CN113288134A - Method and device for training blood glucose classification model, bracelet equipment and processor - Google Patents
Method and device for training blood glucose classification model, bracelet equipment and processor Download PDFInfo
- Publication number
- CN113288134A CN113288134A CN202110489921.9A CN202110489921A CN113288134A CN 113288134 A CN113288134 A CN 113288134A CN 202110489921 A CN202110489921 A CN 202110489921A CN 113288134 A CN113288134 A CN 113288134A
- Authority
- CN
- China
- Prior art keywords
- quaternary
- pulse signals
- matrix
- training
- classification model
- 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.)
- Granted
Links
- 239000008280 blood Substances 0.000 title claims abstract description 116
- 210000004369 blood Anatomy 0.000 title claims abstract description 116
- 238000012549 training Methods 0.000 title claims abstract description 60
- 238000000034 method Methods 0.000 title claims abstract description 55
- 238000013145 classification model Methods 0.000 title claims abstract description 51
- 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 50
- 239000008103 glucose Substances 0.000 title claims abstract description 50
- 239000011159 matrix material Substances 0.000 claims abstract description 113
- 238000000556 factor analysis Methods 0.000 claims abstract description 32
- 238000001914 filtration Methods 0.000 claims abstract description 31
- 230000003287 optical effect Effects 0.000 claims abstract description 30
- 238000012545 processing Methods 0.000 claims abstract description 27
- 238000007637 random forest analysis Methods 0.000 claims abstract description 23
- 230000009467 reduction Effects 0.000 claims abstract description 21
- 238000000605 extraction Methods 0.000 claims abstract description 17
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 13
- 238000001514 detection method Methods 0.000 claims description 7
- 229910003460 diamond Inorganic materials 0.000 claims description 4
- 239000010432 diamond Substances 0.000 claims description 4
- 230000008569 process Effects 0.000 description 12
- 238000010586 diagram Methods 0.000 description 11
- 230000033001 locomotion Effects 0.000 description 11
- 238000004590 computer program Methods 0.000 description 7
- 230000002218 hypoglycaemic effect Effects 0.000 description 6
- 230000008859 change Effects 0.000 description 5
- 230000000694 effects Effects 0.000 description 5
- 208000013016 Hypoglycemia Diseases 0.000 description 4
- 230000006870 function Effects 0.000 description 4
- 201000001421 hyperglycemia Diseases 0.000 description 4
- 230000008901 benefit Effects 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 230000000717 retained effect Effects 0.000 description 3
- 210000001367 artery Anatomy 0.000 description 2
- 210000000988 bone and bone Anatomy 0.000 description 2
- 238000006243 chemical reaction Methods 0.000 description 2
- 210000002808 connective tissue Anatomy 0.000 description 2
- 238000003066 decision tree Methods 0.000 description 2
- 206010012601 diabetes mellitus Diseases 0.000 description 2
- 230000003345 hyperglycaemic effect Effects 0.000 description 2
- 230000031700 light absorption Effects 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 210000003205 muscle Anatomy 0.000 description 2
- 230000001019 normoglycemic effect Effects 0.000 description 2
- 238000005070 sampling Methods 0.000 description 2
- 238000012216 screening Methods 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 210000001519 tissue Anatomy 0.000 description 2
- 230000001131 transforming effect Effects 0.000 description 2
- 210000003462 vein Anatomy 0.000 description 2
- 208000017667 Chronic Disease Diseases 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 150000002303 glucose derivatives Chemical class 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 208000015181 infectious disease Diseases 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000012806 monitoring device Methods 0.000 description 1
- 230000003068 static effect Effects 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
- A61B5/02—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
- A61B5/145—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
- A61B5/145—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
- A61B5/68—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
-
- 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
- A61B5/7267—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems involving training the classification device
-
- 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
- A61B5/7271—Specific aspects of physiological measurement analysis
- A61B5/7275—Determining trends in physiological measurement data; Predicting development of a medical condition based on physiological measurements, e.g. determining a risk factor
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/24323—Tree-organised classifiers
Abstract
The embodiment of the invention provides a method and a device for training a blood glucose classification model, bracelet equipment and a processor, and belongs to the field of signal processing and biomedical treatment. The method for training the blood sugar classification model comprises the following steps: acquiring the blood sugar category of a user and pulse signals of the user respectively detected by four optical sensors; filtering the pulse signals based on quaternary Fourier transform to obtain filtered quaternary pulse signals; performing dimensionality reduction processing on the filtered quaternary pulse signals based on a factor analysis algorithm to obtain a dimensionality-reduced quaternary pulse signal matrix; performing feature extraction on the dimensionality-reduced quaternary pulse signal matrix to obtain a feature matrix; and inputting the feature matrix and the blood sugar category into a random forest model for training and classification to obtain a trained blood sugar classification model. The method of the invention can improve the accuracy of the blood sugar category prediction.
Description
Technical Field
The invention relates to the field of signal processing and biomedical science, in particular to a method and a device for training a blood glucose classification model, bracelet equipment and a processor.
Background
Diabetes is one of three chronic diseases in the world, more than 4 hundred million adults suffer from diabetes globally, and Chinese sufferers live at the first of all countries in the world. However, most of the existing blood glucose detection technologies are performed by taking blood samples through fingertips, and the method has the defects of infection risk, difficulty in observing the trend of blood glucose change and the like. Therefore, a bracelet device including an optical sensor is presented to acquire a photoplethysmogram, i.e., a photoplethysmogram (PPG) signal, and process and analyze the pulse signal, thereby realizing non-invasive blood glucose detection and improving blood glucose classification accuracy.
At present, the effect of a photoelectric detector applied in a bracelet device to acquire a photoplethysmogram (PPG) signal of human body is still not ideal. When wearing the bracelet, the hand movement easily leads to bracelet position change, and the aliasing of signal appears or can't read the signal value, leads to blood sugar category prediction's precision not high enough.
Disclosure of Invention
An object of the embodiments of the present invention is to provide a method and an apparatus for training a blood glucose classification model, a bracelet device, and a processor, so as to solve the problem of low accuracy of blood glucose class prediction.
In order to achieve the above object, a first aspect of the present invention provides a method for training a blood glucose classification model, applied to a bracelet device including four optical sensors, the method including:
acquiring the blood sugar category of a user and pulse signals of the user respectively detected by four optical sensors;
filtering the pulse signals based on quaternary Fourier transform to obtain filtered quaternary pulse signals;
performing dimensionality reduction processing on the filtered quaternary pulse signals based on a factor analysis algorithm to obtain a dimensionality-reduced quaternary pulse signal matrix;
performing feature extraction on the dimensionality-reduced quaternary pulse signal matrix to obtain a feature matrix;
and inputting the feature matrix and the blood sugar category into a random forest model for training and classification to obtain a trained blood sugar classification model.
In the embodiment of the present invention, the filtering the pulse signal based on the quaternary fourier transform to obtain the filtered quaternary pulse signal includes: carrying out quaternary Fourier transform on the pulse signals to obtain frequency domain quaternary pulse signals; filtering the frequency domain quaternary pulse signals to obtain filtered frequency domain quaternary pulse signals; and carrying out inverse quaternary Fourier transform on the filtered frequency domain quaternary pulse signals to obtain time domain quaternary pulse signals.
In the embodiment of the present invention, the filtering the frequency domain quaternary pulse signal to obtain the filtered frequency domain quaternary pulse signal includes: and filtering the frequency domain quaternary pulse signals through a band-pass filter to obtain the filtered frequency domain quaternary pulse signals.
In the embodiment of the present invention, the dimension reduction processing is performed on the filtered quaternary pulse signal based on a factor analysis algorithm to obtain a dimension-reduced quaternary pulse signal matrix, which includes: determining a corresponding matrix according to the filtered quaternary pulse signals; and performing factor analysis on the matrix based on a factor analysis model to obtain a dimension-reduced quaternary pulse signal matrix.
In the embodiment of the present invention, the feature extraction is performed on the dimensionality-reduced quaternary pulse signal matrix to obtain a feature matrix, which includes: performing feature extraction on the quaternary pulse signals in the dimensionality reduced quaternary pulse signal matrix to obtain peak heights, peak widths, peak positions and peak areas of peaks of the quaternary pulse signals; and obtaining a characteristic matrix according to the peak height, the peak width, the peak position and the peak area.
In the embodiment of the invention, the feature matrix and the blood sugar category are input into a random forest model for training and classification to obtain a trained blood sugar classification model, which comprises the following steps: determining a training set according to the feature matrix and a preset proportion; and inputting the training set and the blood sugar category into a random forest model for training and classification to obtain a trained blood sugar classification model.
A second aspect of the invention provides a processor configured to perform the above-described method for training a blood glucose classification model.
A third aspect of the present invention provides an apparatus for training a blood glucose classification model, comprising: a blood glucose detection device for determining a blood glucose category of a user; the pulse detection equipment comprises four optical sensors which are respectively used for detecting pulse signals of a user; and the processor.
The invention provides a bracelet device comprising four optical sensors, wherein each optical sensor comprises a light-emitting diode and a photoelectric detector, and the light-emitting diodes are connected in a square, diamond, circular or oval manner.
A fifth aspect of the invention provides a machine-readable storage medium having stored thereon instructions which, when executed by a processor, cause the processor to perform the above-described method for training a blood glucose classification model.
According to the technical scheme, the blood sugar type of the user and the pulse signals of the user respectively detected by the four optical sensors are obtained, filtering processing is carried out on the pulse signals based on quaternary Fourier transform to obtain filtered quaternary pulse signals, dimension reduction processing is carried out on the filtered quaternary pulse signals based on a factor analysis algorithm to obtain a dimension-reduced quaternary pulse signal matrix, feature extraction is carried out on the dimension-reduced quaternary pulse signal matrix to obtain a feature matrix, and the feature matrix and the blood sugar type are input into a random forest model to be trained and classified to obtain a trained blood sugar classification model. According to the method, the four paths of pulse signals are acquired through the bracelet equipment comprising the four optical sensors, so that signal default values caused by motion artifacts can be reduced; the pulse signals of the four channels are processed by using a quaternary frequency domain denoising method, the original information of the signals can be retained to the maximum extent, filtering is performed by combining a quaternary theory, dimension reduction processing is performed on the pulse signals by using a factor analysis method, relevant characteristics are extracted, and the precision of blood sugar category prediction can be improved.
Additional features and advantages of embodiments of the invention will be set forth in the detailed description which follows.
Drawings
The accompanying drawings, which are included to provide a further understanding of the embodiments of the invention and are incorporated in and constitute a part of this specification, illustrate embodiments of the invention and together with the description serve to explain the embodiments of the invention without limiting the embodiments of the invention. In the drawings:
FIG. 1 schematically illustrates a flow diagram of a method for training a blood glucose classification model in an embodiment of the present invention;
FIG. 2 is a flow chart schematically illustrating the steps of filtering the pulse signal according to an embodiment of the present invention;
fig. 3 is a block diagram schematically illustrating the structure of an apparatus for training a blood glucose classification model according to an embodiment of the present invention.
Detailed Description
The following detailed description of embodiments of the invention refers to the accompanying drawings. It should be understood that the detailed description and specific examples, while indicating embodiments of the invention, are given by way of illustration and explanation only, not limitation.
FIG. 1 schematically shows a flow chart of a method for training a blood glucose classification model in an embodiment of the invention. As shown in fig. 1, in an embodiment of the present invention, a method for training a blood glucose classification model is provided, which is applied to a bracelet device including four optical sensors, and is described by taking the method as an example of being applied to a processor, and the method may include the following steps:
step S102, obtaining the blood sugar category of the user and the pulse signals of the user respectively detected by the four optical sensors.
It is understood that the blood glucose categories may include three categories, hyperglycemic, hypoglycemic and normoglycemic ranges.
Specifically, 4-way photoplethysmogram (PPG) signals are acquired by a bracelet device equipped with four optical sensors (LEDs and photodetectors), whose internal principle is: LED lamp directive skin on the bracelet, the light that reflects back through skin tissue is accepted by photoelectric detector and is converted into the signal of telecommunication, converts digital signal into through AD again, simplifies the process: light → electrical → digital signal.
When LED light is directed to the skin, the absorption of light is substantially constant, like muscles, bones, veins and other connective tissue, but the blood in arteries, unlike it, pulsates with the heart. Thus, the light transmitted or reflected back through the skin is also pulsated. However, due to nonresistible factors, for example, when wearing a bracelet, the movement of a hand easily causes the change of the position of the bracelet, aliasing of signals occurs or signal values cannot be read, so that when reading signals, in order to acquire more effective information and reduce data acquisition errors caused by motion artifacts, 4 LEDs can be designed into a square (not limited to a square) with a distance of 1 cm to read the signals, and the problem of incomplete signal information caused by jitter in the motion process is solved through tight connection between a lamp and a lamp. Respectively marking PPG signals read by 4 LEDs as P1,P2,P3,P4And the blood sugar category (namely the blood sugar high and low conditions) of the user at the moment can be recorded by the blood sugar meter.
And step S104, filtering the pulse signals based on quaternary Fourier transform to obtain filtered quaternary pulse signals.
Specifically, the processor may perform filtering processing on the pulse signals detected by the four optical sensors based on a quaternary fourier transform, so as to obtain filtered quaternary pulse signals.
In one embodiment, as shown in fig. 2, the filtering process is performed on the pulse signal based on the quaternary fourier transform to obtain a filtered quaternary pulse signal, including the following steps S202 to S206:
step S202, quaternary Fourier transform is carried out on the pulse signals to obtain frequency domain quaternary pulse signals.
In particular, a four-channel pulse (PPG) signal P to be read each time blood glucose is measured1,P2,P3,P4The pulse signals are regarded as a real part and an imaginary part of a quaternion vector, and quaternion Fourier transform is carried out to obtain a quaternion pulse (PPG) signal of a frequency domain;
wherein P is1,P2,P3,P4Respectively treated as the real part and i, j, k imaginary part of a quaternion vector, as follows:
h(n)＝P1(n)+P2(n)i+P3(n)j+P4(n)k
wherein P is1Representing PPG signal, P, collected by LED12Representing PPG signal, P, collected by LED23Representing PPG signal, P, collected by LED34Representing the PPG signal collected by LED4, the number of sampling points is N, h (N) is a quaternion vector, where N is 1,2, …, N.
Then by a quaternary fourier transform:
wherein mu is muii+μjj+μkk, satisfying a negative unit condition, i.e. μ, in order to maintain energy conversion2＝-1。
The quaternary PPG signal, converted into the frequency domain, is represented as follows:
H(m)＝P1′(m)+P2′(m)i+P3′(m)j+P4′(m)k
wherein P is1' represents the real value, P, of the LED1 after quaternary Fourier transform2' represents the i imaginary value, P, of the LED2 after quaternary Fourier transform3' represents the imaginary value of j, P, of the LED3 after quaternary Fourier transform4' stands for imaginary k values of the LED4 after quaternary fourier transform, and h (m) is a quaternary vector after fourier transform, where m is 1,2, …, N.
Step S204, filtering the frequency domain quaternary pulse signal to obtain a filtered frequency domain quaternary pulse signal.
Specifically, a quaternary pulse (PPG) signal h (m) obtained by quaternary fourier transform is filtered to obtain a frequency domain quaternary pulse signal in a filtered frequency domain. And a quaternary denoising theory is fused to filter the PPG of the four channels, so that the denoising effect of the signals is greatly improved, and higher classification accuracy is obtained.
In one embodiment, the filtering the frequency domain quaternary pulse signal to obtain a filtered frequency domain quaternary pulse signal includes: and filtering the frequency domain quaternary pulse signals through a band-pass filter to obtain the filtered frequency domain quaternary pulse signals.
Specifically, the processor filters out the higher and/or lower frequency signal components by a band-pass filter to obtain a filtered frequency-domain quaternary pulse signal, for example, components higher than 200Hz and lower than 40Hz in H (m) are filtered out to obtain a filtered frequency-domain quaternary pulse signal H' (m).
And step S206, performing inverse quaternary Fourier transform on the filtered frequency domain quaternary pulse signals to obtain time domain quaternary pulse signals.
And (3) transforming the filtered frequency domain quaternary pulse signal H' (m) by an inverse quaternary Fourier transform:
the quaternary PPG signal, converted into the time domain, is represented as follows:
h′(n)＝P1″(n)+P2″(n)i+P3″(n)j+P4″(n)k
wherein P is1"represents the real value, P, of the LED1 after the quaternary Fourier transform2"represents the i imaginary value, P, of LED2 after quaternary Fourier transform3"represents the imaginary value of j, P, of the LED3 after quaternary Fourier transform4"represents the imaginary value of k after the quaternary inverse fourier transform of LED4, and h' (n) is the quaternary vector after the inverse fourier transform.
Step S106 is followed by step S104, dimension reduction processing is carried out on the filtered quaternary pulse signals based on a factor analysis algorithm, and a dimension-reduced quaternary pulse signal matrix is obtained.
Specifically, the processor performs dimensionality reduction processing on the filtered quaternary pulse signals based on a factor analysis algorithm to obtain a dimensionality reduced quaternary pulse signal matrix, wherein the dimensionality reduction processing belongs to one step of feature screening and aims to extract features after more useful components are obtained. Factor analysis is used to find out features of the same nature in the filtered signal, and the features are classified into one class to realize feature selection.
In one embodiment, the dimension reduction processing is performed on the filtered quaternary pulse signals based on a factor analysis algorithm to obtain a dimension-reduced quaternary pulse signal matrix, which includes: determining a corresponding matrix according to the filtered quaternary pulse signals; and performing factor analysis on the matrix based on a factor analysis model to obtain a dimension-reduced quaternary pulse signal matrix.
In particular, the processor forms the filtered quaternary pulse signal (PPG) signals into a new matrix P ═ P ″1P″2 P″3 P″4]And performing factor analysis on the matrix P' and finding out the most appropriate characteristic to train the model by researching the internal dependency relationship among the characteristic variables. Assuming that F represents the constructed common factor matrix, A is a factor load coefficient matrix, and is obtained by main factor analysis estimation, the dependence degree of the matrix P' on the common factor F is represented, and the larger the value is, the larger the dependence degree is, and the matrix form of the factor model is as follows:
P″＝AF
under a factor analysis model, a matrix P is calculatedTCovariance matrix of P:
here, P ″)TA transpose matrix representing the matrix P, L an N x N matrix representing the payload,
representing compensated NxN matrices, LLTRepresents the ith commonality, where i is 1,2, …, N.
Sorting according to the numerical value of the common factor variance from small to large, and taking the first
Reconstructing common factors, and assuming that a factor load coefficient matrix after reconstruction is A' and the dimensionality is
The reconstructed common factor matrix is F' and the dimensionality is
So the reduced dimension matrix P':
P″′＝A′F′＝[P″′1 P″′2 P″′3 P″′4]
and S108, extracting the characteristics of the four-element pulse signal matrix subjected to dimension reduction to obtain a characteristic matrix.
Specifically, the processor may perform feature extraction on the dimensionality-reduced quaternary pulse signal matrix to obtain a feature matrix containing feature information.
In one embodiment, the feature extraction is performed on the dimensionality-reduced quaternary pulse signal matrix to obtain a feature matrix, and the feature matrix comprises: performing feature extraction on the quaternary pulse signals in the dimensionality reduced quaternary pulse signal matrix to obtain peak heights, peak widths, peak positions and peak areas of peaks of the quaternary pulse signals; and obtaining a characteristic matrix according to the peak height, the peak width, the peak position and the peak area.
In particular, for each PPG signal P ″ 'in the post-dimensionality-reduced quaternary pulse signal matrix'1,P″′2,P″′3,P″′4Extracting characteristics to obtain P'1,P″′2,P″′3,P″′4The peak height, peak width, peak position, and peak area of the signal peak are used as the characteristics of blood glucose classification.
And S110, inputting the feature matrix and the blood sugar category into a random forest model for training and classification to obtain a trained blood sugar classification model.
Specifically, the processor may input the feature matrix and the blood sugar categories (e.g., hyperglycemia and hypoglycemia) into a random forest model to obtain a trained blood sugar classification model.
In one embodiment, inputting the feature matrix and the blood sugar category into a random forest model for training and classification to obtain a trained blood sugar classification model, including: determining a training set according to the feature matrix and a preset proportion; and inputting the training set and the blood sugar category into a random forest model for training and classification to obtain a trained blood sugar classification model.
It will be appreciated that the preset ratio is a size range of the training set, for example 70%.
Specifically, assuming that the training set includes T PPG signals, each PPG signal corresponds to a classification tag, the value of the tag is recorded by a blood glucose meter, for example, the tag can be classified into hyperglycemia and hypoglycemia, a feature matrix X of all filtered and dimensionality-reduced PPG signals, where the matrix size may be T × 4, and the feature matrix X is calculated by 7: and 3, proportionally dividing the training set and the test set, and inputting the training set and the blood sugar category into a random forest model for training and classification to obtain a trained blood sugar classification model. The model is trained and tested by using the unbiased estimation advantage of the random forest, so that the optimal classification effect can be realized.
After the trained blood sugar classification model is obtained, each decision tree in the random forest can be used for judging and classifying, voting is finally carried out through a minority obeying majority principle, the result is used as the final result of classification when the voting result is large, and classification of high and low blood sugar can be achieved.
According to the method for training the blood sugar classification model, the blood sugar type of the user and the pulse signals of the user respectively detected by the four optical sensors are obtained, filtering processing is carried out on the pulse signals based on quaternary Fourier transform to obtain filtered quaternary pulse signals, dimension reduction processing is carried out on the filtered quaternary pulse signals based on a factor analysis algorithm to obtain a dimension-reduced quaternary pulse signal matrix, feature extraction is carried out on the dimension-reduced quaternary pulse signal matrix to obtain a feature matrix, and the feature matrix and the blood sugar type are input into a random forest model to be trained and classified to obtain the trained blood sugar classification model. According to the method, the four paths of pulse signals are acquired through the bracelet equipment comprising the four optical sensors, so that signal default values caused by motion artifacts can be reduced; the pulse signals of the four channels are processed by using a quaternary frequency domain denoising method, the original information of the signals can be retained to the maximum extent, filtering is performed by combining a quaternary theory, dimension reduction processing is performed on the pulse signals by using a factor analysis method, relevant characteristics are extracted, and the precision of blood sugar category prediction can be improved.
Fig. 3 is a block diagram schematically illustrating the structure of an apparatus for training a blood glucose classification model according to an embodiment of the present invention. As shown in fig. 3, in an embodiment of the present invention, there is provided an apparatus 300 for training a blood glucose classification model, comprising: a blood glucose detecting device 310, a pulse detecting device 320, and a processor 330, wherein:
a blood glucose detecting device 310 for determining a blood glucose category of the user.
The pulse detecting device 320 includes four optical sensors for detecting pulse signals of the user, respectively.
A processor 330 configured to: acquiring the blood sugar category of a user and pulse signals of the user respectively detected by four optical sensors; filtering the pulse signals based on quaternary Fourier transform to obtain filtered quaternary pulse signals; performing dimensionality reduction processing on the filtered quaternary pulse signals based on a factor analysis algorithm to obtain a dimensionality-reduced quaternary pulse signal matrix; performing feature extraction on the dimensionality-reduced quaternary pulse signal matrix to obtain a feature matrix; and inputting the feature matrix and the blood sugar category into a random forest model for training and classification to obtain a trained blood sugar classification model.
It is understood that the blood glucose categories may include three categories, hyperglycemic, hypoglycemic and normoglycemic ranges.
In particular, a 4-way photoplethysmogram (PPG) signal is acquired by a pulse detection device 320, i.e. a bracelet device equipped with four optical sensors (LEDs and photodetectors), whose internal principle is: LED lamp directive skin on the bracelet, the light that reflects back through skin tissue is accepted by photoelectric detector and is converted into the signal of telecommunication, converts digital signal into through AD again, simplifies the process: light → electrical → digital signal.
When LED light is directed to the skin, the absorption of light is substantially constant, like muscles, bones, veins and other connective tissue, but the blood in arteries, unlike it, pulsates with the heart. Thus, the light transmitted or reflected back through the skin is also pulsated. However, due to nonresistible factors, for example, when wearing a bracelet, the movement of a hand easily causes the change of the position of the bracelet, aliasing of signals occurs or signal values cannot be read, so that when reading signals, in order to acquire more effective information and reduce data acquisition errors caused by motion artifacts, 4 LEDs can be designed into a square (not limited to a square) with a distance of 1 cm to read the signals, and the problem of incomplete signal information caused by jitter in the motion process is solved through tight connection between a lamp and a lamp. Respectively marking PPG signals read by 4 LEDs as P1,P2,P3,P4And may record the user's blood glucose category (i.e., blood glucose high-low) at that moment via a blood glucose monitoring device 310 (e.g., a blood glucose meter).
Specifically, the processor 330 may perform filtering processing on the pulse signals detected by the four optical sensors based on a quaternary fourier transform, resulting in filtered quaternary pulse signals. The processor 330 performs dimensionality reduction processing on the filtered quaternary pulse signals based on a factor analysis algorithm to obtain a dimensionality reduced quaternary pulse signal matrix, wherein the dimensionality reduction processing belongs to one step of feature screening and is used for performing feature extraction after more useful components are obtained. Factor analysis is used to find out features of the same nature in the filtered signal, and the features are classified into one class to realize feature selection. The processor 330 may perform feature extraction on the dimensionality-reduced quaternary pulse signal matrix to obtain a feature matrix containing feature information. The processor 330 may input the feature matrix and blood glucose categories (e.g., hyperglycemia and hypoglycemia) into a random forest model to obtain a trained blood glucose classification model.
The aforesaid a device for training blood sugar classification model, the user's pulse signal that detects respectively through the blood sugar classification and four optical sensor that acquire the user, carry out filtering process to the pulse signal based on quaternary Fourier transform, obtain the quaternary pulse signal after the filtration, carry out dimensionality reduction processing to the quaternary pulse signal after the filtration based on factor analysis algorithm again, obtain the quaternary pulse signal matrix after the dimensionality reduction, carry out the feature extraction to the quaternary pulse signal matrix after the dimensionality reduction, obtain the feature matrix, input random forest model with feature matrix and blood sugar classification and train the classification, obtain the blood sugar classification model that trains. The device acquires four paths of pulse signals through bracelet equipment comprising four optical sensors, so that signal default values caused by motion artifacts can be reduced; the pulse signals of the four channels are processed by using a quaternary frequency domain denoising method, the original information of the signals can be retained to the maximum extent, filtering is performed by combining a quaternary theory, dimension reduction processing is performed on the pulse signals by using a factor analysis method, relevant characteristics are extracted, and the precision of blood sugar category prediction can be improved.
In one embodiment, the processor 330 is further configured to: carrying out quaternary Fourier transform on the pulse signals to obtain frequency domain quaternary pulse signals; filtering the frequency domain quaternary pulse signals to obtain filtered frequency domain quaternary pulse signals; and carrying out inverse quaternary Fourier transform on the filtered frequency domain quaternary pulse signals to obtain time domain quaternary pulse signals.
In particular, a four-channel pulse (PPG) signal P to be read each time blood glucose is measured1,P2,P3,P4The pulse signals are regarded as a real part and an imaginary part of a quaternion vector, and quaternion Fourier transform is carried out to obtain a quaternion pulse (PPG) signal of a frequency domain;
wherein P is1,P2,P3,P4Respectively treated as the real part and i, j, k imaginary part of a quaternion vector, as follows:
h(n)＝P1(n)+P2(n)i+P3(n)j+P4(n)k
wherein P is1Representing PPG signal, P, collected by LED12Representing PPG signal, P, collected by LED23Representing PPG signal, P, collected by LED34Representing the PPG signal collected by LED4, the number of sampling points is N, h (N) is a quaternion vector, where N is 1,2, …, N.
Then by a quaternary fourier transform:
wherein mu is muii+μjj+μkk, satisfying a negative unit condition, i.e. μ, in order to maintain energy conversion2＝-1。
The quaternary PPG signal, converted into the frequency domain, is represented as follows:
H(m)＝P1′(m)+P2′(m)i+P3′(m)j+P4′(m)k
wherein P is1' represents the real value, P, of the LED1 after quaternary Fourier transform2' represents the i imaginary value, P, of the LED2 after quaternary Fourier transform3' represents the imaginary value of j, P, of the LED3 after quaternary Fourier transform4' stands for imaginary k values of the LED4 after quaternary fourier transform, and h (m) is a quaternary vector after fourier transform, where m is 1,2, …, N.
And filtering a quaternary pulse (PPG) signal H (m) obtained through quaternary Fourier transform to obtain a frequency domain quaternary pulse signal on a filtered frequency domain. And a quaternary denoising theory is fused to filter the PPG of the four channels, so that the denoising effect of the signals is greatly improved, and higher classification accuracy is obtained.
And (3) transforming the filtered frequency domain quaternary pulse signal H' (m) by an inverse quaternary Fourier transform:
the quaternary PPG signal, converted into the time domain, is represented as follows:
h′(n)＝P1″(n)+P2″(n)i+P3″(n)j+P4″(n)k
wherein P is1"represents the real value, P, of the LED1 after the quaternary Fourier transform2"represents the i imaginary value, P, of LED2 after quaternary Fourier transform3"represents the imaginary value of j, P, of the LED3 after quaternary Fourier transform4"represents the imaginary value of k after the quaternary inverse fourier transform of LED4, and h' (n) is the quaternary vector after the inverse fourier transform.
In one embodiment, the processor 330 is further configured to: and filtering the frequency domain quaternary pulse signals through a band-pass filter to obtain the filtered frequency domain quaternary pulse signals.
Specifically, the processor 330 filters out the higher and/or lower frequency signal components by a band-pass filter to obtain a filtered frequency-domain quaternary pulse signal, for example, components above 200Hz and below 40Hz in H (m) are filtered out to obtain a filtered frequency-domain quaternary pulse signal H' (m).
In one embodiment, the processor 330 is further configured to: determining a corresponding matrix according to the filtered quaternary pulse signals; and performing factor analysis on the matrix based on a factor analysis model to obtain a dimension-reduced quaternary pulse signal matrix.
In particular, the processor 330 combines the filtered quaternary pulse signal (PPG) signals into a new matrix P ″ ═ P ″1 P″2 P″3 P4]And performing factor analysis on the matrix P' and finding out the most appropriate characteristic to train the model by researching the internal dependency relationship among the characteristic variables. Assuming that F represents the constructed common factor matrix, A is a factor load coefficient matrix, and is obtained by main factor analysis estimation, the dependence degree of the matrix P' on the common factor F is represented, and the larger the value is, the larger the dependence degree is, and the matrix form of the factor model is as follows:
P″＝AF
under a factor analysis model, a matrix P is calculatedTCovariance matrix of P:
here, P ″)TA transpose matrix representing the matrix P, L an N x N matrix representing the payload,
representing compensated NxN matrices, LLTRepresents the ith commonality, where i is 1,2, …, N.
Sorting according to the numerical value of the common factor variance from small to large, and taking the first
Reconstructing common factors, and assuming that a factor load coefficient matrix after reconstruction is A' and the dimensionality is
The reconstructed common factor matrix is F' and the dimensionality is
So the reduced dimension matrix P':
P″′＝A′F′＝[P″′1 P″′2 P″′3 P″′4]
in one embodiment, the processor 330 is further configured to: performing feature extraction on the quaternary pulse signals in the dimensionality reduced quaternary pulse signal matrix to obtain peak heights, peak widths, peak positions and peak areas of peaks of the quaternary pulse signals; and obtaining a characteristic matrix according to the peak height, the peak width, the peak position and the peak area.
In particular, for each PPG signal P ″ 'in the post-dimensionality-reduced quaternary pulse signal matrix'1,P″′2,P″′3,P″′4Extracting characteristics to obtain P'1,P″′2,P″′3,P″′4The peak height, peak width, peak position, and peak area of the signal peak are used as the characteristics of blood glucose classification.
In one embodiment, the processor 330 is further configured to: determining a training set according to the feature matrix and a preset proportion; and inputting the training set and the blood sugar category into a random forest model for training and classification to obtain a trained blood sugar classification model.
It will be appreciated that the preset ratio is a size range of the training set, for example 70%.
Specifically, assuming that the training set includes T PPG signals, each PPG signal corresponds to a classification tag, the value of the tag is recorded by a blood glucose meter, for example, the tag can be classified into hyperglycemia and hypoglycemia, a feature matrix X of all filtered and dimensionality-reduced PPG signals, where the matrix size may be T × 4, and the feature matrix X is calculated by 7: and 3, proportionally dividing the training set and the test set, and inputting the training set and the blood sugar category into a random forest model for training and classification to obtain a trained blood sugar classification model. The model is trained and tested by using the unbiased estimation advantage of the random forest, so that the optimal classification effect can be realized.
After the trained blood sugar classification model is obtained, each decision tree in the random forest can be used for judging and classifying, voting is finally carried out through a minority obeying majority principle, the result is used as the final result of classification when the voting result is large, and classification of high and low blood sugar can be achieved.
The device for training the blood sugar classification model comprises a processor and a memory, wherein the processor comprises a kernel, and the kernel calls a corresponding program unit from the memory. The kernel can be set to be one or more, and the precision of the blood sugar classification prediction is improved by adjusting the kernel parameters.
The memory may include volatile memory in a computer readable medium, Random Access Memory (RAM) and/or nonvolatile memory such as Read Only Memory (ROM) or flash memory (flash RAM), and the memory includes at least one memory chip.
Embodiments of the present invention provide a processor configured to execute the method for training a blood glucose classification model according to the above embodiments.
The embodiment of the invention provides bracelet equipment which comprises four optical sensors, wherein each optical sensor comprises a light-emitting diode and a photoelectric detector, and the light-emitting diodes are connected in a square, diamond, circular or oval mode.
In this embodiment, the Light Emitting Diodes (LEDs) may be connected in a square, diamond, circle or oval manner, for example, 4 LEDs are designed to be a square with a distance of 1 cm to read signals, and the problem of incomplete signal information caused by jitter in the movement process is reduced by the close connection between the lamps.
An embodiment of the present invention provides a machine-readable storage medium, which stores instructions thereon, and when executed by a processor, causes the processor to execute the method for training a blood glucose classification model according to the above embodiment.
As will be appreciated by one skilled in the art, embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
In a typical configuration, a computing device includes one or more processors (CPUs), input/output interfaces, network interfaces, and memory.
The memory may include forms of volatile memory in a computer readable medium, Random Access Memory (RAM) and/or non-volatile memory, such as Read Only Memory (ROM) or flash memory (flash RAM). The memory is an example of a computer-readable medium.
Computer-readable media, including both non-transitory and non-transitory, removable and non-removable media, may implement information storage by any method or technology. The information may be computer readable instructions, data structures, modules of a program, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), Static Random Access Memory (SRAM), Dynamic Random Access Memory (DRAM), other types of Random Access Memory (RAM), Read Only Memory (ROM), Electrically Erasable Programmable Read Only Memory (EEPROM), flash memory or other memory technology, compact disc read only memory (CD-ROM), Digital Versatile Discs (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices, or any other non-transmission medium that can be used to store information that can be accessed by a computing device. As defined herein, a computer readable medium does not include a transitory computer readable medium such as a modulated data signal and a carrier wave.
It should also be noted that the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other identical elements in the process, method, article, or apparatus that comprises the element.
The above are merely examples of the present application and are not intended to limit the present application. Various modifications and changes may occur to those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.
Claims (10)
1. A method for training a blood glucose classification model, applied to a bracelet device comprising four optical sensors, characterized in that the method comprises:
acquiring the blood sugar category of a user and pulse signals of the user respectively detected by the four optical sensors;
filtering the pulse signals based on quaternary Fourier transform to obtain filtered quaternary pulse signals;
performing dimensionality reduction processing on the filtered quaternary pulse signals based on a factor analysis algorithm to obtain a dimensionality-reduced quaternary pulse signal matrix;
performing feature extraction on the dimensionality-reduced quaternary pulse signal matrix to obtain a feature matrix;
and inputting the characteristic matrix and the blood sugar category into a random forest model for training and classification to obtain a trained blood sugar classification model.
2. The method for training a blood glucose classification model according to claim 1, wherein the filtering the pulse signals based on a quaternary fourier transform to obtain filtered quaternary pulse signals comprises:
carrying out quaternary Fourier transform on the pulse signals to obtain frequency domain quaternary pulse signals;
filtering the frequency domain quaternary pulse signal to obtain a filtered frequency domain quaternary pulse signal;
and carrying out inverse quaternary Fourier transform on the filtered frequency domain quaternary pulse signals to obtain time domain quaternary pulse signals.
3. The method for training the blood glucose classification model according to claim 2, wherein the filtering the frequency domain quaternary pulse signal to obtain a filtered frequency domain quaternary pulse signal comprises:
and filtering the frequency domain quaternary pulse signals through a band-pass filter to obtain the filtered frequency domain quaternary pulse signals.
4. The method for training a blood glucose classification model according to claim 1, wherein the performing dimension reduction on the filtered quaternary pulse signals based on a factor analysis algorithm to obtain a dimension-reduced quaternary pulse signal matrix comprises:
determining a corresponding matrix according to the filtered quaternary pulse signals;
and performing factor analysis on the matrix based on a factor analysis model to obtain a dimension-reduced quaternary pulse signal matrix.
5. The method for training a blood glucose classification model according to claim 1, wherein the performing feature extraction on the dimensionality-reduced quaternary pulse signal matrix to obtain a feature matrix comprises:
performing feature extraction on the quaternary pulse signals in the dimensionality reduced quaternary pulse signal matrix to obtain peak heights, peak widths, peak positions and peak areas of peaks of the quaternary pulse signals;
and obtaining a characteristic matrix according to the peak height, the peak width, the peak position and the peak area.
6. The method for training the blood sugar classification model according to claim 1, wherein the step of inputting the feature matrix and the blood sugar category into a random forest model for training and classification to obtain a trained blood sugar classification model comprises the following steps:
determining a training set according to the characteristic matrix and a preset proportion;
and inputting the training set and the blood sugar category into a random forest model for training and classification to obtain a trained blood sugar classification model.
7. A processor, characterized in that the processor is configured to perform the method for training a blood glucose classification model according to any one of claims 1 to 6.
8. An apparatus for training a blood glucose classification model, comprising:
a blood glucose detection device for determining a blood glucose category of a user;
the pulse detection equipment comprises four optical sensors which are respectively used for detecting pulse signals of the user; and
the processor of claim 7.
9. A bracelet device comprising four optical sensors, wherein the optical sensors comprise light emitting diodes and photodetectors, and the light emitting diodes are connected in a square, diamond, circular or elliptical manner.
10. A machine-readable storage medium having instructions stored thereon, which when executed by a processor cause the processor to perform the method for training a blood glucose classification model according to any one of claims 1 to 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110489921.9A CN113288134B (en) | 2021-05-06 | 2021-05-06 | Method and device for training blood glucose classification model, bracelet equipment and processor |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110489921.9A CN113288134B (en) | 2021-05-06 | 2021-05-06 | Method and device for training blood glucose classification model, bracelet equipment and processor |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113288134A true CN113288134A (en) | 2021-08-24 |
| CN113288134B CN113288134B (en) | 2022-09-02 |
Family
ID=77320813
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110489921.9A Active CN113288134B (en) | 2021-05-06 | 2021-05-06 | Method and device for training blood glucose classification model, bracelet equipment and processor |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113288134B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116942151A (en) * | 2023-07-25 | 2023-10-27 | 迈德医疗科技(深圳)有限公司 | Blood glucose detection method based on wavelet scattering fusion Markov angle field |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2011206591A (en) * | 2011-07-25 | 2011-10-20 | Toshiba Medical Systems Corp | Non-invasive estimation method of glucose and non-invasive estimation device of glucose |
| US20160143576A1 (en) * | 2013-07-15 | 2016-05-26 | Tel Hashomer Medical Research Infrastructure And Services Ltd. | Mri image fusion methods and uses thereof |
| CN106446777A (en) * | 2016-08-29 | 2017-02-22 | 广东工业大学 | Noninvasive blood sugar data processing method and noninvasive blood sugar data processing system based on convolutional neural network |
| US20170143279A1 (en) * | 2015-11-20 | 2017-05-25 | Tata Consultancy Services Limited | Device and method to detect diabetes in a person using pulse palpation signal |
| CN108852307A (en) * | 2018-05-24 | 2018-11-23 | 重庆理工大学 | A kind of non-contact noninvasive Arteriosclerosis detecting device |
| CN111265200A (en) * | 2020-02-21 | 2020-06-12 | 广东高驰运动科技有限公司 | Wearable physiological signal detection device and detection method |
| CN112182492A (en) * | 2020-09-23 | 2021-01-05 | 广东工业大学 | Signal sparse representation method and device based on discrete quaternion Fourier transform |
-
2021
- 2021-05-06 CN CN202110489921.9A patent/CN113288134B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2011206591A (en) * | 2011-07-25 | 2011-10-20 | Toshiba Medical Systems Corp | Non-invasive estimation method of glucose and non-invasive estimation device of glucose |
| US20160143576A1 (en) * | 2013-07-15 | 2016-05-26 | Tel Hashomer Medical Research Infrastructure And Services Ltd. | Mri image fusion methods and uses thereof |
| US20170143279A1 (en) * | 2015-11-20 | 2017-05-25 | Tata Consultancy Services Limited | Device and method to detect diabetes in a person using pulse palpation signal |
| CN106446777A (en) * | 2016-08-29 | 2017-02-22 | 广东工业大学 | Noninvasive blood sugar data processing method and noninvasive blood sugar data processing system based on convolutional neural network |
| CN108852307A (en) * | 2018-05-24 | 2018-11-23 | 重庆理工大学 | A kind of non-contact noninvasive Arteriosclerosis detecting device |
| CN111265200A (en) * | 2020-02-21 | 2020-06-12 | 广东高驰运动科技有限公司 | Wearable physiological signal detection device and detection method |
| CN112182492A (en) * | 2020-09-23 | 2021-01-05 | 广东工业大学 | Signal sparse representation method and device based on discrete quaternion Fourier transform |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116942151A (en) * | 2023-07-25 | 2023-10-27 | 迈德医疗科技(深圳)有限公司 | Blood glucose detection method based on wavelet scattering fusion Markov angle field |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113288134B (en) | 2022-09-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2021184805A1 (en) | Blood pressure prediction method and device using multiple data sources | |
| CN104224197A (en) | Calculation method for arterial oxygen saturation by using characteristic quantity of dual-wavelength photoelectric volume pulse wave | |
| Banerjee et al. | PhotoECG: Photoplethysmographyto estimate ECG parameters | |
| CN106137219B (en) | The absolute difference adduction of dual wavelength calculates arterial oxygen saturation method and device thereof | |
| CN113288134B (en) | Method and device for training blood glucose classification model, bracelet equipment and processor | |
| Awodeyi et al. | Median filter approach for removal of baseline wander in photoplethysmography signals | |
| CN114492501A (en) | Electroencephalogram signal sample expansion method, medium and system based on improved SMOTE algorithm | |
| WO2012001289A1 (en) | Method and device for detecting and quantifying skin symptoms on an area of the skin | |
| Zanelli et al. | Transfer learning of CNN-based signal quality assessment from clinical to non-clinical PPG signals | |
| Guler et al. | Optimal digital filter selection for remote photoplethysmography (rPPG) signal conditioning | |
| CN113729653A (en) | Human body pulse wave signal acquisition method | |
| Ram et al. | Use of multi-scale principal component analysis for motion artifact reduction of PPG signals | |
| Park et al. | Detection of epileptiform activities in the EEG using neural network and expert system | |
| CN113208613A (en) | Multi-mode BCI (binary coded decimal) timing optimization method based on FHLS (FHLS) feature selection | |
| CN113143261A (en) | Electromyographic signal-based identity recognition system, method and equipment | |
| CN105796051B (en) | Three-dimensional physiology-detecting system and its operating method | |
| EP3160336B1 (en) | Device for processing foetal heart rhythm data, and corresponding method and computer program | |
| Zou et al. | Non-contact real-time heart rate measurement algorithm based on PPG-standard deviation | |
| CN112716469B (en) | Real-time heart rate extraction method and device based on fingertip video | |
| CN114532994A (en) | Automatic detection method for unsupervised electroencephalogram high-frequency oscillation signals based on convolution variational self-encoder | |
| Singh et al. | Automated atrial fibrillation classification based on denoising stacked autoencoder and optimized deep network | |
| CN111783857A (en) | Motor imagery brain-computer interface based on nonlinear network information graph | |
| Shao et al. | A Photoplethysmograph Signal Preprocess Method Based on Wavelet Transform | |
| Huerta-Ruiz et al. | Relationship between PPG Signals and Glucose levels through Chaotic Descriptors and Support Vector Machines | |
| Rekha | A General Evaluation of Dermis Sores Identification using MOR-WAVELET Transforms |
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 |