CN116020007A - Closed-loop artificial pancreas insulin infusion control system - Google Patents

Closed-loop artificial pancreas insulin infusion control system Download PDF

Info

Publication number
CN116020007A
CN116020007A CN202111242326.1A CN202111242326A CN116020007A CN 116020007 A CN116020007 A CN 116020007A CN 202111242326 A CN202111242326 A CN 202111242326A CN 116020007 A CN116020007 A CN 116020007A
Authority
CN
China
Prior art keywords
blood glucose
insulin
infusion
algorithm
risk
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
Application number
CN202111242326.1A
Other languages
Chinese (zh)
Inventor
杨翠军
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.)
Medtrum Technologies Inc
Original Assignee
Medtrum Technologies Inc
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 Medtrum Technologies Inc filed Critical Medtrum Technologies Inc
Priority to CN202111242326.1A priority Critical patent/CN116020007A/en
Publication of CN116020007A publication Critical patent/CN116020007A/en
Pending legal-status Critical Current

Links

Images

Landscapes

  • External Artificial Organs (AREA)

Abstract

The invention discloses a closed-loop artificial pancreas insulin infusion control system, which comprises: the detection module is used for continuously detecting the current blood glucose value G; the program module is connected with the detection module and is used for presetting an rMPC algorithm for converting the blood sugar which is asymmetric in the original physical space into an approximately symmetric blood sugar risk space and a target blood sugar value G in the program module B The rMPC algorithm calculates insulin infusion instructions based on blood glucose risk; and the infusion module is connected with the program module, and the program module controls the infusion module to infuse insulin according to the insulin infusion instruction calculated by the rMPC algorithm. The rMPC algorithm in the system converts the blood sugar which is asymmetric in the original physical space into the blood sugar risk space which is approximately symmetric, so that the characteristics of the recent effect of the change considered by the classical MPC algorithm can be reserved, the asymmetry of blood sugar distribution is considered, the system has the advantages of accuracy and flexibility, and the accurate control of the closed-loop artificial insulin infusion system is realizedAnd (5) preparing.

Description

Closed-loop artificial pancreas insulin infusion control system
Technical Field
The invention mainly relates to the field of medical instruments, in particular to a closed-loop artificial pancreas insulin infusion control system.
Background
The pancreas of normal people can automatically secrete needed insulin/glucagon according to the glucose level in the blood of the human body, so that the reasonable blood sugar fluctuation range is maintained. However, the pancreas function of diabetics is abnormal, and insulin required by human body cannot be normally secreted. Diabetes is a metabolic disease, a life-long disease. The existing medical technology cannot radically cure diabetes, and the occurrence and development of diabetes and complications thereof can be controlled only by stabilizing blood sugar.
Diabetics need to test blood glucose before injecting insulin into the body. The current detection means can continuously detect blood sugar and send blood sugar data to the display device in real time, so that the blood sugar data is convenient for a user to check, and the detection method is called continuous glucose detection (Continuous Glucose Monitoring, CGM). The method needs to attach the detection device to the skin surface, and the probe carried by the detection device is penetrated into subcutaneous tissue fluid to complete detection. According to the blood glucose value detected by CGM, the infusion device inputs the insulin required currently into the skin, thereby forming a closed-loop or semi-closed-loop artificial pancreas.
At present, in order to realize the insulin infusion of closed-loop or semi-closed-loop control, a model-prediction-control (MPC) algorithm predicts the future behavior of insulin pump output under the change of blood sugar by using a prediction model, can conveniently process additional inputs such as meal, exercise and the like, has definite physiological meaning of model parameters, is convenient for personalized setting and optimization, and realizes the regulation and control of insulin infusion, thereby being widely researched and used. However, the MPC algorithm faces the dilemma that an accurate model is difficult to establish and has large calculation amount, and the predicted infusion deviation of the MPC algorithm can occur.
Thus, there is a need in the art for a closed loop artificial pancreatic insulin infusion control system that incorporates an optimized MPC algorithm.
Disclosure of Invention
The embodiment of the invention discloses a closed-loop artificial pancreatic insulin infusion control system, wherein an rMPC algorithm in the system converts blood glucose which is asymmetric in an original physical space into an approximately symmetric blood glucose risk space, and meanwhile, the recent change and the distribution asymmetry of the blood glucose are considered, so that the closed-loop artificial pancreatic insulin gland infusion control system has the advantages of accuracy and flexibility, and the accurate control of the closed-loop artificial pancreatic insulin gland infusion system is realized.
The invention discloses a closed-loop artificial pancreas insulin infusion control system, which comprises: the detection module is used for continuously detecting the current blood glucose value G; program module connected with the detection module, wherein the program module is used for presetting asymmetry in the original physical spacerMPC algorithm for conversion of blood glucose into an approximately symmetrical blood glucose risk space and target blood glucose value G B The rMPC algorithm calculates insulin infusion instructions based on blood glucose risk; and the infusion module is connected with the program module, and the program module controls the infusion module to infuse insulin according to the insulin infusion instruction calculated by the rMPC algorithm.
According to one aspect of the invention, the rMPC algorithm includes a predictive model, a cost function and constraints, the predictive model being:
x t+1 ＝Ax t +BI t
G t ＝Cx t
wherein:
x t+1 a state parameter representing the next moment in time,
Figure BDA0003319932270000021
x t a state parameter representing the current time of day,
Figure BDA0003319932270000022
I t indicating the insulin infusion quantity at the current time;
G t indicating the blood glucose concentration at the current time.
The parameter matrix is as follows:
Figure BDA0003319932270000023
Figure BDA0003319932270000024
C＝[1 0 0]
b 1 ,b 2 ,b 3 k is an a priori value.
The cost function is:
Figure BDA0003319932270000025
wherein:
I′ t+j indicating a change in insulin infusion after step j;
r t+j indicating the blood glucose risk value after the j-th step;
t represents the current time;
n, P is the number of steps in the control time window and the prediction time window, respectively;
r is the weighting coefficient of the insulin component therein.
According to one aspect of the invention, the risk of blood glucose r t+j The calculation is as follows:
Figure BDA0003319932270000031
G t+j indicating the blood glucose level detected at step j.
According to one aspect of the invention, the risk of blood glucose r t+j The relation with the calculation is as follows:
Figure BDA0003319932270000032
G t+j indicating the blood glucose level detected at step j.
According to one aspect of the invention, the risk of blood glucose r t+j The calculation is as follows:
Figure BDA0003319932270000033
wherein,,
r(G t+j )＝10*f(G t+j ) 2
wherein f (G) t+j ) As a conversion function, the formula is:
f(G t+j )＝1.509*[(ln(G t+j )) 1.084 -5.381]
G t+j indicating the blood glucose level detected at step j.
According to one aspect of the invention, the risk of blood glucose r t+j The calculation is as follows:
Figure BDA0003319932270000034
G t+j indicating the blood glucose level detected at step j.
According to one aspect of the invention, the risk of blood glucose r t+j Is defined as: r is |r t+j |＝min(|r t+j |,n)。
According to one aspect of the present invention, the maximum value n is defined to be 0 to 80mg/dL.
According to one aspect of the invention, r t+j The maximum value n is defined to be 60mg/dL.
According to an aspect of the present invention, the blood glucose level G detected at the jth step t+j Greater than the target blood glucose value G B Risk of blood glucose r t+j The calculation is as follows:
r t+j =r(G t+j ),if G t+j ＞G B ，
wherein,,
r(G t+j )＝10*f(G t+j ) 2
wherein f (G) t+j ) As a conversion function, the formula is:
f(G t+j )＝1.509*[(ln(G t+j )) 1.084 -5.381]；
blood glucose level G detected at the j-th step t+j Not greater than the target blood glucose value G B When blood glucose risk r is calculated as:
r t+j ＝G t+j -G B ,if G t+j ≤G B
G t+j indicating the blood glucose level detected at step j.
According to one aspect of the invention, the risk of blood glucose r t+j Is defined as: r is |r t+j |＝min(|r t+j |,n)。
According to one aspect of the present invention, the maximum value n is defined to be 0 to 80mg/dL.
According to one aspect of the invention, r t+j The maximum value n is defined to be 60mg/dL.
According to one aspect of the present invention, the blood glucose level G detected at step j t+j Not greater than the target blood glucose value G B Risk of blood glucose r t+j The calculation is as follows:
r t+j =-r(G t+j ),if G t+j ≤G B
wherein:
r(G t+j )＝10*f(G t+j ) 2
conversion function f (G t+j ) The following steps:
f(G t+j )=1.509*[(ln(G t+j )) 1.084 -5.381]
blood glucose level G detected at the j-th step t+j Greater than the target blood glucose value G B At the time, the blood glucose risk r t+j The calculation is as follows:
r t+j ＝-4.8265*10 4 -4*G t+j 2 +0.45563*G t+j -44.855,if G t+j ＞G B
G t+j indicating the blood glucose level detected at step j.
According to an aspect of the present invention, the blood glucose level G detected when the step j is performed t+j Not greater than the target blood glucose value G B At the time, the blood glucose risk r t+j The calculation is as follows:
r t+j ＝-r(G t+j ),if G t+j ≤G B
wherein:
r(G t+j )＝10*f(G t+j ) 2
conversion function f (G t+j ) The following steps:
f(G t+j )＝1.509*[(ln(G t+j )) 1.084 -5.381]
blood glucose level G detected at the j-th step t+j Greater than the target blood glucose value G B At the time, the blood glucose risk r t+j The calculation is as follows:
Figure BDA0003319932270000051
G t+j indicating the blood glucose level detected at step j.
According to one aspect of the invention, insulin is subjected to a meal insulin and a non-meal insulin differentiation process when calculating the amount of Insulin (IOB) that has not yet been reacted in vivo.
According to an aspect of the present invention, the blood glucose level G detected at the j-th step t+j Not greater than the target blood glucose value G B At the time of
Risk of blood glucose r t+j The calculation is as follows:
r t+j ＝-r(G t+j ),if G t+j ≤G B
wherein:
r(G t+j )=10*f(G t+j ) 2
conversion function f (G t+j ) The following steps:
f(G t+j )=1.509*[(ln(G t+j )) 1.084 -5.381]
blood glucose level G detected at the j-th step t+j Greater than the target blood glucose value G B At the time, the blood glucose risk r t+j The calculation is as follows:
Figure BDA0003319932270000061
G t+j indicating the blood glucose level detected at step j.
According to one aspect of the present invention, the target blood glucose level G B 80-140 mg/dL.
According to one aspect of the present invention, the target blood glucose level G B 110-120 mg/dL.
According to one aspect of the invention, the rMPC algorithm further includes one or more of the following:
(3) Deducting the amount of unabsorbed plasma insulin in the body based on the insulin absorption delay in the artificial pancreas control system
Figure BDA0003319932270000062
Figure BDA0003319932270000063
Wherein,,
I t+j an infusion indication to be sent to the insulin infusion system at step j;
rI c(t+j) an infusion instruction sent to the insulin infusion system at step j after risk conversion;
gamma is the compensation coefficient of the estimated plasma insulin concentration to the algorithm output;
Figure BDA0003319932270000064
estimating for plasma insulin concentration;
(4) deducting the amount of insulin that has not been acted in the body based on the insulin action delay in the artificial pancreas control system
IOB(t+j)：
rI′ t+j ＝rI t+j -IOB(t+j)
rI′ t+j Indicating an infusion instruction sent to the insulin infusion system after deducting the IOB at step j after risk conversion;
rI t+j an infusion instruction sent to the insulin infusion system at step j after risk conversion;
IOB (t+j) represents the amount of insulin that has not been acted upon in vivo at time t+j;
(3) the sensing delay of blood glucose and interstitial fluid glucose concentrations is compensated for using an autoregressive method.
According to one aspect of the invention, the plasma insulin concentration estimate is obtained by an autoregressive method.
According to one aspect of the invention, the compensation factor γ is 0.4-0.6.
According to one aspect of the invention, the compensation coefficient γ is 0.5.
According to one aspect of the invention, the amount of insulin IOB (t+j) that has not been reacted in vivo is obtained from an IOB curve.
According to one aspect of the invention, in calculating the insulin quantity IOB (t+j) that has not yet been acted on in vivo, insulin quantities are subjected to meal insulin and non-meal insulin treatments:
IOB(t+j)＝IOB m,t+j +IOB o,t+j
Figure BDA0003319932270000071
wherein,,
IOB m,t+j indicating the amount of meal insulin that has not been active in the body at time t+j;
IOB o,t+j indicating the amount of non-prandial insulin not yet active in the body at time t+j;
D i (i=2-8) represents the corresponding coefficients of IOB curves corresponding to insulin action times i, respectively;
I m,t+j indicating the meal insulin quantity at time t+j;
I 0,t+j representing the non-meal insulin quantity at time t+j;
IOB (t+j) represents the amount of insulin that has not yet been acted upon in vivo at time t+j.
According to one aspect of the invention, two of the detection module, the program module and the infusion module are connected to each other to form a unitary structure and are adhered to different locations of the skin with the third module.
According to one aspect of the invention, the detection module, the program module and the infusion module are connected to form a unitary structure and are adhered to the same location of the skin.
Compared with the prior art, the technical scheme of the invention has the following advantages:
In the closed-loop artificial pancreatic insulin infusion control system disclosed by the invention, the rMPC algorithm converts the blood glucose which is asymmetric in the original physical space into an approximately symmetric blood glucose risk space, so that the characteristic of the recent effect of the change considered by the classical MPC algorithm is reserved, the asymmetry of blood glucose distribution is considered, the advantages of accuracy and flexibility are achieved, and the accurate control of the closed-loop artificial pancreatic insulin gland infusion system is realized.
Furthermore, the rMPC algorithm can be processed through or in combination with the segmentation weighting, the relative value, the BRGI method and the improved CVGA method respectively, and the target blood sugar concentration or the zero risk point blood sugar concentration or the equal risk point data pair can be flexibly selected according to actual conditions, so that the rMPC algorithm is more robust, and the rMPC algorithm still has slow adjustment capability in a relatively gentle section, so that the closed-loop artificial pancreas can face more complex use situations, and more accurate blood sugar control is realized.
Furthermore, the rMPC algorithm compensates for delays in insulin absorption, insulin onset, and blood glucose and interstitial fluid glucose concentration sensing, resulting in a more reliable output calculated by the rMPC algorithm.
Further, to compensate for insulin onset delay, in the rmcp algorithm, the IOB is processed with meal insulin and other insulin than meal, so that insulin can be cleared more quickly during meal and hyperglycemia, larger insulin output can be obtained, and blood glucose regulation is faster. While approaching the target, the blood glucose regulation is more conservative with a longer insulin action time profile.
Further, the detection module, the program module and the injection molding module are connected to form a whole structure and are adhered to the same position of the skin. The three modules are connected into a whole and stuck at the same position, so that the number of the skin sticking devices of the user is reduced, and the interference of more sticking devices on the movement extension of the user is further reduced; meanwhile, the problem of unsmooth wireless communication between the separation devices is effectively solved, and user experience is further enhanced.
Drawings
FIG. 1 is a schematic diagram of the modular relationship of a closed-loop artificial pancreatic insulin infusion control system in accordance with an embodiment of the invention;
FIG. 2 is a graph comparing the blood glucose of a risk space and an original physical space obtained by a segmentation weighting process and a relative value transformation method according to an embodiment of the present invention;
FIG. 3 is a graph comparing risk space converted by BGRI and CVGA methods to blood glucose in an original physical space according to an embodiment of the present invention;
FIG. 4 is an insulin IOB curve according to one embodiment of the invention;
FIG. 5 is a schematic diagram of four clinically optimal basal rate setting types of a main stream referenced in one embodiment in accordance with the invention;
FIG. 6 is a schematic diagram of the modular relationship of a closed-loop artificial pancreatic insulin infusion control system in accordance with another embodiment of the invention;
FIG. 7 is a schematic diagram of the modular relationship of a closed-loop artificial pancreatic insulin infusion control system in accordance with yet another embodiment of the invention;
FIG. 8 is a schematic diagram of a closed-loop artificial pancreas multi-drug infusion control system module relationship according to another embodiment of the invention;
FIG. 9 is a schematic diagram of a dual drug switch according to one embodiment of the present invention;
fig. 10 is a schematic diagram showing the modular relationship of a closed-loop artificial pancreatic insulin infusion control system according to another embodiment of the invention.
Detailed Description
As mentioned above, in the artificial pancreas of the prior art, the MPC algorithm is difficult to build and has a great amount of calculation, so that the predicted infusion deviation of the MPC algorithm may occur.
In order to solve the problem, the invention provides a closed-loop artificial pancreatic insulin infusion control system, wherein a rMPC algorithm is preset in the system, and the asymmetrical blood sugar in the original physical space is converted into an approximately symmetrical blood sugar risk space, so that the characteristics of the recent effect of the change considered by the classical MPC algorithm are reserved, the asymmetry of blood sugar distribution is considered, and the system has the advantages of accuracy and flexibility, and the accurate control of the closed-loop artificial pancreatic insulin gland infusion system is realized.
Various exemplary embodiments of the present invention will now be described in detail with reference to the accompanying drawings. It should be understood that the relative arrangement of parts and steps, numerical expressions and numerical values set forth in these embodiments should not be construed as limiting the scope of the present invention unless it is specifically stated otherwise.
Furthermore, it should be understood that the dimensions of the various elements shown in the figures are not necessarily drawn to actual scale, e.g., the thickness, width, length, or distance of some elements may be exaggerated relative to other structures for ease of description.
The following description of the exemplary embodiment(s) is merely illustrative, and is in no way intended to limit the invention, its application, or uses. Techniques, methods, and apparatus known to one of ordinary skill in the relevant art may not be discussed in detail herein, but where applicable, should be considered part of the present specification.
It should be noted that like reference numerals and letters refer to like items in the following figures, and thus once an item is defined or illustrated in one figure, no further discussion thereof will be necessary in the following figure description.
FIG. 1 is a schematic diagram of the module relationship of a closed-loop artificial pancreatic insulin infusion control system according to an embodiment of the invention.
The embodiment of the invention discloses a closed-loop artificial pancreas insulin infusion control system which mainly comprises a detection module 100, a program module 101 and an infusion module 102.
The detection module 100 is used for continuously detecting the current blood glucose level of the user. Typically, the detection module 100 is a continuous glucose meter (Continuous Glucose Monitoring, CGM) that can detect the current blood glucose level of the user in real time, monitor the change of blood glucose, and send the current blood glucose level to the program module 101.
The program module 101 is used to control the operation of the detection module 100 and the injection molding block 102. Thus, the program module 101 is connected to the detection module 100 and the injection molding block 102, respectively. Here, the connection includes a conventional electrical connection or a wireless connection.
The infusion module 102 contains the mechanical structure necessary for infusing insulin and is controlled by the program module 101. Based on the current insulin infusion data from the program module 101, the infusion module 102 infuses the currently desired insulin into the user. At the same time, the infusion state of the infusion module 102 can also be fed back into the program module 101 in real time.
The embodiment of the present invention is not limited to the specific positions and connection relations of the detection module 100, the program module 101 and the injection molding block 102, as long as the foregoing functional conditions can be satisfied.
In one embodiment of the present invention, the three are electrically connected to each other to form a unitary structure. Therefore, the three are stuck on the same position of the skin of the user. The three modules are connected into a whole and stuck at the same position, so that the number of the skin sticking devices of the user is reduced, and the interference of more sticking devices on the activities of the user is further reduced; meanwhile, the problem of wireless communication reliability between the separation devices is effectively solved, and user experience is further enhanced.
In another embodiment of the present invention, the program module 101 and the injection molding module 102 are interconnected to form a unitary structure, while the detection module 100 is separately disposed in another structure. At this time, the detection module 100 and the program module 101 mutually transmit wireless signals to achieve connection with each other. Thus, the program module 101 and the infusion module 102 are attached to one location on the user's skin, while the detection module 100 is attached to another location on the user's skin.
As in yet another embodiment of the present invention, the program module 101 and the detection module 100 are interconnected to form the same device, while the infusion module 102 is provided separately in another configuration. The infusion module 102 and the program module 101 transmit wireless signals to each other to effect the connection to each other. Thus, the program module 101 and the detection module 100 may be attached to a certain location on the skin of the user, while the infusion module 102 may be attached to another location on the skin of the user.
As in yet another embodiment of the present invention, the three are disposed in different configurations, respectively. Therefore, the three are respectively stuck on different positions of the skin of the user. At this time, the program module 101 transmits wireless signals to each other with the detection module 100 and the injection molding module 102, respectively, to achieve connection with each other.
It should be noted that, the program module 101 according to the embodiment of the present invention further has functions of storing, recording, accessing a database, and the like, and thus, the program module 101 can be reused. Therefore, not only can the physical condition data of the user be stored, but also the production cost and the use cost of the user are saved. As described above, when the lifetime of the detection module 100 or the infusion module 102 is terminated, the program module 101 may be separated from the detection module 100, the infusion module 102, or both the detection module 100 and the infusion module 102.
In general, the detection module 100, the program module 101 and the infusion module 102 have different service lives. Therefore, when the three are electrically connected to form the same device, the three can be separated from each other. If a certain module ends the service life first, the user can only replace the module, and keep the other two modules to continue to use.
Here, it should be noted that the program module 101 according to the embodiment of the present invention may further include a plurality of sub-modules. Depending on the functions of the sub-modules, the different sub-modules may be respectively provided in different configurations, and are not particularly limited herein as long as the control conditions of the program module 101 can be satisfied.
Specifically, a rPID (risk-proportional-integral-derivative) algorithm for converting the blood glucose asymmetric in the original physical space into the blood glucose risk approximately symmetric in the risk space is preset in the program module 101, the rPID algorithm is obtained by performing conversion processing on the basis of a classical PID (proportional-integral-derivative) algorithm, a specific processing manner will be described in detail below, and the program module 101 controls the infusion module 102 to infuse insulin according to a corresponding infusion instruction calculated by the rPID algorithm.
The classical PID algorithm can be expressed by the following formula:
Figure BDA0003319932270000101
wherein:
K P gain coefficients that are proportional to the portion;
K I is the gain factor of the integrating part;
K D is a differentiating partA divided gain coefficient;
g represents the current blood glucose level;
G B indicating a target blood glucose level;
c represents a constant;
PID (t) represents an infusion indication sent to the insulin infusion system.
Considering the actual distribution characteristics of glucose concentration in diabetics, for example, normal blood sugar ranges from 80 to 140mg/dL, and can be relaxed to 70 to 180mg/dL, general hypoglycemia can reach 20 to 40mg/dL, and hyperglycemia can reach 400 to 600mg/dL.
The distribution of high/low blood sugar has obvious asymmetry in the original physical space, the hyperglycemia risk and the hypoglycemia risk corresponding to the same degree of deviation of blood sugar from the normal range in clinical practice are obviously different, for example, the reduction of 70mg/dL from 120mg/dL to 50mg/dL can be regarded as serious hypoglycemia, the high clinical risk is realized, and emergency measures such as carbohydrate supplementation and the like are needed to be adopted; while an increase of 70mg/dL from 120mg/dL to 190mg/dL is just beyond the normal range, the blood glucose level is not so high for diabetics, and is often reached in daily cases, and no treatment is needed.
Aiming at the asymmetric characteristic of clinical risk of glucose concentration, the blood glucose asymmetric in the original physical space is converted into the blood glucose risk approximately symmetric in the risk space, so that the PID algorithm is more robust.
Correspondingly, the rPID algorithm formula is converted into the following form:
Figure BDA0003319932270000111
wherein:
rPID (t) represents an infusion instruction sent to the insulin infusion system after risk conversion;
r represents a blood glucose risk;
the meaning of the other symbols is as described above.
In order to maintain the stability of PID integral, combining the physiological action of insulin in reducing blood sugarIn one embodiment of the present invention, the input parameter to PID, blood glucose bias ge=g-G B Treatment, e.g. of Ge=G-G B The segmentation weighting process is made as follows:
Figure BDA0003319932270000112
in another embodiment of the present invention, for a blood glucose G greater than the target blood glucose G B The offset of (2) is converted using the relative value as follows:
Figure BDA0003319932270000113
FIG. 2 is a graph comparing blood glucose risk space obtained by piecewise weighting and relative value transformation to the original physical space.
In the original PID algorithm, the blood glucose risks (namely Ge) at the two sides of the target blood glucose value show serious asymmetry consistent with the original physical space, and after the blood glucose risk is converted into the blood glucose risk space, the blood glucose risks at the two sides of the blood glucose target value are approximately symmetrical, so that the integral term can be kept stable, and the rPID algorithm is more robust.
In another embodiment of the invention, there is a fixed zero risk point at risk transition, and data on both sides of the zero risk point is processed. The original parameters corresponding to points greater than zero risk are positive values when being converted into a risk space, and the original parameters corresponding to points less than zero risk are negative values when being converted into the risk space. In particular, the classical glycemic risk index (BGRI) method can be consulted, which is based on clinical practice, regarding that the clinical risk of hypoglycemia of 20mg/dL and hyperglycemia of 600mg/dL is comparable, and blood glucose in the range of 20-600mg/dL is treated as a whole by logarithmization. Setting the blood glucose value corresponding to the zero risk point of the method as a target blood glucose value G B . The risk space conversion formula is as follows:
Figure BDA0003319932270000121
wherein:
r(G)=10*f(G) 2
the transfer function f (G) is as follows:
f(G)=1.509*[(ln(G)) 1.084 -5.381]
in the classical glycemic risk index method, the blood glucose value corresponding to the zero risk point of the method is 112mg/dL. In other embodiments of the present invention, the blood glucose level at the zero risk point may also be adjusted in combination with the risk and data trend of clinical practice, and is not specifically limited herein. Fitting is performed on a risk space of the blood glucose level with the blood glucose level greater than the zero risk point, and the specific fitting mode is not particularly limited.
In another embodiment of the present invention, the improved controlled variable grid analysis Control Variability Grid Analysis (CVGA) method is used, the original CVGA defined zero risk point blood glucose value is 110mg/dL, and the following equal risk blood glucose value data pairs (90 mg/dL,180mg/dL, 70mg/dL,300mg/dL, 50mg/dL,400 mg/dL) are assumed, in the embodiment of the present invention, the actual risk of clinical practice and trend characteristics of data are combined to consider, the actual risk and data are adjusted, the equal risk data pairs (70 mg/dL,300 mg/dL) are corrected to (70 mg/dL,250 mg/dL), and the zero risk point blood glucose value is set as a target blood glucose value G B . And performing polynomial model fitting on the model to obtain risk functions respectively processed at two sides of the zero risk point as follows:
Figure BDA0003319932270000122
and limits the maximum value thereof:
|r|＝min(|r|,n)
wherein the maximum value n is defined to be in the range of 0 to 80mg/dL, preferably, n is defined to be 60mg/dL.
In other embodiments of the invention, the blood glucose value and the equal risk data of the zero risk point can also be adjusted by combining the real risk and the data trend of clinical practice, the specific limitation is not made here, the equal risk point is fitted, and the specific fitting mode is not limited; the specific numerical values used to define the maximum values are also not particularly limited.
Fig. 3 is a graph comparing blood glucose risk converted to risk space by BGRI and CVGA methods to blood glucose in the original physical space.
Similar to the treatment with Zone-MPC, the risk of glycemia after conversion by BGRI and CVGA methods was fairly gentle in the euglycemic range, especially in the range of 80-140 mg/dL. Unlike Zone-MPC, which is completely 0 in the range, the further tuning capability is lost, and rPID risk is gentle in the range, but still has stable and slow tuning capability, so that the blood glucose can be further tuned to the target value, and more precise blood glucose control is realized.
In another embodiment of the present invention, a unified processing manner may be adopted for the data on both sides of the zero risk point, and as in the foregoing embodiment, the BGRI or CVGA method may be adopted for the data on both sides of the zero risk point; different processing methods can also be adopted, such as combining BGRI and CVGA methods, and the same zero risk point blood glucose value, such as target blood glucose value G, can be adopted B When the blood glucose level is smaller than the target blood glucose level G B When the BGRI method is adopted, the blood sugar value is larger than the target blood sugar value G B The CVGA method is adopted, and at the moment:
r＝-r(G),if G≤G B ，
wherein:
r(G)＝10*f(G) 2
the transfer function f (G) is as follows:
f(G)＝1.509*[(ln(G)) 1.084 -5.381]
r＝-4.8265*10 4 -4*G 2 +0.45563*G-44.855,if G＞G B 。
similarly, when the blood glucose level is smaller than the target blood glucose level G B When adopting CVGA method, the blood sugar value is larger than the target blood sugar value G B The BGRI method is adopted, and at this time:
r＝r(G),if G>G B ，
wherein:
r(G)＝10*f(G) 2
the transfer function f (G) is:
f(G)＝1.509*[(ln(G)) 1.084 -5.381]
r＝G-G B ,if G≤G B 。
at the same time, the maximum value can be limited:
|r|＝min(|r|，n)
wherein the maximum value n is defined to be in the range of 0 to 80mg/dL, preferably, n is defined to be 60mg/dL.
In other embodiments of the present invention, the blood glucose level at the zero risk point may be set to the target blood glucose level G B For less than or equal to the target blood glucose value G B Adopts BGRI method for data greater than target blood glucose value G B The data of (a) adopts a deviation amount processing method, such as a segmentation weighting process or a relative value process.
When the piecewise weighting process is employed, at this time:
r＝-r(G),if G≤G B ，
wherein:
r(G)=10*f(G) 2
the transfer function f (G) is:
f(G)＝1.509*[(ln(G)) 1.084 -5.381]
Figure BDA0003319932270000141
when the relative value processing is adopted:
r＝-r(G),if G≤G B ，
wherein:
r(G)＝10*f(G) 2
the fitted symmetric transfer function f (G) is:
f(G)＝1.509*[(ln(G)) 1.084 -5.381]
r＝100*(G-G B )/G,if G＞G B
when the blood sugar values corresponding to the zero risk points are all the target blood sugar values G B In the case of the target blood glucose level G or lower B Adopts segment weighting processing to obtain data of the (C)The relative value processing and CVGA method are identical in processing function, and therefore, when the relative value is equal to or less than the target blood glucose level G B The data of (2) is subjected to a sectional weighting process or a relative value process, and when the data of the blood glucose level greater than the zero risk point is subjected to a BGRI method, the processing result is equivalent to the blood glucose level which is less than or equal to the target blood glucose level G B When adopting CVGA method, the blood sugar value is larger than the target blood sugar value G B When the BGRI method is adopted, the calculation formula is not repeated.
In each of the embodiments of the present invention, the target blood glucose level G B It is preferably 80-140 mg/dL, and the target blood glucose level G B 110-120 mg/dL.
Through the processing mode, the blood glucose asymmetric in the original physical space of the rPID algorithm can be converted into the blood glucose risk approximately symmetric in the risk space, so that the characteristics of simplicity and robustness of the PID algorithm can be maintained, the blood glucose risk control function with clinical value is achieved, and the accurate control of the closed-loop artificial insulin gland infusion system is realized.
There are three major delay effects in a closed loop artificial pancreas control system: insulin absorption delay (about 20 minutes from subcutaneous arrival at blood circulation tissue to about 100 minutes at liver), insulin onset delay (about 30-100 minutes), interstitial fluid glucose concentration and blood glucose sensing delay (about 5-15 minutes). Any attempt to accelerate closed loop responsiveness may result in unstable system behavior and system oscillations. To compensate for insulin absorption delays in a closed-loop artificial pancreas control system, in one embodiment of the invention, an insulin feedback compensation mechanism is introduced. Subtracting from the output the amount of insulin not yet absorbed in the body, a fraction proportional to the estimate of plasma insulin concentration
Figure BDA0003319932270000151
(actual human insulin secretion also signals negative feedback regulation of insulin concentration in plasma). The formula is as follows:
Figure BDA0003319932270000152
wherein:
PID (t) represents an infusion indication sent to the insulin infusion system;
PID c (t) represents a compensated infusion indication sent to an insulin infusion system;
gamma represents the compensation coefficient of the estimated plasma insulin concentration to the algorithm output, and a larger coefficient results in a relatively conservative algorithm and a relatively aggressive coefficient, so in the embodiment of the invention, gamma ranges from 0.4 to 0.6, preferably, gamma is 0.5.
Figure BDA0003319932270000153
Estimates representing plasma insulin concentrations may be obtained by various conventional predictive algorithms, such as calculation directly from infused insulin based on the pharmacokinetic profile of insulin, or by conventional autoregressive methods:
Figure BDA0003319932270000154
wherein:
Figure BDA0003319932270000155
an estimate of plasma insulin concentration representing the current time;
PID c (n-1) represents an output of the band offset at the previous time;
Figure BDA0003319932270000156
an estimate of plasma insulin concentration representing the last time instant;
Figure BDA0003319932270000157
an estimate of plasma insulin concentration representing the last time instant;
K 0 indicating the last time with compensationOutputting coefficients of the part;
K 1 a coefficient representing an estimated portion of the plasma insulin concentration at the previous time;
K 2 a coefficient representing an estimated portion of plasma insulin concentration at the previous time;
wherein the initial value
Figure BDA0003319932270000158
The time intervals can be selected according to actual requirements.
Correspondingly, the compensation output formula after risk conversion by the method is as follows:
Figure BDA0003319932270000159
wherein:
rPIDc (t) represents a compensated infusion indication sent to the insulin infusion system after risk conversion;
rPID (t) represents an infusion instruction sent to the insulin infusion system after risk conversion;
the meaning of the representation of the other characters is as described above.
To compensate for the delay in insulin onset in a closed-loop artificial pancreatic control system, in one embodiment of the invention, insulin IOB (insulin on board) is introduced that has not been functional in the body, and IOB is subtracted from the insulin output to prevent the risk of insulin infusion accumulation, excessive amounts, postprandial hypoglycemia, etc.
Fig. 4 is an insulin IOB curve according to an embodiment of the invention.
From the IOB curve shown in fig. 4, the cumulative residual amount of insulin previously infused can be calculated, and the selection of a particular curve can be determined based on the actual insulin action time of the user.
PID′(t)=PID(t)-IOB(t)
Wherein:
PID' (t) represents an infusion indication sent to the insulin infusion system after subtraction of the IOB;
PID (t) represents an infusion indication sent to the insulin infusion system;
IOB (t) represents the amount of insulin that has not been acted upon in the body at time t.
Correspondingly, the output formula for deducting the amount of insulin which has not been acted in the body after risk conversion by the method is as follows:
rPID′(t)＝rPID(t)-IOB(t)
wherein:
rPID' (t) represents an infusion indication sent to the insulin infusion system after risk conversion that deducts the amount of insulin that has not been functional in the body;
rPID (t) represents an infusion instruction sent to the insulin infusion system after risk conversion;
the meaning of the representation of the other characters is as described above.
To obtain more ideal control effect, the calculation of the IOB is processed as follows m 、IOB o IOBs corresponding to meal insulin and other insulins than meal, respectively. The formula is as follows:
IOB(t)＝IOB m,t +IOB o,t
wherein:
Figure BDA0003319932270000161
wherein:
IOB m,t indicating the amount of meal insulin that has not been active in the body at time t;
IOB o,t Indicating the amount of non-prandial insulin that has not been active in the body at time t;
D i (i=2-8) represents the corresponding coefficients of IOB curves corresponding to insulin action times i, respectively;
I m,t indicating the amount of meal insulin;
I 0,t indicating a non-meal insulin amount;
IOB (t) represents the amount of insulin that has not yet been acted upon in the body at time t.
The distinguishing treatment of meal insulin and non-meal insulin is carried out on the IOB, so that the insulin can be cleared more quickly when the meal and the blood sugar are too high, the greater insulin output can be obtained, and the blood sugar regulation is faster. And when approaching the target, a longer insulin action time curve is adopted, so that insulin is cleared more slowly, and blood sugar regulation is more conservative and stable.
When PID '(t) >0 or rPID' (t) >0, the amount of insulin finally infused is PID '(t) or rPID' (t);
when PID '(t) <0 or rPID' (t) <0, the amount of insulin finally infused is 0.
To compensate for interstitial fluid glucose concentration and sensing delays of blood glucose in a closed-loop artificial pancreas control system, in one embodiment of the present invention, an autoregressive method is employed to compensate, as follows:
Figure BDA0003319932270000171
wherein,,
G SC (n) represents the interstitial fluid glucose concentration at the current moment, i.e. the measurement of the sensing system;
Figure BDA0003319932270000172
Representing an estimated concentration of blood glucose at a previous time;
G SC (n-1) and G SC (n-2) represents the interstitial fluid glucose concentration at the previous time and the previous time, respectively;
K 0 a coefficient representing an estimated concentration portion of blood glucose at the previous time;
K 1 and K 2 The coefficients of interstitial fluid glucose concentration at the previous time and the previous time are shown, respectively.
Wherein, at the initial moment,
Figure BDA0003319932270000173
the blood glucose concentration is estimated through the interstitial fluid glucose concentration, so that the sensing delay of the interstitial fluid glucose concentration and the blood glucose is compensated, the PID algorithm is more accurate, and the rPID algorithm can calculate the actual requirement of a human body on insulin more accurately correspondingly.
In the embodiment of the invention, for the insulin absorption delay, the insulin onset delay and the tissue fluid glucose concentration and the blood glucose sensing delay can be partially or completely compensated, and preferably, all delay factors are considered for complete compensation, so that the rPID algorithm is more accurate.
In another embodiment of the present invention, a rmc (risk-model-prediction-control) algorithm for converting the blood glucose asymmetrical in the original physical space into the blood glucose risk approximately symmetrical in the risk space is preset in the program module 101, the rmc algorithm is obtained by performing conversion processing on the basis of a classical MPC (model-prediction-control) algorithm, and the program module 101 controls the infusion module 102 to infuse insulin according to a corresponding infusion instruction calculated by the rmc algorithm.
The classical MPC algorithm consists of three elements, a predictive model, a cost function and constraints. The classical MPC prediction model is as follows:
x t+1 ＝Ax t +BI t
G t ＝Cx t
wherein:
x t+1 a state parameter representing the next moment in time,
Figure BDA0003319932270000181
x t a state parameter representing the current time of day,
Figure BDA0003319932270000182
I t indicating the insulin infusion quantity at the current time;
G t indicating the blood glucose concentration at the current time.
The parameter matrix is as follows:
Figure BDA0003319932270000183
Figure BDA0003319932270000184
C=[1 0 0]
b 1 ,b 2 ,b 3 k is an a priori value.
The cost function of MPC consists of the sum of squares of the deviations of the output G (blood glucose level) and the sum of the squares of the changes of the input I (insulin quantity). The MPC needs to obtain the minimum solution of the cost function.
Figure BDA0003319932270000185
Wherein:
I′ t+j indicating a change in insulin infusion after step j;
Figure BDA0003319932270000186
indicating the difference between the predicted blood glucose concentration and the target blood glucose value after step j;
t represents the current time;
n, P is the number of steps in the control time window and the prediction time window, respectively;
r is the weighting coefficient of the insulin component therein.
Insulin infusion in step j is I t +I′ t+j 。
In the embodiment of the invention, the time window T is controlled c =30min, prediction time window T p =60 min, the weighting coefficient R of insulin quantity is 11000. Although the control time window used in the calculation was 30min, only the first calculation result of insulin output was used in the actual operation, and the minimum solution of the cost function was recalculated based on the latest blood glucose value obtained after the operation.
In an embodiment of the invention, infusions within a time window are controlledTime step j n ，j n The value of (2) is 0-30 min, preferably 2min. Step number n=t c /j n J ranges from 0 to N.
In other embodiments of the present invention, the control time window, the prediction time window, and the weighting coefficient of the insulin amount may also be selected as other values, which are not particularly limited herein.
As described above, since the distribution of high/low blood sugar (original physical space) has significant asymmetry, the hyperglycemia risk and the hypoglycemia risk corresponding to the same degree of deviation of blood sugar from the normal range in clinical practice will be significantly different, and the asymmetric blood sugar in the original physical space is converted into the blood sugar risk approximately symmetric in the risk space according to the asymmetric characteristics of the clinical risk of glucose concentration, so that the MPC algorithm is more accurate and flexible. The cost function of the rmc algorithm after risk transformation is as follows:
Figure BDA0003319932270000191
wherein,,
r t+j indicating the blood glucose risk value after the j-th step;
I′ t+j indicating the change in insulin infusion after step j.
Converting the deviation of the blood glucose value into corresponding blood glucose risk, wherein the specific conversion mode is consistent with the mode in the rPID algorithm, such as sectional weighting processing and relative value processing; further comprising setting a fixed zero risk point in the risk space, the blood glucose concentration of the zero risk point being settable to a target blood glucose value. Processing data on two sides deviating from zero risk points, such as BGRI and improved CVGA methods; and also includes processing the data on both sides of the deviation from the target blood glucose level in different ways.
Specifically, when the piecewise weighting processing is employed:
Figure BDA0003319932270000192
when the relative value processing is adopted:
Figure BDA0003319932270000201
when the classical glycemic risk index method is used:
Figure BDA0003319932270000202
wherein:
r(G t+j )＝10*f(G t+j ) 2
conversion function f (G t+j ) The following are provided:
f(G t+j )＝1.509*[(ln(G t+j )) 1.084 -5.381]
when the control variability grid analysis method is employed:
Figure BDA0003319932270000203
at the same time, the maximum value is limited:
|r t+j |＝min(|r t+j |,n)
wherein the maximum value n is defined to be in the range of 0 to 80mg/dL, preferably, n is defined to be 60mg/d.
When the blood glucose level is smaller than the target blood glucose level G B When the BGRI method is adopted, the blood sugar value is larger than the target blood sugar value G B When the CVGA method is adopted, the following steps are adopted:
r t+j ＝-r(G t+j ),if G t+j ≤G B
wherein:
r(G t+j )＝10*f(G t+j ) 2
conversion function f (G t+j ) The following are provided:
f(G t+j )=1.509*[(ln(G t+j )) 1.084 -5.381]
r t+j ＝-4.8265*10 4 -4*G t+j 2 +0.45563*G t+j -44.855,if G t+j ＞G B
when the blood glucose level is smaller than the target blood glucose level G B When adopting CVGA method, the blood sugar value is larger than the target blood sugar value G B When the BGRI method is adopted, the following steps are adopted:
r t+j ＝r(G t+j ),if G t+j ＞G B ，
wherein:
r(G t+j )＝10*f(G t+j ) 2
conversion function f (G t+j ) The following are provided:
f(G t+j )＝1.509*[(ln(G t+j )) 1.084 -5.381]
r t+j G t+j -G B ,if G t+j ≤G B 。
at the same time, the maximum value can be limited:
|r t+j |＝min(|r t+j |,n)
wherein the maximum value n is defined to be in the range of 0 to 80mg/dL, preferably, n is defined to be 60mg/dL.
When the blood glucose level is smaller than the target blood glucose level G B When the BGRI method is adopted, the blood sugar value is larger than the target blood sugar value G B When a segmentation weighting method is adopted, the following steps are adopted:
r t+j ＝-r(G t+j ),if G t+j ≤G B
wherein:
r(G t+j )＝10*f(G t+j ) 2
conversion function f (G t+j ) The following are provided:
f(G t+j )＝1.509*[(ln(G t+j )) 1.084 -5.381]
Figure BDA0003319932270000211
when the blood glucose level is smaller than the target blood glucose level G B When the BGRI method is adopted, the blood sugar value is larger than the target blood sugar value G B When the relative value conversion is adopted:
r t+j ＝-r(G t+j ),if G t+j ≤G B
Wherein:
r(G t+j )＝10*f(G t+j ) 2
conversion function f (G t+j ) The following are provided:
f(G t+j )＝1.509*[(ln(G t+j )) 1.084 -5.381]
Figure BDA0003319932270000212
when the pair is less than or equal to the target blood glucose value G B The data of (2) is subjected to a sectional weighting process or a relative value process, and when the data of the blood glucose level greater than the zero risk point is subjected to a BGRI method, the processing result is equivalent to the blood glucose level which is less than or equal to the target blood glucose level G B When adopting CVGA method, the blood sugar value is larger than the target blood sugar value G B When the BGRI method is adopted, the calculation formula is not repeated.
In the above various conversion formulas:
r t+j a blood glucose risk value at step j;
G t+j is the blood glucose level detected in step j.
Target blood glucose level G B It is preferably 80-140 mg/dL, and the target blood glucose level G B 110-120 mg/dL.
The beneficial effects after risk conversion and the comparison of the relationship between blood glucose and blood glucose risk are consistent with those in the rPID algorithm and are not repeated here.
Similarly, to compensate for insulin absorption delay, an insulin feedback compensation mechanism may be used for compensation; to compensate for insulin onset delays, IOB compensation may also be employed; sensing delay of tissue fluid glucose concentration and blood glucose concentration can also adopt autoregressive compensation, and a specific compensation mode is consistent with rPID algorithm, and is specific:
for insulin absorption delay, the compensation formula is as follows:
Figure BDA0003319932270000221
Wherein:
I t+j an infusion indication to be sent to the insulin infusion system at step j;
rI c(t+j) an infusion instruction sent to the insulin infusion system at step j after risk conversion;
gamma represents the compensation coefficient of the estimated plasma insulin concentration to the algorithm output, and a larger coefficient results in a relatively conservative algorithm and a relatively aggressive coefficient, so in the embodiment of the invention, gamma ranges from 0.4 to 0.6, preferably, gamma is 0.5.
Figure BDA0003319932270000222
An estimate of plasma insulin concentration at step j is shown.
For insulin onset delay, the compensation formula is as follows:
rI′ t+j ＝rI t+j -IOB(t+j)
wherein:
rI′ t+j indicating an infusion instruction sent to the insulin infusion system after deducting the IOB at step j after risk conversion;
rI t+j an infusion instruction sent to the insulin infusion system at step j after risk conversion;
IOB (t+j) represents the amount of insulin that has not been acted upon in vivo at time t+j.
Likewise, a meal and a non-meal distinction may also be made for IOB (t+j), where:
IOB(t+j)＝IOB m,t+j +IOB o,t+j
wherein:
Figure BDA0003319932270000231
wherein:
IOB m,t+j indicating the amount of meal insulin that has not been active in the body at time t+j;
IOB o,t+j indicating the amount of non-prandial insulin not yet active in the body at time t+j;
D i (i=2-8) represents the corresponding coefficients of IOB curves corresponding to insulin action times i, respectively;
I m,t+j Indicating the meal insulin quantity at time t+j;
I 0,t+j representing the non-meal insulin quantity at time t+j;
IOB (t+j) represents the amount of insulin that has not yet been acted upon in vivo at time t+j.
When rI' t+j >At 0, the final amount of insulin infused is rI' t+j ；
When rI' t+j <At 0, the final amount of insulin infused is 0.
For sensing delays in interstitial fluid glucose concentration and blood glucose concentration, autoregressive compensation can also be employed, as follows:
Figure BDA0003319932270000232
wherein,,
G SC (t+j) represents interstitial fluid glucose concentration at time t+j, i.e. the measurement of the sensing system;
Figure BDA0003319932270000233
the estimated concentration of blood glucose at time t+j-1;
G SC (t+j-1) and G SC (t+j-2) represents interstitial fluid glucose concentrations at times t+j-1 and t+j-2, respectively;
K 0 a coefficient indicating the estimated concentration portion of blood glucose at time t+j-1;
K 1 and K 2 The coefficients of interstitial fluid glucose concentration at times t+j-1 and t+j-2 are shown, respectively.
Wherein, at the initial moment,
Figure BDA0003319932270000234
the beneficial effects of the various compensation modes are consistent with those of the rPID algorithm and are not repeated here.
In the rmc algorithm, it is preferable to compensate for the delay in insulin action and the delay in sensing of the interstitial fluid glucose concentration and the blood glucose concentration.
In another embodiment of the present invention, a composite artificial pancreas algorithm is preset in the program module 101, the composite artificial pancreas algorithm includes a first algorithm and a second algorithm, and when the detecting module 100 detects the current blood glucose level and sends the current blood glucose level to the program module 101, the first algorithm calculates the first insulin infusion amount I 1 The second algorithm calculates a second insulin infusion amount I 2 First insulin infusion quantity I by composite artificial pancreas algorithm 1 And a second insulin infusion amount I 2 Performing optimization calculation to obtain final insulin infusion quantity I 3 And the final insulin infusion amount I 3 Is sent to the infusion module 102, the infusion module 102 based on the final infusion quantity I 3 Insulin infusion is performed.
The first algorithm and the second algorithm are one of a classical PID algorithm, a classical MPC algorithm, a rMPC algorithm or a rPID algorithm. The rmcp algorithm or the rPID algorithm is an algorithm that converts blood glucose that is asymmetric in the original physical space to blood glucose risk that is approximately symmetric in the risk space. The conversion mode of the blood glucose risk in the rMPC algorithm and the rPID algorithm is as described above.
When I 1 ＝I 2 When I 3 ＝I 1 ＝I 2 ；
When I 1 ≠I 2 When it is, I can be 1 And I 2 The arithmetic mean values of the (a) are respectively substituted into the first algorithm and the second algorithm to re-optimize algorithm parameters, the insulin infusion quantity required at the current moment is calculated through the first algorithm and the second algorithm again after parameter optimization, and if I 1 And I 2 Still not the same, then fetch I again 1 And I 2 Arithmetic mean of (2)Repeating the above process until I 1 And I 2 The same, namely:
(1) solving for first insulin infusion quantity I 1 And a second insulin infusion amount I 2 Average value of (2)
Figure BDA0003319932270000241
Figure BDA0003319932270000242
(2) Will average the value
Figure BDA0003319932270000243
Respectively carrying out the algorithm parameters into a first algorithm and a second algorithm, and adjusting the algorithm parameters;
(3) recalculating the first insulin infusion amount I based on the current blood glucose value, the first algorithm after adjusting the parameters, and the second algorithm 1 And a second insulin infusion amount I 2 ；
(4) Performing cyclic calculation on steps (1) - (3) until I 1 ＝I 2 The final insulin infusion amount I 3 ＝I 1 ＝I 2 。
At this time, when the first algorithm or the second algorithm is the PID or rPID algorithm, the algorithm parameter is K P And K is D ＝T D /K P ，T D Can be taken for 60min-90min, K I ＝T I *K P ，T I It can be taken for 150min-450min. When the first algorithm or the second algorithm is an MPC or an rmmc algorithm, the algorithm parameter is K.
When I 1 ≠I 2 In this case, it is also possible to apply to I 1 And I 2 Weighting, substituting the weighted calculated value into the first algorithm and the second algorithm to re-optimize algorithm parameters, and calculating the insulin infusion amount required at the current moment through the first algorithm and the second algorithm after parameter optimization, if I 1 And I 2 Still not the same, then pair I again 1 And I 2 Weighting, adjusting weighting coefficient, repeating above process until I 1 And I 2 The same, namely:
(1) solving for first insulin infusion quantity I 1 And a second insulin infusion amount I 2 Weighted mean of (2)
Figure BDA0003319932270000244
Wherein alpha and beta are respectively the first insulin infusion quantity I 1 And a second insulin infusion amount I 2 Weighting coefficients of (2);
(2) will weight the mean value
Figure BDA0003319932270000245
Carrying out the algorithm parameters in a first algorithm and a second algorithm, and adjusting the algorithm parameters;
(3) recalculating the first insulin infusion amount I based on the current blood glucose value, the first algorithm after adjusting the parameters, and the second algorithm 1 And a second insulin infusion amount I 2 ；
(4) Performing cyclic calculation on steps (1) - (3) until I 1 ＝I 2 The final insulin infusion amount I 3 ＝I 1 ＝I 2 。
Similarly, when the first algorithm or the second algorithm is a PID or RPID algorithm, the algorithm parameter is K P And K is D ＝T D /K P ，T D Can be taken for 60min-90min, K I ＝T I *K P ，T I It can be taken for 150min-450min. When the first algorithm or the second algorithm is an MPC or an rmmc algorithm, the algorithm parameter is K.
In embodiments of the invention, alpha and beta may be based on the first insulin infusion amount I 1 And a second insulin infusion amount I 2 Is adjusted in size when I 1 ≥I 2 When alpha is less than or equal to beta; when I 1 ≤I 2 When alpha is more than or equal to beta; preferably, α+β=1. In other embodiments of the present invention, α and β may be other ranges, which are not specifically limited herein.
When the calculation results of the two are the same, i.e. I 3 ＝I 1 ＝I 2 It is considered that the insulin infusion amount at the present time can bring the blood glucose level to a desired level. By the processing of the above mode, each calculation The methods are mutually referenced, preferably, the first algorithm and the second algorithm are respectively an rMPC algorithm and an rPID algorithm, which are mutually referenced, so that the accuracy of the output result is further improved, and the result is more feasible and reliable.
In another embodiment of the present invention, the program module 101 is further provided with a memory for storing information such as the body state, blood glucose level, insulin infusion amount, etc. of the user history, and statistical analysis can be performed based on the information in the memory to obtain the statistical analysis result I at the current time 4 When I 1 ≠I 2 Respectively compare I 1 、I 2 And I 4 Calculating the final insulin infusion quantity I 3 Selecting I 1 And I 2 Closer to the statistical analysis result I 4 As a result of the calculation of the final complex artificial pancreas algorithm, i.e. the final insulin infusion I 3 The program module 101 will end the insulin infusion amount I 3 To the infusion device 102 for infusion; namely:
Figure BDA0003319932270000251
by comparison with the historical data, on the other hand, the reliability of the insulin infusion quantity is ensured.
In another embodiment of the present invention, when I 1 And I 2 When the two are inconsistent and have large differences, the blood sugar risk space conversion mode and/or the delay effect compensation mode in the rMPC algorithm and/or the rPID algorithm can be changed to be similar, and then the output result of the composite artificial pancreas algorithm can be finally determined through the arithmetic average value, the weighting treatment or the comparison with the statistical analysis result.
In another embodiment of the invention, the closed-loop artificial pancreas control system further comprises a meal recognition module and a motion recognition module. The usual meal recognition for recognizing whether the user is taking a meal or exercising may be based on the blood glucose change rate and determined by a specific threshold. The blood glucose change rate can be calculated from front and back time points or obtained by linear regression of multiple time points within a period of time, and specifically, when the change rate calculation of the front and back time points is adopted, the calculation formula is as follows:
dG t /dt＝(G t -G t-1 )/Δt
wherein:
G t a blood glucose level indicating the current time;
G t-1 a blood glucose level indicating the last time;
Δt represents the time interval between the current time and the previous time.
When a three-point time change rate calculation formula is adopted, the calculation formula is as follows:
dG t /dt＝(3G t -4G t-1 +G t-2 )/2Δt
wherein:
G t a blood glucose level indicating the current time;
G t-1 a blood glucose level indicating the last time;
G t-2 a blood glucose level indicating the previous time;
Δt represents the time interval between the current time and the previous time.
The raw continuous glucose data may also be filtered or smoothed prior to calculating the blood glucose rate of change. The threshold value can be set to be 1.8mg/mL-3mg/mL, or can be set in a personalized way.
Similar to meal recognition, since exercise causes a rapid decrease in blood glucose, exercise recognition may also be based on the rate of change of blood glucose and determined by a specific threshold. The calculation of the blood glucose rate of change may also be as previously described and the threshold may be personalized. To more quickly determine the occurrence of motion, the closed-loop artificial pancreatic insulin infusion control system also includes a motion sensor (not shown). The motion sensor is used to automatically detect physical activity of the user and the program module 101 may receive physical activity status information. The motion sensor can automatically and accurately sense the physical activity state of the user, and send the activity state parameters to the program module 101, so that the output reliability of the composite artificial pancreas algorithm under the motion scene is improved.
The motion sensor may be provided in the detection module 100, the program module 101 or the injection molding block 102. Preferably, in an embodiment of the present invention, the motion sensor is provided in the program module 101.
It should be noted that, the embodiments of the present invention do not limit the number of motion sensors and the setting positions of the plurality of motion sensors, as long as the conditions that the motion sensors sense the activity status of the user can be satisfied.
The motion sensor includes a three-axis acceleration sensor or a gyroscope. The triaxial acceleration sensor or gyroscope can sense the activity intensity, the activity mode or the body posture of the body more accurately. Preferably, in the embodiment of the present invention, the motion sensor is a combination of a triaxial acceleration sensor and a gyroscope.
It should be noted that, in the calculation process, the blood glucose risk conversion modes adopted by the rmcp algorithm and the rPID algorithm may be the same or different, and the compensation modes about the delay effect may be the same or different, and the calculation process may also be adjusted according to the actual situation.
In another embodiment of the present invention, an adaptive unit for adjusting the gain factor of the algorithm according to the weight of the user is also provided in the program module 101. In some inventive embodiments, the infusion module 102 or the program module 101 may indicate the user's daily insulin demand DIR. In another embodiment of the present invention, the DIR may be calculated from the body weight BW, in particular, the DIR is proportional to BW, i.e. dir=e×bw, where e is the body weight adjustment factor.
For a type-I diabetes patient, the weight adjustment coefficient e can select the average value of the crowd to be 0.53U/kg, and can also select personalized treatment by combining the exercise habit of the patient, such as selecting a lower weight adjustment coefficient for a professional exercise patient, such as 0.4U/kg; patients who are less involved in exercise choose a higher weight adjustment factor, such as 0.6U/kg. For type II diabetics, the pancreatic secretion function and insulin resistance condition can be combined, and the personalized weight adjustment coefficient can be selected in a large range, such as 0.1-1.5U/kg, and the common range is 0.6-1.1U/kg.
In one embodiment of the present invention, the algorithm preset in the program module 101 is a classical PID algorithm or an rPID algorithm, where the gain factor kp=dir/(bw×m) of the proportional part of the algorithm, m is the weight compensation factor of the user, and the value is 50-500, preferably, m is 135.
Gain factor K of integral part in PID algorithm or rPID algorithm I And gain coefficient K of differential part D Can be converted into Kp-related coefficients, e.g. K D ＝T D /K P ，T D Can be taken for 60min-90min, K I ＝T I *K P ，T I It can be taken for 150min-450min. T (T) D 、T I Big then algorithm is aggressive, otherwise relatively conservative. Different coefficient settings may be used during daytime and night sleep, for example, a smaller time parameter may be selected at night.
In another embodiment of the present invention, the algorithm preset in the program module 101 is a classical MPC algorithm or an rmc algorithm, the gain factor K of which is BW dependent:
Figure BDA0003319932270000271
wherein:
c is a safety coefficient;
s is a clinical experience coefficient;
e is the weight adjustment coefficient, U/Kg.
The safety factor c can be selected between 1.25 and 3 according to the risk of hypoglycemia at night; the values of the clinical experience coefficients s may be 1500, 1700, 1800, 2000, 2200, 2500, etc., and may be adjusted according to clinical results, without specific limitation. In a preferred embodiment of the invention, the clinical experience coefficient s is 1700. The weight adjustment coefficient e has the value range as described above.
In both of the foregoing embodiments, the gain factor Kp of the PID algorithm or rPID algorithm and the gain factor K of the MPC algorithm or rMPC algorithm may also be adjusted by introducing a factor Sb (t) that is related to the basal insulin demand, and correspondingly, K' P ＝K P *Sb(t)，K′=K*Sb(t)。
The coefficient Sb (t) of basal insulin demand correlation is the ratio of basal insulin demand B (t) at t to the average value Ba of basal insulin quantities throughout the day, i.e., sb (t) =b (t)/Ba. Wherein ba=y is DIR/24, y is the basic insulin quantity compensation coefficient, the value is 0.1-5, the crowd mean value of the coefficient is 0.47, children are slightly smaller, for example, 0.3-0.4 can be taken.
The daily basal insulin quantity average value Ba can be calculated according to the actual basal rate setting of the user. the basal insulin demand B (t) at t can be set according to four types of clinical optimal basal rate settings of the main stream. Fig. 5 is a main stream of clinically optimal basal rate setting types, four in number, from the references Holterhus, P.M., J.Bokelmann, et al (2013), "Predicting the Optimal Basal Insulin Infusion Pattern in Children and Adolescents on Insulin samples," Diabetes Care 36 (6): 1507-1511 ], where the horizontal axis is time, 24 hours a day, and the vertical axis is the relative deviation of basal insulin demand from the mean of basal insulin quantity Ba over the day, mostly between 0.5 and 1.5.
B (t) can also refer to a basic rate segmentation setting commonly used in clinic, such as a three-segment setting, as follows:
(1) when time t is 0 to 4 am, B (t) =0.5 DIR/48;
(2) when time t is 4 a.m. to 10 a.m., B (t) =1.5 DIR/48;
(3) when time t is 10 am to 0 am, B (t) =dir/48.
In other inventive embodiments, B (t) may also be calculated based on a base rate setting known and appropriate to the user.
In the embodiment of the present invention, sb (t) ranges from 0.2 to 2, preferably from 0.5 to 1.5. By introducing coefficients Sb (t) related to the basic insulin demands of different time periods, the gain coefficients are adjusted along with the change of time, so that the insulin demands of users in different time periods are met, and the accuracy of closed-loop control is further improved.
In the embodiment of the invention, the conversion mode of the rPID algorithm and the rMPC algorithm for converting the blood glucose which is asymmetric in the original physical space into the blood glucose risk which is approximately symmetric in the risk space, the compensation mode of various delays and the beneficial effects are not repeated here as described above. Meanwhile, the calculation results of each algorithm can be further processed, and the further processing mode, the beneficial effects and the like of the calculation results are not repeated here.
Fig. 6 is a schematic diagram of the modular relationship of a closed-loop artificial pancreatic insulin infusion control system in accordance with another embodiment of the invention.
In the embodiment of the invention, the closed-loop artificial pancreatic insulin infusion control system mainly comprises a detection module 100, an infusion module 102 and an electronic module 103.
The detection module 100 is used for continuously detecting the real-time blood glucose level of the user. Typically, the detection module 100 is a continuous glucose meter (Continuous Glucose Monitoring, CGM) that can detect blood glucose levels in real time and monitor blood glucose changes and send current blood glucose levels to the infusion module 102 and the electronic module 103.
The infusion module 102 contains the necessary mechanical structure for infusing insulin and also includes elements such as an infusion processor 1021 that execute a first algorithm and is controlled by the electronic module 103. The infusion module 102 calculates a first insulin infusion amount I currently required by a first algorithm after receiving the current blood glucose value sent by the detection module 100 1 And calculate the first insulin infusion quantity I 1 To the electronic module 103.
The electronic module 103 is used to control the operation of the detection module 100 and the injection molding block 102. Thus, the electronic module 103 is connected to the detection module 100 and the injection molding block 102, respectively. Here, the electronic module 103 is an external electronic device such as a mobile phone or a handset, and thus the connection means wireless connection. The electronic module 103 includes a second processor, in this embodiment of the present invention, the second processor is an element such as the electronic processor 1031 capable of executing a second algorithm and a third algorithm, and the electronic module 103 calculates a second insulin infusion amount I currently required by the second algorithm after receiving the current blood glucose level sent by the detection module 100 2 . Here, the first and second algorithms used by the electronic module 103 and the infusion module 102 to calculate the amount of insulin currently required are not the same.
The electronic module 103 receives the first insulin infusion amount I sent by the infusion module 102 1 Thereafter, the first insulin infusion quantity I is further subjected to a third algorithm 1 And a second insulin infusion amount I 2 Performing optimization calculation to obtain final insulin infusion quantity I 3 And the final insulin infusion amount I 3 Is sent to the infusion module 102, and the infusion module 102 infuses the currently required insulin I into the body of the user 3 . At the same time, the infusion status of the infusion module 102 can also be fed back into the electronic module 103 in real time. The specific optimization is as described above. Namely:
when I 1 ＝I 2 When I 3 ＝I 1 ＝I 2 ；
When I 1 ≠I 2 In this case, the electronic module 103 further substitutes the arithmetic average value or the weighted value of the two values into the algorithm to recalculate the current insulin infusion amount I 1 And I 2 If the data are not the same, repeating the above process until I 3 ＝I 1 ＝I 2 The method comprises the following steps:
(1) solving for the first insulin infusion quantity I 1 And said second insulin infusion amount I 2 Average value of (2)
Figure BDA0003319932270000291
Figure BDA0003319932270000292
(2) Will average the value
Figure BDA0003319932270000293
Carrying out the first algorithm and the second algorithm, and adjusting algorithm parameters;
(3) recalculating the first insulin infusion amount I based on the current blood glucose value, the first algorithm after adjusting the parameters, and the second algorithm 1 And a second insulin infusion amount I 2 ；
(4) Performing cyclic calculation on steps (1) - (3) until I 1 ＝I 2 Final insulin infusion quantity I 3 ＝I 1 ＝I 2 。
Or:
(1) solving for the first insulin infusion quantity I 1 And a second insulin infusion amount I 2 Weighted mean of (2)
Figure BDA0003319932270000294
Wherein alpha and beta are respectively the first insulin infusion quantity I 1 And said second insulin infusion amount I 2 Weighting coefficients of (2);
(2) will weight the mean value
Figure BDA0003319932270000295
Carrying out the first algorithm and the second algorithm, and adjusting algorithm parameters;
(3) Recalculating a first insulin infusion amount I based on the current blood glucose value, the first algorithm after adjusting the parameters, and the second algorithm 1 And a second insulin infusion amount I 2 ；
(4) Performing cyclic calculation on steps (1) - (3) until I 1 ＝I 2 The final insulin infusion amount I 3 ＝I 1 ＝I 2 。
When the two are different, the electronic module 103 can also perform statistical analysis on the two and the historical information based on the physical state, blood glucose level, insulin infusion amount and the like of the user at each past moment to obtain a statistical analysis result I at the current moment 4 Comparing and selecting I 1 And I 2 Closer to the statistical analysis result I 4 As a final insulin infusion quantity I 3 The electronic module 103 will end the final insulin infusion amount I 3 To the infusion device 102 for infusion; namely:
Figure BDA0003319932270000301
in the embodiment of the present invention, the history information of the user may be stored in the electronic module 103, or may be stored in a cloud management system (not shown), where the cloud management system is connected to the electronic module 103 wirelessly.
Fig. 7 is a schematic diagram of the modular relationship of a closed-loop artificial pancreatic insulin infusion control system in accordance with a further embodiment of the invention.
In the embodiment of the invention, the closed-loop artificial pancreatic insulin infusion control system mainly comprises a detection module 100, an infusion module 102 and an electronic module 103.
The detection module 100 is used for continuously detecting the real-time blood glucose level of the user. Typically, the detection module 100 is a continuous glucose meter (Continuous Glucose Monitoring, CGM) that can detect blood glucose levels in real time and monitor blood glucose changes, and the current blood glucose level is only sent to the infusion module 102. The detection module 100 further includes a second processor, in this embodiment of the present invention, the second processor is an element capable of executing a second algorithm, such as the detection processor 1001, and the detection module 100 directly calculates the second insulin infusion amount I through the second algorithm after detecting the real-time blood glucose level 2 And calculate a second insulin infusion quantity I 2 To the electronic module 103.
The infusion module 102 calculates the first insulin infusion amount I by a first algorithm after receiving the current blood glucose value sent by the detection module 100 as previously described 1 And infusing a first insulin infusion quantity I 1 To the electronic module 103. Here, the first algorithm and the second algorithm used by the detection module 103 and the infusion module 102 to calculate the amount of insulin are not identical.
The electronic module 103 receives the first insulin infusion quantity I from the detection module 100 and the infusion module 102 respectively 1 And a second insulin infusion amount I 2 Thereafter, the first insulin infusion quantity I is further subjected to a third algorithm 1 And a second insulin infusion amount I 2 Performing optimization calculation to obtain final insulin infusion quantity I 3 And the final insulin infusion amount I 3 Is sent to the infusion module 102, and the infusion module 102 infuses the currently required insulin I into the body of the user 3 . At the same time, the infusion status of the infusion module 102 can also be fed back into the electronic module 103 in real time. The specific optimization is as described above.
In the above two embodiments of the present invention, the infusion processor 1021 initially calculates the first insulin infusion amount I after the detection module 100 detects the current blood glucose level 1 Second, secondPreliminary calculation of the second insulin infusion quantity I by the processor (e.g., electronic processor 1031 and detection processor 1001) 2 And will I 1 And I 2 Sending the final insulin infusion amount I to the electronic module 103, further optimizing the final insulin infusion amount I by the electronic module 103 3 And is sent to the infusion module 102 for insulin infusion to improve the accuracy of the infusion instructions.
In the above two embodiments of the present invention, the first algorithm and the second algorithm are one of the classical PID algorithm, the classical MPC algorithm, the rmc algorithm or the rmpid algorithm, and the advantages of the calculation using the rmpid or rmc algorithm are as described above, and the advantages of the further optimization method are not repeated here.
The embodiment of the present invention is not limited to the specific location and connection relationship between the detection module 100 and the injection molding block 102, as long as the foregoing functional conditions can be satisfied.
In one embodiment of the invention, the two are electrically connected to each other to form a unitary structure and are adhered to the skin of the user at the same location. The two modules are connected into a whole and stuck at the same position, so that the number of the skin sticking devices of the user is reduced, and the interference of more sticking devices on the movement extension of the user is further reduced; meanwhile, the problem of unsmooth wireless communication between the separation devices is effectively solved, and user experience is further enhanced.
As in yet another embodiment of the present invention, both are separately provided in different structures and separately adhered to different locations of the user's skin. At this time, the detection module 100 and the injection molding block 102 transmit wireless signals to each other to be connected to each other.
Fig. 8 is a schematic diagram of the modular relationship of a closed-loop artificial pancreas multi-drug infusion control system according to another embodiment of the invention.
The closed-loop artificial pancreatic insulin infusion control system in the embodiment of the invention mainly comprises the detection module 100, the program module 101 and the infusion module 102, wherein the infusion module 102 has a multi-drug infusion function, the drug can be a drug combination for adjusting and controlling blood sugar of diabetics, the metabolite is glucose (glucose), the main drug is a hypoglycemic drug such as insulin (insulin) and the like, and other combined drugs are hypoglycemic drugs with opposite effects, such as glucagon (glucagon) and the like, cortisol (cortisol) and the like, auxin (growth hormone) and the like, epinephrine (epinephrine) and the like, glucose and the like, and dextrin analogues (such as pramlinine) and the like with similar effects.
Infusion module 102 may infuse hypoglycemic drugs and/or hypoglycemic drugs into the user's body in accordance with the hypoglycemic drug infusion instructions and/or the hypoglycemic drug infusion instructions issued by program module 101. The hypoglycemic drug and the hypoglycemic drug can be infused through different drug pipelines respectively, and also can be infused through the same drug pipeline at different times, and the specific drug pipeline design is not limited here.
Fig. 9 is a schematic diagram of a dual drug infusion switch according to two embodiments of the present invention.
In one embodiment of the invention, the hypoglycemic drug infusion instructions and/or the current hypoglycemic drug infusion instructions are an estimate of G by comparing blood glucose concentrations P And target blood glucose level G B And the blood glucose concentration estimate G P The estimation can be performed according to a prediction model of the rmc or other suitable blood glucose prediction algorithms; the hypoglycemic drug infusion data and/or the hypoglycemic drug infusion data may be calculated by the rmcp algorithm or the rPID algorithm or the composite artificial pancreas algorithm described previously. Specific:
when G P ≥G B At this time, the infusion module 102 begins to infuse the data I for the hypoglycemic agent calculated according to the rMPC algorithm or the rPID algorithm or the compound artificial pancreas algorithm t Performing blood glucose lowering medicine infusion;
When G P ＜G B At this time, the infusion module 102 begins to infuse the data D of the glycemic agent calculated according to the rMPC algorithm or the rPID algorithm or the composite artificial pancreas algorithm t Performing blood sugar increasing drug infusion;
it should be noted that, in the embodiment of the present invention, I b Indicating that the blood sugar is controlled to the target blood sugar value G without interference B Hypoglycemic medicine for infusionAmount, when G P ＝G B When I t ＝I b When G P ＞G B At the time, along with the infusion of the hypoglycemic drug, G P Further decrease, I t And also decreases. When the infusion module 102 has only one set of drug infusion lines, when G P ＜G B When, i.e. I t ＜I b At this time, the infusion module 102 begins the blood glucose-increasing drug infusion, data D t The rMPC algorithm, the rPID algorithm or the composite artificial pancreas algorithm can be used for calculating, and simultaneously stopping the infusion of the hypoglycemic medicament, so that the hypoglycemic medicament and the hypoglycemic medicament are prevented from affecting each other due to antagonism. When the infusion module 102 has at least two sets of drug infusion lines, when 0.ltoreq.I t ＜I b When the blood sugar-increasing medicine is infused, the blood sugar-increasing medicine can be infused continuously, and the occurrence of hypoglycemia can be effectively prevented; when I t When less than 0, the infusion of the hypoglycemic agent is stopped and only the hypoglycemic agent is infused.
In another embodiment of the present invention, the hypoglycemic agent infusion instructions and/or the current hypoglycemic agent infusion instructions may be directly by comparing the required amount I of hypoglycemic agent t And target hypoglycemic drug amount I b Is carried out, the demand I of the hypoglycemic medicine t And target hypoglycemic drug amount I b The calculation may be performed by the rMPC algorithm or rPID algorithm described previously or by a compound artificial pancreas algorithm. Specific: when the infusion module 102 has at least two sets of drug infusion lines:
when I t ≥I b At this time, the infusion module 102 begins to infuse the data I for the hypoglycemic agent calculated according to the rMPC algorithm or the rPID algorithm or the compound artificial pancreas algorithm t Performing blood glucose lowering medicine infusion;
when 0 is less than or equal to I t ＜I b When the blood sugar-increasing medicine is infused, the blood sugar-increasing medicine can be infused continuously, the occurrence of hypoglycemia can be effectively prevented, and the blood sugar-increasing medicine infuses data I t And blood glucose increasing drug infusion data D t The calculation can be performed by the rMPC algorithm or the rPID algorithm or the composite artificial pancreas algorithm.
When I t When the blood sugar level is less than 0, stopping the infusion of the blood sugar level reducing medicine and only infusing the blood sugar level increasing medicine, and infusing data D by the blood sugar level increasing medicine t The calculation can be performed by an rMPC algorithm or an rPID algorithm or a composite artificial pancreas algorithm.
When the infusion module 102 has only one set of drug infusion lines:
when I t When not less than 0, the infusion module 102 starts to calculate the blood glucose lowering medicine infusion data I according to the rMPC algorithm or the rPID algorithm or the composite artificial pancreas algorithm t Performing blood glucose lowering medicine infusion;
when I t When less than 0, the infusion of the hypoglycemic agent is stopped and only the hypoglycemic agent is infused.
Preferably, in an embodiment of the present invention, the hypoglycemic agent is insulin and the hypoglycemic agent is glucagon.
It should be noted that in the above embodiment, the calculation modes of the blood glucose lowering drug infusion data and the glucagon infusion data in each stage may be the same or different, and preferably, the same algorithm architecture is adopted to calculate, so as to ensure the consistency of the basic conditions during calculation and make the calculation result more accurate. More preferably, the method adopts a composite artificial pancreas algorithm for calculation, and fully utilizes the advantages of the rPID algorithm and the rMPC algorithm to face complex situations, so that the blood sugar control level is more ideal.
Fig. 10 is a schematic diagram of the closed-loop artificial pancreatic insulin infusion control system module relationship according to another embodiment of the invention.
In an embodiment of the present invention, the closed-loop artificial pancreatic insulin infusion control system disclosed in the embodiment of the present invention mainly includes a detection module 200 and an infusion module 202. The detection module 200 is used for continuously detecting the current blood glucose level of the user. In general, the detection module 100 is a continuous glucose detector (Continuous Glucose Monitoring, CGM) that can detect the current blood glucose level of a user in real time and monitor the change of blood glucose; the detection module 200 further includes a detection processing unit 2001, an algorithm for calculating an insulin infusion amount is preset in the detection processing unit 2001, and when the detection module 200 detects a current blood glucose level of a user, the detection processing unit 2001 calculates an insulin amount required by the user through the preset algorithm and sends the insulin amount required by the user to the infusion module 202.
The infusion module 202 contains the mechanical structure necessary to infuse insulin and an electronic transceiver that receives the user's insulin amount information from the detection module 200. Based on the current insulin infusion data from the detection module 200, the infusion module 202 infuses the currently desired insulin into the user. At the same time, the infusion status of the infusion module 102 can also be fed back into the detection module 200 in real time.
In the embodiment of the present invention, the algorithm for calculating insulin infusion preset in the detection processing unit 2001 is one of a classical PID algorithm, a classical MPC algorithm, a rmc algorithm, a rmpid algorithm or a composite artificial pancreas algorithm, and the method and the beneficial effects of calculating by using the rmpid algorithm, the rmc algorithm or the composite artificial pancreas algorithm are not repeated here as described above.
The embodiment of the present invention does not limit the specific location and connection relationship between the detection module 2100 and the injection molding block 202, as long as the aforementioned functional conditions can be satisfied.
In one embodiment of the invention, the two are electrically connected to each other to form a unitary structure and are adhered to the skin of the user at the same location. The two modules are connected into a whole and stuck at the same position, so that the number of the skin sticking devices of the user is reduced, and the interference of more sticking devices on the movement extension of the user is further reduced; meanwhile, the problem of unsmooth wireless communication between the separation devices is effectively solved, and user experience is further enhanced.
As in yet another embodiment of the present invention, both are separately provided in different structures and separately adhered to different locations of the user's skin. At this time, the detection module 200 and the injection molding block 202 transmit wireless signals to each other to achieve connection with each other.
In summary, the invention discloses a closed-loop artificial pancreatic insulin infusion control system, wherein an rMPC algorithm is preset in the system, and asymmetrical blood sugar in an original physical space is converted into an approximately symmetrical blood sugar risk space, so that the characteristics of the traditional MPC algorithm considering the recent effect of change are reserved, the asymmetry of blood sugar distribution is considered, the system has the advantages of accuracy and flexibility, and the accurate control of the closed-loop artificial pancreatic insulin gland infusion system is realized.
While certain specific embodiments of the invention have been described in detail by way of example, it will be appreciated by those skilled in the art that the above examples are for illustration only and are not intended to limit the scope of the invention. It will be appreciated by those skilled in the art that modifications may be made to the above embodiments without departing from the scope and spirit of the invention. The scope of the invention is defined by the appended claims.

Claims (26)

1. A closed loop artificial pancreatic insulin infusion control system, comprising:
The detection module is used for continuously detecting the current blood glucose value G;
a program module connected with the detection module, wherein the program module is preset with an rMPC algorithm for converting the blood glucose which is asymmetric in the original physical space into the blood glucose risk which is approximately symmetric in the risk space and a target blood glucose value G B The rMPC algorithm calculates insulin infusion instructions based on blood glucose risk; and
and the infusion module is connected with the program module and used for controlling the infusion module to infuse insulin according to the insulin infusion instruction calculated by the rMPC algorithm.
2. The closed loop artificial pancreatic insulin infusion control system of claim 1 wherein said rmc algorithm includes a predictive model, a cost function and constraints, said predictive model being:
x t+1 ＝Ax t +BI t
G t ＝Cx t
wherein:
x t+1 a state parameter representing the next moment in time,
Figure FDA0003319932260000011
x t a state parameter representing the current time of day,
Figure FDA0003319932260000012
I t indicating the insulin infusion quantity at the current time;
G t indicating the blood glucose concentration at the current time.
The parameter matrix is as follows:
Figure FDA0003319932260000013
Figure FDA0003319932260000014
C＝[1 0 0]
b 1 ，b 2 ，b 3 k is an a priori value.
The cost function is:
Figure FDA0003319932260000015
wherein:
I′ t+j indicating a change in insulin infusion after step j;
r t+j indicating the blood glucose risk value after the j-th step;
t represents the current time;
n, P is the number of steps in the control time window and the prediction time window, respectively;
r is the weighting coefficient of the insulin component therein.
3. The closed loop artificial pancreatic insulin infusion control system of claim 2 wherein said blood glucose risk r t+j The calculation is as follows:
Figure FDA0003319932260000021
G t+j indicating the blood glucose level detected at step j.
4. The closed loop artificial pancreatic insulin infusion control system of claim 2 wherein said blood glucose risk r t+j The relation with the calculation is as follows:
Figure FDA0003319932260000022
G t+j indicating the blood glucose level detected at step j.
5. The closed loop artificial pancreatic insulin infusion control system according to claim 2, wherein the blood glucose risk r t+j The calculation is as follows:
Figure FDA0003319932260000023
wherein,,
r(G t+j )＝10*f(G t+j ) 2
wherein f (G) t+j ) As a conversion function, the formula is:
f(G t+j )＝1.509*[(ln(G t+j )) 1.084 -5.381]
G t+j indicating the blood glucose level detected at step j.
6. The closed loop artificial pancreatic insulin infusion control system of claim 5 wherein said blood glucose risk r t+j The calculation is as follows:
Figure FDA0003319932260000031
G t+j is shown inAnd (3) detecting the blood glucose value in the j-th step.
7. The closed loop artificial pancreatic insulin infusion control system of claim 6 wherein said blood glucose risk r t+j Is defined as: r is |r t+j |＝min(|r t+j |，n)。
8. The closed loop artificial pancreatic insulin infusion control system according to claim 7, wherein the defined maximum value n takes a value of 0-80 mg/dL.
9. The closed loop artificial pancreatic insulin infusion control system according to claim 8, wherein the defined maximum value n takes a value of 60mg/dL.
10. The closed loop artificial pancreatic insulin infusion control system according to claim 2, wherein the blood glucose value G detected at the j-th step t+j Greater than the target blood glucose value G B At the time, the blood glucose risk r t+j The calculation is as follows:
r t+j ＝r(G t+j )，if G t+j ＞G B ，
wherein,,
r(G t+j )＝10*f(G t+j ) 2
wherein f (G) t+j ) As a conversion function, the formula is:
f(G t+j )＝1.509*[(ln(G t+j )) 1.084 -5.381]；
blood glucose level G detected at the j-th step t+j Not greater than the target blood glucose value G B When the blood glucose risk r is calculated as:
r t+j ＝G t+j -G B ，if G t+j ≤G B
G t+j indicating the blood glucose level detected at step j.
11. The closed-loop artificial pancreatic insulin of claim 10Infusion control system, characterized in that said maximum value of the blood glucose risk r is defined as: r is |r t+j |＝min(|r t+j |，n)。
12. The closed loop artificial pancreatic insulin infusion control system according to claim 11, wherein the defined maximum value n takes a value of 0-80 mg/dL.
13. The closed loop artificial pancreatic insulin infusion control system according to claim 12, wherein the defined maximum value n takes a value of 60mg/dL.
14. The closed loop artificial pancreatic insulin infusion control system according to claim 2, wherein the blood glucose value G detected at the j-th step t+j Not greater than the target blood glucose value G B At the time, the blood glucose risk r t+j The calculation is as follows:
r t+j ＝-r(G t+j )，if G t+j ≤G B
wherein:
r(G t+j )＝10*f(G t+j ) 2
conversion function f (G t+j ) The following steps:
f(G t+j )＝1.509*[(ln(G t+j )) 1.084 -5.381]
blood glucose level G detected at the j-th step t+j Greater than the target blood glucose value G B At the time, the blood glucose risk r t+j The calculation is as follows:
r t+j ＝-4.8265*10 4 -4*G t+j 2 +0.45563*G t+j -44.855，if G t+j ＞G B
G t+j indicating the blood glucose level detected at step j.
15. The closed loop artificial pancreatic insulin infusion control system according to claim 2, wherein the blood glucose value G detected when step j t+j Not greater than the target blood glucose value G B When the blood sugar is presentRisk r t+j The calculation is as follows:
r t+j ＝-r(G t+j )，if G t+j ≤G B
wherein:
r(G t+j )＝10*f(G t+j ) 2
conversion function f (G t+j ) The following steps:
f(G t+j )＝1.509*[(ln(G t+j )) 1.084 -5.381]
blood glucose level G detected at the j-th step t+j Greater than the target blood glucose value G B At the time, the blood glucose risk r t+j The calculation is as follows:
Figure FDA0003319932260000041
G t+j indicating the blood glucose level detected at step j.
16. The closed loop artificial pancreatic insulin infusion control system according to claim 2, wherein the blood glucose value G detected at the j-th step t+j Not greater than the target blood glucose value G B At the time, the blood glucose risk r t+j The calculation is as follows:
r t+j ＝-r(G t+j )，if G t+j ≤G B
wherein:
r(G t+j )＝10*f(G t+j ) 2
conversion function f (G t+j ) The following steps:
f(G t+j )＝1.509*[(ln(G t+j )) 1.084 -5.381]
blood glucose level G detected at the j-th step t+j Greater than the target blood glucose value G B At the time, the blood glucose risk r t+j The calculation is as follows:
Figure FDA0003319932260000051
G t+j indicating the blood glucose level detected at step j.
17. The closed loop artificial pancreatic insulin infusion control system according to any one of claims 1-16, wherein the target blood glucose value G B 80-140 mg/dL.
18. The closed loop artificial pancreatic insulin infusion control system according to any one of claim 17, wherein the target blood glucose value G B 110-120 mg/dL.
19. The closed loop artificial pancreatic insulin infusion control system according to any one of claims 1-16, wherein the rmc algorithm further includes one or more of the following:
(1) deducting the amount of unabsorbed plasma insulin in the body based on the insulin absorption delay in the artificial pancreas control system
Figure FDA0003319932260000052
Figure FDA0003319932260000053
Wherein,,
I t+j an infusion indication to be sent to the insulin infusion system at step j;
rI c(t+j) an infusion indication representing a transmission to the insulin infusion system at step j after the risk conversion;
gamma represents the compensation coefficient of the estimated plasma insulin concentration to the algorithm output;
Figure FDA0003319932260000061
an estimate of plasma insulin concentration at step j;
(2) subtracting the amount of insulin IOB (t+j) that has not been acted in the body from the insulin action delay in the artificial pancreas control system:
rI′ t+j ＝rI t+j -IOB(t+j)
rI′ t+j Indicating an infusion instruction sent to the insulin infusion system after deducting the IOB at step j after risk conversion;
rI t+j an infusion instruction sent to the insulin infusion system at step j after risk conversion;
IOB (t+j) represents the amount of insulin that has not been acted upon in vivo at time t+j;
(3) the sensing delay of blood glucose and interstitial fluid glucose concentrations is compensated for using an autoregressive method.
20. The closed loop artificial pancreatic insulin infusion control system of claim 19 wherein said plasma insulin concentration estimate is obtained by an autoregressive method.
21. The closed loop artificial pancreatic insulin infusion control system of claim 19 wherein said compensation coefficient γ is 0.4-0.6.
22. The closed loop artificial pancreatic insulin infusion control system of claim 21 wherein said compensation coefficient γ is 0.5.
23. The closed loop artificial pancreatic insulin infusion control system according to claim 19, wherein the in vivo not yet functional insulin amount IOB (t) is obtained by IOB curve.
24. The closed loop artificial pancreatic insulin infusion control system according to claim 19, wherein in calculating the in vivo not yet functional insulin amount IOB (t), insulin amounts are subjected to meal insulin and non-meal insulin treatments:
IOB(t+j)＝IOB m，t+j +IOB o，t+j
Figure FDA0003319932260000062
Wherein,,
IOB m，t+j indicating the amount of meal insulin that has not been active in the body at time t+j;
IOB o，t+j indicating the amount of non-prandial insulin not yet active in the body at time t+j;
D i (i=2-8) represents the corresponding coefficients of IOB curves corresponding to insulin action times i, respectively;
I m，t+j indicating the meal insulin quantity at time t+j;
I 0，t+j representing the non-meal insulin quantity at time t+j;
IOB (t+j) represents the amount of insulin that has not yet been acted upon in vivo at time t+j.
25. The single-sided drive closed-loop artificial pancreas according to claim 1, wherein two of the detection module, the program module and the infusion module are connected to each other to form a unitary structure and are adhered to different locations of the skin with a third module.
26. The single-sided drive closed-loop artificial pancreas according to claim 1, wherein the detection module, the program module and the infusion module are connected to form a unitary structure and adhered to the same location of the skin.
CN202111242326.1A 2021-10-25 2021-10-25 Closed-loop artificial pancreas insulin infusion control system Pending CN116020007A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202111242326.1A CN116020007A (en) 2021-10-25 2021-10-25 Closed-loop artificial pancreas insulin infusion control system

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202111242326.1A CN116020007A (en) 2021-10-25 2021-10-25 Closed-loop artificial pancreas insulin infusion control system

Publications (1)

Publication Number Publication Date
CN116020007A true CN116020007A (en) 2023-04-28

Family

ID=86074801

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202111242326.1A Pending CN116020007A (en) 2021-10-25 2021-10-25 Closed-loop artificial pancreas insulin infusion control system

Country Status (1)

Country Link
CN (1) CN116020007A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116999649A (en) * 2023-07-21 2023-11-07 北京理工大学 Artificial pancreas model prediction controller for realizing blood sugar non-offset tracking

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116999649A (en) * 2023-07-21 2023-11-07 北京理工大学 Artificial pancreas model prediction controller for realizing blood sugar non-offset tracking

Similar Documents

Publication Publication Date Title
CN107715230B (en) Insulin pump individuation configuration optimization system based on cloud big data
CN116020001A (en) Closed-loop artificial pancreas insulin infusion control system
KR20150097731A (en) Method and system for a hybrid control-to-target and control-to-range model predictive control of an artificial pancreas
Boiroux et al. Overnight control of blood glucose in people with type 1 diabetes
CN116139367A (en) Self-adaptive insulin infusion method and system
Dansa et al. Blood glucose regulation through bihormonal non-singular terminal sliding mode controller
CN116020007A (en) Closed-loop artificial pancreas insulin infusion control system
CN116020004A (en) Closed-loop artificial pancreas insulin infusion control system
CN116020003A (en) Closed-loop artificial pancreas insulin infusion control system
CN116020002A (en) Closed-loop artificial pancreas insulin infusion control system
CN116020006A (en) Closed-loop artificial pancreas insulin infusion control system
CN116020005A (en) Closed-loop artificial pancreas insulin infusion personalized control system
Kaveh et al. Higher order sliding mode control for blood glucose regulation
Ata et al. Model based control of artificial pancreas under meal disturbances
CN116020013A (en) Full closed loop artificial pancreas medicine infusion control system
CN116020012A (en) Closed-loop artificial pancreas insulin infusion control system
WO2023070247A1 (en) Closed-loop artificial pancreas insulin infusion control system
WO2023070253A1 (en) Closed-loop artificial pancreas insulin infusion control system
WO2023070252A1 (en) Closed-loop artificial pancreas insulin multi-drug control system
WO2023070251A1 (en) Closed-loop artificial pancreas insulin infusion control system
WO2023070246A1 (en) Closed-loop artificial pancreas insulin infusion control system
WO2023072306A1 (en) Closed-loop artificial pancreatic insulin infusion control system
WO2018042147A1 (en) Automatic closed-loop glucose control with an adaptive meal bolus calculator
CN116052822A (en) Method for simulating blood glucose loss value
WO2024082651A1 (en) Adaptive closed-loop control method based on linear model

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