CN106803009A - A kind of near infrared no-wound blood glucose meter correcting algorithm based on manifold learning and RBF neural - Google Patents
A kind of near infrared no-wound blood glucose meter correcting algorithm based on manifold learning and RBF neural Download PDFInfo
- Publication number
- CN106803009A CN106803009A CN201510832642.2A CN201510832642A CN106803009A CN 106803009 A CN106803009 A CN 106803009A CN 201510832642 A CN201510832642 A CN 201510832642A CN 106803009 A CN106803009 A CN 106803009A
- Authority
- CN
- China
- Prior art keywords
- datas
- rbf neural
- dimension
- manifold learning
- blood glucose
- 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.)
- Pending
Links
- 239000008280 blood Substances 0.000 title claims abstract description 24
- 210000004369 blood Anatomy 0.000 title claims abstract description 24
- 230000001537 neural effect Effects 0.000 title claims abstract description 17
- 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 description 11
- 239000008103 glucose Substances 0.000 title claims description 11
- 241001269238 Data Species 0.000 claims abstract description 22
- 238000000034 method Methods 0.000 claims abstract description 11
- 230000009467 reduction Effects 0.000 claims abstract description 7
- 238000002835 absorbance Methods 0.000 claims abstract description 6
- 230000005540 biological transmission Effects 0.000 abstract description 2
- 230000006870 function Effects 0.000 description 6
- 238000005259 measurement Methods 0.000 description 5
- 239000011159 matrix material Substances 0.000 description 4
- 206010052428 Wound Diseases 0.000 description 2
- 208000027418 Wounds and injury Diseases 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 206010012601 diabetes mellitus Diseases 0.000 description 2
- 238000002329 infrared spectrum Methods 0.000 description 2
- 230000003595 spectral effect Effects 0.000 description 2
- 239000000126 substance Substances 0.000 description 2
- 241000282412 Homo Species 0.000 description 1
- 206010048038 Wound infection Diseases 0.000 description 1
- 238000001467 acupuncture Methods 0.000 description 1
- 230000032683 aging Effects 0.000 description 1
- 235000013405 beer Nutrition 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 210000000624 ear auricle Anatomy 0.000 description 1
- 235000006694 eating habits Nutrition 0.000 description 1
- 238000012417 linear regression Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 210000002569 neuron Anatomy 0.000 description 1
- 230000008520 organization Effects 0.000 description 1
- 238000003909 pattern recognition Methods 0.000 description 1
- 230000005180 public health Effects 0.000 description 1
- 238000011946 reduction process Methods 0.000 description 1
- 238000001228 spectrum Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 238000012549 training Methods 0.000 description 1
Classifications
-
- G06F19/30—
Landscapes
- Health & Medical Sciences (AREA)
- Engineering & Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)
Abstract
The present invention relates to a kind of calibration model algorithm of near-infrared transmission method non-invasive blood sugar instrument.The algorithm carries out nonlinear extensions and Nonlinear Dimension Reduction on the basis of langbobier law, and calibration model is set up with RBF neural.Comprise the following steps：First 5 absorbance datas of wavelength obtained with LED photoelectric receiving tubes are carried out with nonlinear extensions, extended mode includes that Polynomial Expansion and logarithm extend；To totally 25 dimension data after initial data and extension, (Locally Linear Embedding, LLE) algorithm is locally linear embedding into in manifold learning dimensionality reduction theory, 25 dimension datas are reduced to 10 dimensions；5 sensitive wave length data are selected from 10 dimension datas with successive projection algorithm (Successive Projections Algorithm, SPA)；Calibration model is set up with RBF neural to 5 wavelength datas.
Description
Technical field
Chemical Measurement field the invention belongs to set up calibration model near infrared spectrum, and in particular to computer skill
The methods such as art, Chemical Measurement, pattern-recognition, machine learning.
Background technology
With the aggravation of modern humans' life style, the change of eating habit and aging population, onset diabetes rate
It is substantially increased in recent years, oneself is influenceed as one of serious public health problem caused by diabetes and its complication.It is more next
More diabetics need repeatedly to be timed internal blood sugar concentration or monitored at any time according to physical condition daily.Typically
Patient needs to detect daily 4-6 blood sugar, and show the blood glucose meter sold of in the market be mostly invasive, it is necessary to having an acupuncture treatment takes blood, make
Patient is subjected to pain, and it is also possible to cause wound infection, is unfavorable for the frequent detection of blood sugar.For invasive blood sugar test
Major defect, people start to be transferred to notice efficiently, easily on noninvasive dynamics monitoring.
The non-invasive measurement of human blood glucose concentration, can be with near-infrared transmission method measurement ear-lobe position, according to multiple wavelength
Absorbance estimate blood glucose concentration value.But organization of human body is extremely complex, various composition influences each other in blood, based on lambert
The correcting algorithm of Beer law, such as multiple linear regression, PLS method, it is impossible to describe human body spectral absorbance
Nonlinear characteristic, cause to be unable to reach the estimated accuracy of blood sugar concentration the accuracy standard of practicality.
The content of the invention
In view of the limitation of near infrared spectrum calibration model linear method, human body is directed to it is an object of the invention to provide one kind
The non-linear spectral correcting algorithm of measurement of blood sugar concentration.The algorithm by extending the methods such as non-linear factor and Nonlinear Dimension Reduction,
Adaptability of the calibration model to human body is enhanced, the estimated accuracy of blood sugar concentration is effectively increased.
Technical method of the invention, comprises the following steps：
Step one：5 absorbance datas of wavelength to being obtained with LED- photoelectric receiving tubes carry out nonlinear extensions, extended mode
Including the extension of whole binomials and logarithm extension, such as X12, X22, X32, X42, X52, X1*X2, X1*X3, X1*X4, X1*X5, X2*
X3, X2*X4, log (X1) etc.；
Step 2：To total 25 dimension datas of data after initial data and extension, with the part in manifold learning dimensionality reduction theory
25 dimension datas, are reduced to 10 dimensions by linearly embedding (Locally linear embedding, LLE) algorithm；In LLE reduction process
In, it is necessary to be optimized to parameter k and d；Wherein k is Neighbourhood parameter, and d is the intrinsic dimension of sample, the equal round numbers of k and d；This method
The optimal value of k and d is chosen using grid data service；
Step 3：With successive projection algorithm (Successive Projections Algorithm, SPA) from 10 dimension datas
In select most sensitive 5 wavelength datas predicted to blood sugar concentration；
Step 4：Calibration model, the selection of RBF neural input layer are set up with RBF neural to 5 wavelength datas
It it is 5, hidden node selection is 10, output node layer selection is 1.
Brief description of the drawings
Fig. 1 is the flow chart of the near infrared no-wound blood glucose meter correcting algorithm based on manifold learning and RBF neural.
Specific embodiment
In order to make the purpose , technical scheme and advantage of the present invention be clearer, below in conjunction with accompanying drawing, the present invention is entered
Row is further described.It should be noted that specific embodiment described herein is only used to explain the present invention, and without
It is of the invention in limiting.
In step（Fig. 1）, 5 absorbance datas of wavelength to being obtained with LED- photoelectric receiving tubes carry out non-linear expansion
Exhibition, extended mode is that what logarithm extension whole binomials extend, including：X12, X22, X32, X42, X52, X1*X2, X1*X3, X1*
X4, X1*X5, X2*X3, X2*X4, X2*X5, X3*X4, X3*X5, X4*X5, log (X1), log (X2), log (X3), log (X4),
Log (X5), altogether 25 dimension data.By extension, non-linear factor is added in model, enhances model to nonlinear system
Adaptability.
In step 2（Fig. 1）, total 25 dimension datas of data after initial data and extension are managed with manifold learning dimensionality reduction
25 dimension datas, are reduced to 10 dimensions by the Local Liner Prediction in；The step of Local Liner Prediction is：
（1）Calculate any two points x in sample XiAnd xjBetween Euclidean distance dx(i, j), then distance matrix is Dij=dx(i,j)；
（2）According to DijFind out in sample set X apart from xiK nearest point；
（3）It is with expression formulaIt is object function（Wherein）, calculate each point xiWith its point of proximityLinear reconstruction coefficient；
（4）Known ωij, it is with expression formulaObject function, calculates low-dimensional mapping, i.e. solution matrix M=
（I-W）T（I-W）The 2nd to d+1 minimal eigenvalue corresponding characteristic vector Y, wherein I is unit matrix, and W is N × N's
Square formation, if xiAnd xjIt is point of proximity, Wij=ωij, otherwise Wij=0。
To close to the number k and intrinsic dimension d of sample, this method chooses the optimal value of k and d using grid data service.
In step 3（Fig. 1）, to dimensionality reduction after 10 dimension datas, therefrom select to predict most sensitive 5 ripples to blood sugar concentration
The step of data long, SPA, is as follows：
（1）Initialization：n=1（First time iteration）, the optional column vector x in light spectrum matrixi, it is designated as xk(0)（That is k (0)=j）；
（2）Set S is defined as：, i.e., it is not chosen also into wavelength chain
Column vector, x is calculated respectivelyjTo the projection vector of vector in S
（3）Record the sequence number of maximal projection
（4）Using maximum projection as lower whorl projection vector
（5）N=n+1, if n<N, returns to（2）Continue to project.
N × K is so obtained to wavelength combination, to every a pair of xk(0)Calibration model is set up in the combination determined with N respectively, makes
With the quality of prediction RMSEP established models to judge.Select the RMSEP, the x corresponding to it of minimumk(0) *And N*It is as optimal
Wavelength combination.
In step 4（Fig. 1）, calibration model is set up with RBF neural to 5 wavelength datas.Multivariate interpolation
RBF neural has outstanding discrete data interpolation characteristic, can provide best approximation function, its network structure domain and multilayer
Forward direction type network is similar to, and is, to type network, to be made up of input layer, hidden layer, output layer before a kind of 3 layers, and hidden layer neuron is passed
Delivery function is the non-negative nonlinear function to central point radial symmetric and decay, is become from input sheaf space to the space of hidden layer
It is linear to change, and is also linear from implicit sheaf space to output layer spatial alternation.RBF has simple structure, training speed
Hurry up, Function approximation capabilities and classification capacity are strong, in the absence of local optimum problem the advantages of.RBF neural input in the present invention
Node layer selection is 5, and hidden node selection is 10, and output node layer selection is 1.RBF（RBF）Form is：
。
Above is better embodiment of the invention, but protection scope of the present invention not limited to this.It is any to be familiar with this area
Technical staff disclosed herein technical scope in, without conversion or replacement that creative work is expected, should all cover
Within protection scope of the present invention.Therefore the protection domain that protection scope of the present invention should be limited by claim is defined.
Claims (5)
1. a kind of near infrared no-wound blood glucose meter correcting algorithm based on manifold learning and RBF neural, it is characterised in that including
Following steps：
Step one：5 absorbance datas of wavelength to being obtained with LED- photoelectric receiving tubes carry out nonlinear extensions, extended mode
Extended including Polynomial Expansion and logarithm, such as X12,, X1*X2, log (X1) etc.；
Step 2：To total 25 dimension datas of data after initial data and extension, with the part in manifold learning dimensionality reduction theory
25 dimension datas, are reduced to 10 dimensions by linearly embedding (Locally Linear Embedding, LLE) algorithm；
Step 3：With successive projection algorithm (Successive Projections Algorithm, SPA) from 10 dimension datas
In select 5 sensitive wave length data；
Step 4：Calibration model is set up with RBF neural to 5 wavelength datas.
2. the near infrared no-wound blood glucose meter correcting algorithm based on manifold learning and RBF neural according to claim 1,
It is characterized in that：In step one, the nonlinear extensions mode to initial data is quadratic polynomial extension and logarithm extension.
3. the near infrared no-wound blood glucose meter correcting algorithm based on manifold learning and RBF neural according to claim 1,
It is characterized in that：In step 2, it is locally linear embedding into and 25 dimension datas is reduced to 10 dimensions, the parameter being locally linear embedding into drops in LLE
, it is necessary to be optimized to parameter k and d during dimension, wherein k is Neighbourhood parameter, and d is the intrinsic dimension of sample, the equal round numbers of k and d
This method chooses the optimal value of k and d using grid data service.
4. the near infrared no-wound blood glucose meter correcting algorithm based on manifold learning and RBF neural according to claim 1,
It is characterized in that：In step 3, selected from 10 dimension datas with SPA methods and 5 most sensitive number of wavelengths are predicted to blood sugar concentration
According to.
5. the near infrared no-wound blood glucose meter correcting algorithm based on manifold learning and RBF neural according to claim 1,
It is characterized in that：In step 4, the selection of RBF neural input layer is 5, and hidden node selection is 10, output layer
Node selection is 1.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510832642.2A CN106803009A (en) | 2015-11-26 | 2015-11-26 | A kind of near infrared no-wound blood glucose meter correcting algorithm based on manifold learning and RBF neural |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510832642.2A CN106803009A (en) | 2015-11-26 | 2015-11-26 | A kind of near infrared no-wound blood glucose meter correcting algorithm based on manifold learning and RBF neural |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106803009A true CN106803009A (en) | 2017-06-06 |
Family
ID=58975779
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510832642.2A Pending CN106803009A (en) | 2015-11-26 | 2015-11-26 | A kind of near infrared no-wound blood glucose meter correcting algorithm based on manifold learning and RBF neural |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106803009A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112450919A (en) * | 2020-11-11 | 2021-03-09 | 云南省第一人民医院 | Blood glucose monitoring device for evaluating nighttime hypoglycemia of type 2 diabetes patients |
| CN113598763A (en) * | 2021-08-05 | 2021-11-05 | 重庆大学 | Non-invasive blood glucose detection device and method based on MIC-PCA-NARX correction algorithm |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20020193671A1 (en) * | 2000-08-21 | 2002-12-19 | Ciurczak Emil W. | Near infrared blood glucose monitoring system |
| CN102198004A (en) * | 2010-03-25 | 2011-09-28 | 葛歆瞳 | Noninvasive near-infrared electronic blood-glucose meter |
| CN105044022A (en) * | 2015-08-06 | 2015-11-11 | 黑龙江大学 | Method for rapidly nondestructively detecting wheat hardness based on near infrared spectrum technology and application |
-
2015
- 2015-11-26 CN CN201510832642.2A patent/CN106803009A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20020193671A1 (en) * | 2000-08-21 | 2002-12-19 | Ciurczak Emil W. | Near infrared blood glucose monitoring system |
| CN102198004A (en) * | 2010-03-25 | 2011-09-28 | 葛歆瞳 | Noninvasive near-infrared electronic blood-glucose meter |
| CN105044022A (en) * | 2015-08-06 | 2015-11-11 | 黑龙江大学 | Method for rapidly nondestructively detecting wheat hardness based on near infrared spectrum technology and application |
Non-Patent Citations (3)
| Title |
|---|
| JYOTI YADAV ET.AL: "Noninvasive Prediction of Glucose by Near-Infrared Diffuse Reflectance Spectroscopy", 《CLINICAL CHEMISTRY》 * |
| 刘创: "近红外血糖无创检测校正模型研究", 《中国优秀博硕士学位论文全文数据库 (硕士)》 * |
| 成忠 等: "连续投影算法及其在小麦近红外光谱波长选择中的应用", 《光谱学与光谱分析》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112450919A (en) * | 2020-11-11 | 2021-03-09 | 云南省第一人民医院 | Blood glucose monitoring device for evaluating nighttime hypoglycemia of type 2 diabetes patients |
| CN112450919B (en) * | 2020-11-11 | 2022-11-04 | 云南省第一人民医院 | Blood glucose monitoring device for evaluating nighttime hypoglycemia of type 2 diabetes patients |
| CN113598763A (en) * | 2021-08-05 | 2021-11-05 | 重庆大学 | Non-invasive blood glucose detection device and method based on MIC-PCA-NARX correction algorithm |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104287706B (en) | A kind of health status real time monitor and diagnosis and treatment commending system | |
| CN110944577B (en) | Method and system for detecting blood oxygen saturation | |
| US8986207B2 (en) | Systems and methods for providing sensor arrays for detecting physiological characteristics | |
| US8457706B2 (en) | Estimation of a physiological parameter using a neural network | |
| EP4093270A1 (en) | Method and system for personalized prediction of infection and sepsis | |
| US20250325230A1 (en) | Blood glucose prediction method and device based on optical signal features and metabolic thermal features | |
| WO2011041216A1 (en) | Systems and method for identifying non-corrupted signal segments for use in determining physiological parameters | |
| US20250261866A1 (en) | Blood pressure detection apparatus based on arteriolar photoplethysmogram | |
| CN111358453A (en) | Blood pressure classification prediction method and device | |
| US20240194298A1 (en) | Method for classifying quality of biological sensor data | |
| Premalatha et al. | Wireless IoT and cyber-physical system for health monitoring using honey badger optimized least-squares support-vector machine | |
| CN106446765A (en) | Health state evaluation system based on multidimensional physiological big data depth learning | |
| Ma et al. | A prediction method for transport stress in meat sheep based on GA-BPNN | |
| CN120225115A (en) | Device, system and method for blood glucose monitoring | |
| Cheng et al. | Study of a noninvasive blood glucose detection model using the near-infrared light based on SA-NARX | |
| CN106803009A (en) | A kind of near infrared no-wound blood glucose meter correcting algorithm based on manifold learning and RBF neural | |
| Ong et al. | Characteristic wavelength optimization for partial least squares regression using improved flower pollination algorithm | |
| US11903699B2 (en) | Method and apparatus for light-weight, non-invasive, point of care diabetes screening device | |
| CN112788200B (en) | Method and device for determining frequency spectrum information, storage medium and electronic device | |
| CN114869276A (en) | Noninvasive hemoglobin concentration detection method and system | |
| Li et al. | Prediction of total volatile basic nitrogen (TVB‐N) in fish meal using a metal‐oxide semiconductor electronic nose based on the VMD‐SSA‐LSTM algorithm | |
| Baciu et al. | Anomaly detection in multi-wavelength photoplethysmography using lightweight machine learning algorithms | |
| Sambo et al. | A Bayesian network for probabilistic reasoning and imputation of missing risk factors in type 2 diabetes | |
| TW202531254A (en) | Signal quality assessement system and blood glucose level prediction system based on photoplethysmography | |
| CN117558441A (en) | A non-invasive blood glucose prediction method, system and chip |
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 | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20170606 |
|
| WD01 | Invention patent application deemed withdrawn after publication |