CN113662538A - Non-invasive blood glucose detection method based on time-frequency domain comprehensive analysis - Google Patents
Non-invasive blood glucose detection method based on time-frequency domain comprehensive analysis Download PDFInfo
- Publication number
- CN113662538A CN113662538A CN202110973721.0A CN202110973721A CN113662538A CN 113662538 A CN113662538 A CN 113662538A CN 202110973721 A CN202110973721 A CN 202110973721A CN 113662538 A CN113662538 A CN 113662538A
- Authority
- CN
- China
- Prior art keywords
- analysis
- characteristic parameters
- blood glucose
- frequency domain
- time
- 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- Withdrawn
Links
- 239000008280 blood Substances 0.000 title claims abstract description 84
- 210000004369 blood Anatomy 0.000 title claims abstract description 84
- WQZGKKKJIJFFOK-GASJEMHNSA-N Glucose Natural products OC[C@H]1OC(O)[C@H](O)[C@@H](O)[C@@H]1O WQZGKKKJIJFFOK-GASJEMHNSA-N 0.000 title claims abstract description 44
- 239000008103 glucose Substances 0.000 title claims abstract description 44
- 238000004458 analytical method Methods 0.000 title claims abstract description 38
- 238000001514 detection method Methods 0.000 title claims abstract description 26
- 238000000034 method Methods 0.000 claims abstract description 24
- 238000010219 correlation analysis Methods 0.000 claims abstract description 12
- 238000002474 experimental method Methods 0.000 claims abstract description 6
- 230000002068 genetic effect Effects 0.000 claims abstract description 5
- 238000007410 oral glucose tolerance test Methods 0.000 claims abstract description 5
- 238000007621 cluster analysis Methods 0.000 claims abstract description 4
- 238000003062 neural network model Methods 0.000 claims abstract description 4
- 238000013186 photoplethysmography Methods 0.000 claims description 17
- 238000013528 artificial neural network Methods 0.000 claims description 9
- 230000003595 spectral effect Effects 0.000 claims description 9
- 238000012549 training Methods 0.000 claims description 9
- 238000012360 testing method Methods 0.000 claims description 7
- 230000029058 respiratory gaseous exchange Effects 0.000 claims description 4
- 238000012795 verification Methods 0.000 claims description 4
- 238000001914 filtration Methods 0.000 claims description 3
- 230000006870 function Effects 0.000 claims description 3
- 238000010801 machine learning Methods 0.000 claims description 3
- 230000035772 mutation Effects 0.000 claims description 3
- 238000007781 pre-processing Methods 0.000 claims description 3
- 238000000513 principal component analysis Methods 0.000 claims description 3
- 108090000623 proteins and genes Proteins 0.000 claims description 3
- 238000010187 selection method Methods 0.000 claims description 3
- 238000001228 spectrum Methods 0.000 claims description 3
- 238000010200 validation analysis Methods 0.000 claims description 2
- 238000010183 spectrum analysis Methods 0.000 abstract 1
- 238000010586 diagram Methods 0.000 description 6
- 239000000126 substance Substances 0.000 description 4
- 230000003287 optical effect Effects 0.000 description 3
- XUMBMVFBXHLACL-UHFFFAOYSA-N Melanin Chemical compound O=C1C(=O)C(C2=CNC3=C(C(C(=O)C4=C32)=O)C)=C2C4=CNC2=C1C XUMBMVFBXHLACL-UHFFFAOYSA-N 0.000 description 2
- 238000002795 fluorescence method Methods 0.000 description 2
- 230000031700 light absorption Effects 0.000 description 2
- 238000005259 measurement Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 210000001519 tissue Anatomy 0.000 description 2
- 238000004497 NIR spectroscopy Methods 0.000 description 1
- 238000001069 Raman spectroscopy Methods 0.000 description 1
- 230000005856 abnormality Effects 0.000 description 1
- 210000002159 anterior chamber Anatomy 0.000 description 1
- 210000001742 aqueous humor Anatomy 0.000 description 1
- 235000013405 beer Nutrition 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 150000001875 compounds Chemical class 0.000 description 1
- 238000002790 cross-validation Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000003111 delayed effect Effects 0.000 description 1
- 206010012601 diabetes mellitus Diseases 0.000 description 1
- 239000003814 drug Substances 0.000 description 1
- 229940079593 drug Drugs 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000003912 environmental pollution Methods 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 239000012530 fluid Substances 0.000 description 1
- 230000005182 global health Effects 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 208000015181 infectious disease Diseases 0.000 description 1
- 208000030159 metabolic disease Diseases 0.000 description 1
- 238000004476 mid-IR spectroscopy Methods 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 230000035515 penetration Effects 0.000 description 1
- 238000004867 photoacoustic spectroscopy Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
Images
Classifications
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/145—Measuring characteristics of blood in vivo, e.g. gas concentration, pH value; Measuring characteristics of body fluids or tissues, e.g. interstitial fluid, cerebral tissue
- A61B5/14532—Measuring characteristics of blood in vivo, e.g. gas concentration, pH value; Measuring characteristics of body fluids or tissues, e.g. interstitial fluid, cerebral tissue for measuring glucose, e.g. by tissue impedance measurement
-
- 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
- A61B5/0205—Simultaneously evaluating both cardiovascular conditions and different types of body conditions, e.g. heart and respiratory condition
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/145—Measuring characteristics of blood in vivo, e.g. gas concentration, pH value; Measuring characteristics of body fluids or tissues, e.g. interstitial fluid, cerebral tissue
- A61B5/1455—Measuring characteristics of blood in vivo, e.g. gas concentration, pH value; Measuring characteristics of body fluids or tissues, e.g. interstitial fluid, cerebral tissue using optical sensors, e.g. spectral photometrical oximeters
-
- 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
- A61B5/7264—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems
-
- 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
Landscapes
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Animal Behavior & Ethology (AREA)
- Pathology (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Heart & Thoracic Surgery (AREA)
- Medical Informatics (AREA)
- Molecular Biology (AREA)
- Surgery (AREA)
- Veterinary Medicine (AREA)
- Public Health (AREA)
- Signal Processing (AREA)
- Physiology (AREA)
- Artificial Intelligence (AREA)
- Psychiatry (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Cardiology (AREA)
- Optics & Photonics (AREA)
- Power Engineering (AREA)
- Emergency Medicine (AREA)
- Evolutionary Computation (AREA)
- Fuzzy Systems (AREA)
- Mathematical Physics (AREA)
- Spectroscopy & Molecular Physics (AREA)
- Pulmonology (AREA)
- Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)
Abstract
The invention discloses a non-invasive blood sugar detection method based on time-frequency domain comprehensive analysis, which comprises the following steps: 1) acquiring a fingertip photoplethysmographic pulse wave signal, and simultaneously carrying out an OGTT experiment to acquire and detect blood glucose concentration; 2) denoising the photoplethysmographic signals by using a wavelet threshold method and a curve fitting method; 3) extracting a representative waveform using cluster analysis; 4) extracting characteristic parameters with physiological significance by time domain analysis; 5) carrying out spectrum analysis on the signal by adopting fast Fourier transform, and advancing frequency domain characteristic parameters; 6) extracting time domain and frequency domain characteristic parameters related to blood sugar through correlation analysis; 7) establishing a genetic algorithm by using the acquired characteristic parameters and the real blood glucose value, and optimizing a BP neural network model; 8) the error between the blood glucose value and the actual blood glucose value is predicted through Parkers error grid analysis, wherein 80.3 percent of prediction results are in a clinically accurate area in an area A, and 19.7 percent of prediction results are in a clinically acceptable area in an area B.
Description
Technical Field
The invention belongs to the technical field of biological signal processing, and particularly relates to a non-invasive blood glucose detection method based on time-frequency domain comprehensive analysis.
Background
Diabetes is one of the fastest growing global health emergencies in the 21 st century, and the cause of death is a number of complications, usually caused by metabolic disorders. Currently, there is no cure method, and the main treatment method is to frequently detect blood sugar and control blood sugar level with drug assistance. In the detection of blood glucose, invasive detection causes pain with a concomitant risk of infection. Secondly, purchase disposable test paper and blood taking needle for a long time, not only increase the detection cost and can cause environmental pollution, in addition, unable continuous blood sugar that detects still can easily miss effective information. The minimally invasive method has the disadvantages that the change of the glucose concentration in the tissue fluid is slightly delayed relative to the change of the glucose concentration in the blood, and the position is easy to deviate, thereby bringing about measurement errors.
At present, noninvasive detection is a hot point of research in international academia at present, and an optical method is more applied to the aspect of noninvasive blood glucose detection technology. The detection position of the Raman spectroscopy is the anterior chamber of the eye, so that the incident light intensity is limited, and the influence of fluorescence interference of the skin and surface melanin needs to be solved; the detection position of the polarized light optical rotation method is an eye, the problem of measurement safety exists, and the concentration of aqueous humor in front of the eye and the concentration of blood glucose in blood have delay performance; photoacoustic spectroscopy is susceptible to interference from the environment of use and the stability of the instrument, which is stronger than that caused by fluctuations in blood glucose concentration; the scattering phenomenon of the fluorescence method has a large influence on the fluorescence method, and the skin color and the skin thickness can influence the fluorescence effect; the optical coherence imaging method is sensitive to physiological changes, tissue composition changes and motion abnormalities; mid-infrared spectroscopy has poor skin penetration and is susceptible to background interference from other compounds; the near infrared spectroscopy has a weak signal and is greatly influenced by physiological background and individual difference.
Disclosure of Invention
In order to overcome the defects of the prior art, the invention provides a non-invasive blood sugar detection method based on time-frequency domain comprehensive analysis, which combines a photoplethysmography and a near-infrared detection technology, performs time-frequency domain comprehensive analysis on photoplethysmography, extracts characteristic parameters related to blood sugar change for modeling, predicts the blood sugar value by processing a pulse wave signal and performs error analysis. The detection method has the characteristics of no wound, small error and high precision.
In order to achieve the purpose, the invention adopts the technical scheme that:
a non-invasive blood sugar detection method based on time-frequency domain comprehensive analysis comprises the following steps:
step 3, extracting representative waveforms from the pure photoplethysmography signals of a plurality of periods obtained in the step 2 by using cluster analysis, and eliminating the influence of distorted waveforms on time domain and frequency domain analysis of the sphygmometer waves;
step 5, performing fast Fourier transform on the representative waveform obtained in the step 3 to obtain frequency spectrum information and frequency domain characteristic parameters, and extracting characteristic parameters related to blood sugar;
step 7, performing machine learning modeling by using the obtained characteristic parameters and the actually measured blood sugar values, establishing a genetic algorithm optimized BP neural network model, predicting the blood sugar concentration, taking the invasively collected blood sugar values as true values, and evaluating the prediction error of the model;
and 8, using international general Parkers grid error analysis to predict the error and clinical risk of the blood sugar value and the actual blood sugar value.
Further, 3 characteristic points and characteristic parameters with physiological significance are extracted in the step 4, and the ending time t of the single-cycle pulse wavea’Dominant wave amplitude hbMajor wave rise time tbAnd the start time t of the dicrotic waveeRatio t of2。
Further, the amplitudes of the frequencies around 1.17Hz and 2.15Hz in the step 5 are respectively expressed as Af1、Af2A isf1、Af2As the characteristic parameters extracted by frequency domain analysis, because the correlations between the frequency components of different individuals and the blood sugar fluctuate within a certain range, the first six peak frequencies with certain correlations with the blood sugar and the corresponding amplitudes thereof are extracted, the spectral peak information is analyzed by using principal component analysis, and the main information of 6 spectral peaks is represented by using two principal components PC1 and PC2 as the frequency domain characteristic parameters.
Further, the correlation analysis described in step 6 is used to determine the degree of correlation between two variables, the variables satisfy normal distribution, and the correlation analysis is performed on the extracted characteristic parameters and the true blood glucose value, and the formula is as follows:
wherein: x and Y are two variable factors for correlation analysis, and Cov (X and Y) is the covariance between X and Y variables; d (X), D (Y) is the variance of X and Y; the selected time domain characteristic parameters comprise: end time t of single-cycle pulse wavea’Dominant wave amplitude hbMajor wave rise time tbAnd the start time t of the dicrotic waveeRatio t of2(ii) a The frequency domain characteristic parameters are as follows: amplitude A of frequencies around 1.17Hzf1Amplitude A of a frequency around 2.15Hzf2Two principal components PC1, PC2 representing spectral peak information.
Further, the specific implementation of step 7 is:
adjusting and optimizing the initial weight value, the threshold value and the coding mode of the neural network, calculating the fitness value of each individual by taking the absolute error sum of model prediction as the fitness value, selecting an operation selection proportion selection method, forming a new population by using individuals with better fitness, obtaining a new individual by adopting a real number intersection method, selecting a certain gene of the individual for mutation to obtain a new individual, judging whether the result meets a constraint condition or not, returning to the step of determining the fitness function if the result does not meet the constraint condition, recording the optimal individual fitness if the result meets the constraint condition, then using the optimal individual generated by evolution as the initial weight value and the threshold value of the neural network to predict the blood glucose concentration, using the blood glucose value collected initiatively as a true value, evaluating the model prediction error, using cross-folding verification to determine the model parameters, and dividing 355 groups of data into 3: 1: 1, using 213 training sets, 71 validation sets and 71 test sets to train models, putting 71 test sets into the model with the minimum training error, and analyzing the model prediction result.
Compared with the prior art, the invention has the following beneficial effects:
the invention provides a non-invasive blood sugar detection method based on time-frequency domain comprehensive analysis, which is a new blood sugar detection method. Combining photoplethysmography and near infrared detection technology, so that transmitted light passing through the fingertip contains substance information in blood; the OGTT experiment is carried out, so that the concentration change of glucose in blood is obvious and the concentrations of other substances are kept unchanged in a period of time, and the influence of other substances in the blood on light absorption is avoided; and the influence of external factors on the absorption of light is avoided as much as possible through time and frequency comprehensive analysis. Data of 85 volunteers are collected in the experiment, and 355 groups of data are analyzed and processed, so that the established model has universality.
Drawings
FIG. 1 is a flow chart of a non-invasive blood glucose detection method based on time-frequency domain comprehensive analysis according to the present invention.
Fig. 2 is a waveform of the photoplethysmography denoised in step 2 in the noninvasive blood glucose detection method based on time-frequency domain comprehensive analysis according to the present invention.
Fig. 3 is a waveform of the photoplethysmography including distorted waveforms preprocessed in step 2 in the noninvasive blood glucose sensing method based on time-frequency domain comprehensive analysis according to the present invention.
FIG. 4 is a schematic diagram of characteristic points of the time domain analysis of the photoplethysmography pulse wave in the non-invasive blood glucose detection method based on the time-frequency domain comprehensive analysis.
FIG. 5 is a schematic diagram of a frequency domain analysis of photoplethysmography in a non-invasive blood glucose detecting method based on time-frequency domain analysis.
FIG. 6 is a flowchart of the BP neural network optimized by the genetic algorithm in the present invention.
FIG. 7 is a diagram illustrating the training result of the BP neural network according to the present invention.
FIG. 8 is a diagram of an error analysis of a blood glucose prediction result according to the present invention.
Detailed Description
The present invention will be described in further detail with reference to the accompanying drawings and specific embodiments.
The collection of the photoplethysmography is based on the lambert beer law, and the emergent intensity of light passing through a certain medium is influenced by the concentration of a light-absorbing substance and the thickness of the medium. The preliminarily acquired pulse wave signals contain noises such as myoelectricity interference, motion artifact, respiration, power frequency interference and the like.
As shown in fig. 1, a non-invasive blood glucose detecting method based on time-frequency domain comprehensive analysis includes the following steps:
step 3, extracting a representative waveform from the pure photoplethysmography signals of a plurality of periods obtained in the step 2 by using cluster analysis, and eliminating the influence of the distorted waveform on the analysis of the photoplethysmography time domain and the frequency domain as shown in the figure 3;
step 5, performing fast Fourier transform on the photoplethysmogram waveform obtained in the step 3 to obtain frequency spectrum information and frequency domain characteristic parameters; characteristic parameters relevant to blood sugar are extracted. As shown in FIG. 5, the amplitudes of frequencies around 1.17Hz and 2.15Hz are denoted Af1、Af2A isf1、Af2As a characteristic parameter extracted by frequency domain analysis. Because the correlations between the frequency components of different individuals and the blood sugar fluctuate within a certain range, the first six peak frequencies with certain correlation with the blood sugar and the corresponding amplitudes thereof are extracted, the spectral peak information is analyzed by using principal component analysis, and the two principal components PC1 and PC2 are used as frequency domain characteristic parameters to represent the main information of 6 spectral peaks;
and 6, performing correlation analysis on the extracted characteristic parameters and the real blood glucose values, finding that the extracted characteristic parameters and the real blood glucose values meet normal distribution, and showing that the extracted characteristic parameters are closely related to the real blood glucose values and can be used for establishing a subsequent blood glucose model. The specific formula of the correlation analysis is as follows:
wherein: x and Y are two variable factors for correlation analysis, and Cov (X and Y) is the covariance between X and Y variables; d (X), D (Y) is the variance of X and Y. The selected time domain characteristic parameters comprise: end time t of single-cycle pulse wavea’Dominant wave amplitude hbMajor wave rise time tbAnd the start time t of the dicrotic waveeRatio t of2(ii) a The frequency domain characteristic parameters are as follows: amplitude A of frequencies around 1.17Hzf1Amplitude A of a frequency around 2.15Hzf2Two principal components PC1, PC2 representing spectral peak information;
and 7, performing machine learning modeling by using the obtained characteristic parameters and the actually measured blood sugar value, establishing a genetic algorithm optimized BP neural network model, adjusting and optimizing the initial weight, the threshold and the coding mode of the neural network as shown in FIG. 6, calculating the fitness value of each individual by taking the sum of absolute errors predicted by the model as the fitness value, selecting an operation selection proportion selection method, and forming a new population by the individuals with better fitness. And obtaining a new individual by adopting a real number crossing method, and then selecting a certain gene of a certain individual for mutation to obtain the new individual. Judging whether the result meets the constraint condition, if not, returning to a determined fitness function, if so, recording the fitness of the optimal individual, then taking the optimal individual generated by evolution as the initial weight and the threshold value of a neural network to predict the blood glucose concentration, taking the invasively collected blood glucose value as the true value, evaluating the prediction error of the model, determining the model parameters by using cross validation of ten folds, and dividing 355 groups of data into 3: 1: 1, training models of 213 groups of training sets, 71 groups of verification sets and 71 groups of test sets by using the training sets and the verification sets, putting the 71 groups of test sets into a model with the minimum training error to obtain the actual blood glucose value of the output result of the GA-BP neural network and the predicted blood glucose value of the model, wherein the actual blood glucose value and the predicted blood glucose value of the model are shown in figure 7, and analyzing the predicted result of the model;
and 8, using international general Parkers grid error analysis to predict the error and clinical risk of the blood sugar value and the actual blood sugar value, wherein 80.3 percent of prediction results are in a clinically accurate area in an area A and 19.7 percent of prediction results are in a clinically acceptable area in an area B, and the results show that the noninvasive blood sugar detection technology has feasibility.
While specific embodiments of the invention have been described above, it will be appreciated by those skilled in the art that this is by way of example only, and that the scope of the invention is defined by the appended claims. Various changes and modifications to these embodiments may be made by those skilled in the art without departing from the spirit and scope of the invention, and these changes and modifications are within the scope of the invention.
Claims (5)
1. A non-invasive blood sugar detection method based on time-frequency domain comprehensive analysis is characterized by comprising the following steps:
step 1, collecting human body pulse wave signals by using a finger-clipped photoelectric sensor, wherein the obtained photoplethysmographic pulse wave signals also contain noises such as myoelectricity interference, motion artifact, respiration, power frequency interference and the like, and meanwhile, carrying out an OGTT experiment, and collecting fingertip blood glucose concentration by using a portable glucometer;
step 2, preprocessing the photoplethysmography signals obtained in the step 1 by a wavelet threshold method and a curve fitting method, filtering noise, baseline drift and accidental errors, and obtaining relatively pure photoplethysmography signals;
step 3, extracting representative waveforms from the pure photoplethysmography signals of a plurality of periods obtained in the step 2 by using cluster analysis, and eliminating the influence of distorted waveforms on time domain and frequency domain analysis of the sphygmometer waves;
step 4, performing time domain analysis on the representative waveform obtained in the step 3, and extracting characteristic points and characteristic parameters with physiological significance; detecting an extreme point in the pulse wave by using a differential threshold method, determining the position range of the occurrence of an inflection point by setting a time threshold, determining the inflection point by combining the characteristics of a slope near the inflection point, and when the slope changes from negative to zero to positive, wherein the point with the fastest curvature change is the inflection point: namely the wave trough; conversely, when the slope goes from positive to zero to negative, the point where the curvature changes most rapidly is the inflection point: namely the wave crest;
step 5, performing fast Fourier transform on the representative waveform obtained in the step 3 to obtain frequency spectrum information and frequency domain characteristic parameters, and extracting characteristic parameters related to blood sugar;
step 6, carrying out correlation analysis on the extracted characteristic parameters and the real blood sugar value;
step 7, performing machine learning modeling by using the obtained characteristic parameters and the actually measured blood sugar values, establishing a genetic algorithm optimized BP neural network model, predicting the blood sugar concentration, taking the invasively collected blood sugar values as true values, and evaluating the prediction error of the model;
and 8, using international general Parkers grid error analysis to predict the error and clinical risk of the blood sugar value and the actual blood sugar value.
2. The noninvasive blood glucose detection method of claim 1, wherein 3 physiologically significant feature points and parameters are extracted in step 4, the end time t of monocycle pulse wave isa’Dominant wave amplitude hbMajor wave rise time tbAnd the start time t of the dicrotic waveeRatio t of2。
3. The method of claim 1, wherein the amplitudes of frequencies around 1.17Hz and 2.15Hz in step 5 are respectively denoted as Af1、Af2A isf1、Af2As the characteristic parameters extracted by frequency domain analysis, because the correlations between the frequency components of different individuals and the blood sugar fluctuate within a certain range, the first six peak frequencies with certain correlations with the blood sugar and the corresponding amplitudes thereof are extracted, the spectral peak information is analyzed by using principal component analysis, and the main information of 6 spectral peaks is represented by using two principal components PC1 and PC2 as the frequency domain characteristic parameters.
4. The noninvasive blood glucose detection method based on time-frequency domain comprehensive analysis of claim 1, wherein the correlation analysis of step 6 is used to determine the degree of correlation between two variables, the variables satisfy normal distribution, and the correlation analysis is performed on the extracted characteristic parameters and the true blood glucose value, and the formula is as follows:
wherein: x and Y are two variable factors for correlation analysis, and Cov (X and Y) is the covariance between X and Y variables; d (X), D (Y) is the variance of X and Y; the selected time domain characteristic parameters comprise: end time t of single-cycle pulse wavea’Dominant wave amplitude hbMajor wave rise time tbAnd the start time t of the dicrotic waveeRatio t of2(ii) a The frequency domain characteristic parameters are as follows: amplitude A of frequencies around 1.17Hzf1Amplitude A of a frequency around 2.15Hzf2Two principal components PC1, PC2 representing spectral peak information.
5. The non-invasive blood glucose detecting method based on time-frequency domain comprehensive analysis according to claim 1, wherein the specific implementation of the step 7 is:
adjusting and optimizing the initial weight value, the threshold value and the coding mode of the neural network, calculating the fitness value of each individual by taking the absolute error sum of model prediction as the fitness value, selecting an operation selection proportion selection method, forming a new population by using individuals with better fitness, obtaining a new individual by adopting a real number intersection method, selecting a certain gene of the individual for mutation to obtain a new individual, judging whether the result meets a constraint condition or not, returning to the step of determining the fitness function if the result does not meet the constraint condition, recording the optimal individual fitness if the result meets the constraint condition, then using the optimal individual generated by evolution as the initial weight value and the threshold value of the neural network to predict the blood glucose concentration, using the blood glucose value collected initiatively as a true value, evaluating the model prediction error, using cross-folding verification to determine the model parameters, and dividing 355 groups of data into 3: 1: 1, using 213 training sets, 71 validation sets and 71 test sets to train models, putting 71 test sets into the model with the minimum training error, and analyzing the model prediction result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110973721.0A CN113662538A (en) | 2021-08-24 | 2021-08-24 | Non-invasive blood glucose detection method based on time-frequency domain comprehensive analysis |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110973721.0A CN113662538A (en) | 2021-08-24 | 2021-08-24 | Non-invasive blood glucose detection method based on time-frequency domain comprehensive analysis |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113662538A true CN113662538A (en) | 2021-11-19 |
Family
ID=78545561
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110973721.0A Withdrawn CN113662538A (en) | 2021-08-24 | 2021-08-24 | Non-invasive blood glucose detection method based on time-frequency domain comprehensive analysis |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113662538A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114528888A (en) * | 2022-04-25 | 2022-05-24 | 广东玖智科技有限公司 | PPG signal clustering center acquisition method and device and PPG signal processing method and device |
| CN115472222A (en) * | 2022-11-02 | 2022-12-13 | 杭州链康医学检验实验室有限公司 | Single cell transcriptome RNA pollution identification method, medium and equipment |
| CN116559143A (en) * | 2023-05-15 | 2023-08-08 | 西北大学 | Method and system for analyzing composite Raman spectrum data of glucose component in blood |
| CN116803344A (en) * | 2023-07-27 | 2023-09-26 | 迈德医疗科技(深圳)有限公司 | Blood glucose classification method and system based on multi-norm clustering and double-layer discrete network |
-
2021
- 2021-08-24 CN CN202110973721.0A patent/CN113662538A/en not_active Withdrawn
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114528888A (en) * | 2022-04-25 | 2022-05-24 | 广东玖智科技有限公司 | PPG signal clustering center acquisition method and device and PPG signal processing method and device |
| CN114528888B (en) * | 2022-04-25 | 2022-07-12 | 广东玖智科技有限公司 | PPG signal clustering center acquisition method and device and PPG signal processing method and device |
| CN115472222A (en) * | 2022-11-02 | 2022-12-13 | 杭州链康医学检验实验室有限公司 | Single cell transcriptome RNA pollution identification method, medium and equipment |
| CN116559143A (en) * | 2023-05-15 | 2023-08-08 | 西北大学 | Method and system for analyzing composite Raman spectrum data of glucose component in blood |
| CN116803344A (en) * | 2023-07-27 | 2023-09-26 | 迈德医疗科技(深圳)有限公司 | Blood glucose classification method and system based on multi-norm clustering and double-layer discrete network |
| CN116803344B (en) * | 2023-07-27 | 2024-02-13 | 迈德医疗科技(深圳)有限公司 | Blood glucose classification method and system based on multi-norm clustering and double-layer discrete network |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113662538A (en) | Non-invasive blood glucose detection method based on time-frequency domain comprehensive analysis | |
| EP3358485A1 (en) | General noninvasive blood glucose prediction method based on timing analysis | |
| Delbeck et al. | Non-invasive monitoring of blood glucose using optical methods for skin spectroscopy—Opportunities and recent advances | |
| Zuo et al. | Comparison of three different types of wrist pulse signals by their physical meanings and diagnosis performance | |
| US20160367198A1 (en) | Apparatus and method for detecting and removing artifacts in optically acquired biological signals | |
| JP5676432B2 (en) | Method and system for non-invasive and optical detection of blood glucose using spectral data analysis | |
| CN110575181A (en) | Near infrared spectrum noninvasive blood glucose detection network model training method | |
| JP2005043343A (en) | Non-invasive method of detecting analyte concentration using hyperosmotic fluid | |
| CN101683260B (en) | Rhythmicity physiological signal optimization method and device | |
| Roy et al. | On-device reliability assessment and prediction of missing photoplethysmographic data using deep neural networks | |
| US20220378380A1 (en) | Method and system for determining cardiovascular parameters | |
| Guo | Noninvasive blood glucose measurement based on NIR spectrums and double ANN analysis | |
| Jean Effil et al. | Wavelet scattering transform and long short-term memory network-based noninvasive blood pressure estimation from photoplethysmograph signals | |
| CN112274120A (en) | Noninvasive arteriosclerosis detection method and device based on one-way pulse wave | |
| Gupta et al. | In-vivo estimation of glucose level using PPG signal | |
| Awelisah et al. | The effect of spectral photoplethysmography amplification and its application in dynamic spectrum for effective noninvasive detection of blood components | |
| CN116186502B (en) | Multi-mode visual nerve function detection method and system thereof | |
| Nampoothiri et al. | Comparison of infrared and red photoplethysmography signals for non-calibrated non-invasive blood glucose monitoring | |
| Li et al. | A dynamic spectrum optimized single-trail extraction method for noninvasive analysis of human blood components | |
| CN115979995A (en) | Blood lactic acid content detection model establishing method and detection method | |
| Rosati et al. | Feature selection applied to the time-frequency representation of muscle near-infrared spectroscopy (NIRS) signals: characterization of diabetic oxygenation patterns | |
| CN113171061B (en) | Noninvasive vascular function assessment method, noninvasive vascular function assessment device, noninvasive vascular function assessment equipment and noninvasive vascular function assessment medium | |
| Zhang et al. | Non-invasive blood glucose detection using NIR based on GA and SVR | |
| Slapničar et al. | Blood pressure estimation with a wristband optical sensor | |
| US20050027183A1 (en) | Method for non-invasive monitoring of blood and tissue glucose |
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: 20211119 |
|
| WW01 | Invention patent application withdrawn after publication |