regression tree is the average value of leaf nodes where the sample points reach, a large number of random trees are established to form a random forest, and the final prediction result of corresponding sample systolic pressure and diastolic pressure is the average value of all CART regression tree prediction results.
Regarding a support vector machine prediction model, taking relevant characteristics of signals such as PPG, PAT and the like as data input, and setting the epsilon value of a sensitivity coefficient to be 10 -2 Optimizing and selecting the penalty factor C and the RBF kernel function parameter sigma; and searching by using a cross-validation grid search method, and finally selecting the parameter combination which minimizes the training model error.
(3) Repeating the operation of the step (2), and so on, repeating the operation for N times in total, to obtain a prediction value set { lightgbm_pred_1, lightgbm_pred_2, }, a prediction value set { random forest_pred_1, random forest_pred_2, }, random forest_pred_n }, a prediction value set { svm_pred_1, svm_pred_2, svm_pred_n }, of the LightGBM model.
(4) And (3) taking the predicted values obtained by the three base models in the step (3) as a data set of a second layer of the Stacking model, wherein the characteristic dimension is 3. And taking the multiple linear regression model as a second layer of the Stacking model, and performing multiple linear regression on the new data set obtained for the three base models of the first layer of the Stacking model and the corresponding systolic pressure and diastolic pressure based on the least square method idea to obtain optimal regression parameters so as to realize the prediction of blood pressure.
Thus, the construction of the Stacking model is completed.
And 106, predicting the systolic pressure and the diastolic pressure of the target blood pressure detection personnel through the trained Stacking model.
The blood pressure condition of the target blood pressure detection personnel is extracted and monitored through the trained Stacking model, and similarly, the PPG signal and the ECG signal of the target blood pressure detection personnel are acquired through a PPG signal acquisition device and an ECG signal acquisition device on the intelligent bracelet, and according to the steps 102-104, the spectrum amplitude information and the phase information of the PPG signal and the pulse wave arrival time PAT are acquired and are input into the trained Stacking model, so that the diastolic pressure and the systolic pressure predicted by the trained Stacking model are obtained.
In microcirculation, pressure pulse represents flow, volume pulse wave represents pressure of blood flow, so that the photoelectric volume pulse wave PPG signal and blood pressure have strong correlation, and along with changes of physical conditions and age, waveforms of the photoelectric volume pulse wave PPG signal in a time domain and harmonic power spectrums in a frequency domain have obvious differences, so that pressure research through capacitance pulse wave has important significance, but precious PPG waveform information is lost in the current related research of predicting the blood pressure based on PPG and ECG signals, the blood pressure of a patient in calm state is required to be collected in advance before prediction, and the blood pressure prediction can be realized through parameter pre-fitting of the blood pressure of a small human body in calm state, so that the method has great inconvenience. According to the non-invasive blood pressure extraction method provided by the invention, a single-period PPG signal obtained by combining a plurality of period PPG signals is used as a research object, frequency decomposition is carried out on the single-period PPG signal through fast Fourier transform extracted by a time axis, the problem of researching the relation between different PPG signals and blood pressure is converted into the relation between the frequency spectrum amplitude and the phase of the PPG signal and the blood pressure, the time interval between the peak value of the ECG signal and the point with the maximum gradient of the rising edge of the PPG signal is defined as pulse wave arrival time PAT, the frequency spectrum amplitude and the phase of waveforms of the pulse wave arrival time PAT and the PPG signal can completely represent the physical characteristics of a tested person, the pulse wave arrival time PAT and the frequency spectrum amplitude and the phase of the PPG signal are used as the input of a Stacking model to train the Stacking model, and then the trained Stacking model is used for carrying out blood pressure prediction, so that accurate judgment on the blood pressure can be realized, and the technical problems of larger error and low accuracy of the existing non-invasive blood pressure extraction method are solved. Meanwhile, the invention does not need to collect the blood pressure of the patient in calm before prediction, does not need to perform parameter pre-fitting through the blood pressure of the small human body in calm state, can realize blood pressure prediction, and improves convenience.
Example 2
Referring to fig. 2, another embodiment of a method for non-invasive blood pressure extraction is provided in the present invention, comprising:
step 201, synchronously collecting a PPG signal and an ECG signal of a tested person through a PPG signal collecting device and an ECG signal collecting device on the intelligent bracelet.
Step 201 in this embodiment corresponds to step 101 in embodiment 1, and will not be described here again.
Step 202, inputting the PPG signal into an FIR filter for filtering, and filtering out the high-frequency component of the dicrotic peak of the PPG signal.
Step 203, calculating the number of periods of the PPG signal, and combining the PPG signals of the preset number of periods.
And 204, performing fast Fourier transform on the PPG signals after the period combination to obtain spectrum amplitude information and phase information of the PPG signals.
The discrete Fourier transform is performed on the PPG signal to find that the frequency of the PPG signal is mainly distributed within 5Hz, so that for a fixed-length photoelectric volume signal, the embodiment of the invention designs an FIR filter with the band-pass frequency of 2Hz, the band-stop frequency of 4Hz and the stop band attenuation of 80dB, and high-frequency parts such as a dicrotic peak in the PPG signal are filtered, so that the influence of the dicrotic peak on the calculation of the cycle number in the PPG signal with a certain time can be eliminated, and the calculation of the PPG signal cycle number is realized. However, the FIR filter can generate a time delay effect, and meanwhile, a direct current component of 3s is added to the PPG signal to offset the time delay of the FIR filter, so that the influence of the time delay on the calculation of the number of PPG signal periods at a given time is eliminated.
The baseline drift removal processing is shown in fig. 4, and the baseline of the PPG signal is aligned in a linear mapping mode, so that signal distortion caused by nonlinear mapping is avoided, and signal information is better protected. Since the original data is affected by the hardware device and thus the signal data has irregular oscillation, baseline drift removal processing is required to be performed on the PPG signal through a sliding window according to the number of periods, the average period length T is calculated based on the calculation result of the number of periods, and the sliding window size is set to 0.6T. The maximum value and the minimum value of the sequence in each window are found through sliding window movement, the difference between the maximum value and the minimum value in a specific time is calculated to be D, if the detected maximum value is larger than 3/4*D and the minimum value is smaller than 1/3*D, the detected maximum value and the detected minimum value can be preliminarily identified as the maximum value and the minimum value of the periodic sequence, and then the second screening is carried out to carry out final judgment, and the removed mispositioning condition is shown in fig. 3, so that the condition is often caused by that the detected maximum values and the detected minimum values of two adjacent sliding windows are smaller in the time axis direction, and in fact, the two adjacent maximum points are all the values near the same wave crest or wave trough, but not adjacent wave crests. Therefore, the invention calculates the time difference between the maximum value and the minimum value of the detected adjacent maximum values in pairs, if the time difference is smaller than 0.6T, the time difference is considered as the error positioning, if the time difference is the error positioning of the maximum value, the time difference is larger between the maximum value and the minimum value, and if the time difference is the minimum value, the time difference is the smaller between the maximum value and the minimum value. After the minimum point coordinates of each period are obtained, the base lines of the PPG signals are aligned in a linear mapping mode, the average coordinates of the trough longitudinal axis coordinates of each period are taken as the longitudinal axis coordinates of the base lines, the difference between the trough longitudinal axes at two ends of the same period is proportioned, the displacement amount of the offset is sequentially used as the point longitudinal axis direction lock on the period, and the baseline removing drift of the PPG signals is removed based on the operation line of sight, as shown in fig. 4.
And in order to avoid the loss of information of the dicrotic peak caused by cycle combination, when the PPG signals are subjected to cycle combination, the distances from the dicrotic peak trough in each cycle to the trough of the adjacent PPG signal cycle are respectively calculated and averaged, then the average value is used as the standard distance from the baseline to the starting point and the end point of a single cycle after the cycle combination, the cycle length of the PPG signals measured in a given time can be unified in a mode of combining interpolation or sampling, and finally the PPG signals subjected to cycle combination are subjected to fast Fourier transformation to obtain the frequency amplitude and the phase information of the PPG signals.
The invention realizes noise filtering of the PPG signals in a period combining mode, and the noise has randomness, and can not only influence the oscillation amplitude of the PPG signals, but also can influence the period width of the PPG signals to a certain extent, so that a simple low-pass filter or a band-pass filter can not well solve the noise problem, and therefore, the invention combines a plurality of PPG signals with the baseline drift removed in a period and averages the noise. Considering that small differences exist between adjacent periods and loss of the information of the dicrotic wave needs to be avoided in an averaging process, the trough position of the dicrotic wave is obtained through positioning in a forward differential mode, the distance between the trough position and the trough position of the adjacent PPG signals is calculated respectively and averaged respectively, the average value is taken as the standard distance between the initial point and the final point of a single period of the baseline after period combination, period normalization is sequentially carried out from the first PPG signal period, under the same sampling frequency, if the distance between the two ends of the trough position of the dicrotic wave in the period is greater than the standard distance, the distance length is compressed to the standard distance in a sampling mode, if the distance between the two ends of the trough position of the dicrotic wave in the period is less than the standard distance, the distance length is amplified to the standard distance in a three-spline difference mode, and denoising processing of the PPG signals can be realized through the step, and the single PPG signals after a plurality of periods are combined are obtained.
The single-period PPG signal obtained by combining a plurality of period PPG signals in about 5s of a tested person is taken as a study object, and the problem of researching the relation between different PPG waveforms and blood pressure is converted into the relation between the spectral amplitude, the phase and the blood pressure of the study PPG signal by carrying out frequency decomposition on the single-period PPG signal through fast Fourier transform of a time axis, as shown in figure 6, the frequency with higher energy is selected as the input for extracting the blood pressure according to the attenuation condition of the single-period PPG signal.
Step 205, calculating pulse wave arrival time PAT according to the PPG signal and the ECG signal, wherein pat=ecg peak -PPG D-max ,ECG peak PPG for a point in time corresponding to the peak of the ECG signal located by a sliding window D-max Is the horizontal axis time point corresponding to the point where the rising edge of the PPG signal has the maximum gradient obtained by the first-order forward difference.
In addition to processing the spectral amplitude and phase of the PPG signal at the input of the diastolic and systolic pressures, a pulse wave arrival time PAT signal is also introduced, which is defined as having the maximum on the ECG peak and PPG rising edgeThe time interval between points of gradient, as shown in fig. 5, the specific formula is defined as: pat=ecg peak -PPG D-max ,ECG peak PPG for a point in time corresponding to the peak of the ECG signal located by a sliding window D-max Is the horizontal axis time point corresponding to the point where the rising edge of the PPG signal has the maximum gradient obtained by the first-order forward difference. ECG peak detection based on PPG maximum detection, the thresholds of maximum and minimum are set to 2/3T and 1/3T respectively according to the ECG characteristics.
Step 206, inputting the spectrum amplitude information and the phase information of the PPG signal and the pulse wave arrival time PAT as training data sets into a Stacking model for training to obtain a trained Stacking model, wherein a first layer of the Stacking model integrates a random forest prediction model, a support vector machine prediction model and a LightGBM prediction model, and a second layer is a multiple linear regression model;
step 207, predicting the systolic pressure and the diastolic pressure of the target blood pressure detecting person through the trained Stacking model.
Based on the idea of integrated learning, the diastolic pressure and the systolic pressure are predicted by fusing a plurality of regression base models through a Stacking model. The Stacking model is divided into two layers, the first layer consists of a random forest prediction model, a support vector machine prediction model and a LightGBM prediction model, 3 machine learning models, as shown in fig. 7, PAT and PPG signal characteristics are learned and predicted, a prediction result is used as input data of the second layer, in order to reduce the risk of over fitting, the second layer carries out second regression on the data obtained by the first layer of the Stacking model through a simple multiple linear regression model, optimal regression parameters are obtained, and finally, the prediction of systolic pressure and diastolic pressure is realized.
Example 3
Embodiments of a non-invasive blood pressure extraction system are provided, comprising:
the signal acquisition module is used for synchronously acquiring the PPG signal and the ECG signal of the tested person through the PPG signal acquisition device and the ECG signal acquisition device on the intelligent bracelet;
the signal period processing module is used for calculating the period number of the PPG signals and combining the PPG signals with the preset period number;
the signal conversion module is used for carrying out fast Fourier transform on the PPG signals after the period combination to obtain spectrum amplitude information and phase information of the PPG signals;
a PAT calculation module for calculating pulse wave arrival time PAT according to the PPG signal and the ECG signal, wherein pat=ecg peak -PPG D-max ,ECG peak PPG for a point in time corresponding to the peak of the ECG signal located by a sliding window D-max A horizontal axis time point corresponding to a point with the maximum gradient on the rising edge of the PPG signal obtained through first-order forward difference;
the model training module is used for inputting the spectrum amplitude information and the phase information of the PPG signal and the pulse wave arrival time PAT as training data sets into a Stacking model for training to obtain a trained Stacking model, wherein a random forest prediction model, a support vector machine prediction model and a LightGBM prediction model are integrated on a first layer of the Stacking model, and a multiple linear regression model is arranged on a second layer of the Stacking model;
and the blood pressure detection module is used for predicting the systolic pressure and the diastolic pressure of the target detection blood pressure personnel through the trained Stacking model.
Further comprises:
and the filtering module is used for inputting the PPG signal into the FIR filter for filtering and filtering out the high-frequency component of the dicrotic peak of the PPG signal.
The signal period processing module is specifically used for:
calculating the period number of the PPG signal, setting a sliding window with preset length according to the period number of the PPG signal, performing baseline drift removal processing on the PPG signal, and aligning the PPG signal baseline in a linear mapping mode;
and merging PPG signals of a preset number of periods, respectively calculating the distances between the baselines and the troughs of adjacent PPG signal periods by taking the trough of the dicrotic peak of each period as a baseline when merging, calculating a distance average value, taking the distance average value as the standard distance between the baselines and the starting point and the finishing point of a single period after period merging, and unifying the period lengths of the PPG signals measured in a given time.
The band pass frequency of the FIR filter is 2Hz, the band reject frequency is 4Hz, and the stop band attenuation is 80dB.
The non-invasive blood pressure extraction system provided by the invention is used for executing the non-invasive blood pressure extraction system in the non-invasive blood pressure extraction method embodiment, and can obtain the same technical effects as those of the non-invasive blood pressure extraction method embodiment, and the detailed description is omitted.
The above embodiments are only for illustrating the technical solution of the present invention, and not for limiting the same; although the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims (8)
1. A method of non-invasive blood pressure extraction, comprising:
the PPG signal acquisition device and the ECG signal acquisition device on the intelligent bracelet synchronously acquire the PPG signal and the ECG signal of the tested person;
calculating the number of periods of the PPG signal, and combining the PPG signals with the preset number of periods;
performing fast Fourier transform on the PPG signals subjected to the period combination to obtain spectrum amplitude information and phase information of the PPG signals;
calculating pulse wave arrival time from PPG signal and ECG signalPATWherein, the method comprises the steps of, wherein,,/>for the point in time corresponding to the peak of the ECG signal located by means of a sliding window +.>A horizontal axis time point corresponding to a point with the maximum gradient on the rising edge of the PPG signal obtained through first-order forward difference;
spectral amplitude information and phase information of PPG signal and pulse wave arrival timePATInputting a Stacking model as a training data set for training to obtain a trained Stacking model, wherein a first layer of the Stacking model is a random forest prediction model, a support vector machine prediction model and a LightGBM prediction model, and a second layer of the Stacking model is a multiple linear regression model;
predicting the systolic pressure and the diastolic pressure of a target blood pressure detection person through a trained Stacking model;
calculating the number of periods of the PPG signal, combining the PPG signals of the preset number of periods, and comprising:
calculating the period number of the PPG signal, setting a sliding window with preset length according to the period number of the PPG signal, performing baseline drift removal processing on the PPG signal, and aligning the PPG signal baseline in a linear mapping mode;
and merging PPG signals of a preset number of periods, respectively calculating the distances between the baselines and the troughs of adjacent PPG signal periods by taking the trough of the dicrotic peak of each period as a baseline when merging, calculating a distance average value, taking the distance average value as the standard distance between the baselines and the starting point and the finishing point of a single period after period merging, and unifying the period lengths of the PPG signals measured in a given time.
2. The method of non-invasive blood pressure extraction according to claim 1, wherein after obtaining the PPG signal and the ECG signal, before calculating the number of periods of the PPG signal, further comprising:
the PPG signal is input into an FIR filter for filtering, and the high-frequency component of the dicrotic peak of the PPG signal is filtered.
3. The method of non-invasive blood pressure extraction according to claim 2, wherein the FIR filter has a bandpass frequency of 2Hz, a bandstop frequency of 4Hz, and a stopband attenuation of 80dB.
4. A non-invasive blood pressure extraction method according to claim 3, wherein when the PPG signal is input to the FIR filter for filtering, a direct current component of 3s is added to the PPG signal to cancel the delay of the FIR filter.
5. The method of claim 1, wherein the sliding window has a preset length of 0.6t, and t is an average period length calculated from the number of periods of the PPG signal.
6. A non-invasive blood pressure extraction system, comprising:
the signal acquisition module is used for synchronously acquiring the PPG signal and the ECG signal of the tested person through the PPG signal acquisition device and the ECG signal acquisition device on the intelligent bracelet;
the signal period processing module is used for calculating the period number of the PPG signals and combining the PPG signals with the preset period number;
the signal conversion module is used for carrying out fast Fourier transform on the PPG signals after the period combination to obtain spectrum amplitude information and phase information of the PPG signals;
a PAT calculation module for calculating pulse wave arrival time according to the PPG signal and the ECG signalPATWherein, the method comprises the steps of, wherein,,/>for the point in time corresponding to the peak of the ECG signal located by means of a sliding window +.>A horizontal axis time point corresponding to a point with the maximum gradient on the rising edge of the PPG signal obtained through first-order forward difference;
model training module for training spectral amplitude information and phase information of PPG signal, and pulse wave arrival timePATAs a training data setTraining in a Stacking model to obtain a trained Stacking model, wherein a first layer of the Stacking model is a random forest prediction model, a support vector machine prediction model and a LightGBM prediction model, and a second layer of the Stacking model is a multiple linear regression model;
the blood pressure detection module is used for predicting the systolic pressure and the diastolic pressure of the target detection blood pressure personnel through the trained Stacking model;
the signal period processing module is specifically used for:
calculating the period number of the PPG signal, setting a sliding window with preset length according to the period number of the PPG signal, performing baseline drift removal processing on the PPG signal, and aligning the PPG signal baseline in a linear mapping mode;
and merging PPG signals of a preset number of periods, respectively calculating the distances between the baselines and the troughs of adjacent PPG signal periods by taking the trough of the dicrotic peak of each period as a baseline when merging, calculating a distance average value, taking the distance average value as the standard distance between the baselines and the starting point and the finishing point of a single period after period merging, and unifying the period lengths of the PPG signals measured in a given time.
7. The non-invasive blood pressure extraction system of claim 6, further comprising:
and the filtering module is used for inputting the PPG signal into the FIR filter for filtering and filtering out the high-frequency component of the dicrotic peak of the PPG signal.
8. The non-invasive blood pressure extraction system of claim 7, wherein the FIR filter has a bandpass frequency of 2Hz, a bandstop frequency of 4Hz, and a stopband attenuation of 80dB.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111033773.6A CN113576438B (en) | 2021-09-03 | 2021-09-03 | Non-invasive blood pressure extraction method and system |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111033773.6A CN113576438B (en) | 2021-09-03 | 2021-09-03 | Non-invasive blood pressure extraction method and system |
Publications (2)
Publication Number | Publication Date |
---|---|
CN113576438A CN113576438A (en) | 2021-11-02 |
CN113576438B true CN113576438B (en) | 2024-03-05 |
Family
ID=78240918
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202111033773.6A Active CN113576438B (en) | 2021-09-03 | 2021-09-03 | Non-invasive blood pressure extraction method and system |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN113576438B (en) |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114027804A (en) * | 2021-12-03 | 2022-02-11 | 科思技术(温州)研究院 | Pulse condition diagnosis method, device and readable storage medium |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105943005A (en) * | 2016-06-01 | 2016-09-21 | 合肥芯福传感器技术有限公司 | Non-invasive blood pressure detection method based on mixing of photoelectric green-light pulses and electrocardiogram |
KR20190056858A (en) * | 2017-11-17 | 2019-05-27 | 가천대학교 산학협력단 | System and method for estimating blood pressure based on deep learning |
CN110141249A (en) * | 2019-06-18 | 2019-08-20 | 张远 | Woundless blood sugar monitoring method, system, equipment and medium based on PPG signal |
CN111973165A (en) * | 2020-08-14 | 2020-11-24 | 北京航空航天大学 | Linear and nonlinear mixed non-invasive continuous blood pressure measuring system based on PPG |
CN113242716A (en) * | 2019-02-01 | 2021-08-10 | 深圳市大耳马科技有限公司 | Method, equipment and system for monitoring arrhythmia event |
-
2021
- 2021-09-03 CN CN202111033773.6A patent/CN113576438B/en active Active
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105943005A (en) * | 2016-06-01 | 2016-09-21 | 合肥芯福传感器技术有限公司 | Non-invasive blood pressure detection method based on mixing of photoelectric green-light pulses and electrocardiogram |
KR20190056858A (en) * | 2017-11-17 | 2019-05-27 | 가천대학교 산학협력단 | System and method for estimating blood pressure based on deep learning |
CN113242716A (en) * | 2019-02-01 | 2021-08-10 | 深圳市大耳马科技有限公司 | Method, equipment and system for monitoring arrhythmia event |
CN110141249A (en) * | 2019-06-18 | 2019-08-20 | 张远 | Woundless blood sugar monitoring method, system, equipment and medium based on PPG signal |
CN111973165A (en) * | 2020-08-14 | 2020-11-24 | 北京航空航天大学 | Linear and nonlinear mixed non-invasive continuous blood pressure measuring system based on PPG |
Also Published As
Publication number | Publication date |
---|---|
CN113576438A (en) | 2021-11-02 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106413534B (en) | Continuous blood pressure measuring device, measuring model establishing method and system | |
CN106037694B (en) | A kind of continuous blood pressure measurer based on pulse wave | |
CN110537910B (en) | Coronary heart disease noninvasive screening system based on electrocardio and heart sound signal joint analysis | |
CN104382571B (en) | A kind of measurement blood pressure method and device based on radial artery pulse wave conduction time | |
CN111493850A (en) | Blood pressure measuring method and device based on deep neural network | |
CN103027690B (en) | Hypoperfusion oxyhemoglobin saturation measuring method based on self-correlation modeling method | |
CN109907752A (en) | A kind of cardiac diagnosis and monitoring method and system of the interference of removal motion artifacts and ecg characteristics detection | |
CN112089405B (en) | Pulse wave characteristic parameter measuring and displaying device | |
EP3133985B1 (en) | A method and a device for non invasive blood pressure measurement | |
CN110236508A (en) | A kind of non-invasive blood pressure continuous monitoring method | |
CN109124610A (en) | A kind of anti-interference method and device of non-invasive blood pressure measurement | |
CN103876733A (en) | System and method for phase synchronization analysis of cardiopulmonary system | |
CN107157461A (en) | Noninvasive continuous BP measurement method based on photoplethysmographic | |
CN110881967A (en) | Non-invasive multi-segment peripheral arterial vessel elastic function detection method and instrument thereof | |
CN113576438B (en) | Non-invasive blood pressure extraction method and system | |
CN110840428B (en) | Noninvasive blood pressure estimation method based on one-dimensional U-Net network | |
CN111973165A (en) | Linear and nonlinear mixed non-invasive continuous blood pressure measuring system based on PPG | |
CN111839488A (en) | Non-invasive continuous blood pressure measuring device and method based on pulse wave | |
CN104068841B (en) | A kind of measuring method and device measuring Indices of Systolic Time parameter | |
CN111297340A (en) | Movement state monitoring method based on combination of GPS and heart rate | |
US20090156949A1 (en) | Medical device with real-time physiological signal analysis function | |
Pan et al. | Improved blood pressure estimation using photoplethysmography based on ensemble method | |
CN108294736A (en) | Continuous BP measurement system and measurement method | |
CN111166322A (en) | Electrocardiosignal characteristic wave extraction method | |
CN113100735B (en) | Thrombus detection device and method based on venous blood flow graph |
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 | ||
GR01 | Patent grant | ||
GR01 | Patent grant |