CN111643771A - Closed-loop insulin infusion system based on adaptive generalized predictive control - Google Patents

Closed-loop insulin infusion system based on adaptive generalized predictive control Download PDF

Info

Publication number
CN111643771A
CN111643771A CN202010351347.6A CN202010351347A CN111643771A CN 111643771 A CN111643771 A CN 111643771A CN 202010351347 A CN202010351347 A CN 202010351347A CN 111643771 A CN111643771 A CN 111643771A
Authority
CN
China
Prior art keywords
adaptive
value
blood sugar
reference curve
blood glucose
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN202010351347.6A
Other languages
Chinese (zh)
Other versions
CN111643771B (en
Inventor
金浩宇
刘文平
余丽玲
陈婷
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.)
Guangdong Food and Drugs Vocational College
Original Assignee
Guangdong Food and Drugs Vocational College
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
Priority claimed from CN201910365168.5A external-priority patent/CN110124151A/en
Priority claimed from CN201910363842.6A external-priority patent/CN109925568A/en
Priority claimed from CN201910365159.6A external-priority patent/CN109925569A/en
Priority claimed from CN201910363857.2A external-priority patent/CN110124150A/en
Application filed by Guangdong Food and Drugs Vocational College filed Critical Guangdong Food and Drugs Vocational College
Publication of CN111643771A publication Critical patent/CN111643771A/en
Application granted granted Critical
Publication of CN111643771B publication Critical patent/CN111643771B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61M—DEVICES FOR INTRODUCING MEDIA INTO, OR ONTO, THE BODY; DEVICES FOR TRANSDUCING BODY MEDIA OR FOR TAKING MEDIA FROM THE BODY; DEVICES FOR PRODUCING OR ENDING SLEEP OR STUPOR
    • A61M5/00—Devices for bringing media into the body in a subcutaneous, intra-vascular or intramuscular way; Accessories therefor, e.g. filling or cleaning devices, arm-rests
    • A61M5/14—Infusion devices, e.g. infusing by gravity; Blood infusion; Accessories therefor
    • A61M5/142—Pressure infusion, e.g. using pumps
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61M—DEVICES FOR INTRODUCING MEDIA INTO, OR ONTO, THE BODY; DEVICES FOR TRANSDUCING BODY MEDIA OR FOR TAKING MEDIA FROM THE BODY; DEVICES FOR PRODUCING OR ENDING SLEEP OR STUPOR
    • A61M5/00—Devices for bringing media into the body in a subcutaneous, intra-vascular or intramuscular way; Accessories therefor, e.g. filling or cleaning devices, arm-rests
    • A61M5/14—Infusion devices, e.g. infusing by gravity; Blood infusion; Accessories therefor
    • A61M5/168—Means for controlling media flow to the body or for metering media to the body, e.g. drip meters, counters ; Monitoring media flow to the body
    • A61M5/16804—Flow controllers
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61M—DEVICES FOR INTRODUCING MEDIA INTO, OR ONTO, THE BODY; DEVICES FOR TRANSDUCING BODY MEDIA OR FOR TAKING MEDIA FROM THE BODY; DEVICES FOR PRODUCING OR ENDING SLEEP OR STUPOR
    • A61M5/00—Devices for bringing media into the body in a subcutaneous, intra-vascular or intramuscular way; Accessories therefor, e.g. filling or cleaning devices, arm-rests
    • A61M5/14—Infusion devices, e.g. infusing by gravity; Blood infusion; Accessories therefor
    • A61M5/168—Means for controlling media flow to the body or for metering media to the body, e.g. drip meters, counters ; Monitoring media flow to the body
    • A61M5/16877—Adjusting flow; Devices for setting a flow rate
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B13/00—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
    • G05B13/02—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
    • G05B13/04—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators
    • G05B13/042—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators in which a parameter or coefficient is automatically adjusted to optimise the performance
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B13/00—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
    • G05B13/02—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
    • G05B13/04—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators
    • G05B13/048—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators using a predictor
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61M—DEVICES FOR INTRODUCING MEDIA INTO, OR ONTO, THE BODY; DEVICES FOR TRANSDUCING BODY MEDIA OR FOR TAKING MEDIA FROM THE BODY; DEVICES FOR PRODUCING OR ENDING SLEEP OR STUPOR
    • A61M5/00—Devices for bringing media into the body in a subcutaneous, intra-vascular or intramuscular way; Accessories therefor, e.g. filling or cleaning devices, arm-rests
    • A61M5/14—Infusion devices, e.g. infusing by gravity; Blood infusion; Accessories therefor
    • A61M5/142—Pressure infusion, e.g. using pumps
    • A61M2005/14208—Pressure infusion, e.g. using pumps with a programmable infusion control system, characterised by the infusion program
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61M—DEVICES FOR INTRODUCING MEDIA INTO, OR ONTO, THE BODY; DEVICES FOR TRANSDUCING BODY MEDIA OR FOR TAKING MEDIA FROM THE BODY; DEVICES FOR PRODUCING OR ENDING SLEEP OR STUPOR
    • A61M5/00—Devices for bringing media into the body in a subcutaneous, intra-vascular or intramuscular way; Accessories therefor, e.g. filling or cleaning devices, arm-rests
    • A61M5/14—Infusion devices, e.g. infusing by gravity; Blood infusion; Accessories therefor
    • A61M5/142—Pressure infusion, e.g. using pumps
    • A61M2005/14288—Infusion or injection simulation
    • A61M2005/14292—Computer-based infusion planning or simulation of spatio-temporal infusate distribution
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61M—DEVICES FOR INTRODUCING MEDIA INTO, OR ONTO, THE BODY; DEVICES FOR TRANSDUCING BODY MEDIA OR FOR TAKING MEDIA FROM THE BODY; DEVICES FOR PRODUCING OR ENDING SLEEP OR STUPOR
    • A61M2205/00—General characteristics of the apparatus
    • A61M2205/33—Controlling, regulating or measuring
    • A61M2205/3331—Pressure; Flow
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61M—DEVICES FOR INTRODUCING MEDIA INTO, OR ONTO, THE BODY; DEVICES FOR TRANSDUCING BODY MEDIA OR FOR TAKING MEDIA FROM THE BODY; DEVICES FOR PRODUCING OR ENDING SLEEP OR STUPOR
    • A61M2205/00—General characteristics of the apparatus
    • A61M2205/35—Communication
    • A61M2205/3546—Range
    • A61M2205/3553—Range remote, e.g. between patient's home and doctor's office
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61M—DEVICES FOR INTRODUCING MEDIA INTO, OR ONTO, THE BODY; DEVICES FOR TRANSDUCING BODY MEDIA OR FOR TAKING MEDIA FROM THE BODY; DEVICES FOR PRODUCING OR ENDING SLEEP OR STUPOR
    • A61M2230/00—Measuring parameters of the user
    • A61M2230/20—Blood composition characteristics
    • A61M2230/201—Glucose concentration

Landscapes

  • Health & Medical Sciences (AREA)
  • Engineering & Computer Science (AREA)
  • Public Health (AREA)
  • Anesthesiology (AREA)
  • Biomedical Technology (AREA)
  • Heart & Thoracic Surgery (AREA)
  • Hematology (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Animal Behavior & Ethology (AREA)
  • General Health & Medical Sciences (AREA)
  • Vascular Medicine (AREA)
  • Veterinary Medicine (AREA)
  • Physics & Mathematics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Computation (AREA)
  • Medical Informatics (AREA)
  • Software Systems (AREA)
  • Artificial Intelligence (AREA)
  • General Physics & Mathematics (AREA)
  • Automation & Control Theory (AREA)
  • Fluid Mechanics (AREA)
  • External Artificial Organs (AREA)
  • Infusion, Injection, And Reservoir Apparatuses (AREA)

Abstract

The invention discloses a closed-loop insulin infusion system based on self-adaptive generalized predictive control, which comprises a blood sugar value detection module, a controller, an insulin pump and a program module, wherein the blood sugar value detection module is used for collecting blood sugar values, and the output end of the blood sugar value detection module is in signal connection with the input end of the controller; the insulin pump is used for infusing insulin, and the input end of the insulin pump is in signal connection with the output end of the controller; the program module is executed by the controller, acquires the blood sugar value of the user through the blood sugar value detection module, and adjusts the infusion rate of the insulin pump by combining the CARIMA model, the minimum variance control algorithm, the self-adaptive reference curve and the self-adaptive softening factor. The self-adaptive reference curve adopted in the invention adjusts the slope of the reference curve according to the condition of a user, and the self-adaptive softening factor flexibly adjusts the value of the self-adaptive reference curve according to different blood sugar values, thereby realizing the real-time regulation and control of the infusion rate of the insulin pump and better realizing the personalized treatment.

Description

Closed-loop insulin infusion system based on adaptive generalized predictive control
Technical Field
The invention relates to the field of insulin pump infusion amount estimation, in particular to a closed-loop insulin infusion system based on adaptive generalized predictive control.
Background
About 4.22 million adults worldwide have diabetes and cause death of 150 million people each year. Type 1 diabetes, also known as insulin-dependent diabetes mellitus, is primarily characterized by apoptosis of insulin-secreting cells in the pancreatic region and inadequate insulin secretion. Although the exact cause of type 1 diabetes is not clear, the blood glucose level of diabetic patients can be effectively controlled by daily insulin infusion, thereby improving the quality of life of the patients. The artificial pancreas can effectively detect the blood sugar level of a diabetic patient, and accurately calculate the infusion time and the infusion amount of insulin. The artificial pancreas mainly comprises three parts: continuous blood glucose monitoring (CGM), intelligent control systems and insulin pumps.
The intelligent control system is the core of the whole artificial pancreas, directly determines the accuracy and effectiveness of blood sugar control, and various control theories are applied to the intelligent control system of the artificial pancreas at present, such as a proportional-derivative-integral control algorithm, a model predictive control algorithm, a fuzzy logic control algorithm, a generalized predictive control algorithm and the like.
Proportional Integral Derivative (PID) is a control algorithm widely used in the industrial field. The control algorithm has been successfully transplanted into an artificial pancreas system and shows a good closed-loop control effect. The intelligent control system based on PID design has a simple structure, less parameter setting and higher robustness. However, the PID controller in the artificial pancreas requires the addition of gain adjustments and feed-forward operations and requires the patient to manually enter meal information. The above operation severely limits the application of the PID control algorithm and brings more inconvenience to the blood glucose control of the patient.
Fuzzy Logic (FL) control algorithms are widely used in the field of intelligent control, and are also widely used in artificial pancreas systems. The design principle and the process mainly comprise the following steps: 1. the method comprises the steps of establishing an expert knowledge base by summarizing and analyzing experience accumulated in long-term practice of clinical diabetes treatment; 2. the basic theory and method of fuzzy mathematics are used, the conditions and operations of the clinical treatment experience rules are expressed by fuzzy sets, and the fuzzy control rules and relevant information (such as clinical evaluation indexes and the like) are stored in an expert knowledge base as knowledge; 3. and obtaining the adaptive insulin infusion dosage parameter by fuzzy reasoning according to the blood glucose data monitored in real time. The FL control algorithm builds experience knowledge accumulated in long-term practice of clinical diabetes treatment as an expert knowledge base by relying on clinical experts, converts the experience knowledge into a fuzzy logic control rule, accords with the conventional experience of clinical treatment, and is easy to understand by clinicians. However, the fuzzy logic control algorithm is adopted, the fuzzy rule is summarized and the fuzzy membership function is selected mainly by depending on experience, so that the method has high subjectivity and has the defects of difficulty in dividing a fuzzy interval and untimely response, and great obstruction is brought to the development and further popularization of fuzzy control.
Model Predictive Control (MPC) is also widely used in artificial pancreas intelligent Control systems. The algorithm uses detailed models to describe the behavior of the dynamic system. The artificial pancreas intelligent control system based on the MPC can accurately predict the influence of dining and effectively regulate the blood sugar level of the diabetic. But MPC controllers rely heavily on the accuracy of the model. The human body metabolism model is very complex and has large calculation amount, and a perfect model which can accurately describe the dynamic relation of blood sugar-insulin is still not available up to now.
The design of an artificial pancreas intelligent Control system based on Generalized Predictive Control (GPC) is the current research focus. As an adaptive control algorithm, the generalized predictive control overcomes the model-dependent defect of the MPC algorithm and can automatically adjust the parameters of the control model in the absence of initial conditions and system descriptions. The artificial pancreas intelligent controller based on GPC can predict future blood sugar change by reading blood sugar information detected by CGM, and can acquire ideal insulin infusion rate by tracking reference curve and minimum variance control. Specifically, it has the following characteristics: firstly, adopting a Controlled Auto-Regressive integrated moving-Average (CARIMA) based on Controlled Auto-Regressive integration to predict blood sugar; second, the consideration of the weighting of the control increment in the objective function; thirdly, remote forecast of the output is utilized; and fourthly, controlling the introduction of a time domain length concept. Compared with the existing proportional calculus control algorithm, fuzzy logic control algorithm and model control algorithm, the generalized predictive control is easy to realize, and an expert database is not required to be built and a model is not required to be built according to abundant clinical medical knowledge.
The main problem faced by artificial pancreas at present is the problem of personalized treatment of diabetic patients. For example, juvenile and pediatric diabetics have lower insulin sensitivity and require higher insulin infusion rates than adult diabetics. Meanwhile, the food intake of three meals a day of the diabetic patient is different and fluctuates, and the artificial pancreas intelligent control system is required to carry out real-time judgment and fine adjustment. Generalized predictive control has unique advantages in addressing individualized treatment of artificial pancreas. This is because the generalized predictive control is an adaptive algorithm whose model parameters can be adjusted on-line in real time. The reference curve is a tracking object of the generalized predictive control, and directly determines the effect of the control. While adults, adolescents and children have different insulin sensitivities, it is difficult to achieve good control among the various populations if a uniform reference curve is used. But the use of different reference curves for different populations increases the complexity of the system and reduces the ease of use. And the softening factor directly determines the tracking speed of the generalized predictive control on the reference curve. The lower softening factor can ensure that the generalized predictive control has higher tracking speed, but influences the robustness of the system. Although the artificial pancreas system can respond to the change of blood sugar quickly, the artificial pancreas system has the defect of poor system stability. Conversely, a higher softening factor may ensure the robustness of the system, but sacrifices the tracking speed of the system. The artificial pancreas system is stable but has slow response to the fluctuation of blood sugar. Due to the different insulin sensitivity, adult, adolescent and child patients need to adopt different values of the softening factor respectively, and even the meal size of the same patient at different time needs to be finely adjusted.
Disclosure of Invention
In order to better realize the individualized treatment problem of the artificial pancreas system, the patent provides a closed-loop insulin infusion system based on adaptive generalized predictive control, further optimizes a reference curve and a softening factor in the closed-loop insulin infusion system on the basis of the traditional generalized predictive control, designs an adaptive softening factor strategy, can automatically generate an appropriate softening factor value according to the real-time blood sugar fluctuation of an individual, thereby enhancing the individualized treatment effect of the artificial pancreas, and provides an adaptive reference curve strategy, which can automatically generate a reference curve suitable for the insulin infusion system according to the real-time blood sugar fluctuation.
The technical scheme of the invention is as follows:
a closed-loop insulin infusion system based on adaptive generalized predictive control comprises a blood sugar value detection module, a controller, an insulin pump and a program module, wherein,
the blood sugar value detection module is used for collecting the blood sugar value of a user, and the output end of the blood sugar value detection module is in signal connection with the input end of the controller;
the insulin pump is used for infusing insulin to a user, and the input end of the insulin pump is in signal connection with the output end of the controller;
the program module is executed by the controller, acquires the blood sugar value of the user through the blood sugar value detection module, and adjusts the infusion rate of the insulin pump by combining the CARIMA model, the minimum variance control algorithm, the self-adaptive reference curve and the self-adaptive softening factor.
In a preferred embodiment, the program module is executed by a controller, and includes the following steps:
step S1: collecting the current blood sugar value of the user through a blood sugar value detection module;
step S2: obtaining a predicted value of future blood sugar change of the user through a CARIMA model and a loss-of-image equation according to the current blood sugar value of the user;
step S3: calculating the insulin injection rate increment of the insulin pump by a minimum variance control algorithm according to the predicted value of future blood sugar change of the user; the reference curve used by the minimum variance control algorithm is an adaptive reference curve, the softening factor used by the minimum variance control algorithm is an adaptive softening factor, the adaptive reference curve can adjust the slope of the reference curve according to the blood glucose level of a user, and the adaptive softening factor can adjust the value of the adaptive softening factor according to the change of the blood glucose value;
step S4: based on the idea of closed-loop control, the insulin injection rate increment of the insulin pump is iteratively optimized.
In a preferred embodiment, in step S2, the CRIMA model is represented by the following formula (1);
A(z-1)y(k)＝B(z-1)u(k-1)+C(z-1)ξ(k)/Δ(1)
wherein,
A(z-1) Is a coefficient of y (k), A (z)-1)＝1+a1z-1+…+anaz-naDetermining the time range of the past blood sugar value and the weight of the blood sugar value at each moment, which are included in the CRIMA model in the prediction process;
B(z-1) Is the coefficient of u (k-1), B (z)-1)＝b0+b1z-1+…+bnbz-nbDetermining the time range of the past insulin infusion rate value and the weight of the insulin infusion rate at each moment, wherein the past insulin infusion rate value is included in the CRIMA model in the prediction process;
C(z-1) Coefficient of ξ (k), C (z)-1)＝1+c1z-1+…+cncz-ncDetermining the time range and the weight of white noise which is included in the CRIMA model in the prediction process;
wherein y (k) represents the blood glucose level of the user at time k, and the unit is mg/dl;
the u (k-1) is the insulin injection rate at the k-1 moment, and the unit is pmol/step;
the xi (k) is white noise with the mean value of zero;
said DELTA ═ 1-z-1) Represents an integration factor;
z is as described-1For backward shifting operators, e.g. z-1y(k)＝y(k-1)；A(z-1) The coefficient y (k) determines the time range of the past blood glucose level and the weight of the blood glucose level at each time point, which are included in the CRIMA model during the prediction process.
And na, nb and nc represent the order of the model, respectively determine the time range of the past blood sugar value, the time range of the past insulin infusion rate value and the time range of white noise of the CRIMA model, and preferably, the values are 3, 5 and 1 respectively.
A is described1～ana，b0～bnbAnd c1～cnzAll the model parameters can be optimized on line in real time, the weight of the blood glucose value at each past moment, the weight of the past insulin infusion rate value and the weight of white noise which are brought into the CRIMA model are respectively determined, and initial values can be given when the model is initialized; here, an initial value a is given1＝1，a2＝0，a3＝0，b1＝0.5，b2＝0.5，b3＝0.5，b4＝0.5，b5＝0.5，c1＝1。
In order to predict the blood sugar change in the leading j steps, a chartlet equation is introduced; the charpy equation is shown in the following formula (2):
1＝Ej(z-1)A(z-1)Δ+Fj(z-1)z-j(2)
wherein,
Ej(z-1) Is A (z)-1) Coefficient of Δ, Ej(z-1)＝ej0+ej1z-1+…+ej-1z-j+1；
Fj(z-1) Is z-jCoefficient of (A), Fj(z-1)＝fj0+fj1z-1+…+fjnz-n；
Wherein, E isj(z-1) And Fj(z-1) Is composed of model parameters A (z)-1) And a polynomial uniquely determined for the prediction step j; a (z) as described-1) And A (z) in the formula (1)-1) The same physical quantity and the same numerical value; said DELTA ═ 1-z-1) Represents an integration factor; z is as described-1Represents a back shift operator; z is a radical of-j+1An operator representing a step of backward shift j-1; z is a radical of-jAn operator representing a step of backward shift j; z is as described-nIndicating a backward shiftn steps of operators; said ej0～ej-1And fj0～cjnAll parameters are parameters which can be optimized online in real time, wherein j is a prediction step size, n represents a maximum prediction length, and n is 8. The prediction step j is 1,2 … n.
In a preferred embodiment, the specific operation of obtaining the predicted value of the future blood glucose change of the user according to the current blood glucose value of the user by the CARIMA model and the loss-of-image equation in step S2 is as follows:
multiplying both sides of the formula (1) by E simultaneously through a CARIMA model and a chartlet equationj(z-1) After Δ, the predicted value of the blood glucose level at time k, which is advanced by j steps, can be obtained by equation (2) as:
y(k+j)＝Gj(z-1)Δu(k+j-1)+Fj(z-1)y(k)(j＝1，2…n)(3)
Gj(z-1)＝Ej(z-1)B(z-1)
in the formula (3), y (k + j) represents the predicted value of the blood glucose level of the user in j steps ahead of the time k, and the unit is mg/dl;
the delta u (k + j-1) represents the increment of the insulin injection rate of the insulin pump at the time k, and the unit is pmol/step;
said Gj(z-1) Is Ej(z-1) And B (z)-1) Represents the weight of the insulin injection rate increment between the time k and the time k + j-1 in predicting the blood glucose value of the time k ahead by j steps;
y (k) represents the blood glucose level of the user at time k, and the unit is mg/dl;
said Ej(z-1) And Fj(z-1) Both represent coefficients of formula (3) and E in formula (2)j(z-1) And Fj(z-1) Having the same coefficient function and the same numerical value, E of formula (3)j(z-1) And Fj(z-1) Is obtained by solving the following formula (1) and formula (2), wherein Ej(z-1) Is B (z)-1) Coefficient of (A), Fj(z-1) A coefficient of y (k);
b (z) as defined-1) The time range of past insulin infusion rate values and the weight of the insulin infusion rate at each time instant incorporated into the CRIMA model during prediction are determined for the coefficient of u (k-1) in equation (1).
In a preferred embodiment, the step S3 of calculating the insulin injection rate increment of the insulin pump by the minimum variance control algorithm model according to the predicted value of the future blood glucose change of the user comprises the following steps:
Figure BDA0002472036080000061
in the formula (4), Deltau (k + j-1) represents the increment of the insulin injection rate of the insulin pump in the time advance step as pmol/step; j is a control target, and when the J takes the minimum value, the value of delta u (k + J-1) can be determined;
the n represents the maximum prediction length, and is 8;
the lambda is a control weighting factor, and lambda is 3;
the m represents the control length of insulin infusion, and m is 5;
y (k + j) represents the predicted value of the blood sugar value of the user which leads the user by j steps at the time k, and the unit is mg/dl;
w (k + j) is composed of blood glucose value y (k) at time k and reference curve yrCalculated, specifically expressed by the following formula:
w(k+j)＝αjy(k)+(1-αj)yr(j＝1，2，…，n)
equation (5) above can be further expressed in vector form:
W＝Qy(k)+Myr
said yrRepresenting an adaptive reference curve; the W is expressed by the following formula:
W＝[w(k+1)，w(k+2)，…，w(k+n)]T
said Q is expressed by the formula:
Q＝[α，α2，…，αn]T
the alpha represents an adaptive softening factor;
said M is expressed by the formula:
M＝[1-α，1-α2，…，1-αn]T。
in a preferred scheme, the adaptive reference curve sets different descending slopes of the adaptive reference curve according to parameters of the current blood glucose concentration and by combining the variation of the current blood glucose concentration; if the blood sugar concentration is lower than the set threshold value, the descending slope of the self-adaptive reference curve is 0.
In a preferred embodiment, the adaptive reference curve includes the following steps:
scheme 1: if the rate of change of blood glucose is within a predetermined time period>k0mg/dl/step, defined as the blood sugar rising sharply; if the rate of change of blood glucose is between k1mg/dl/step～k0mg/dl/step, which is defined as slow fluctuation of blood sugar; if the rate of change of blood glucose is less than k1mg/dl/step, defined as a sharp drop in blood glucose;
wherein, k is1A preset value is set artificially; k is as described0A preset value which is artificially set, and k0>k1，k0>0mg/dl/step,k1<0mg/dl/step；
And (2) a flow scheme: the following judgments were made:
if the blood glucose concentration is higher than ζ1While the falling slope of the adaptive reference curve is set to k4mg/dl/step; zeta of1The unit is a preset value which is set manually and is mg/dl; k is as described4A preset value, k, set manually4≥0mg/dl/step；
If the blood glucose concentration is between ζ2～ζ1When, the zeta2Preset value, ζ, set manually2<ζ1(ii) a According to the change condition of blood sugar, the following selections are carried out:
if there is a rapid rise in blood glucose, the falling gradient of the adaptive reference curve is set to k5mg/dl/step; k is as described5Preset value set for human；
If the condition of slow fluctuation of the blood sugar exists, the descending slope of the adaptive reference curve is set as k6mg/dl/step; k is as described6A preset value is set artificially;
if the blood sugar is sharply reduced, the descending slope of the self-adaptive reference curve is set to be 0; k is as described5>k6≥0mg/dl/step；
If the blood glucose concentration is between ζ3～ζ2When, the zeta3Preset value, ζ, set manually2>ζ3(ii) a According to the change condition of blood sugar, the following selections are carried out:
if there is a rapid rise in blood glucose, the falling gradient of the adaptive reference curve is set to k7mg/dl/step; k is as described7A preset value is set artificially;
if the condition of slow fluctuation of the blood sugar exists, the descending slope of the adaptive reference curve is set as k8mg/dl/step; k is as described8A preset value is set artificially;
if the blood sugar is sharply reduced, the descending slope of the self-adaptive reference curve is set to be 0; k is as described7>k8K is not less than 0mg/dl/step5≥k7，k6≥k8；
If the blood glucose concentration is lower than ζ3Then the adaptive reference curve is fixed to be set at a constant ζ, i.e., a slope of 0 mg/dl/step.
In a preferred embodiment, the obtaining of the adaptive softening factor comprises the following steps:
step A: artificially setting blood glucose expectation value
Figure BDA0002472036080000081
The unit is mg/dl;
and B: calculating the deviation degree u of the blood sugar value of the current user and the expected blood sugar value according to the blood sugar value y (k) of the current user k; the degree of deviation u is expressed by the following formula:
Figure BDA0002472036080000082
wherein y (k) is the blood glucose concentration of the user at the moment k, and the unit is mg/dl;
said
Figure BDA0002472036080000083
Is the blood glucose expectation in mg/dl;
and C: calculating a variation Δ y of a current blood glucose level of the user, the variation Δ y being expressed by the following equation:
Δy＝y(k)-y(k-1)；
wherein y (k) is the blood glucose concentration of the user at the moment k, and the unit is mg/dl;
y (k-1) represents the blood glucose level of the user at the previous time, and has a unit of mg/dl.
Step D: calculating an adaptive softening factor alpha through the deviation u and the variation delta y; the self-adaptive softening factor alpha is expressed by the following formula:
α＝u-|Δy|(u≥1)。
in a preferred embodiment, in the step S4, the "iteratively optimizing the insulin injection rate increment of the insulin pump based on the idea of closed-loop control" specifically includes the following sub-steps:
step S4-1: taking the insulin injection rate increment of the insulin pump as an input value of a CARIMA model, and updating the predicted value of future blood sugar change of the user;
step S4-2: updating the insulin injection rate increment of the insulin pump by taking the updated predicted value of the future blood glucose change of the user as the input of a minimum variance control algorithm; the reference curve used by the minimum variance control algorithm is an adaptive reference curve, the softening factor used by the minimum variance control algorithm is an adaptive softening factor, the adaptive reference curve can adjust the slope of the reference curve according to the condition of a user, and the adaptive softening factor can adjust the value of the adaptive softening factor according to the change of the blood sugar value;
step S4-3: and circularly executing S4-1-S4-2 to realize iterative optimization of the insulin injection rate increment of the insulin pump.
In a preferred embodiment, the insulin infusion device further comprises a communication module, and the communication module is electrically connected with the controller in a bidirectional mode.
In the preferred scheme, the communication module is used for transmitting the blood sugar value and the insulin injection rate increment of the user to the remote system/medical worker system, so that the medical worker can conveniently monitor the blood sugar change of the user at any time.
In a preferred embodiment, the program modules further comprise a telemedicine access function, and the telemedicine access function comprises the following contents:
if medical personnel through user's blood sugar value and insulin pump's insulin injection rate increment, when the control insulin infusion needs to be intervened in the judgement, can input insulin pump's the corresponding instruction code of the infusion volume through communication module, this instruction code has the priority, and the controller can be preferred according to medical personnel's instruction code, controls insulin pump's infusion rate.
Compared with the prior art, the technical scheme of the invention has the beneficial effects that:
1. compared with the existing proportional calculus control, fuzzy logic control and model predictive control, the invention is easy to build and does not need to manually input meal information;
2. according to the invention, a CARIMA prediction model, a minimum variance control algorithm, closed-loop feedback correction and parameter rolling optimization are adopted, so that the accuracy of prediction and the effectiveness of control are ensured;
3. the invention adopts the self-adaptive reference curve, and the strategy considers the difference of blood sugar fluctuation among different individuals and can generate different reference curves for different individuals in real time, thereby realizing the personalized insulin infusion amount.
4. The invention uses the self-adaptive softening factor, and the strategy enhances the personalized treatment effect of the artificial pancreas and simultaneously ensures the robustness and the sensitivity of a control system. When the blood sugar value is stable, the method can ensure higher robustness; when the blood sugar value is rapidly increased or decreased, the invention can improve the response sensitivity, rapidly track the reference curve and effectively regulate the insulin infusion rate.
Drawings
Fig. 1 is a schematic structural diagram of the embodiment.
Fig. 2 is a control schematic diagram of the embodiment.
Fig. 3 is a schematic diagram of an adaptive reference curve according to an embodiment.
FIG. 4 is a diagram of an adaptive softening factor according to an embodiment.
Fig. 5 is an experimental result of a conventional generalized predictive control algorithm.
Fig. 6 shows the results of an experiment using only the adaptive reference curve.
FIG. 7 shows the experimental results using only adaptive softening factors
FIG. 8 shows the experimental results of the examples
Detailed Description
The drawings are for illustrative purposes only and are not to be construed as limiting the patent;
for the purpose of better illustrating the embodiments, certain features of the drawings may be omitted, enlarged or reduced, and do not represent the size of an actual product;
it will be understood by those skilled in the art that certain well-known structures in the drawings and descriptions thereof may be omitted.
The technical solution of the present invention is further described below with reference to the accompanying drawings and examples.
Example 1
As shown in fig. 1 and 2, a closed-loop insulin infusion system based on adaptive generalized predictive control includes a blood glucose value detection module, a controller, an insulin pump, a TF card and a program module, wherein,
the chip model of the controller is an ARM920T chip;
the blood sugar value detection module is used for collecting the blood sugar value of a user, and the output end of the blood sugar value detection module is in signal connection with the input end of the ARM920T chip;
the insulin pump is used for infusing insulin to a user, and the input end of the insulin pump is in signal connection with the output end of the ARM920T chip;
the program module is stored in the TF card and executed by the ARM920T chip, and comprises the following steps:
step S1: obtaining a predicted value of future blood sugar change of the user through a CARIMA model and a loss-of-image equation according to the current blood sugar value of the user;
the CRIMA model is described below;
A(z-1)y(k)＝B(z-1)u(k-1)+C(z-1)ξ(k)/Δ(1)
wherein,
A(z-1) Is a coefficient of y (k), A (z)-1)＝1+a1z-1+…+anaz-na；
B(z-1) Is the coefficient of u (k-1), B (z)-1)＝b0+b1z-1+…+bnbz-nb；
C(z-1) Coefficient of ξ (k), C (z)-1)＝1+c1z-1+…+cncz-nc；
Wherein y (k) represents blood glucose level of user at time k in mg/dl, u (k-1) represents insulin injection rate at time k-1 in mg/dl, ξ (k) represents white noise with mean value of zero, and Δ ═ 1-z-1) Represents an integration factor; z is as described-1A backward shift operator; and na, nb and nc represent the order of the model, and the values are respectively 3, 5 and 1. A is described1～ana，b0～bnbAnd c1～cnzAll are model parameters which can be optimized on line in real time, and are respectively given initial values a1＝1，a2＝0，a3＝0，b1＝0.5，b2＝0.5，b3＝0.5，b4＝0.5，b5＝0.5，c1＝1。
In order to predict the blood sugar change in the leading j steps, a chartlet equation is introduced; the charpy equation is shown in the following formula (2):
1＝Ej(z-1)A(z-1)Δ+Fj(z-1)z-j(2)
Ej(z-1)＝ej0+ej1z-1+…+ej-1z-j+1
Fj(z-1)＝fj0+fj1z-1+…+fjnz-n
wherein E isj(z-1) Is A (z)-1) Coefficient of Δ, Ej(z-1)＝ej0+ej1z-1+…+ej-1z-j+1；
Fj(z-1) Is z-jCoefficient of (A), Fj(z-1)＝fj0+fj1z-1+…+fjnz-n；
Wherein, E isj(z-1)Ej(z-1) And Fj(z-1) Is composed of model parameters A (z)-1) And a polynomial uniquely determined for the prediction step j; a (z) as described-1) And A (z) in the formula (1)-1) The same physical quantity and the same numerical value; said DELTA ═ 1-z-1) Represents an integration factor; z is as described-j+1An operator representing a step of backward shift by j-1, said z-1Represents a back shift operator; z is as described-jAn operator representing a step of backward shift j; z is as described-nAn operator representing a step of backward movement by n; said ej0～ej-1And fj0～cjnAll parameters are parameters which can be optimized online in real time, wherein j is a prediction step size, n represents a maximum prediction length, and n is 8. The prediction step j is 1,2 … n.
Through the CARIMA model and the charpy equation, the following equation is obtained:
y(k+j)＝Gj(z-1)Δu(k+j-1)+Fj(z-1)y(k)(j＝1,2…n)(3)
Gj(z-1)＝Ej(z-1)B(z-1)
in the formula (3), y (k + j) represents the predicted value of the blood glucose level of the user in j steps ahead of the time k, and the unit is mg/dl; the delta u (k + j-1) represents the increment of the insulin injection rate of the insulin pump at the time k, and the unit is pmol/step;
said Gj(z-1) Is Ej(z-1) And B (z)-1) Represents the weight of the insulin injection rate increment between the time k and the time k + j-1 in predicting the blood glucose value of the time k ahead by j steps;
y (k) represents the blood glucose level of the user at time k, and the unit is mg/dl;
said Ej(z-1) And Fj(z-1) Each represents a coefficient of formula (3) wherein Ej(z-1) Is B (z)-1) Coefficient of (A), Fj(z-1) A coefficient of y (k);
b (z) as defined-1) Is the coefficient of u (k-1) in formula (1).
Step S2: calculating the insulin injection rate increment of the insulin pump by a minimum variance control algorithm according to the predicted value of the future blood sugar change of the user in the step S1; wherein the minimum variance control algorithm uses a softening factor that is an adaptive softening factor;
Figure BDA0002472036080000121
in the formula (4), the delta u (k + j-1) represents that the increment of the insulin injection rate of the insulin pump leads the k time by j-1 step is pmol/step; j is a control target, and when the J takes the minimum value, the value of delta u (k + J-1) can be determined;
the n represents the maximum prediction length, and is 8;
the lambda is a control weighting factor, and lambda is 3;
the m represents the control length of insulin infusion, and m is 5;
the w (k + j) is expressed by the following formula:
w(k+j)＝αjy(k)+(1-αj)yr(j＝1，2，…，n)
the above formula (5) can be further expressed in the form of a vector
W＝Qy(k)+Myr
Said yrRepresenting an adaptive reference curve; w is expressed by the following formula:
W＝[w(k+1)，w(k+2)，…，w(k+n)]T
said Q is expressed by the formula:
Q＝[α，α2，…，αn]T
the alpha represents an adaptive softening factor;
said M is expressed by the formula:
M＝[1-α，1-α2，…，1-αn]T；
as shown in fig. 3, the adaptive reference curve includes the following procedures:
scheme 1: within 1 minute, a sharp rise in blood glucose is defined as if the rate of change in blood glucose is >5 mg/dl; if the rate of blood sugar change is between-5 mg/dl and 5mg/dl, the slow fluctuation of the blood sugar is defined; if the rate of change of the blood sugar is less than-5 mg/dl, defining that the blood sugar is sharply reduced;
and (2) a flow scheme: the following judgments were made:
if the blood sugar concentration is higher than 180mg/dl, the descending slope of the self-adaptive reference curve is set to be 20 mg/dl/step;
if the blood sugar concentration is between 160mg/dl and 180 mg/dl; according to the change condition of blood sugar, the following selections are carried out:
if the blood sugar is in a sharp rise, the descending slope of the self-adaptive reference curve is set to be 10 mg/dl/step;
if the condition of slow fluctuation of the blood sugar exists, the descending slope of the self-adaptive reference curve is set to be 5 mg/dl/step;
if the blood sugar is sharply reduced, the descending slope of the self-adaptive reference curve is set to be 0;
if the blood sugar concentration is between 130mg/dl and 160 mg/dl; according to the change condition of blood sugar, the following selections are carried out:
if the blood sugar is in a sharp rise, the descending slope of the self-adaptive reference curve is set to be 5 mg/dl/step;
if the condition of slow fluctuation of the blood sugar exists, the descending slope of the self-adaptive reference curve is set to be 1 mg/dl/step;
if the blood sugar is sharply reduced, the descending slope of the self-adaptive reference curve is set to be 0 mg/dl/step;
if the blood sugar concentration is lower than 130mg/dl, the self-adaptive reference curve is fixedly set to be 130 mg/dl/step;
as shown in fig. 4, the adaptive softening factor includes the following:
step A: setting a glycemic expectation
Figure BDA0002472036080000131
And B: calculating the deviation u between the current blood glucose level of the user and the expected blood glucose value according to the current blood glucose level y (k) of the user at the time k, wherein the deviation u is expressed by the following formula:
Figure BDA0002472036080000141
and C: calculating a variation Δ y of the current blood glucose level of the user, the variation Δ y being expressed by the following expression:
Δy＝y(k)-y(k-1)；
step D: calculating an adaptive softening factor alpha by the deviation u and the variation Δ y, the adaptive softening factor alpha being expressed by the following formula:
α＝u-|Δy|；
for example, if the blood glucose level y (k) at time k-1 of the patient is 168mg/dl and the blood glucose level y (k) at time k is 170mg/dl., the blood glucose change amount Δ y at time k is 2mg/dl., and the expected blood glucose value is
Figure BDA0002472036080000142
The blood glucose deviation u at time k is 1.31, and the adaptive softening factor α is 1.31-20.58. As can be seen from the reference curve design of FIG. 3, if the blood glucose concentration is between 160mg/dl and 180mg/dl and the blood glucose slowly fluctuates (-5 mg/dl)<Δy<5 mg/dl-), then adaptiveThe falling slope of the reference curve was set to 5 mg/dl. So yr＝y(k)-5mg/dl＝165mg/dl。w(k+j)＝0.06j×170mg/dl+1-0.06j× 160mg/dl, e.g. w (k +1) ═ 0.58 × 170mg/dl +0.42 × 165 mg/dl-167.9 mg/dl, w (k +2) ═ 0.58 mg/dl2×170mg/dl+(1-0.582)×165mg/dl＝166.68mg/dl。
Step S3: based on the idea of closed-loop control, carrying out iterative optimization on the insulin injection rate increment of the insulin pump;
s3-1: taking the insulin injection rate increment of the insulin pump as an input value of a CARIMA model, and updating the predicted value of future blood sugar change of the user;
s3-2: updating the insulin injection rate increment of the insulin pump by taking the updated predicted value of the future blood glucose change of the user as the input of a minimum variance control algorithm; the reference curve used by the minimum variance control method is an adaptive reference curve, the softening factor used by the minimum variance control algorithm is an adaptive softening factor, the adaptive reference curve can adjust the slope of the reference curve according to the condition of a user, and the adaptive softening factor can adjust the value of the adaptive softening factor according to the change of the blood sugar value;
s3-3: and circularly executing S3-1-S3-2 to realize iterative optimization of the insulin injection rate increment of the insulin pump.
Test environment of the present embodiment:
this example was implanted into the U.S. FDA approved diabetes simulation therapy test software T1DMS that can replace animal experiments, and the algorithms were performance tested. Software T1DMS is the only diabetes treatment testing software approved by the FDA in the united states that can be used in place of animal experiments. The software includes 100 virtual adult diabetic patients, 100 adolescent patients and 100 pediatric patient models and provides virtual CGMS and insulin pumps. In the test process, the blood sugar control effect of the insulin pump can be observed only by implanting the control algorithm into the test platform, selecting a test object and setting a meal plan and monitoring indexes.
Experimental results for this example:
as shown in FIG. 3, when blood glucose rises sharply, for example, to 180mg/dl, the reference curve in the example takes a higher slope value; when the blood sugar fluctuates between 130mg/dl and 160mg/dl, the reference curve in the embodiment adopts a lower slope value; when blood glucose drops sharply, for example, to 180mg/dl, the slope value of the reference curve in the example will be set to 0mg/dl.
As shown in FIG. 4, when the blood sugar level fluctuates around 130mg/dl, the embodiment adopts a higher value of the adaptive softening factor, and the embodiment has low sensitivity and high robustness to the blood sugar fluctuation; when significant increases or decreases in blood glucose occur, such as a significant increase in blood glucose to around 170mg/dl, embodiments may use lower values of the adaptive softening factor, where embodiments have greater sensitivity to blood glucose fluctuations, may quickly track the reference curve and quickly adjust the rate of insulin infusion.
As shown in fig. 5, 6, 7 and 8, the experimental results of adolescents with diabetes. FIG. 5 shows the effect of blood glucose control based on a conventional generalized predictive control algorithm with a softening factor set at a fixed value of 0.5 and a falling slope of the reference curve set at a fixed value of 5mg/dl/step (66%); fig. 6 shows the results of an experiment using only the adaptive reference curve, with the softening factor set at a fixed value of 0.5 (93%); FIG. 7 shows the results of an experiment using only the adaptive softening factor, with the falling slope of the reference curve set at a fixed value of 5mg/dl/step (95%); fig. 8 shows the blood glucose control effect of this example (97%). The test results clearly show that the present embodiment can stabilize the blood glucose level of the user more than the first three algorithms.
The terms describing positional relationships in the drawings are for illustrative purposes only and are not to be construed as limiting the patent;
for example, the terms "ARM 920T chip" and "TF card" are only an example of the embodiments, and all components/assemblies capable of achieving similar effects belong to the protection scope of the present patent.
It should be understood that the above-described embodiments of the present invention are merely examples for clearly illustrating the present invention, and are not intended to limit the embodiments of the present invention. Other variations and modifications will be apparent to persons skilled in the art in light of the above description. For example, different adaptation curves may be set for users of different age groups (adult patients, adolescent patients and pediatric patients). Or different adaptive softening factor calculation models, such as an exponential model or a logarithmic model, can be set for users (adults, teenagers and children) in different age groups, so that the adaptive softening factors are more suitable for the patients and more beneficial to stabilizing the blood sugar values. When the patient uses the embodiment, the selection is carried out, so that the self-adaptive reference curve and the self-adaptive softening factor are more suitable for the patient per se, and a better treatment effect is achieved. Alternatively, if the controller chip has a memory function, the program module may be stored in the controller, and an external memory module (such as a TF card) is not necessarily required for storing the program. Or, a communication module (such as a 4G communication module) can be added on the basis of the embodiment, and the communication module can transmit the blood sugar value of the user and the infusion rate of the insulin pump to a remote system/medical worker system, so that the medical worker can conveniently monitor the blood sugar change of the user at any time; if medical personnel through user's blood glucose value and insulin pump's infusion rate, when judging need intervene control insulin infusion, can input insulin pump's infusion volume's corresponding instruction code through communication module, this instruction code has the priority, ARM920T chip can be preferred according to medical worker's instruction code, insulin injection rate increment.
And are neither required nor exhaustive of all embodiments. Any modification, equivalent replacement, and improvement made within the spirit and principle of the present invention should be included in the protection scope of the claims of the present invention.

Claims (12)

1. A closed-loop insulin infusion system based on adaptive generalized predictive control, characterized by: comprises a blood sugar value detection module, a controller, an insulin pump and a program module, wherein,
the blood sugar value detection module is used for collecting the blood sugar value of a user, and the output end of the blood sugar value detection module is in signal connection with the input end of the controller;
the insulin pump is used for infusing insulin to a user, and the input end of the insulin pump is in signal connection with the output end of the controller;
the program module is executed by the controller, acquires the blood sugar value of the user through the blood sugar value detection module, and adjusts the infusion rate of the insulin pump by combining the CARIMA model, the minimum variance control algorithm, the self-adaptive reference curve and the self-adaptive softening factor.
2. The closed-loop insulin infusion system based on the adaptive generalized predictive control of claim 1, wherein: the program module is executed by the controller and comprises the following steps:
step S1: collecting the current blood sugar value of the user through a blood sugar value detection module;
step S2: obtaining a predicted value of future blood sugar change of the user through a CARIMA model and a loss-of-image equation according to the current blood sugar value of the user;
step S3: calculating the insulin injection rate increment of the insulin pump by a minimum variance control algorithm according to the predicted value of future blood sugar change of the user; the reference curve used by the minimum variance control algorithm is an adaptive reference curve, the softening factor used by the minimum variance control algorithm is an adaptive softening factor, the adaptive reference curve can adjust the slope of the reference curve according to the blood glucose condition of a user, and the adaptive softening factor can adjust the value of the adaptive softening factor according to the change of the blood glucose value;
step S4: based on the idea of closed-loop control, the insulin injection rate increment of the insulin pump is iteratively optimized.
3. The closed-loop insulin infusion system based on the adaptive generalized predictive control of claim 2, wherein: in step S2, the CRIMA model is represented by the following formula (1);
A(z-1)y(k)＝B(z-1)u(k-1)+C(z-1)ξ(k)/Δ (1)
wherein,
A(z-1) Is a coefficient of y (k), A (z)-1)＝1+a1z-1+…+anaz-na；
B(z-1) Is the coefficient of u (k-1), B (z)-1)＝b0+b1z-1+…+bnbz-nb；
C(z-1) Coefficient of ξ (k), C (z)-1)＝1+c1z-1+…+cncz-nc；
Wherein y (k) represents the blood glucose level of the user at time k in mg/dl, u (k-1) is the insulin injection rate at time k-1 in pmol/step, ξ (k) is white noise with a mean value of zero, and Δ ═ 1-z-1) Represents an integration factor; z is as described-1A backward shift operator; na, nb and nc represent the order of the model; a is described1～ana，b0～bnbAnd c1～cnzAll the model parameters can be optimized on line in real time, and initial values can be given when the model is initialized;
in order to predict the blood sugar change in the leading j steps, a chartlet equation is introduced; the charpy equation is shown in the following formula (2):
1＝Ej(z-1)A(z-1)Δ+Fj(z-1)z-j(2)
wherein,
Ej(z-1) Is A (z)-1) Coefficient of Δ, Ej(z-1)＝ej0+ej1z-1+…+ej-1z-j+1；
Fj(z-1) Is z-jCoefficient of (A), Fj(z-1)＝fj0+fj1z-1+…+fjnz-n；
Wherein, E isj(z-1) And Fj(z-1) Is formed by model parameters A (z)-1) And a polynomial uniquely determined for the prediction step j; a is described(z-1) And A (z) in the formula (1)-1) The same physical quantity and the same numerical value; said DELTA ═ 1-z-1) Represents an integration factor; z is as described-1Represents a back shift operator; z is as described-j+1An operator representing a step of backward shift by j-1, said z-jAn operator representing a step of backward shift j; z is as described-nAn operator representing a step of backward movement by n; said ej0～ej-1And fj0～fjnAre parameters that can be optimized online in real time, where j is the prediction step size, n is the maximum prediction step size, n is 8, j is 1,2 … n.
4. A closed-loop insulin infusion system based on adaptive generalized predictive control as claimed in claim 3, characterized in that: in step S2, the specific operation equation for obtaining the predicted value of the future blood glucose change of the user through the CARIMA model and the loss-of-image equation according to the current blood glucose value of the user is as follows:
multiplying both sides of the formula (1) by Ej(z-1) After Δ, the predicted value of the blood glucose level at time k, which is advanced by j steps, can be obtained by equation (2) as:
y(k+j)＝Gj(z-1)Δu(k+j-1)+Fj(z-1)y(k) (j＝1，2…n) (3)
Gj(z-1)＝Ej(z-1)B(z-1)
in the formula (3), y (k + j) represents the predicted value of the blood glucose level of the user in j steps ahead of the time k, and the unit is mg/dl; the delta u (k + j-1) represents the increment of the insulin injection rate of the insulin pump at the time of k + j-1, and the unit is pmol/step; said Gj(z-1) Is Ej(z-1) And B (z)-1) Represents the weight of the insulin injection rate increment between the time k and the time k + j-1 in predicting the blood glucose value of the time k ahead by j steps; y (k) represents the blood glucose level of the user at time k, and the unit is mg/dl; said Ej(z-1) And Fj(z-1) Both represent coefficients of formula (3) and E in formula (2)j(z-1) And Fj(z-1) The values are the same, wherein Ej(z-1) Is B (z)-1) Coefficient of (A), Fj(z-1) A coefficient of y (k); b (z) as defined-1) Is the coefficient of u (k-1) in formula (1).
5. A closed-loop insulin infusion system based on adaptive generalized predictive control as claimed in claim 3, characterized in that: the specific operation equation for calculating the insulin injection rate increment of the insulin pump by the minimum variance control algorithm according to the predicted value of the future blood glucose change of the user in the step S3 is as follows:
Figure FDA0002472036070000031
in the formula (4), the delta u (k + j-1) represents the increment of the insulin injection rate of the insulin pump when the k moment leads by j-1 step, and the unit is pmol/step;
j is a control target, and when the J takes the minimum value, the value of delta u (k + J-1) can be determined;
the n represents the maximum prediction length;
said λ represents a control weighting factor;
m represents the control length of insulin infusion;
y (k + j) represents the predicted value of the blood sugar value of the user which leads the user by j steps at the time k, and the unit is mg/dl;
w (k + j) is composed of blood glucose value y (k) at time k and reference curve yrCalculated, specifically expressed by the following formula:
w(k+j)＝αjy(k)+(1-αj)yr(j＝1,2,…,n)
equation (5) above can be further expressed in vector form:
W＝Qy(k)+Myr
said yrRepresenting an adaptive reference curve; the W is expressed by the following formula:
W＝[w(k+1),w(k+2),…,w(k+n)]T
said Q is expressed by the formula:
Q＝[α,α2,…,αn]T
the alpha represents an adaptive softening factor, and is more than or equal to 0 and less than or equal to 1;
said M is expressed by the formula:
M＝[1-α,1-α2,…,1-αn]T。
6. a closed-loop insulin infusion system based on adaptive generalized predictive control according to any of claims 1 to 5, characterized by: the adaptive reference curve comprises the following contents:
the self-adaptive reference curve sets different descending slopes of the self-adaptive reference curve according to the parameters of the current blood glucose concentration and by combining the variable quantity of the current blood glucose concentration; if the blood sugar concentration is lower than the set threshold value, the descending slope of the self-adaptive reference curve is 0.
7. The closed-loop insulin infusion system based on the adaptive generalized predictive control of claim 6, wherein: the self-adaptive reference curve comprises the following procedures:
scheme 1: during the time period which is preset in advance,
if the rate of change of blood glucose>k0mg/dl/step, defined as the blood sugar rising sharply;
if the rate of change of blood glucose is between k1mg/dl/step～k0mg/dl/step, which is defined as slow fluctuation of blood sugar;
if the rate of change of blood glucose is less than k1mg/dl/step, defined as a sharp drop in blood glucose;
wherein, k is1A preset value is set artificially; k is0A preset value which is artificially set, and k0>k1，k0>0mg/dl/step,k1<0mg/dl/step；
And (2) a flow scheme: the following judgments were made:
if the blood glucose concentration is higher than ζ1While the falling slope of the adaptive reference curve is set to k4mg/dl/step; what is needed isζ mentioned1The unit is a preset value which is set manually and is mg/dl; k is as described4A preset value, k, set manually4≥0mg/dl/step；
If the blood glucose concentration is between ζ2～ζ1When, the zeta2Preset value, ζ, set manually2<ζ1(ii) a According to the change condition of blood sugar, the following selections are carried out:
if there is a rapid rise in blood glucose, the falling gradient of the adaptive reference curve is set to k5mg/dl/step; k is as described5A preset value is set artificially;
if the condition of slow fluctuation of the blood sugar exists, the descending slope of the adaptive reference curve is set as k6mg/dl/step; k is as described6A preset value is set artificially;
if the blood sugar is sharply reduced, the descending slope of the self-adaptive reference curve is set to be 0; k is as described5>k6≥0mg/dl/step；
If the blood glucose concentration is between ζ3～ζ2When, the zeta3Preset value, ζ, set manually2>ζ3(ii) a According to the change condition of blood sugar, the following selections are carried out:
if there is a rapid rise in blood glucose, the falling gradient of the adaptive reference curve is set to k7mg/dl/step; k is as described7A preset value is set artificially;
if the condition of slow fluctuation of the blood sugar exists, the descending slope of the adaptive reference curve is set as k8mg/dl/step; k is as described8A preset value is set artificially;
if the blood sugar is sharply reduced, the descending slope of the self-adaptive reference curve is set to be 0; k is as described7>k8K is not less than 0mg/dl/step5≥k7，k6≥k8；
If the blood glucose concentration is lower than ζ3Then the adaptive reference curve is fixed to be set at a constant ζ, i.e., a slope of 0 mg/dl/step.
8. A closed-loop insulin infusion system based on adaptive generalized predictive control according to any of claims 1 to 5, characterized by: the obtaining of the self-adaptive softening factor comprises the following steps:
step A: artificially setting blood glucose expectation value
Figure FDA0002472036070000051
The unit is mg/dl;
and B: calculating the deviation degree u of the blood sugar value of the current user and the expected blood sugar value according to the blood sugar value y (k) of the current user k;
and C: calculating the current variation amount delta y of the blood glucose value of the user;
step D: calculating an adaptive softening factor alpha by the deviation u and the variation delta y, wherein the adaptive softening factor alpha is expressed by the following formula:
α＝u-|Δy|(u≥1)。
9. a closed loop insulin infusion system based on adaptive generalized predictive control according to claim 8, characterized by: in the step B, the deviation u is expressed by the following formula:
Figure FDA0002472036070000052
y (k) is the blood glucose concentration of the user at the moment k, and the unit is mg/dl;
said
Figure FDA0002472036070000053
The expected blood glucose value is given in mg/dl.
10. A closed loop insulin infusion system based on adaptive generalized predictive control according to claim 8, characterized by: in the step C, the change Δ y in blood glucose level is expressed by the following equation:
Δy＝y(k)-y(k-1)
y (k) is the blood glucose concentration of the user at the moment k, and the unit is mg/dl;
y (k-1) represents the blood glucose level of the user at the previous time, and has a unit of mg/dl.
11. A closed loop insulin infusion system based on adaptive generalized predictive control according to claim 2, 3, 4, 5, 7, 9 or 10 characterized by: in the step S4, the "iteratively optimizing the insulin injection rate increment of the insulin pump based on the idea of closed-loop control" specifically includes the following sub-steps:
step S4-1: taking the insulin injection rate increment of the insulin pump as an input value of a CARIMA model, and updating the predicted value of future blood sugar change of the user;
step S4-2: updating the insulin injection rate increment of the insulin pump according to the updated predicted value of the future blood glucose change of the user as the input of a minimum variance control algorithm model; the reference curve used by the minimum variance control algorithm is an adaptive reference curve, the softening factor used by the minimum variance control algorithm is an adaptive softening factor, the adaptive reference curve can adjust the slope of the reference curve according to the condition of a user, and the adaptive softening factor can adjust the value of the adaptive softening factor according to the change of the blood sugar value;
step S4-3: and circularly executing S4-1-S4-2 to realize iterative optimization of the insulin injection rate increment of the insulin pump.
12. The closed-loop insulin infusion system based on the adaptive generalized predictive control of claim 6, wherein: in the step S4, the "iteratively optimizing the insulin injection rate increment of the insulin pump based on the idea of closed-loop control" specifically includes the following sub-steps:
step S4-1: taking the insulin injection rate increment of the insulin pump as an input value of a CARIMA model, and updating the predicted value of future blood sugar change of the user;
step S4-2: updating the insulin injection rate increment of the insulin pump according to the updated predicted value of the future blood glucose change of the user as the input of a minimum variance control algorithm model; the reference curve used by the minimum variance control algorithm is an adaptive reference curve, the softening factor used by the minimum variance control algorithm is an adaptive softening factor, the adaptive reference curve can adjust the slope of the reference curve according to the condition of a user, and the adaptive softening factor can adjust the value of the adaptive softening factor according to the change of the blood sugar value;
step S4-3: and circularly executing S4-1-S4-2 to realize iterative optimization of the insulin injection rate increment of the insulin pump.
CN202010351347.6A 2019-04-30 2020-04-28 Closed-loop insulin infusion system based on adaptive generalized predictive control Active CN111643771B (en)

Applications Claiming Priority (8)

Application Number Priority Date Filing Date Title
CN201910365168.5A CN110124151A (en) 2019-04-30 2019-04-30 A kind of generalized predictive control closed-loop insulin infusion system based on adaptive reference curve strategy
CN2019103651596 2019-04-30
CN201910363842.6A CN109925568A (en) 2019-04-30 2019-04-30 It is a kind of based on adaptive softening because of the generalized predictive control infusion of insulin amount calculation method of substrategy
CN2019103651685 2019-04-30
CN201910365159.6A CN109925569A (en) 2019-04-30 2019-04-30 A kind of generalized predictive control infusion of insulin amount calculation method based on adaptive reference curve strategy
CN2019103638426 2019-04-30
CN2019103638572 2019-04-30
CN201910363857.2A CN110124150A (en) 2019-04-30 2019-04-30 It is a kind of based on adaptive softening because of the generalized predictive control closed-loop insulin infusion system of substrategy

Publications (2)

Publication Number Publication Date
CN111643771A true CN111643771A (en) 2020-09-11
CN111643771B CN111643771B (en) 2020-12-29

Family

ID=72343603

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010351347.6A Active CN111643771B (en) 2019-04-30 2020-04-28 Closed-loop insulin infusion system based on adaptive generalized predictive control

Country Status (1)

Country Link
CN (1) CN111643771B (en)

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112402731A (en) * 2020-10-10 2021-02-26 广东食品药品职业学院 Closed-loop insulin infusion system for preventing hypoglycemia phenomenon
CN112402732A (en) * 2020-10-10 2021-02-26 广东食品药品职业学院 Insulin infusion amount control method based on adaptive control weighting factor strategy
CN116159208A (en) * 2021-11-24 2023-05-26 上海微创生命科技有限公司 Artificial pancreas control method, readable storage medium and blood glucose management system
CN116504355A (en) * 2023-04-27 2023-07-28 广东食品药品职业学院 Closed-loop insulin infusion control method, device and storage medium based on neural network
CN116807464A (en) * 2023-08-30 2023-09-29 武汉联影智融医疗科技有限公司 Blood glucose control method and device based on artificial pancreas system and electronic equipment
CN117919546A (en) * 2024-01-26 2024-04-26 广东食品药品职业学院 Feedback linearization closed-loop insulin infusion control method and system based on neural network

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20120265125A1 (en) * 2008-09-12 2012-10-18 K&Y Corporation Infusion Pump System
CN104958077A (en) * 2015-07-24 2015-10-07 珠海福尼亚医疗设备有限公司 Intelligent control close-loop artificial pancreas system
CN106860955A (en) * 2017-01-21 2017-06-20 广东食品药品职业学院 Method based on fuzzy self-adaption proportional integral derivative control insulin pump closed loop infusion

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20120265125A1 (en) * 2008-09-12 2012-10-18 K&Y Corporation Infusion Pump System
CN104958077A (en) * 2015-07-24 2015-10-07 珠海福尼亚医疗设备有限公司 Intelligent control close-loop artificial pancreas system
CN106860955A (en) * 2017-01-21 2017-06-20 广东食品药品职业学院 Method based on fuzzy self-adaption proportional integral derivative control insulin pump closed loop infusion

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
LIU WENPING等: "Improved Generalized Predictive Control Algorithm for Blood Glucose Control of Type 1 Diabetes", 《ARTIFICIAL ORGANS》 *
余丽玲等: "闭环式人工胰腺血糖控制算法的研究进展", 《中国医疗器械信息》 *

Cited By (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112402731A (en) * 2020-10-10 2021-02-26 广东食品药品职业学院 Closed-loop insulin infusion system for preventing hypoglycemia phenomenon
CN112402732A (en) * 2020-10-10 2021-02-26 广东食品药品职业学院 Insulin infusion amount control method based on adaptive control weighting factor strategy
CN116159208A (en) * 2021-11-24 2023-05-26 上海微创生命科技有限公司 Artificial pancreas control method, readable storage medium and blood glucose management system
CN116159208B (en) * 2021-11-24 2024-03-15 上海微创生命科技有限公司 Artificial pancreas control method, readable storage medium and blood glucose management system
CN116504355A (en) * 2023-04-27 2023-07-28 广东食品药品职业学院 Closed-loop insulin infusion control method, device and storage medium based on neural network
CN116504355B (en) * 2023-04-27 2024-04-02 广东食品药品职业学院 Closed-loop insulin infusion control method, device and storage medium based on neural network
CN116807464A (en) * 2023-08-30 2023-09-29 武汉联影智融医疗科技有限公司 Blood glucose control method and device based on artificial pancreas system and electronic equipment
CN116807464B (en) * 2023-08-30 2024-01-26 武汉联影智融医疗科技有限公司 Blood sugar control device based on artificial pancreas system
CN117919546A (en) * 2024-01-26 2024-04-26 广东食品药品职业学院 Feedback linearization closed-loop insulin infusion control method and system based on neural network

Also Published As

Publication number Publication date
CN111643771B (en) 2020-12-29

Similar Documents

Publication Publication Date Title
CN111643771B (en) Closed-loop insulin infusion system based on adaptive generalized predictive control
CN110124150A (en) It is a kind of based on adaptive softening because of the generalized predictive control closed-loop insulin infusion system of substrategy
US8690820B2 (en) Automatic insulin pumps using recursive multivariable models and adaptive control algorithms
AU2024202178A1 (en) Closed loop control of physiological glucose
JP2020142112A (en) Glucose control system with automatic adaptation of glucose target
Hajizadeh et al. Plasma-insulin-cognizant adaptive model predictive control for artificial pancreas systems
JP7019127B2 (en) Insulin assessment based on reinforcement learning
CN112402731B (en) Closed-loop insulin infusion system for preventing hypoglycemia
CN106860955A (en) Method based on fuzzy self-adaption proportional integral derivative control insulin pump closed loop infusion
JP2019534066A (en) Automated system for controlling a patient&#39;s blood glucose level
JP2019528913A (en) Automated system for controlling a patient&#39;s blood glucose level
CN109925568A (en) It is a kind of based on adaptive softening because of the generalized predictive control infusion of insulin amount calculation method of substrategy
Panigrahi et al. Adaptive controller design based on grasshopper optimisation technique for BG regulation in TIDM patient
CN110124151A (en) A kind of generalized predictive control closed-loop insulin infusion system based on adaptive reference curve strategy
Patra et al. An automatic artificial pancreas based on AOA-VPTIDF control algorithm
CN116504355B (en) Closed-loop insulin infusion control method, device and storage medium based on neural network
Li et al. Fuzzy-PID control for the regulation of blood glucose in diabetes
WO2023092908A1 (en) Artificial pancreas control method, readable storage medium, and blood glucose management system
Patra et al. An adaptive control algorithm for blood glucose regulation in Type-I Diabetes Mellitus patients
Sylvester et al. Back stepping SMC for blood glucose control of type-1 diabetes mellitus patients
CN109925569A (en) A kind of generalized predictive control infusion of insulin amount calculation method based on adaptive reference curve strategy
CN112402732B (en) Insulin infusion quantity control method based on self-adaptive control weighting factor strategy
Rashid et al. Plasma insulin cognizant predictive control for artificial pancreas
WO2023070245A1 (en) Closed-loop artificial pancreas insulin infusion control system
Dief et al. PID Controller Tuning Using Multi-Objective Ant Colony Optimization for Blood Glucose Level of a Diabetic Patient

Legal Events

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