CN105403705A - Continuous blood glucose monitoring equipment comprising blood glucose classification function fault detection module - Google Patents

Continuous blood glucose monitoring equipment comprising blood glucose classification function fault detection module Download PDF

Info

Publication number
CN105403705A
CN105403705A CN201510889922.7A CN201510889922A CN105403705A CN 105403705 A CN105403705 A CN 105403705A CN 201510889922 A CN201510889922 A CN 201510889922A CN 105403705 A CN105403705 A CN 105403705A
Authority
CN
China
Prior art keywords
blood glucose
spe
test
data
class
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
Application number
CN201510889922.7A
Other languages
Chinese (zh)
Other versions
CN105403705B (en
Inventor
赵春晖
宋广健
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Zhejiang University ZJU
Original Assignee
Zhejiang University ZJU
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Zhejiang University ZJU filed Critical Zhejiang University ZJU
Priority to CN201510889922.7A priority Critical patent/CN105403705B/en
Publication of CN105403705A publication Critical patent/CN105403705A/en
Application granted granted Critical
Publication of CN105403705B publication Critical patent/CN105403705B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • GPHYSICS
    • G01MEASURING; TESTING
    • G01NINVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N33/00Investigating or analysing materials by specific methods not covered by groups G01N1/00 - G01N31/00
    • G01N33/48Biological material, e.g. blood, urine; Haemocytometers
    • G01N33/50Chemical analysis of biological material, e.g. blood, urine; Testing involving biospecific ligand binding methods; Immunological testing
    • G01N33/66Chemical analysis of biological material, e.g. blood, urine; Testing involving biospecific ligand binding methods; Immunological testing involving blood sugars, e.g. galactose

Abstract

The invention discloses continuous blood glucose monitoring equipment comprising a blood glucose classification function fault detection module. According to the equipment, a blood glucose PCA monitoring chart is established to perform fault detection on a continuous blood glucose monitoring instrument so that accurate blood glucose signals can be provided. Due to the influences (such as changes of diets and the emotion of a patient) of external source input, the blood glucose fluctuation condition of the patient can change, and correspondingly the SPE value of the blood glucose signals can change obviously; besides, obvious nonlinearity exists between the blood glucose signals, and thus it is impossible to establish a unified and effective control limit for detecting whether a blood glucose meter has a fault or not. In the equipment, the blood glucose signals are divided into multiple classes, and it is ensured that signals in each class are locally linearized and the blood glucose fluctuation conditions in each class are similar, so that it is ensured that fault detection of the blood glucose meter can be better performed in each class, more accurate and credible data are provided for subsequent blood glucose prediction, control and the like, and the equipment has great significance.

Description

A kind of continuous blood sugar monitoring equipment comprising blood sugar classification feature fault detection module
Technical field
The invention belongs to the research field of blood glucose level data process and analysis, particularly relate to a kind of continuous blood sugar monitoring equipment comprising blood sugar classification feature fault detection module.
Background technology
In order to manage Monitoring Blood Glucose level, must measure blood sugar level, what adopt at present is continuous blood sugar monitoring equipment.Along with the fast development of continuous blood sugar monitoring (CGM) equipment, more high-precision control is become possible, the in advance detection of real-time CGM system to high/low blood sugar plays an important role.Just warning can be produced by the relation between more current measured value and high/low blood sugar threshold values, and the high hypoglycemia warning of reporting to the police especially night is timely particularly important for diabetic.But as a rule, various fault can occur CGM system, thus cause the blood glucose information that provides no longer reliable, the glycemic therapeutic decision-making of the high hypoglycemia blood glucose alert that even can lead to errors and mistake.Therefore, the online fault of CGM blood glucose meter seems particularly important.
The present invention is based on PCA surveillance map, whether broken down by the situation of change and then detection CGM blood glucose meter analyzing blood sugar correlation properties.In actual applications, the correlation properties of blood sugar change by the external source imports such as diet affect.In addition, exist significantly non-linear between blood glucose signal, therefore cannot set up an effective pca model of unification and carry out fault detect.By PCA iteration, by blood glucose level data by correlation properties classification, local linearization can be realized simultaneously, thus improves the performance of PCA surveillance map
Summary of the invention
The object of the invention is to the deficiency for conventional P CA surveillance map, propose a kind of continuous blood sugar monitoring equipment comprising blood sugar classification feature fault detection module
The object of the invention is to be achieved through the following technical solutions: a kind of continuous blood sugar monitoring equipment comprising blood sugar classification feature fault detection module, this equipment comprises: for gathering blood sugar for human body information, exports the sensor of blood sugar available signal; For carrying out the signal amplifier amplifying process to the output signal of sensor; Simulating signal for exporting signal amplifier carries out the single-chip microcomputer of digital conversion; Digital signal for exporting single-chip microcomputer carries out the fault detection module of data processing, and fault detection module can be integrated in single-chip microcomputer, also can be used alone; For the display that blood sugar monitoring value SPE fault detection module exported shows; For the storer that data store; The process that described fault detection module carries out data processing comprises the following steps:
(1) blood glucose level data pre-service: the continuous blood sugar signal combination that the single-chip microcomputer obtained with certain sampling period Δ t exports is become one dimension time series data xg 1 × l, wherein xg represents the blood glucose signal detected, l is number of samples;
(2) pca model is set up for continuous blood sugar signal: using a continuous m blood glucose signal as the input of pca model multidimensional, i.e. x (k)=[xg (k-m+1), xg (k-m+2) ... xg (k)] t
T=XP
E=X-XPP T
Wherein X=[x (1), x (2) ... x (n)] tthe training matrix that the blood glucose level data tieed up by n × m forms; P is that m × r ties up load matrix, and r is the pivot number retained, r≤m; T is that n × r ties up score matrix, for blood glucose level data matrix X is in the projection of principal component space; E=[e (1), e (2) ..., e (n)] tfor the residual matrix of n × m dimension.By statistic SPE, blood glucose level data is monitored:
SPE(i)=e(i) Te(i)
The control limit of SPE can be asked for by following formula is approximate:
S P E ~ gχ h , α 2
Wherein g=v/2m, h=2m 2/ v, m are the average of whole SPE, and v is the variance of whole SPE, and α is the degree of confidence of control line.
(3) by the method for PCA iteration, the blood glucose signal of patient is classified, and set up corresponding pca model;
(4) according to the classification results of step (3), set up the PCA surveillance map of the online fault detect of blood glucose meter, specifically comprise following sub-step:
(4.1) blood glucose level data imported into is online substituted into all kinds of pca model respectively, calculates SPE value:
e test,i T=x test T-x test TP iP i T
SPE new,i=e test,i Te test,i
Wherein x testthe blood sugar test data that m × 1 of current time input is tieed up, P ithe load matrix of the i-th class pca model, and e test, ithe residual vector that test data substitutes into the i-th class pca model gained, SPE test, ie test, icorresponding SPE value;
(4.2) testing result is determined:
result test,i=SPE test,i<SPE_Limit i
result test=result test,1∨result test,2∨...∨result test,k
Wherein result test, itest data testing result, wherein result under the i-th class pca model test, i=1 represents testing result normally, result test, i=0 represents testing result extremely; ∨ is the logical operation of getting union; Result testfor final detection result, be the union of testing result under all class pca models, the failure detection result when continuous some moment is exception, then can illustrate that blood glucose meter breaks down.
Further, described step (3) specifically comprises following sub-step:
(3.1) blood glucose level data when enough blood glucose meter normally work is chosen as training data for certain given patient, classify for glucose; And by these training datas as step (2) sprawls into n × m dimension data matrix X of n sampled point composition.Assuming that whole training data belongs to same class (i.e. the first kind), that is:
X 1=X=[x(1),x(2),...x(n)] T
(3.2) utilize the method for PCA iteration that blood sugar is divided into some classes, concrete steps are as follows:
(3.2.1) pca model is set up according to step (2) data matrix to the first kind:
E 1=X 1-X 1P 1P 1 T
SPE 1(i)=e 1(i) Te 1(i)
S P E _ Limit 1 = g 1 &chi; h 1 , &alpha; 2
(3.2.2) limit SPE_Limit is controlled by pca model 1can by data X 1be divided into two classes, namely SPE control limit below with SPE control limit more than wherein import Equations of The Second Kind into, and as the new first kind, namely X 1 = X 1 , < S P E _ Limit 1 ;
(3.2.3) step (3.2.1) and (3.2.2) is repeated until meet first kind stopping criterion for iteration;
(3.2.4) according to step (3.2.1)---(3.2.3), Equations of The Second Kind and all kinds of afterwards can be obtained, until all classes all stop iteration and do not have new class to separate, then judged classification;
(3.2.5) all kinds of final classification results is utilized to set up all kinds of final pca model;
Further, in described step (3.2.3), described stopping criterion for iteration judges to comprise linear decision and the judgement of SPE dispersion degree.
Whether described linear decision is based on pca model, be linear relationship between blood glucose signal after inspection-classification:
A) the i-th class X is extracted ipivot T i, and carry out K mean cluster by X ibe divided into m region;
B) calculate the correlation matrix in each region respectively, be designated as R (h), h=1,2 ..., m.R can be calculated according to the average of each variable and variance (h)in the confidence limit of each element, be expressed in matrix as:
Wherein r i j ( h ) L = exp ( 2 ( &zeta; i j ( h ) - &epsiv; ) ) - 1 exp ( 2 ( &zeta; i j ( h ) - &epsiv; ) ) + 1 With r i j ( h ) U = exp ( 2 ( &zeta; i j ( h ) + &epsiv; ) ) - 1 exp ( 2 ( &zeta; i j ( h ) + &epsiv; ) ) + 1 Be respectively the related coefficient of the i-th variable and jth variable lower bound and the upper bound, with c αcritical value when be standardized normal distribution insolation level being α, n (h)it is the sample number in h region;
C) particle swarm optimization algorithm is utilized to obtain a kth eigenvalue λ of correlation matrix kbound with meet:
&lambda; k max = arg m a x &Delta;R m a x &lambda; k ( R ( h ) + &Delta;R m a x )
&lambda; k m i n = arg m a x &Delta;R m i n &lambda; k ( R ( h ) + &Delta;R min )
And calculate this norm border of not Luo Beini crow limit of residual matrix: and wherein n is that pca model retains pivot number, Δ R maxwith Δ R minin element be the fluctuation of off-diagonal element in R respectively, for determining maximal value and minimum value
D) this norm of not Luo Beini crow of the residual matrix in more each region whether drop in border, if all norms all fall to being within the boundary, then illustrate that data are linear data;
The judgement of described SPE dispersion degree and correlation properties judge:
If the SPE of data meets in class:
var ( SPE i m i ) < 0.5
And only have a small amount of samples point to exceed SPE control limit, that is:
X i , > S P E _ Limit 1 X i &le; &alpha;
Then illustrate that the correlation properties of data in class are similar to, namely blood glucose fluctuation situation is identical.Wherein m ibe the average of the i-th class SPE, α is the degree of confidence of control line.
The invention has the beneficial effects as follows: blood sugar can be divided into some classes according to correlation properties different between glucose signal by the continuous blood sugar monitoring equipment containing blood sugar classification feature fault detection module proposed by the invention, carry out fault detect respectively again, to improve the accuracy of detection of PCA surveillance map, thus promote the confidence level of blood glucose signal.The present invention is easy to implement, for the research of blood sugar process and analysis specifies new direction.
Accompanying drawing explanation
Fig. 1 is the structured flowchart of continuous blood sugar monitoring equipment of the present invention;
Fig. 2 is the process flow diagram of blood sugar classification in continuous blood sugar monitoring equipment of the present invention;
Fig. 3 is the realization flow figure of online fault detect in continuous blood sugar monitoring equipment of the present invention;
Fig. 4 is the SPE surveillance map of original PCA;
Fig. 5 is the SPE surveillance map by PCA Iterative classification, (a) first kind SPE surveillance map (b) Equations of The Second Kind SPE surveillance map (c) the 3rd class SPE surveillance map;
Fig. 6 is the testing result (normal condition) of PCA Iterative classification, (a) euglycemia signal (b) final detection result (c) first kind SPE surveillance map (d) Equations of The Second Kind SPE surveillance map (e) the 3rd class SPE surveillance map;
Fig. 7 is the testing result (failure condition) of PCA Iterative classification, (a) fault blood glucose signal (b) final detection result (c) first kind SPE surveillance map (d) Equations of The Second Kind SPE surveillance map (e) the 3rd class SPE surveillance map.
Embodiment
Below in conjunction with the drawings and specific embodiments, the present invention is described in further detail.
As shown in Figure 1, a kind of continuous blood sugar monitoring equipment comprising blood sugar classification feature fault detection module provided by the invention, this equipment comprises: for gathering blood sugar for human body information, exports the sensor of blood sugar available signal; For carrying out the signal amplifier amplifying process to the output signal of sensor; Simulating signal for exporting signal amplifier carries out the single-chip microcomputer of digital conversion; Digital signal for exporting single-chip microcomputer carries out the fault detection module of data processing, and fault detection module can be integrated in single-chip microcomputer, also can be used alone; For the display that blood sugar monitoring value SPE fault detection module exported shows; For the storer that data store; The process that described fault detection module carries out data processing comprises the following steps:
Step 1: blood glucose level data pre-service
By with the sampling period for the continuous blood sugar monitor signal that 5 minutes obtain is combined into one dimension time series data xg 1 × lwherein, y represents the blood glucose signal detected, l is sample number: in this example, have the sampled signal of three group objects, sampling period is 5 minutes, and the 1st group is teenager's group, and the 2nd group is adult group, 3rd group is children's group, often organize 2 people, three groups of totally 6 people, the sampled signal of each object comprises the data of five days.Add spike class (spike) fault of single time data exception respectively and continue for some time forfeiture susceptibility class (losesensitivity) fault of data exception;
Step 2: pca model is set up for continuous blood sugar signal: using a continuous m blood glucose signal as the input of pca model dimension multidimensional, i.e. x (k)=[xg (k-m+1), xg (k-m+2) ... xg (k)] t, make m=7
T=XP
E=X-XPP T
Wherein X=[x (1), x (2) ... x (n)] tthe training matrix that the blood glucose level data tieed up by n × m forms; P is that m × r ties up load matrix, and r is the pivot number retained, r≤m; T is that n × r ties up score matrix, for blood glucose level data matrix X is in the projection of principal component space; E=[e (1), e (2) ..., e (n)] tfor the residual matrix of n × m dimension.By statistic SPE, blood glucose level data is monitored:
SPE(i)=e(i) Te(i)
The control limit of SPE can be asked for by following formula is approximate:
S P E ~ g&chi; h , &alpha; 2
Wherein g=v/2m, h=2m 2/ v, m are the average of whole SPE, and v is the variance of whole SPE, and α is the degree of confidence of control line.
Step 3: by the method for PCA iteration, the blood glucose signal of patient is classified, and set up corresponding pca model;
(3.1) blood glucose level data when choosing that blood glucose meter normally worked in a day for certain given patient is as training data, and (l=288) classifies for glucose; And by these training datas as step (2) sprawls into n × m dimension data matrix X of n sampled point composition.Assuming that whole training data belongs to same class (i.e. the first kind), that is:
X 1=X=[x(1),x(2),...x(n)] T
(3.2) utilize the method for PCA iteration that blood sugar is divided into some classes, concrete steps are as follows:
(3.2.1) pca model is set up according to step (2) data matrix to the first kind:
E 1=X 1-X 1P 1P 1 T
SPE 1(i)=e 1(i) Te 1(i)
S P E _ Limit 1 - g 1 &chi; h 1 , &alpha; 2
(3.2.2) limit SPE_Limit is controlled by pca model 1can by data X 1be divided into two classes, namely SPE control limit below with SPE control limit more than wherein import Equations of The Second Kind into, and as the new first kind, namely X 1 = X 1 , < S P E _ Limit 1 ;
(3.2.3) step (3.2.1) and (3.2.2) is repeated until meet first kind stopping criterion for iteration (referring to step 3.3);
(3.2.4) according to step (3.2.1)---(3.2.3), Equations of The Second Kind and all kinds of afterwards can be obtained, until all classes all stop iteration and do not have new class to separate, then judged classification;
(3.2.5) all kinds of final classification results is utilized to set up all kinds of final pca model;
(3.3) PCA stopping criterion for iteration judges to comprise linear decision and the judgement of SPE dispersion degree.
(3.3.1) linear decision is based on pca model, and after inspection-classification, whether blood glucose signal linearly promotes:
A) the i-th class X is extracted ipivot T i, and carry out K mean cluster by X ibe divided into m region;
B) calculate the correlation matrix in each region respectively, be designated as R (h), h=1,2 ..., m.R can be calculated according to the average of each variable and variance (h)in the confidence limit of each element, be expressed in matrix as:
Wherein r i j ( h ) L = exp ( 2 ( &zeta; i j ( h ) - &epsiv; ) ) - 1 exp ( 2 ( &zeta; i j ( h ) - &epsiv; ) ) + 1 With r i j ( h ) U = exp ( 2 ( &zeta; i j ( h ) + &epsiv; ) ) - 1 exp ( 2 ( &zeta; i j ( h ) + &epsiv; ) ) + 1 Be respectively the related coefficient of the i-th variable and jth variable lower bound and the upper bound, with c αcritical value when be standardized normal distribution insolation level being α, n (h)it is the sample number in h region;
C) particle swarm optimization algorithm is utilized to obtain a kth eigenvalue λ of correlation matrix kbound with meet:
&lambda; k m a x = arg m a x &Delta;R m a x &lambda; k ( R ( h ) + &Delta;R m a x )
&lambda; k min = arg m a x &Delta;R min &lambda; k ( R ( h ) + &Delta;R m i n )
And calculate this norm border of not Luo Beini crow limit of residual matrix: and wherein n is that pca model retains pivot number, Δ R maxwith Δ R minin element be the fluctuation of off-diagonal element in R respectively, for determining maximal value and minimum value
D) this norm of not Luo Beini crow of the residual matrix in more each region whether drop in border, if all norms all fall to being within the boundary, then illustrate that data are linear data;
The judgement of described SPE dispersion degree and correlation properties judge:
If the SPE of data meets in class:
var ( SPE i m i ) < 0.5
And only have a small amount of samples point to exceed SPE control limit, that is:
X i , > S P E _ Limit 1 X i &le; &alpha;
Then illustrate that the correlation properties of data in class are similar to, namely blood glucose fluctuation situation is identical.Wherein m ibe the average of the i-th class SPE, α is the degree of confidence of control line.
(3.3.2) judgement of SPE dispersion degree and correlation properties judge: if the SPE of data meets in class:
var ( SPE i m i ) < 0.5
And only have a small amount of samples point to exceed SPE control limit, that is:
X i , > S P E _ Limit 1 X i &le; &alpha;
Then illustrate that the correlation properties of data in class are similar to, namely blood glucose fluctuation situation is identical.Wherein m iit is the average of the i-th class SPE.
(4) according to the classification results of step (3), set up the PCA surveillance map of the online fault detect of blood glucose meter, specifically comprise following sub-step:
(4.1) blood glucose level data imported into is online substituted into all kinds of pca model respectively, calculates SPE value:
e test,i T=x test T-x test TP iP i T
SPE new,i=e test,i Te test,i
Wherein x testthe blood sugar test data that m × 1 of current time input is tieed up, P ithe load matrix of the i-th class pca model, and e test, ithe residual vector that test data substitutes into the i-th class pca model gained, SPE test, ie test, icorresponding SPE value;
(4.2) testing result of test data is determined:
result test,i=SPE test,i<SPE_Limit i
result test=result test,1∨result test,2∨...∨result test,k
Wherein result test, itest data testing result, wherein result under the i-th class pca model test, i=1 represents testing result normally, result test, i=0 represents testing result extremely; ∨ is the logical operation of getting union; Result testfor final detection result, be the union of testing result under all class pca models, the failure detection result when continuous some moment is exception, then can illustrate that blood glucose meter breaks down.
(4.3) adopt following two indices as the evaluation index of detection perform:
(4.3.1) rate of false alarm
Wrong report sample point is that this is normal sample point but alarm of breaking down; Rate of false alarm is lower then illustrates that detection perform is better;
(4.3.2) rate of failing to report
Failing to report sample point is that this is fault sample point but alarm of not breaking down; Rate of failing to report is lower then illustrates that detection perform is better;
As can be seen from table 1 table 2, the PCA surveillance map of Based PC A Iterative classification will significantly better than original PCA surveillance map, and rate of false alarm, rate of failing to report all significantly decrease.
Table 1 for 3 groups (teenager's group, adult group and children's groups) the blood sugar sampled data of totally 6 objects adopt the rate of false alarm Comparative result (result means standard deviation represents) of fault detect of classification PCA and original PCA respectively
Object # Classification PCA Original PCA
1 1.51 5.41
2 1.42 3.37
3 5.14 5.05
4 1.33 5.14
5 0.27 4.61
6 4.52 6.56
Means standard deviation 2.36±1.8 5.02±0.95
Table 2 for 3 groups (teenager's group, adult group and children's groups) the blood sugar sampled data of totally 6 objects adopt the rate of failing to report Comparative result (result means standard deviation represents) of fault detect of classification PCA and original PCA respectively

Claims (3)

1. comprise a continuous blood sugar monitoring equipment for blood sugar classification feature fault detection module, it is characterized in that, this equipment comprises: for gathering blood sugar for human body information, exports the sensor of blood sugar available signal; For carrying out the signal amplifier amplifying process to the output signal of sensor; Simulating signal for exporting signal amplifier carries out the single-chip microcomputer of digital conversion; Digital signal for exporting single-chip microcomputer carries out the fault detection module of data processing, and fault detection module can be integrated in single-chip microcomputer, also can be used alone; For the display that blood sugar monitoring value SPE fault detection module exported shows; For the storer that data store; The process that described fault detection module carries out data processing comprises the following steps:
(1) blood glucose level data pre-service: the continuous blood sugar signal combination that the single-chip microcomputer obtained with certain sampling period Δ t exports is become one dimension time series data xg 1 × l, wherein xg represents the blood glucose signal detected, l is number of samples;
(2) pca model is set up for continuous blood sugar signal: using a continuous m blood glucose signal as the input of pca model multidimensional, i.e. x (k)=[xg (k-m+1), xg (k-m+2) ... xg (k)] t
T=XP
E=X-XPP T
Wherein X=[x (1), x (2) ... x (n)] tthe training matrix that the blood glucose level data tieed up by n × m forms; P is that m × r ties up load matrix, and r is the pivot number retained, r≤m; T is that n × r ties up score matrix, for blood glucose level data matrix X is in the projection of principal component space; E=[e (1), e (2) ..., e (n)] tfor the residual matrix of n × m dimension.By statistic SPE, blood glucose level data is monitored:
SPE(i)=e(i) Te(i)
The control limit of SPE can be asked for by following formula is approximate:
S P E ~ g&chi; h , &alpha; 2
Wherein g=v/2m, h=2m 2/ v, m are the average of whole SPE, and v is the variance of whole SPE, and α is the degree of confidence of control line.
(3) by the method for PCA iteration, the blood glucose signal of patient is classified, and set up corresponding pca model;
(4) according to the classification results of step (3), set up the PCA surveillance map of the online fault detect of blood glucose meter, specifically comprise following sub-step:
(4.1) blood glucose level data imported into is online substituted into all kinds of pca model respectively, calculates SPE value:
e t e s t , i T = x t e s t T - x t e s t T P i P i T
SPE new,i=e test,i Te test,i
Wherein x testthe blood sugar test data that m × 1 of current time input is tieed up, P ithe load matrix of the i-th class pca model, and e test, ithe residual vector that test data substitutes into the i-th class pca model gained, SPE test, ie test, icorresponding SPE value;
(4.2) testing result is determined:
result test,i=SPE test,i<SPE_Limit i
result test=result test,1∨result test,2∨...∨result test,k
Wherein result test, itest data testing result, wherein result under the i-th class pca model test, i=1 represents testing result normally, result test, i=0 represents testing result extremely; ∨ is the logical operation of getting union; Result testfor final detection result, be the union of testing result under all class pca models, the failure detection result when continuous some moment is exception, then can illustrate that blood glucose meter breaks down.
2. a kind of continuous blood sugar monitoring equipment comprising blood sugar classification feature fault detection module according to claim 1, it is characterized in that, described step (3) specifically comprises following sub-step:
(3.1) blood glucose level data when enough blood glucose meter normally work is chosen as training data for certain given patient, classify for glucose; And by these training datas as step (2) sprawls into n × m dimension data matrix X of n sampled point composition.Assuming that whole training data belongs to same class (i.e. the first kind), that is:
X 1=X=[x(1),x(2),...x(n)] T
(3.2) utilize the method for PCA iteration that blood sugar is divided into some classes, concrete steps are as follows:
(3.2.1) pca model is set up according to step (2) data matrix to the first kind:
E 1 = X 1 - X 1 P 1 P 1 T
SPE 1(i)=e 1(i) Te 1(i)
S P E _ Limit 1 = g 1 &chi; h 1 , &alpha; 2
(3.2.2) limit SPE_Limit is controlled by pca model 1can by data X 1be divided into two classes, namely SPE control limit below with SPE control limit more than wherein import Equations of The Second Kind into, and as the new first kind, namely X 1 = X 1 , < S P E _ Limit 1 ;
(3.2.3) step (3.2.1) and (3.2.2) is repeated until meet first kind stopping criterion for iteration;
(3.2.4) according to step (3.2.1)-(3.2.3), Equations of The Second Kind and all kinds of afterwards can be obtained, until all classes all stop iteration and do not have new class to separate, then judged classification;
(3.2.5) all kinds of final classification results is utilized to set up all kinds of final pca model.
3. a kind of continuous blood sugar monitoring equipment comprising blood sugar classification feature fault detection module according to claim 2, is characterized in that, in described step (3.2.3), described stopping criterion for iteration judges to comprise linear decision and SPE dispersion degree judges.
Whether described linear decision is based on pca model, be linear relationship between blood glucose signal after inspection-classification:
A) the i-th class X is extracted ipivot T i, and carry out K mean cluster by X ibe divided into m region;
B) calculate the correlation matrix in each region respectively, be designated as R (h), h=1,2 ..., m.R can be calculated according to the average of each variable and variance (h)in the confidence limit of each element, be expressed in matrix as:
Wherein r i j ( h ) L = exp ( 2 ( &zeta; i j ( h ) - &epsiv; ) ) - 1 exp ( 2 ( &zeta; i j ( h ) - &epsiv; ) ) + 1 With r i j ( h ) U = exp ( 2 ( &zeta; i j ( h ) + &epsiv; ) ) - 1 exp ( 2 ( &zeta; i j ( h ) + &epsiv; ) ) + 1 Be respectively the related coefficient of the i-th variable and jth variable lower bound and the upper bound, with c αcritical value when be standardized normal distribution insolation level being α, n (h)it is the sample number in h region;
C) particle swarm optimization algorithm is utilized to obtain a kth eigenvalue λ of correlation matrix kbound with meet:
&lambda; k m a x = arg m a x &Delta;R m a x &lambda; k ( R ( h ) + &Delta;R m a x )
&lambda; k m i n = arg m a x &Delta;R m i n &lambda; k ( R ( h ) + &Delta;R m i n )
And calculate this norm border of not Luo Beini crow limit of residual matrix: and wherein n is that pca model retains pivot number, Δ R maxwith Δ R minin element be the fluctuation of off-diagonal element in R respectively, for determining maximal value and minimum value
D) this norm of not Luo Beini crow of the residual matrix in more each region whether drop in border, if all norms all fall to being within the boundary, then illustrate that data are linear data;
The judgement of described SPE dispersion degree and correlation properties judge:
If the SPE of data meets in class:
var ( SPE i m i ) < 0.5
And only have a small amount of samples point to exceed SPE control limit, that is:
X i , > S P E _ Limit 1 X i &le; &alpha;
Then illustrate that the correlation properties of data in class are similar to, namely blood glucose fluctuation situation is identical.Wherein m ibe the average of the i-th class SPE, α is the degree of confidence of control line.
CN201510889922.7A 2015-12-05 2015-12-05 Continuous blood glucose monitoring equipment comprising blood glucose classification function fault detection module Active CN105403705B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201510889922.7A CN105403705B (en) 2015-12-05 2015-12-05 Continuous blood glucose monitoring equipment comprising blood glucose classification function fault detection module

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201510889922.7A CN105403705B (en) 2015-12-05 2015-12-05 Continuous blood glucose monitoring equipment comprising blood glucose classification function fault detection module

Publications (2)

Publication Number Publication Date
CN105403705A true CN105403705A (en) 2016-03-16
CN105403705B CN105403705B (en) 2017-04-12

Family

ID=55469314

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201510889922.7A Active CN105403705B (en) 2015-12-05 2015-12-05 Continuous blood glucose monitoring equipment comprising blood glucose classification function fault detection module

Country Status (1)

Country Link
CN (1) CN105403705B (en)

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107137093A (en) * 2017-04-20 2017-09-08 浙江大学 A kind of continuous blood sugar monitoring device for including abnormal plasma glucose probability alarm
CN109171754A (en) * 2018-07-27 2019-01-11 清华-伯克利深圳学院筹备办公室 Training method, device, terminal and the storage medium of blood glucose prediction model
CN110379503A (en) * 2019-07-30 2019-10-25 东北大学 A kind of online fault detection and diagnosis system based on continuous blood sugar monitoring system
CN110448306A (en) * 2019-07-30 2019-11-15 东北大学 A kind of online fault detection and diagnosis method based on continuous blood sugar monitoring system
WO2020052147A1 (en) * 2018-09-11 2020-03-19 清华大学合肥公共安全研究院 Monitoring device fault detection method and apparatus
CN116889395A (en) * 2023-08-24 2023-10-17 迈德医疗科技(深圳)有限公司 Noninvasive blood glucose meal classification method and system based on CATPCA

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20070176867A1 (en) * 2006-01-31 2007-08-02 Abbott Diabetes Care, Inc. Method and system for providing a fault tolerant display unit in an electronic device
WO2013186104A1 (en) * 2012-06-12 2013-12-19 Roche Diagnostics Gmbh Medical device and method for displaying medical data by a medical device
CN103824137A (en) * 2014-03-20 2014-05-28 北京信息科技大学 Multi-condition fault prediction method for complex mechanical equipment
CN103853152A (en) * 2014-03-21 2014-06-11 北京工业大学 Batch process failure monitoring method based on AR-PCA (Autoregressive Principal Component Analysis)
US20140218211A1 (en) * 2011-11-23 2014-08-07 Abbott Diabetes Care Inc. Mitigating Single Point Failure of Devices in an Analyte Monitoring System and Methods Thereof
CN104921736A (en) * 2015-06-08 2015-09-23 浙江大学 Continuous blood glucose monitoring device comprising parameter estimation function filtering module

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20070176867A1 (en) * 2006-01-31 2007-08-02 Abbott Diabetes Care, Inc. Method and system for providing a fault tolerant display unit in an electronic device
US20140218211A1 (en) * 2011-11-23 2014-08-07 Abbott Diabetes Care Inc. Mitigating Single Point Failure of Devices in an Analyte Monitoring System and Methods Thereof
WO2013186104A1 (en) * 2012-06-12 2013-12-19 Roche Diagnostics Gmbh Medical device and method for displaying medical data by a medical device
CN103824137A (en) * 2014-03-20 2014-05-28 北京信息科技大学 Multi-condition fault prediction method for complex mechanical equipment
CN103853152A (en) * 2014-03-21 2014-06-11 北京工业大学 Batch process failure monitoring method based on AR-PCA (Autoregressive Principal Component Analysis)
CN104921736A (en) * 2015-06-08 2015-09-23 浙江大学 Continuous blood glucose monitoring device comprising parameter estimation function filtering module

Cited By (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107137093A (en) * 2017-04-20 2017-09-08 浙江大学 A kind of continuous blood sugar monitoring device for including abnormal plasma glucose probability alarm
CN107137093B (en) * 2017-04-20 2019-06-07 浙江大学 A kind of continuous blood sugar monitoring device comprising abnormal plasma glucose probability alarm
CN109171754A (en) * 2018-07-27 2019-01-11 清华-伯克利深圳学院筹备办公室 Training method, device, terminal and the storage medium of blood glucose prediction model
CN109171754B (en) * 2018-07-27 2022-07-05 清华-伯克利深圳学院筹备办公室 Training method and device of blood sugar prediction model, terminal and storage medium
WO2020052147A1 (en) * 2018-09-11 2020-03-19 清华大学合肥公共安全研究院 Monitoring device fault detection method and apparatus
US11385122B2 (en) 2018-09-11 2022-07-12 Hefei Institute For Public Safety Research, Tsinghua University Method and device for detecting fault of monitoring device
CN110379503A (en) * 2019-07-30 2019-10-25 东北大学 A kind of online fault detection and diagnosis system based on continuous blood sugar monitoring system
CN110448306A (en) * 2019-07-30 2019-11-15 东北大学 A kind of online fault detection and diagnosis method based on continuous blood sugar monitoring system
CN110448306B (en) * 2019-07-30 2023-02-07 东北大学 Online fault detection and diagnosis method based on continuous blood glucose monitoring system
CN110379503B (en) * 2019-07-30 2023-08-25 东北大学 Online fault detection and diagnosis system based on continuous blood glucose monitoring system
CN116889395A (en) * 2023-08-24 2023-10-17 迈德医疗科技(深圳)有限公司 Noninvasive blood glucose meal classification method and system based on CATPCA
CN116889395B (en) * 2023-08-24 2024-02-13 迈德医疗科技(深圳)有限公司 Noninvasive blood glucose meal classification method and system based on CATPCA

Also Published As

Publication number Publication date
CN105403705B (en) 2017-04-12

Similar Documents

Publication Publication Date Title
CN105403705A (en) Continuous blood glucose monitoring equipment comprising blood glucose classification function fault detection module
US6556951B1 (en) System and method for intelligent quality control of a process
CN111143438B (en) Workshop field data real-time monitoring and anomaly detection method based on stream processing
CN108255151A (en) A kind of evaluation system of automatic driving vehicle
CN107436277B (en) The single index data quality control method differentiated based on similarity distance
CN106080655A (en) Detection method, device and the train that a kind of train axle temperature is abnormal
CN104921736A (en) Continuous blood glucose monitoring device comprising parameter estimation function filtering module
CN104596780B (en) Diagnosis method for sensor faults of motor train unit braking system
GB2491564A (en) Method of system monitoring
CN107137093B (en) A kind of continuous blood sugar monitoring device comprising abnormal plasma glucose probability alarm
CN203101018U (en) Tunnel ventilation model test measuring system
CN110569876A (en) Non-invasive load identification method and device and computing equipment
CN115876258A (en) Livestock and poultry breeding environment abnormity monitoring and alarming system based on multi-source data
CN113177646A (en) Power distribution equipment online monitoring method and system based on self-adaptive edge proxy
CN113053063A (en) Mobile terminal-based disaster online disposal flow implementation method
CN201017232Y (en) Industry process non-linearity failure diagnosis device based on fisher
CN108319664A (en) A kind of dam and the recognition methods of project security monitoring data error and system
US11331024B2 (en) Calibration method for critical point of mental fatigue based on self-organized criticality
TWI752638B (en) Method and system for detection of driving anomaly
CN104216397B (en) Failure recognition and detection method for intelligent drive axle system
CN106195247B (en) A kind of control system of speed variator based on big data management mode
CN103197060A (en) Autonomous recognition, positioning and quantitative detecting system and detecting method for colloidal gold immunochromatographic boxes
CN206656749U (en) Heat-engine plant sensor fault diagnosis system
CN110243598A (en) Processing method, device and the storage medium of train bearing temperature
CN105389475B (en) A kind of electric power factory equipment fault detection method based on WAVELET PACKET DECOMPOSITION

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant