CN115645679A

CN115645679A - Self-adaptive closed-loop control method based on linear model

Info

Publication number: CN115645679A
Application number: CN202211282532.XA
Authority: CN
Inventors: 彭璨
Original assignee: Shandong Sijisuanli Technology Co ltd
Current assignee: Shandong Sijisuanli Technology Co ltd; Shenzhen Guiji Sensing Technology Co ltd
Priority date: 2022-10-19
Filing date: 2022-10-19
Publication date: 2023-01-31

Abstract

The present disclosure provides a linear model-based adaptive closed-loop control method, which is a control method for controlling a target physiological parameter of a target by using a chemical substance, and comprises: the method comprises the steps of obtaining preset parameters, obtaining target physiological parameters of a target and activity residual parameters of chemical substances, calculating sensitivity coefficients of the chemical substances based on the target physiological parameters and the activity residual parameters, calculating speed increasing coefficients of the target physiological parameters based on the target physiological parameters and the activity residual parameters, wherein the speed increasing coefficients are used for representing the influence degree of factors except the chemical substances on the target physiological parameters, calculating safe doses and basic doses of the chemical substances based on the preset parameters, the target physiological parameters, the activity residual parameters, the sensitivity coefficients and the speed increasing coefficients, and calculating target doses of the chemical substances based on the safe doses and the basic doses. In this case, it is possible to adaptively adjust the physiological parameter related to the target physiological parameter and reduce the input parameter.

Description

Self-adaptive closed-loop control method based on linear model

Technical Field

The present disclosure relates generally to the field of medical machines, and more particularly to a linear model based adaptive closed-loop control method.

Background

Diabetes is a group of metabolic diseases characterized by hyperglycemia, and in diabetic patients, due to insulin secretion deficiency and/or impaired biological action thereof, has symptoms such as hyperglycemia, unstable blood sugar, dyspepsia, and the like. In this case, chronic damage can occur to various tissues (particularly, eye, kidney, heart, blood vessel, nerve). In some cases, diabetic patients are treated by ingestion of medications to relieve symptoms. For example, a type one diabetic patient may avoid hyperglycemia by injecting insulin, and some type one diabetic patients may also use an insulin pump for auxiliary treatment, however, since the increase or decrease of blood sugar is related to various factors, the change of blood sugar is difficult to be accurately controlled, and for the treatment method of injecting insulin, a situation of continuous hyperglycemia due to insufficient injection amount or hypoglycemia due to excessive injection amount is easy to occur, thereby endangering the health of the diabetic patient. When the medicine is taken, in order to reduce adverse reactions caused by excessive or insufficient intake of the medicine, the intake of the medicine is regulated by a closed-loop algorithm.

Currently, commonly used closed-loop algorithms include traditional PID algorithms, which require input of a plurality of parameters with ambiguous physiological meaning and thus require individualized adjustment, or action model algorithms, which require input of carbon water intake that is difficult to quantify accurately, for example, chinese patent application publication No. CN 113453619A discloses a safety tool for making decision support suggestions for users of continuous blood glucose monitoring systems, which requires receiving a plurality of input data items affecting the diabetic condition of users of continuous blood glucose monitors, the plurality of input data items including nutritional data of food or beverages (equivalent to carbon water intake). Therefore, the existing closed-loop algorithm for regulating and controlling the intake of the medicine still has a higher use threshold, and the input of the intake of carbon water which is difficult to accurately quantify is also not beneficial to daily use of ordinary users.

Disclosure of Invention

The present disclosure has been made in view of the above-described state of the art, and an object of the present disclosure is to provide an adaptive closed-loop control method based on a linear model, which is capable of adaptively adjusting a physiological parameter related to a target physiological parameter while reducing input parameters.

Therefore, the present disclosure provides a linear model-based adaptive closed-loop control method, which is a control method for controlling a target physiological parameter of a target by using a chemical substance, and includes: the method comprises the steps of obtaining preset parameters, obtaining target physiological parameters of a target and activity residual parameters of the chemical substances, calculating sensitivity coefficients of the chemical substances based on the target physiological parameters and the activity residual parameters, wherein the sensitivity coefficients are used for representing the influence degree of the chemical substances on the target physiological parameters, calculating speed-increasing coefficients of the target physiological parameters based on the target physiological parameters and the activity residual parameters, the speed-increasing coefficients are used for representing the influence degree of factors except the chemical substances on the target physiological parameters, calculating safe doses and basic doses of the chemical substances based on the preset parameters, the target physiological parameters, the activity residual parameters, the sensitivity coefficients and the speed-increasing coefficients, and calculating target doses of the chemical substances based on the safe doses and the basic doses.

In this case, the sensitivity coefficient and the acceleration coefficient can be adaptively adjusted, so that the control accuracy is improved, and meanwhile, the acceleration coefficient can be used for representing the influence degree of factors other than chemical substances on the target physiological parameter, and the acceleration coefficient is obtained by calculating the target physiological parameter and the activity margin parameter, so that the input parameters (such as the carbohydrate intake of the target object) can be reduced, and the use threshold of the adaptive closed-loop control method is further reduced.

In addition, in the adaptive closed-loop control method according to the embodiment of the present disclosure, optionally, the preset parameter includes physiological information of the target, a kind of the chemical substance, a control step size, a unit of the target physiological parameter, a safety threshold of the target physiological parameter, and an infusion accuracy of the chemical substance. In this case, the parameters can be adaptively adjusted in advance according to the preset parameters, and the control accuracy can be improved.

In addition, in the adaptive closed-loop-control method according to the embodiment of the present disclosure, optionally, the preset parameter is obtained by data import or manual input. Under the condition, the preset parameters can be conveniently obtained through the data importing mode, the use threshold is further reduced, and the preset parameters can be conveniently modified through the manual input mode.

In addition, in the adaptive closed-loop-control method according to the embodiment of the present disclosure, optionally, an infusion dose of the chemical substance is entered into the target by way of infusion, and the infusion dose is related to the target dose and the infusion accuracy. In this case, the calculated target dose can be matched with the infusion accuracy of the actual device, resulting in an infusion dose that can be infused to the target within the accuracy range.

Further, in the adaptive closed-loop-control method according to the embodiment of the present disclosure, optionally, the infusion dose is metered by an infusion device. In this case, because the actual infusion has uncertainty, the infusion device is used for metering to obtain the infusion dosage, so that the actual dosage of the chemical substance infused into the target can be more accurately obtained, and the subsequent regulation and control precision can be improved conveniently.

In addition, in the adaptive closed-loop control method according to the embodiments of the present disclosure, the target physiological parameter is optionally acquired by a continuous blood glucose monitoring device or a blood glucose meter. In this case, the target physiological parameter can be acquired in different ways.

In addition, in the adaptive closed-loop control method according to the embodiment of the present disclosure, optionally, the target physiological parameter and the remaining activity parameter of the chemical substance of the target at a plurality of time nodes including a target node and other nodes are obtained, and the remaining activity parameter of the target node is obtained based on the remaining activity parameters of the other nodes. In this case, the parameters such as the activity margin parameter, the speed increasing factor, and the sensitivity factor do not change much in a short time. When parameters such as an activity margin parameter, a speed-up coefficient or a sensitivity coefficient of one time node are calculated, the accuracy of calculation can be improved by using the parameters of a plurality of time nodes.

In addition, in the adaptive closed-loop-control method according to the embodiment of the present disclosure, optionally, the sensitivity coefficient located at the target node is obtained by minimizing a loss function, where the loss function includes an error term and a regular term, the error term is used to represent a difference between the target physiological parameter obtained by measurement and the target physiological parameter obtained by calculation, and the regular term is used to represent a difference degree of the sensitivity coefficient of an adjacent time node. In this case, the accuracy of the sensitivity coefficient of each time node can be improved by using the error term, and meanwhile, because the sensitivity coefficient in a short time under a normal condition does not change too much, the stability and the accuracy of the sensitivity coefficient can also be improved by using the regular term.

In addition, in the adaptive closed-loop control method according to the embodiment of the present disclosure, optionally, the sensitivity coefficient of the target node is obtained based on the sensitivity coefficient of the other node, the target physiological parameter of the other node, the parameter of the surplus activity of the other node, and the target physiological parameter of the target node. In this case, the sensitivity coefficient of the target node can be calculated by using a part of parameters acquired in previous time nodes, so that adaptive adjustment of the sensitivity coefficient is realized, and meanwhile, the sensitivity coefficient generally does not change too much in adjacent time nodes, so that the stability and accuracy of the sensitivity coefficient can be improved by using the adjacent time nodes.

In addition, in the adaptive closed-loop control method according to the embodiment of the present disclosure, the rate-increase coefficient of the target node may be obtained based on a difference between the target physiological parameter obtained by measurement and the target physiological parameter obtained by calculation, which are located in the target node, and the rate-increase coefficient of the other node. In this case, the speed increase coefficient of the target node can be calculated using a part of the parameters acquired in the previous time node, so that adaptive adjustment of the speed increase coefficient is realized, and the accuracy of the speed increase coefficient can be improved.

According to the present disclosure, it is possible to provide a linear model-based adaptive closed-loop control method capable of reducing input parameters and adaptively adjusting a physiological parameter related to a target physiological parameter.

Drawings

Embodiments of the present disclosure will now be explained in further detail by way of example only with reference to the accompanying drawings, in which:

fig. 1 is a schematic diagram illustrating an application scenario of an adaptive closed-loop control method based on a linear model according to an embodiment of the present disclosure.

Fig. 2 is a schematic diagram illustrating a two-compartment dynamical model according to an embodiment of the present disclosure.

Fig. 3 is a flow chart diagram illustrating a linear model based adaptive closed-loop control method according to an embodiment of the present disclosure.

Fig. 4 is a schematic diagram illustrating an application scenario of acquiring a target physiological parameter of a target according to an embodiment of the present disclosure.

Fig. 5 is a graph illustrating a target physiological parameter according to an embodiment of the present disclosure.

Fig. 6 is a block diagram illustrating a structure of a linear model-based adaptive closed-loop control system according to an embodiment of the present disclosure.

Fig. 7 is a flow chart illustrating an adaptive closed-loop control system based on a linear model according to an embodiment of the present disclosure.

Fig. 8a is a simulation diagram illustrating the application of the naive PID method to adults in accordance with an embodiment of the present disclosure.

Fig. 8b is a simulation diagram illustrating that the pidfb method according to the embodiment of the present disclosure is applied to an adult.

Fig. 8c is a simulation diagram illustrating the application of the adaptive closed-loop-control method according to the embodiment of the present disclosure to an adult.

Fig. 9a is a simulation diagram illustrating the application of the naive PID method according to the embodiment of the disclosure to a teenager.

Fig. 9b is a simulation diagram illustrating that the pidfb method according to the embodiment of the present disclosure is applied to a teenager.

Fig. 9c is a simulation diagram illustrating the application of the adaptive closed-loop control method according to the embodiment of the disclosure to a teenager.

Fig. 10a is a simulation diagram illustrating the application of the naive PID method to a child according to an embodiment of the disclosure.

Fig. 10b is a simulation diagram illustrating the application of the pidfb method according to the embodiment of the present disclosure to a child.

Fig. 10c is a simulation diagram illustrating the application of the adaptive closed-loop-control method according to the embodiment of the present disclosure to a child.

Fig. 11 is a simulation diagram illustrating the application of the optimized adaptive closed-loop-control method according to the embodiment of the disclosure to a child.

Fig. 12a is a simulation diagram illustrating an application of the adaptive closed-loop-control method with infinite infusion accuracy to an adult according to an embodiment of the present disclosure.

Fig. 12b is a simulation diagram illustrating the application of the adaptive closed-loop-control method with limited infusion accuracy to an adult according to an embodiment of the present disclosure.

Detailed Description

Hereinafter, preferred embodiments of the present disclosure will be described in detail with reference to the accompanying drawings. In the following description, the same components are denoted by the same reference numerals, and redundant description thereof is omitted. The drawings are schematic and the ratio of the dimensions of the components and the shapes of the components may be different from the actual ones.

It is noted that the terms "comprises," "comprising," and "having," and any variations thereof, in this disclosure, for example, a process, method, system, article, or apparatus that comprises or has a list of steps or elements is not necessarily limited to those steps or elements expressly listed, but may include or have other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.

The embodiment of the disclosure relates to a linear model-based adaptive closed-loop control method, which utilizes a speed-increasing coefficient to represent the degree of influence of factors except for representing chemical substances on target physiological parameters, and simultaneously performs adaptive adjustment on a sensitive coefficient and the speed-increasing coefficient. In this case, it is possible to adaptively adjust the physiological parameter related to the target physiological parameter and reduce the input parameter.

The embodiment of the disclosure relates to an adaptive closed-loop control system based on a linear model, and the adaptive closed-loop control method is used for controlling a target physiological parameter of a target. In this case, the physiological parameter related to the target physiological parameter can be adaptively adjusted, the stability of the target physiological parameter of the target can be improved, and the input parameter can be reduced.

(background description-independent overview)

In some examples, referring to fig. 1, the adaptive closed-loop control method may be a control method of controlling a target physiological parameter of the target 2 with a chemical substance. In some examples, the linear model-based adaptive closed-loop-control method referred to in this document may be referred to as an LMPID method, a control method, or an adaptive closed-loop-control method.

In some examples, an infusion dose of the chemical may be obtained based on an adaptive closed-loop-control method and infused to the target 2 through the infusion device 20. The present disclosure is not so limited and chemicals may also enter target 2 through a variety of routes. For example, depending on the composition of the chemical substance, the chemical substance may also be delivered to the target 2 by oral, sublingual, rectal, transmucosal, inhalation, injection, etc. Preferably, the chemical substance may be infused to the target 2 by the infusion device 20, in which case the chemical substance can be delivered to the target 2 quickly and the time to effect the chemical substance is accelerated.

In some examples, the infusion device 20 may be an infusion device 20 fixed to the target 2 and automatically infused based on an infusion dose of the chemical substance obtained by an adaptive closed-loop-control method. In some examples, the infusion device 20 may also be a syringe, and the target 2 or other person may manually infuse the chemical to the target 2 based on the infusion dose of the chemical obtained by the adaptive closed-loop-control method. Preferably, the infusion device 20 may be an insulin pump (or artificial pancreas), in which case a chemical substance (e.g. insulin) can be conveniently delivered, while the closed loop insulin pump system can be composed in combination with the continuous blood glucose monitoring means 10a (CGM) and the processing means 30.

In some examples, the target 2 may be an animal, for example, the target 2 may be an animal such as a human, an orangutan, or a mouse. In some examples, target 2 may be a patient with an endocrine disease, e.g., target 2 may be a patient with a defect in a secretory gland (an endocrine gland such as adrenal gland, thyroid gland, pancreas, or pituitary). In this case, since a patient with endocrine diseases may have abnormal secretion function and/or structure of endocrine glands or endocrine tissues, which may cause imbalance of chemical substances for regulation in vivo, and thus imbalance of target physiological parameters, the adaptive closed-loop control method according to the embodiments of the present disclosure may control the chemical substances, and thus the target physiological parameters in a suitable range.

In some examples, target 2 may be a patient with diabetes, hypopituitarism, thyroid disease, or obesity. Preferably, the target 2 may be a patient suffering from diabetes, in particular a patient with type 1 diabetes (insulin dependent diabetes mellitus). In this case, since type 1 diabetes patients have difficulty in forming insulin, and are likely to cause hyperglycemia or large fluctuations in blood glucose, blood glucose can be effectively stabilized using the adaptive closed-loop control method according to the embodiment of the present disclosure.

In some examples, the chemical may be a substance having a special effect on the body, which is directly secreted into the blood by endocrine organs or tissues of humans and animals, for example, the chemical may be a hormone (including: the chemical may be pituitary hormone, insulin, glucagon, calcitonin, parathyroid hormone, etc.), preferably, the chemical may be insulin.

In some examples, the chemical substance may be a synthetic hormone. In some examples, the chemical may be a chemical that has a modulating effect on blood glucose.

In some examples, the target physiological parameter may refer to a blood glucose concentration of the target 2. In some examples, the target physiological parameter may also refer to a substance in the blood that has a particular effect on the body, for example, the target physiological parameter may also be a hormone or the like that is partially related to the target physiological health.

In order to better describe the adaptive closed-loop control method, the adaptive closed-loop control method according to the embodiment of the present disclosure will be described below, taking an example in which the target 2 is a diabetic patient, the chemical substance is insulin, and the target physiological parameter is blood glucose concentration. However, the adaptive closed-loop control method according to the embodiment of the present disclosure is not limited to this, and the adaptive closed-loop control method according to the embodiment of the present disclosure is also applicable to other cases where the target physiological parameter of the target 2 is controlled by a chemical substance.

Fig. 2 is a schematic diagram illustrating a two-compartment dynamics model according to an embodiment of the present disclosure.

In some examples, the adaptive closed-loop-control method may be based on a glucodynamic model. In some examples, the glycemic model may be expressed as:

wherein Δ G (k + 1) represents the change of the target physiological parameter (i.e. the change of the blood glucose concentration) at the (k + 1) th time node, G (k + 1) represents the target physiological parameter (i.e. the blood glucose concentration) at the (k + 1) th time node,

indicating the sensitivity coefficient of the chemical substance (i.e. the insulin sensitivity coefficient), I _eff Represents the parameter of the activity margin of the chemical substance (i.e. active insulin), and alpha represents the rate-increase coefficient of the target physiological parameter.

In some examples, the glycemic dynamics model is a linear model, and thus the adaptive closed-loop-control method may also be referred to as a linear model-based adaptive closed-loop-control method. In this case, the linear model can conveniently reflect the rule of the glucodynamic model at adjacent time nodes (short time intervals), so that the calculation cost can be reduced, the calculation speed can be increased, and the timeliness of the adaptive closed-loop control method can be improved.

In some examples, a metabolic model of insulin within target 2 may be established using a two-chamber model. In some examples, referring to fig. 2, the two-compartment model may include a central compartment r1 and a peripheral compartment r2. The insulin in the central chamber r1 may be the insulin infused through a vein and entered into the atrioventricular space, and the insulin in the peripheral chamber r2 may be the insulin diffused from the atrioventricular space into the blood vessel, which may also be called active insulin.

In some examples, the two-compartment model may be expressed as,

wherein t may represent time, I _p Can represent the amount of insulin in the central compartment r1, I _sc The infusion dose of insulin (i.e., the insulin infused to the target) may be represented, τ may represent the time at which the insulin reaches a peak, and h (t) may represent the shock response of the system.

In some examples, based on the two-compartment model, the metabolic model may be expressed as:

I _eff (k+1)＝K ₀ I _sc (k)+K ₁ I _eff (k)-K ₂ I _eff (k-1)，

K ₀ ＝1-K ₁ +K ₂ ，

where K may represent the kth time node, Δ t may represent the time interval between two adjacent time nodes, Δ t may also be referred to as a control step, K ₀ ，K ₁ And K ₂ Parameters related to diffusion rate in insulin at different locations can be separately expressed.

In some examples, K ₀ ，K ₁ And K ₂ May be a constant obtained based on the control step size and the time at which insulin reaches the peak, in other words, K ₀ ，K ₁ And K ₂ Can be obtained by preset parameters.

In some examples, as described above, the adaptive closed-loop-control method may be established based on a glucodynamic model and a metabolic model.

In some examples, referring to fig. 3, an adaptive closed-loop-control method may include: acquiring preset parameters (step S101), acquiring target physiological parameters of the target 2 and the remaining activity parameters of the chemical substance (step S103), calculating the sensitivity coefficient of the chemical substance based on the target physiological parameters and the remaining activity parameters (step S105), calculating the rate-increasing coefficient of the target physiological parameters based on the target physiological parameters and the remaining activity parameters (step S107), calculating the safe dose and the basic dose of the chemical substance based on the preset parameters, the target physiological parameters, the remaining activity parameters, the sensitivity coefficient and the rate-increasing coefficient (step S109), and calculating the target dose of the chemical substance based on the safe dose and the basic dose (step S111).

In this case, the sensitivity coefficient and the acceleration coefficient can be adaptively adjusted, so that the control accuracy is improved, and meanwhile, the acceleration coefficient can be used for representing the influence degree of factors other than chemical substances on the target physiological parameter, and the acceleration coefficient is obtained by calculating the target physiological parameter and the activity margin parameter, so that the input parameters (such as the carbohydrate intake of the target 2) can be reduced, and the use threshold of the adaptive closed-loop control method is further reduced.

(step S101)

In some examples, as described above, in step S101, preset parameters may be acquired.

In some examples, the preset parameters may include at least one of physiological information of the target 2, a kind of the chemical substance, a control step size, a unit of the target physiological parameter, a safety threshold of the target physiological parameter, and an infusion accuracy of the chemical substance. In this case, the parameters can be adaptively adjusted in advance according to the preset parameters, and the control accuracy can be improved.

In some examples, the preset parameters may include physiological information of the target 2. In some examples, the physiological information may include readily available medical information such as gender, age, weight, disease course, and the like. In some examples, the physiological information may also include family history and medication history. In this case, since the sensitivity coefficient of the chemical substance has a large correlation with the physiological information of the individual of the target 2 among the different targets 2, the initial value of the sensitivity coefficient can be estimated by acquiring the physiological information.

In some examples, the preset parameter may include a type of chemical substance, for example the preset parameter may include a type of insulin. In this case, since the time at which different kinds of chemical substances reach the peak in blood differs, the time at which the chemical substances reach the peak in blood can be set based on the kind of the chemical substances.

In some examples, the preset parameter may include a control step size, and the control step size may be an interval between two adjacent time nodes. In this case, the input control step size can be adjusted by the time node interval, so that different control step sizes can be selected for different targets 2. For example, if the fluctuation of the target physiological parameter of the target 2 is relatively stable, the control step length may be selected to be larger, and if the fluctuation of the target physiological parameter of the target 2 is relatively severe, the control step length may be selected to be shorter. In some examples, the appropriate control step size may also be selected based on other reasons.

In some examples, the preset parameter may include units of a target physiological parameter, for example, where the target physiological parameter is blood glucose concentration, the target physiological parameter may be millimoles per liter (mmol/l) or milligrams per deciliter (mg/dl). In this case, due to different usage habits, people in different regions may use different units, so that the unit for inputting the target physiological parameter can facilitate the usage habits of different people, reduce the conversion difficulty, and improve the universality of the adaptive closed-loop control method.

In some examples, the preset parameters may include a safety threshold for the target physiological parameter. In this case, upon subsequent acquisition of an infusion dose of the chemical substance, an appropriate infusion dose of the chemical substance can be obtained based on the safety threshold, so that the target physiological parameter can be controlled within a safe range.

In some examples, the preset parameter may include an infusion accuracy of the chemical substance. In this case, since different infusion devices 20 may have different infusion accuracies, the same infusion device 20 may also set different infusion accuracies, and the dosage of the chemical substance actually infused to the target 2 is related to the infusion accuracy of the infusion device 20, and may not be completely the same as the target dosage calculated by the adaptive closed-loop control method, and by obtaining the infusion accuracy of the chemical substance, the accuracy of the chemical substance actually infused to the target 2 can be determined, and thus the dosage of the chemical substance actually infused to the target 2 can be better determined, and thus the calculation by the adaptive closed-loop control method is facilitated.

In some examples, the preset parameters are obtained by means of data import or manual input. For example, a part of the preset parameters may be obtained by data import, and a part of the preset parameters may be obtained by manual input. Under the condition, the preset parameters can be conveniently obtained through the data importing mode, the use threshold is further reduced, and the preset parameters can be conveniently modified through the manual input mode.

In some examples, a portion of the preset parameters may be entered by a healthcare worker. In this case, the medical staff can set the preset parameters based on the professional knowledge and experience. In some examples, preset parameters may also be input by target 2. In this case, the target 2 can adaptively adjust the preset parameters based on the situation of the individual.

In some examples, a portion of the preset parameters may be obtained via internal parameters of the infusion device 20, such as by invoking internal parameters of the infusion device 20, or via technical specifications or instructions for use of the device.

(step S103)

Fig. 4 is a schematic diagram illustrating an application scenario of acquiring a target physiological parameter of a target 2 according to an embodiment of the present disclosure. Fig. 5 is a graph illustrating a target physiological parameter according to an embodiment of the present disclosure.

In some examples, step S103 may be to obtain a target physiological parameter of the target 2 and a parameter of a margin of activity of the chemical substance. In this case, the margin of activity parameter of the chemical substance can be determined, and since the chemical substance which still maintains activity may be present in the target 2, the margin of activity parameter of the chemical substance can be taken into account when the target dose is obtained by using the adaptive closed-loop control method, so that the target physiological parameter exceeding a safe range (for example, the blood glucose concentration is lower than a safe threshold) due to the target dose being too large can be reduced.

In some examples, the balance-of-activity parameter for a chemical may be understood in the following manner, the balance-of-activity parameter being the amount (or concentration) of the chemical that remains active in target 2, or the amount (or concentration) of the chemical that is still effective and can be adjusted for the target physiological parameter.

In some examples, referring to fig. 4, the target physiological parameter may be obtained by the continuous blood glucose monitoring device 10a (CGM) or the blood glucose meter 10b, in other words, the obtaining apparatus 10 may be the continuous blood glucose monitoring device 10a or the blood glucose meter 10b. In this case, the target physiological parameter can be acquired in different ways.

In some examples, the target physiological parameter may be continuously acquired with the continuous blood glucose monitoring device 10 a. In this case, the use of the continuous blood glucose monitoring device 10a can improve the convenience of acquiring the target physiological parameter, while facilitating the processing of a large number of target physiological parameters by the processing device 30. In some examples, the continuous blood glucose monitoring device 10a may be disposed in an arm, abdomen, or thigh, among other locations.

In some examples, blood of the target 2 may also be collected through a blood collection needle (e.g., collecting blood from a fingertip, palm, or arm, etc.), and a target physiological parameter in the finger blood may be measured using the blood glucose meter 10b. Preferably, blood (finger blood) of the fingertip of the target 2 may be collected. In this case, since the fingertip has a highly dense capillary network, the change in blood glucose concentration in the body can be reflected quickly, and the control of blood glucose concentration can be facilitated.

In some examples, referring to fig. 5, the target physiological parameter and the residual activity parameter of the chemical substance of the target 2 may be acquired at a plurality of time nodes, which may be consecutive time nodes. In some examples, the plurality of time nodes includes a target node and other nodes, and the activity margin parameter of the target node is obtained based on activity margin parameters of the other nodes. In this case, the parameters such as the parameter of the activity margin, the rate-increasing coefficient, and the sensitivity coefficient do not change much in a short time. When parameters such as an activity margin parameter, a speed-up coefficient or a sensitivity coefficient of one time node are calculated, the calculation accuracy can be improved by using the parameters of a plurality of time nodes.

In some examples, referring to fig. 5, the other nodes may be nodes before the target node. For example, the target node may be the (k + 1) th time node, and the other nodes may be the (k + 1) th time node and the (k-1) th time node. It should be noted that the same time node may be the target node or other nodes. Specifically, one time node corresponds to one control step, and in one control step, the 5 th time node may be a target node, and the 4 th time node and the 3 rd time node may be other nodes. In the next control step, the 6 th time node may be a target node, and the 5 th and 4 th time nodes may be other nodes.

In some examples, the time node may include not less than 2 other nodes and 1 target node in one control step, and preferably, the time node may include 2 other nodes and 1 target node in one control step. In this case, a large number of other nodes can represent a complex algorithm model, and the large number of other nodes can reduce the complexity of the algorithm and reduce the calculation cost.

In some examples, the activity margin parameter may satisfy the metabolic model described above:

I _eff (k+1)＝K ₀ I _sc (k)+K ₁ I _eff (k)-K ₂ I _eff (k-1)，

in this case, the active insulin of the target node can be obtained based on the active insulin, the infusion dose of insulin of the other node.

(step S105)

In some examples, step S105 may calculate a sensitivity coefficient of the chemical based on the target physiological parameter and the parameter of the activity margin.

In some examples, the sensitivity factor may be used to characterize the degree of influence of the chemical substance on the target physiological parameter, and in the case where the chemical substance is insulin, the sensitivity factor may also be referred to as an insulin sensitivity factor. In some examples, the sensitivity coefficient is negative, the smaller the sensitivity coefficient, the greater the degree of influence of the chemical substance on the target physiological parameter; the greater the sensitivity coefficient (closer to 0), the less the chemical substance has an effect on the target physiological parameter. In other words, the greater the absolute value of the coefficient of sensitivity, the greater the degree of influence of the chemical substance on the target physiological parameter; the smaller the absolute value of the sensitivity coefficient, the smaller the degree of influence of the chemical substance on the target physiological parameter.

In some examples, the sensitivity coefficient may be obtained iteratively through a formula. In this case, the sensitivity coefficient of each time node can be obtained by using formula iteration, so that the sensitivity coefficient of adaptive adjustment can be obtained, and the control precision can be further improved.

In some examples, the initial value of the sensitivity coefficient may be actively input or calculated by the processing device 30 based on preset parameters. In some examples, the initial value of the sensitivity coefficient may be obtained by a 500-rule, a 1500-rule, or a 1800-rule, among others. In this case, the initial value of the sensitivity coefficient can be obtained conveniently, and then the sensitivity coefficient of each time node can be obtained by iteration through a formula.

In some examples, the formula for calculating the sensitivity coefficient may be obtained by a loss function. Specifically, the sensitivity coefficient at the target node may be obtained by minimizing a loss function, where the loss function includes an error term and a regular term, the error term is used to represent a difference between the target physiological parameter obtained by measurement and the target physiological parameter obtained by calculation, and the regular term is used to represent a difference degree between the sensitivity coefficients of adjacent time nodes. In this case, the accuracy of the sensitivity coefficient of each time node can be improved by using the error term, and meanwhile, because the sensitivity coefficient in a short time under a normal condition does not change too much, the stability and the accuracy of the sensitivity coefficient can also be improved by using the regular term.

In some examples, the sensitivity coefficient of the target node may be obtained based on the sensitivity coefficient of the other node, the target physiological parameter of the other node, the activity margin parameter of the other node, and the target physiological parameter of the target node. In this case, the sensitivity coefficient of the target node can be calculated by using a part of parameters acquired in previous time nodes, so that adaptive adjustment of the sensitivity coefficient is realized, and meanwhile, the sensitivity coefficient generally does not change too much in adjacent time nodes, so that the stability and accuracy of the sensitivity coefficient can be improved by using the adjacent time nodes.

The manner in which the sensitivity factor is obtained is further described below, and in some examples, the loss function may satisfy the formula:

wherein the content of the first and second substances,

may represent a loss function, G (k) may represent a target physiological parameter (i.e., blood glucose concentration) at the kth time node, α (k-1) may represent a rate-increase coefficient at the kth-1 time node,

can represent the sensitivity coefficient (i.e. insulin sensitivity coefficient) of the chemical substance at the kth time node, I _eff (k-1) may represent the activity margin parameter (i.e., active insulin) of the chemical substance at the (k-1) th time node, μmay represent a regularization term coefficient for adjusting the action ratio of the error term and regularization term,

the predicted value of the coefficient of sensitivity of the chemical substance at the k-1 th time node (i.e. the predicted value of the insulin sensitivity coefficient) can be represented,

it is possible to represent the error term in,

a regularization term may be represented.

It should be noted that the forms of the error term and the error term are not limited to this, and all the expressions that can be used to represent the difference between the target physiological parameter obtained by measurement and the target physiological parameter obtained by calculation may be used as the error term, for example, expressions such as the ratio, the difference, or the absolute value (or square, cube, or multiple power) of the difference between the target physiological parameter obtained by measurement and the target physiological parameter obtained by calculation may be used as the error term, and expressions that can be used to represent the degree of difference between the sensitivity coefficients of adjacent time nodes may also be used as the regular term, for example, expressions such as the ratio, the difference, or the absolute value (or square, cube, or multiple power) of the difference between the sensitivity coefficients of adjacent time nodes may be used as the regular term.

In some examples, the regularization term coefficients may be set empirically or as default data set for processing device 30.

In some examples, the predicted value of the coefficient of sensitivity for the chemical may satisfy the formula:

wherein the content of the first and second substances,

show order

And (4) minimizing.

In some examples, a first iterative formula for the coefficient of sensitivity may be obtained based on a formula that a predicted value of the coefficient of sensitivity for the chemical satisfies. In some examples, the first and second pairs can be connected by a common pair

The derivation mode obtains a first iterative formula of the sensitivity coefficient. In some examples, the iteration step size in the first iteration formula may be set based on experience.

In some examples, the sensitivity coefficient obtained by the calculation (i.e., the predicted value of the sensitivity coefficient) may be calculated as the sensitivity coefficient of another step.

(step S107)

In some examples, step S107 may calculate a rate-increase coefficient of the target physiological parameter based on the target physiological parameter and the activity margin parameter. In some examples, the rate of increase factor, which may also be referred to as a blood glucose increase when the chemical is insulin, may be used to characterize the degree of effect of factors other than the chemical on the target physiological parameter, and may be used to characterize the intake of carbohydrate, endogenous glucoseGlucose is secreted or not

Explained the degree of influence of factors such as insulin sensitivity change on blood glucose concentration. In this case, since the acceleration factor can be used to represent the effects of various factors, and the calculation of the acceleration factor does not require the input of carbon (described later), the input of parameters can be reduced, and the accuracy of the model can be improved.

In some examples, the rate-increase coefficient of the target node is obtained based on a difference between a target physiological parameter obtained by measurement and the target physiological parameter obtained by calculation at the target node and rate-increase coefficients of other nodes. In this case, the speed increase coefficient of the target node can be calculated using a part of the parameters acquired in the previous time node, so that adaptive adjustment of the speed increase coefficient is realized, and the accuracy of the speed increase coefficient can be improved.

The manner in which the speed-increasing factor is obtained is further described below, and in some examples, the speed-increasing factor may satisfy the formula:

wherein the content of the first and second substances,

may represent a predicted value of the rate increase coefficient at the kth time node,

may represent a calculated value of the blood glucose concentration at the kth time node (i.e. a predicted value of the blood glucose concentration),

it may represent the difference between the predicted value of the blood glucose concentration and the measured value of the blood glucose concentration at the kth time node.

It is noted that the predicted value of the blood glucose concentration and the measured value of the blood glucose concentration may be calculated byThe following means understand that the prediction value of blood glucose concentration

Is an intermediate parameter appearing in step S109, and may be obtained based on the blood glucose concentration at the k-1 th time node, the predicted value of the insulin sensitivity coefficient, the active insulin, the acceleration coefficient, and the blood glucose dynamics model, the measured value of the blood glucose concentration is the blood glucose concentration obtained by the continuous blood glucose monitoring device 10a or the blood glucose meter 10b described above, and the blood glucose concentration in the body of the target 2 and the blood glucose concentration involved in the derivation process in the other steps may be the measured value of the blood glucose concentration, unless otherwise specified.

In some examples, referring to the result of the reduction of the formula satisfied by the speed-increasing coefficient, the speed-increasing coefficient may satisfy a second iterative formula:

in other words, the rate of increase may represent the sum of the differences between the predicted values of the blood glucose concentration and the measured values of the blood glucose concentration in the different control steps (i.e. in the different time nodes).

In some examples, the initial value of the speed increasing coefficient may be 0, but the present disclosure is not limited thereto, and the speed increasing coefficient may also be an arbitrary value.

(step S109)

In some examples, in step S109, a safe dose and a base dose of the chemical substance may be calculated based on the preset parameter, the target physiological parameter, the remaining activity parameter, the sensitivity coefficient, and the rate-increase coefficient, the safe dose may refer to a dose at which the target physiological parameter does not fall below a safe threshold value for a long time after the chemical substance is infused into the target 2, and the base dose may refer to a dose at which the target physiological parameter can reach the target value for a short time (e.g., one control step) after the chemical substance is infused into the target 2. In this case, after the chemical substance is infused, the chemical substance with activity is present in each time node during the metabolic process of the chemical substance, and under the action of the chemical substance with activity in a plurality of time nodes, the adjustment on the target physiological parameter is overlapped, so that the target physiological parameter may exceed a safe range (for example, be lower than a safe threshold), and the safe dose and the basic dose are calculated at the same time, so that the difference between the two doses can be determined, and the safety and the adjustment effect of the infused dose can be ensured.

In some examples, in calculating the base dose, the activity margin parameter for the chemical substance may satisfy the formula according to a hemodynamic model:

wherein, G _d (k + 1) may represent a target value of the target physiological parameter in the (k + 1) th time node, which may be obtained by calculation in some examples. For example, the target value can be obtained by calculating the target physiological parameter and the control step size of the kth time node. In some examples, the target value may also be obtained by human input.

In some examples, in calculating the base dose, the activity margin parameter of the chemical substance may also satisfy the formula according to the hemodynamic model:

wherein λ and ρ can represent adjusting parameters. In this case, since

There may be a case of approaching 0, leading to I _eff (k) The degree of change is large, and the addition of the adjusting parameters can stabilize I _eff (k) So that the subsequently calculated base dose can be controlled to be within a suitable range. In some examples, the tuning parameters may be selected or adjusted empirically.

In some examples, the base dose may be obtained from a metabolic model of insulin and a calculation formula of a margin of activity parameter of the chemical.

In some examples, the adjustment parameter ρ may be used to adjust the base dose, for example, in the case of adjusting the blood glucose concentration by using an adaptive closed-loop control method, the same infusion dose may have different degrees of influence on adults and children, and a safe and appropriate dose for adults may have a large influence on children (e.g., cause hypoglycemia), so that the dose of children may be proportionally reduced, and thus, the value of ρ may be in the range of 0 to 1, and different ρ may be set for different populations. Under the condition, the adaptability of the self-adaptive closed-loop control method to different crowds can be improved.

In some examples, the sustained effect of the chemical that remains active on the target physiological parameter may be calculated first in the calculation of the safe dose. Taking insulin as an example, a metabolic model based on insulin, as well as historical insulin infusion information, may predict future changes in the amount of active insulin over time.

In some examples, the total effect on blood glucose concentration of insulin infused to target 2 may be obtained using future active insulin and insulin sensitivity coefficients, and in particular, after insulin is infused to target 2 at the k-1 time node, the effect on blood glucose concentration of active insulin may be integrated and the total effect on blood glucose concentration obtained.

In some examples, the sensitivity factor may be constant when calculating the total influence of the infusion of insulin on the blood glucose concentration at the k-1 st time node, in which case, since one control step corresponds to one time node, each control step may calculate the total influence of active insulin on the blood glucose concentration, so that the glucokinetic model established in the single control step may be appropriately simplified, and when calculating the total influence of active insulin on the blood glucose concentration in one control step, the sensitivity factor may be constant, which may reduce the calculation cost and increase the calculation speed.

In some examples, in calculating the safe dose, the safe dose may be obtained from the total effect of active insulin on blood glucose concentration and a glycemic dynamics model, without considering the rate-of-increase factor.

In some examples, during the process of calculating the safe dose, a PID (Proportional Integral Derivative) control principle may be further added in consideration of a speed-increasing coefficient, specifically, a PID term may be constructed by using the speed-increasing coefficient, in which case, the accuracy and stability of the adaptive closed-loop control method can be improved.

(step S111)

In some examples, a target dose of the chemical may be calculated based on the safe dose and the base dose in step S111. In this case, the calculation results of the safe dose and the base dose can be simultaneously considered and an appropriate target dose can be obtained.

In some examples, the smaller of the safe dose and the basal dose may be the target dose. In this case, the safety of the adaptive closed-loop control method can be improved.

In some examples, as described above, an infusion dose of chemical may be entered into the target 2 by way of infusion, the infusion dose being related to the target dose and infusion accuracy. In some examples, the infusion dose may be calculated with a downward cut. For example, in the infusion device 20, the infusion accuracy is 0.1U/time, the target dose obtained by calculation is 6.47U, and the infusion dose may be 6.4U. In this case, the calculated target dose can be matched to the infusion accuracy of the actual device, resulting in an infusion dose that can be infused to the target 2 within an accuracy range.

In some examples, the infusion dose may be metered by the infusion device 20. In this case, since there is uncertainty in the actual infusion, the infusion device 20 is used to measure and obtain the infusion dose, so that the actual dose of the chemical substance infused into the target 2 can be obtained more accurately, and the accuracy of subsequent regulation and control can be improved.

Fig. 6 is a block diagram showing the structure of the linear model-based adaptive closed-loop control system 1 according to the embodiment of the present disclosure. Fig. 7 is a flowchart showing the linear model-based adaptive closed-loop control system 1 according to the embodiment of the present disclosure.

As mentioned above, the present disclosure also relates to an adaptive closed-loop control system 1 based on a linear model, which controls a target physiological parameter of a target 2 by using the adaptive closed-loop control method according to the embodiment of the present disclosure. In this case, the physiological parameter related to the target physiological parameter can be adaptively adjusted, the stability of the target physiological parameter of the target 2 is improved, and the input parameter is reduced.

In some examples, referring to fig. 6, the linear model-based adaptive closed-loop control system 1 may include an infusion device 20, a processing arrangement 30, and an acquisition device 10.

In some examples, referring to fig. 7, the acquisition device 10 may be used to acquire a target physiological parameter of the target 2.

In some examples, referring to fig. 7, the infusion device 20 may infuse a chemical to the target 2 to adjust a target physiological parameter of the target 2.

In some examples, referring to fig. 7, the processing arrangement 30 may calculate the dosage of the chemical substance that the infusion device 20 needs to infuse based on the pre-set coefficients entered at the previous stage and the target physiological parameter acquired by the acquisition device 10.

In some examples, referring to fig. 7, after the processing device 30 obtains the target physiological parameter, the sensitivity coefficient and the activity margin parameter may be calculated, and the safe dose and the base dose may be calculated based on the sensitivity coefficient and the activity margin parameter, and the target dose may be calculated based on the safe dose and the base dose, and sent to the infusion device 20.

In some examples, referring to fig. 7, the infusion device 20 may infuse an infusion dose of a chemical to the target 2 based on the target dose.

In some examples, the processing device 30 may obtain the target dosage and control the infusion apparatus 20 to infuse the chemical substance using an adaptive closed-loop control method according to embodiments of the present disclosure.

(simulation results)

Fig. 8a is a simulation diagram illustrating the application of the naive PID method to adults in accordance with an embodiment of the present disclosure. Fig. 8b is a simulation diagram illustrating that the pidfb method according to the embodiment of the present disclosure is applied to an adult. Fig. 8c is a simulation diagram illustrating the application of the adaptive closed-loop-control method according to the embodiment of the present disclosure to an adult. Fig. 9a is a simulation diagram illustrating the application of the naive PID method according to the embodiment of the disclosure to a teenager.

Fig. 9b is a simulation diagram illustrating that the pidfb method according to the embodiment of the present disclosure is applied to a teenager. Fig. 9c is a simulation diagram illustrating the application of the adaptive closed-loop control method according to the embodiment of the disclosure to a teenager. Fig. 10a is a simulation diagram illustrating the application of the naive PID method to a child according to an embodiment of the disclosure. Fig. 10b is a simulation diagram illustrating the application of the pidfb method according to the embodiment of the present disclosure to a child. Fig. 10c is a simulation diagram illustrating the application of the adaptive closed-loop-control method according to the embodiment of the present disclosure to a child.

The simulation schematic diagram is a schematic diagram obtained by a Simglucose model, simulation objects comprise adults, young people and children, a Dexcom and Insule continuous blood glucose monitoring device 10a (CGM) is used in the simulation, and an initial value of an insulin sensitivity coefficient can be set through a 1800 rule.

In some examples, a naive PID method may refer to a control method that takes historical blood glucose data and insulin injection data as inputs, does not require the declaration of carbohydrate, and uses PID alone, a pidfb method may be a control method that combines a PID method and insulin feedback, and LMAPID may refer to an adaptive closed-loop control method to which embodiments of the present disclosure are directed.

Referring to fig. 8 a-10 c, the adaptive closed-loop control method according to the embodiment of the present disclosure is superior to the naive PID method and pidfb method in simulation results, regardless of adults, adolescents or children.

Referring to fig. 10c, when the adaptive closed-loop control method is applied to children, the simulation results of target 2 of #008 and #003 present a risk of hypoglycemia. In this regard, the adaptive closed-loop-control method may be optimized, for example, the target dose of insulin may be reduced proportionally for children. In particular, the target dose can be reduced by an equal ratio of 50% to 90%, in which case the risk of hypoglycemia can be reduced.

Referring to fig. 11, the optimized adaptive closed-loop control method (target dose reduction by 70%) effectively reduces the risk of hypoglycemia.

Fig. 12a is a simulation diagram illustrating an application of the adaptive closed-loop-control method with infinite infusion accuracy to an adult according to an embodiment of the present disclosure. Fig. 12b is a simulation diagram illustrating the application of the adaptive closed-loop-control method with limited infusion accuracy to an adult according to an embodiment of the present disclosure.

It should be noted that in an adaptive closed-loop control method with infinite infusion accuracy, the target dose may be the same as the infusion dose. In an adaptive closed-loop control method with limited infusion accuracy, the infusion dose may be calculated based on the infusion accuracy and the target dose using the method described above. Fig. 12b shows the infusion accuracy is 0.1U/time. Referring to fig. 12a and 12b, the simulation result of the adaptive closed-loop control method with infinite infusion accuracy is substantially the same as the simulation result of the adaptive closed-loop control method with infusion accuracy of 0.1U/time, which illustrates that the adaptive closed-loop control method according to the embodiment of the present disclosure has low sensitivity to infusion accuracy within a certain accuracy range, in other words, can maintain a good control effect even when the adaptive closed-loop control method according to the embodiment of the present disclosure is used in an infusion apparatus 20 with low accuracy within a certain accuracy range.

While the present disclosure has been described in detail in connection with the drawings and examples, it should be understood that the above description is not intended to limit the disclosure in any way. Those skilled in the art can make modifications and variations to the present disclosure as needed without departing from the true spirit and scope of the disclosure, which fall within the scope of the disclosure.

Claims

1. A linear model-based adaptive closed-loop control method is a control method for controlling a target physiological parameter of a target by using a chemical substance, and is characterized by comprising the following steps:

the preset parameters are obtained and the preset parameters are obtained,

acquiring a target physiological parameter of the target and an activity margin parameter of the chemical substance,

calculating a sensitivity coefficient of the chemical substance based on the target physiological parameter and the activity margin parameter, wherein the sensitivity coefficient is used for representing the influence degree of the chemical substance on the target physiological parameter,

calculating a rate-increasing coefficient of the target physiological parameter based on the target physiological parameter and the activity margin parameter, the rate-increasing coefficient being used for representing the degree of influence of factors other than the chemical substance on the target physiological parameter,

calculating a safe dose and a basic dose of the chemical substance based on the preset parameter, the target physiological parameter, the activity margin parameter, the sensitivity coefficient and the rate-increase coefficient,

calculating a target dose of the chemical based on the safe dose and the basal dose.

2. The adaptive closed-loop-control method according to claim 1,

the preset parameters include physiological information of the target, a kind of the chemical substance, a control step size, a unit of the target physiological parameter, a safety threshold of the target physiological parameter, and an infusion precision of the chemical substance.

3. The adaptive closed-loop-control method according to claim 1,

the preset parameters are obtained through data import or manual input.

4. The adaptive closed-loop-control method according to claim 1,

an infusion dose of the chemical into the target is caused by infusion, the infusion dose being related to the target dose and the infusion accuracy.

5. The adaptive closed-loop-control method according to claim 4,

the infusion dose is metered by an infusion device.

6. The adaptive closed-loop-control method according to claim 1,

the target physiological parameter is collected by a continuous blood glucose monitoring device or a blood glucose meter.

7. The adaptive closed-loop-control method according to claim 1,

acquiring the target physiological parameters and the activity residual parameters of the chemical substances of the target at a plurality of time nodes, wherein the plurality of time nodes comprise a target node and other nodes, and acquiring the activity residual parameters of the target node based on the activity residual parameters of the other nodes.

8. The adaptive closed-loop-control method according to claim 7,

obtaining the sensitivity coefficient located at the target node by minimizing a loss function, wherein the loss function comprises an error term and a regular term, the error term is used for representing the difference between the target physiological parameter obtained by measurement and the target physiological parameter obtained by calculation, and the regular term is used for representing the difference degree of the sensitivity coefficient of the adjacent time node.

9. The adaptive closed-loop-control method according to claim 8,

obtaining the sensitivity coefficient of the target node based on the sensitivity coefficient of the other node, the target physiological parameter of the other node, the activity margin parameter of the other node, and the target physiological parameter of the target node.

10. The adaptive closed-loop-control method according to claim 7,

and obtaining the speed-increasing coefficient of the target node based on the difference between the target physiological parameter obtained by measurement and the target physiological parameter obtained by calculation and the speed-increasing coefficients of the other nodes.