JPS631060B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to an improvement in an artificial pancreas equipped with an insulin injection device based on a predetermined insulin injection program. This invention relates to an artificial pancreas device that automatically performs conversion in response to changes in the characteristics of a blood sugar regulating mechanism to achieve adaptive control of blood sugar levels.

Various methods and devices are known for artificially injecting insulin into patients with insulin secretion dysfunction, such as diabetic patients, based on a fixed program. The present applicant has also published Japanese Patent Application Laid-Open No. 53-47195 (Japanese Patent Publication No. 58-44382) regarding this type of artificial pancreas.
A patent application was filed as .

Traditionally, insulin therapy for diabetic patients consists of 1
This is often done by subcutaneously injecting depot insulin once a day. However, since blood sugar levels or insulin secretion in living bodies fluctuate based on certain biorhythms, it is impossible to maintain normal blood sugar levels even if a large amount of insulin is administered once a day. Furthermore, with current treatment methods, blood sugar levels fluctuate widely, and the progress of diabetic vascular lesions cannot be prevented.

Therefore, we continuously measure blood sugar using a glucose sensor, etc., calculate the appropriate amount of insulin to inject based on the obtained blood sugar level, and then, based on this data, administer insulin at a predetermined amount under the action of a metering pump. Various artificial pancreases have been proposed that employ a so-called batch method insulin injection amount control method in which insulin is injected at time intervals.

However, in this type of artificial pancreas,
It has been difficult to administer an appropriate amount of insulin based on ever-changing blood sugar levels.

Therefore, the applicant filed the application earlier.
- In No. 44382, exogenous insulin is injected into diabetic patients in response to ever-changing blood sugar levels, and the method is based on a specific program or calculation formula in order to reproduce the same sugar metabolic state as seen in healthy individuals. proposed a method of injecting insulin. That is, the plasma insulin concentration at a predetermined time t
IRI (t) is the objective variable, blood sugar level BG (t) (mg/100
ml) and rate of change in blood sugar level ΔBG (t) (mg/100
The relational expression when ml・min) is used as an independent variable is obtained by the following equation.

IRI(t)=aBG(t)+bΔBG(t)+c...(1) However, a, b, and c are constant numbers and are unique values for each individual.

When exogenous insulin is administered intravenously to a diabetic patient with low or absent endogenous insulin secretion, the relationship between plasma insulin concentration IRI (t) and insulin injection volume IIR (insulin injection volume per minute) is as follows. It becomes like the expression.

θ・d IRI(t)/dt=IIR−θ・F・D・IRI(
t)...(2) However, IRI(t): Plasma insulin concentration (μU/
ml) θ: Insulin space [body weight x 15.7/100 (ml)] D: Rate of insulin uptake into tissues per minute (min ^-1 ) F: Diffusion constant (dimensionless) IIR: Insulin injection volume (μU/min ) Therefore, when calculating the insulin injection amount by eliminating IRI(t), which cannot be measured in a short time, from the above equations (1) and (2), the following equation is obtained.

IIR=Kp・BG(t)+Kd・ΔBG(t)+Kc...(3) However, Kp=F・D・θ・a (comparative action coefficient) Kd=(a+F・D・b)・θ(differential action coefficient) Kc = F・D・θ・c・(Basal insulin secretion constant) Based on the equation (3) obtained in this way, an arithmetic circuit that performs this calculation is provided, and this is used to continuously control the blood glucose level. By connecting and arranging a glucose sensor that detects blood pressure, an insulin injection device, and a controller that controls the insulin injection device based on the calculated value calculated by the calculation circuit, it is possible to maintain a plasma insulin concentration that is always physiologically compatible. Insulin injection can be achieved according to the conditions, and as a result, the injection amount is 1/2 to 1/3 that of conventional subcutaneous injection,
An artificial pancreas was proposed that has the characteristic of not causing hypoglycemia.

However, the above-mentioned artificial pancreas cannot be used in various pathological conditions of diabetes, especially when insulin resistance is shown due to infection, when endogenous insulin secretion ability exists or when insulin sensitivity is high, and even when insulin resistance is high, such as in unstable diabetes. To apply this to cases where sensitivity changes, etc., after extracting the individual characteristics of blood sugar control ability after a certain period of application, the parameters of the insulin injection program (a, b, c shown in equations (1) to (3) above) ) must be changed manually to perform adaptive control of blood sugar. However, such manual changes are extremely troublesome to control when changes in hormone-metabolic dynamics occur over a long period of time, and automatic control is desired in order to further improve adaptability and control performance. Therefore, in the present invention, in order to cope with changes in the patient's metabolic dynamics, the parameters of the insulin infusion program are automatically converted over time based on a certain algorithm in response to changes in the characteristics of the blood sugar regulation mechanism. , provides an artificial pancreas that executes a parameter self-conversion program designed to adaptively control blood sugar levels.

As a result of various studies, the present inventors found that in order to inject insulin in response to changes in individual characteristics,
Convert the parameters a, b, and c in equation (1) above to obtain a ^* ,
Since it is necessary to create an algorithm to calculate b ^* and c ^* , we first administered oral glucose loads of 50 g and 100 g to 48 healthy subjects, respectively, in order to obtain the blood glucose response characteristic curve seen in healthy subjects as a control target. When blood was collected over time and blood sugar levels were determined, the first
The results shown in the figure were obtained. That is, the first
The figure illustrates the blood sugar level and the rate of change in blood sugar at each point in time after the blood sugar level reaches its peak value. From the results shown in FIG. 1, for example, a linear relational expression indicated by a broken line can be derived.

ΔBG(t)=−0.0176・BG(t)+1.091 (4) Therefore, from the above equation (4), the blood sugar level when ΔBG(t)=0 is 60 mg/100ml.

Next, from the measured values obtained in this way, 80
With mg/100ml as the target value for blood sugar control, linear relational expressions (5) to (7) between the two were obtained as shown by the solid line in FIG.

ΔBGp(t)=-1.5 [BG(t)＞165]...(5) ΔBGp(t)=-0.0176・BG(t)+1.408 [165≧BG(t)≧55]...(6) ΔBGp(t)=0.44 [55>BG(t)...(7) However, ΔBGp(t): Target value of the rate of change in blood sugar per minute at time (t) Therefore, the rate of change in the measured blood sugar ΔBG(t)
From the measured blood sugar level BG(t), the above formula (5)
The difference between the target value ΔBGp(t) and the rate of change in blood sugar determined by applying ~(7) is a constant value L (e.g. 0.3~0.5
mg/100ml・min) or more, from the following formula, a ^* ,
Find b ^* and c ^* .

1/SI _i = 1/SI _i-1 + [ΔBG(t)−ΔBGp(t)]
/W...(8) a ^* =a/SI _i ...(9) b ^* =b/SI _i ...(10) c ^* =c/SI _i ...(11) However, SI _i : Parameter conversion Insulin sensitivity index SI _i-1 : Insulin sensitivity index W before parameter conversion: Weight coefficient Note that SI _i =1 during normoglycemic control.

In this way, the control target for adaptive control was set as the blood sugar response curve seen in healthy individuals, and the blood sugar level at which ΔBG(t) = 0 was measured based on the obtained blood sugar response curve, and it was found to be 60 mg/100 ml. However, the blood sugar level control target was set at 80 to prevent the onset of hypoglycemia.
mg/100ml, and the control target for the rate of change in blood sugar [ΔBGp(t)] is set to 110
±55mg/100ml (M±2SD)], ΔBGp between blood sugar levels 55-165mg/100ml
(t) is calculated from the linear relationship between the blood sugar level and the rate of change of blood sugar in a healthy person, ΔBGp(t) = -0.0176・BG(t)+1.408
In addition, if the blood sugar level is 55mg/100ml or less, ΔBGp(t) = 0.44, and if the blood sugar level is 165mg/100ml.
As a result of setting ΔBGp(t)=-1.5 in the case of ml or more,
Blood glucose lowering rate is approximately 90 when a small amount of insulin is continuously injected at 5B to 100B [B is the basal insulin infusion rate, usually 225μU/Kg・min] to control blood sugar in ketosis or diabetic coma in pancreatectomized diabetic dogs and diabetic patients.
It was found that there was good agreement with mg/100ml/hr.

Therefore, in the present invention, a glucose sensor that continuously measures blood sugar levels, a calculation circuit that calculates an insulin injection amount according to the measured blood sugar level, an insulin injection device, and a calculation value calculated by the calculation circuit are provided. In an artificial pancreas equipped with a controller that controls an insulin injection device, the calculation circuit includes a calculation circuit that calculates an insulin injection amount based on a calculation formula, and a conversion circuit that self-converts parameters of the calculation formula at a predetermined time, Furthermore, a circuit is provided for detecting a deviation value between the measured blood sugar level and the rate of change in blood sugar level determined by a relational expression that determines a target value for blood sugar control obtained from the blood sugar response characteristics of a healthy person. The present invention is characterized in that the parameters of the arithmetic expression are self-converted by the conversion circuit when the arithmetic expression reaches a predetermined value, and the insulin injection amount is calculated by the calculation circuit based on the converted arithmetic expression.

In the present invention, when converting parameters, the problem is how far the measured value deviates from the control target value to be considered non-physiological. Statistically,
Based on the linear relational expression shown in Figure 1, deviations outside the 95% confidence band should be considered abnormal, but the variation is large, so applying that value makes it difficult to perform constant control based on individual characteristics. becomes. Therefore, the difference between the target value ΔBGp(t) of the rate of change in blood sugar and the rate of change ΔBG(t) with respect to the blood glucose level at the time of measurement is 0.3 to 0.5.
When it was set to be non-physiological when it exceeded mg/100ml/min, and the parameters were set to be converted, a physiological blood sugar response curve could be obtained.

In this case, the change in insulin sensitivity of the individual or during a certain pathological condition is the deviation of the blood glucose change rate at that point from the target blood glucose change rate, that is, [ΔBG
(t)−ΔBGp(t)]. Therefore, the insulin sensitivity index at each point in the blood glucose response curve observed in healthy subjects is set to 1, and [ΔBG
(t) - ΔBGp(t)] is multiplied by a certain weighting coefficient (W) to obtain the individual's insulin sensitivity index. Based on this index, insulin injection parameters a, b, and c are calculated using equations (9) to (11). As a result of conversion, the weighting coefficient W
By setting the value to 0.5 to 1.0, results with good followability can be obtained; when it is less than 0.5, overcontrol occurs, and when it is more than 1.0, control is impossible.

In addition, taking into account the time delay until the onset of insulin action after insulin injection, a certain period of time after parameter change is set as the refractory time for parameter conversion, which is approximately 15
It is preferable to set the time to ~20 minutes.

Although it is possible to operate the parameter self-conversion program through the above operations, in order to fully utilize the functions of this program, processing of noise in blood glucose measurement becomes a problem. Therefore, based on the fact that the change in the blood sugar level BGi for i minutes ago closely matches the hyperbolic tangent, this was applied to create a filter. That is, the following equation However, k: amplitude of hyperbolic tangent function ω: coefficient of fitting Perform regression calculation using least squares method with k as variable,
The indicator of variation RI is a constant value R (mg/100ml)
An algorithm was created to perform parameter conversion only in the following cases. However, in determining the constant value R, it is necessary to take into account the fact that the variation in blood sugar level measurement is small when the blood sugar level is low and becomes large when the blood sugar level is high, and to correct it based on the blood sugar level. Therefore, by setting the dispersion index RI using the following formula, the blood sugar level 500
The range of variation around mg/100ml is 100mg/100ml.
5 times that of ml can be tolerated.

The functions of each program related to the operation of the parameter self-conversion program described above are summarized as shown in FIG. 2 as a computer flowchart of the algorithm, and the contents thereof will be described in detail later.

Therefore, another embodiment of the present invention is a parameter self-conversion program that grasps an individual's ability to regulate blood sugar over time, automatically converts the parameters of an insulin injection program under a certain algorithm, and controls blood sugar. When executing, determine the deviation width from the control target value for which parameter conversion should be performed,
To provide an artificial pancreas device capable of adaptively controlling blood sugar in response to any dynamic changes by determining the refractory time until the onset of insulin action after insulin injection, and processing noise in blood sugar measurement. .

Next, embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

FIG. 3 shows the basic configuration of the present invention, in which the blood glucose level of a diabetic patient or hypoglycemic patient 10 with insulin secretion dysfunction is measured using a glucose sensor 12.
The blood glucose level signal measured at 2 is configured to be transmitted to the controller 14. Next, the output of the controller 14 operates the arithmetic circuit 16 in which a program for calculating the amount of insulin to be injected is incorporated, and calculates the amount of insulin to be injected into the living body. It is configured to drive and control the insulin injection pump 18 to inject an appropriate amount of insulin from the insulin reservoir 20 to the patient 10.

Hereinafter, in the artificial pancreas of the present invention, the amount of insulin is calculated based on the blood sugar level measured by the glucose sensor 12 based on the following equation.

IIR=Kp・BG(t)+Kd・ΔBG(t)+Kc...(3) However, Kp=F・D・θ・a (proportional action coefficient) Kd=(a+F・D・b)・θ(differential action coefficient) Kc = F・D・θ・c (basal insulin secretion constant) IIR...Insulin injection amount (μU/min) θ...Insulin space [body weight x 15.7/100 (ml)] D...Insulin taken into tissue every minute F...Diffusion constant ( ^{dimensionless} ) ΔBG(t)...Rate of change in blood sugar level at time t (mg/100ml・min) a, b, c... Constants specific to the patient's medical condition ( Parameter) In this case, the basal insulin injection rate B is
It was set to 225μU/Kg・min.

In addition, in order to inject insulin in response to changes in the patient's individual characteristics, the parameters a in equation (3) above,
b and c are converted under the following conditions.

(1) Setting the control target value [ΔBGp(t)] for the rate of change in blood sugar a. When the blood sugar level is 165 mg/100 ml or more ΔBGp (t) = -1.5 ......(5) b When the blood sugar level is 55 to 165 mg/100 ml ΔBGp(t)=-0.0176・BG(t)+1.408...(6) c When blood sugar level is 55mg/100ml or less ΔBGp(t)=0.44...(7) (2) Between the measured value and control target value Deviation setting |ΔBG(t)−ΔBGp(t)|≧L...(14) L=0.3~0.5 In this case, it is determined that it is non-physiological, and the parameters are converted. (3) Set the filter to remove noise. setting When the index RI of this variation is less than a certain value R, the parameters (a, b, c or Kp,
Kd, Kc) is converted.

(4) Setting the refractory time The refractory time after a parameter change is set to 15 to 20 minutes until the next parameter conversion becomes possible.

Next, we apply the above algorithm to PANAFACOMU−
200 mini-computer and the validity of parameter conversion using a pancreatectomized diabetic dog. In these experimental examples, the basal insulin infusion rate B was 225 μU/Kg・min
And so.

Experimental Example 1 (Glucose control during experimental ketoacidosis) Pancreatectomized diabetic dogs were given 3-OHBA (3-oxybutyric acid;
3-hydroxy butyric acid) was continuously injected at 0.08 mM/Kg·min over 5 hours of the experiment. The blood sugar response characteristics in this case are shown in FIG. Plasma 3-
OHBA level is 5.43-5.95mM/, arterial blood PH
It was 7.206.

First, when 3-OHBA is not injected, the parameters of the insulin infusion program are a=0.1, b=0.4, c
When blood sugar control is performed with = 0, a blood sugar response characteristic curve as shown by the broken line in Figure 4 is obtained, and as a result, the blood sugar level gradually decreases from the previous value of 400 mg/100 ml, and after 3 and a half hours it reaches 100 mg/100 ml. It reached 100ml. Furthermore, when blood sugar control is performed using the same insulin injection program parameters as described above during 3-OHBA injection, a blood sugar response characteristic curve as shown by the dashed-dotted line in Figure 4 is obtained, resulting in blood sugar levels. Even after 5 hours, the insulin level remained high at 250 to 260 mg/100 ml, indicating a low insulin resistance value.

Next, when blood sugar control was attempted under the operation of the self-conversion program, a blood sugar response characteristic curve as shown by the solid line in FIG. 4 was obtained. The parameters of the insulin infusion program in this case are a=
0.1, b=0.4, c=0, and in the self-conversion program L=0.3, W=0.5, R=1.0, RP=20,
It was set as ω = 0.1. The parameters were automatically converted 30 minutes after the start of insulin infusion, and as shown in the figure, parameter conversion was performed a total of 4 times during the experiment, increasing the insulin infusion amount and lowering the blood glucose level to 3-OHBA.
It was possible to approximate the non-infusion case (blood sugar response characteristic curve). SI _i in this case is SI ₁ (30 minutes) = 0.55, SI ₂ (60 minutes) = 0.42, SI ₃ at each time point.
(100 minutes) = 0.33, SI ₄ (120 minutes) = 0.24, and the parameters of the insulin infusion program (a = 0.1, b
= 0.4, C = 0) at each time point, respectively (a ^* = 0.18, b ^* = 0.73, c ^* = 0), (a ^* = 0.24, b ^*
= 0.95, c ^* = 0), (a ^* = 0.31, b ^* = 1.22, c ^* =
0), (a ^* =0.41, b ^* =1.64, c ^* =0). Also, when comparing the amount of insulin injected, 3-
3- compared to 823×B/5 hours without OHBA injection.
If the insulin resistance during OHBA injection approaches the blood glucose response characteristic curve when 3-OHBA is not injected under the self-conversion program, 2.314×B/
It took 5 hours.

Experimental Example 2 (Blood Glucose Control During Loading) In the same manner as in Experimental Example 1, epinephrine (epinephrine) was continuously injected at a rate of 0.1 μg/Kg·min over 5 hours to a pancreatectomized diabetic dog. The blood sugar response characteristics in this case are shown in FIG.

First, when no adrenaline is injected, the parameters of the insulin infusion program are a=0.1, b=0.4, c
When blood sugar control was carried out with =0, a blood sugar responsiveness curve as shown by the broken line in FIG. 5 was obtained. Furthermore, if blood sugar control is performed using the same parameters of the insulin injection program as described above during adrenaline injection, a blood sugar response characteristic curve as shown by the dashed-dotted line in FIG. 5 will be obtained, and this was confirmed to be impossible.

Next, when blood sugar control was attempted under the operation of the self-conversion program, a blood sugar response characteristic curve as shown by the solid line in FIG. 5 was obtained. The parameters of the insulin infusion program in this case are a=
0.1, b=0.4, c=0, and in the self-conversion program L=0.5, W=1, R=1.0, RP=2.0,
It was set as ω = 0.1. 20 minutes after starting insulin infusion,
After another 80 minutes, the parameters were automatically converted, making it possible to bring the blood sugar level closer to that without epinephrine injection (blood sugar response characteristic curve). In this case
SI _i is SI ₁ (20 minutes) = 0.66, SI ₂ at each time point
(80 minutes) = 0.46, and the parameters of the insulin infusion program (a = 0.1, b = 0.4, c = 0) are:
At each time point, (a ^* = 0.15, b ^* =
0.61, c ^* = 0), (a ^* = 0.22, b ^* = 0.88, c ^* = 0)
Converted to . The amount of insulin injected at this time is 660 × B / 5 hours when epinephrine is not injected, whereas when insulin resistance is controlled under the self-conversion program operation during epinephrine injection,
It was 1.420×B/5 hours.

From the above experimental example, the parameters at the start of insulin injection are a = 0.1, b = 0.4, and c = 0, respectively, and the parameter self-conversion program is L = 0.5,
By setting W = 1.0, R = 1.0, RP = 20, and ω = 0.1, the parameters will be automatically converted according to the degree of insulin resistance of the individual (pancreatectomized dog), and the amount of insulin injected will be adjusted. It was possible to bring the blood sugar level close to the blood sugar level when 3-OHBA or epinephrine was not injected. In addition, by applying this algorithm, the amount of 3-OHBA injection was 28 times that of non-injection, and the amount of adrenaline injection was 28 times that of non-injection.
It was confirmed that 2.2 times the amount of insulin injected was required, and quantitative evaluation of insulin resistance was possible.

In the following experimental example, 3-OHBA was continuously injected during the experimental period, and although this differs from the clinical trial, the presence of ketone bodies led to a decrease in blood sugar and poor insulin sensitivity, necessitating an increase in the amount of insulin injected. For example, it is known that insulin resistance is often observed during insulin treatment for diabetic coma, and that at least 8 U/hr. is required during low-dose insulin infusion therapy. However, in the clinical cases mentioned above, it is common to see improvement in glucose metabolism by insulin injection, as well as disappearance of ketone bodies and correction of PH.
Resistance usually decreases over time. Therefore, the results of the above-mentioned experimental examples are in good agreement with the results in clinical cases where, during prolonged ketosis, insulin resistance is shown despite a decrease in blood sugar levels, and a larger amount of insulin is required. It is thought that the insulin sensitivity of an individual is closely related to the height of blood sugar level and the amount of insulin injected at each time point.

As is clear from the foregoing, the present invention is applied to an insulin injection program incorporated into bedside artificial pancreatic β cells, and the parameters of the program are changed over time under a certain algorithm in the characteristics of the blood sugar regulation mechanism. If the parameter self-conversion program is configured to self-convert immediately in response to the change in blood glucose level, the parameter self-conversion program can calculate the target blood glucose level for the rate of change in blood sugar during blood sugar control, which is determined from the relational expression obtained from the relationship between the blood sugar level and the rate of change in blood sugar level in a healthy person. The deviation from the rate of change can be taken as an index of the individual's insulin sensitivity, and the insulin infusion rate can be automatically changed, and the parameter self-conversion program obtained in this way can be applied to pancreatectomized diabetic dogs using 3-OHBA.
Or, as a result of adapting to the insulin resistance model during continuous infusion of epinephrine, it is possible to adaptively control blood sugar during insulin resistance.Furthermore, when pancreatectomized diabetic dogs are injected with 3-OHBA or epinephrine continuously, they show insulin resistance and non-diabetic dogs. It became clear that 2.8 times and 2.2 times the amount of insulin needed to be injected, respectively, to achieve the same blood sugar control as during injection.

Therefore, according to the present invention, it is possible to automatically change the parameters of an insulin infusion program using changes in insulin sensitivity at each point in time as an index, and to perform adaptive control of blood sugar. Not only can it be applied to quantification, but it can also be a powerful means for elucidating blood sugar regulation mechanisms during various pathological conditions.

The preferred embodiment of the present invention has been described above. In this embodiment, when creating the parameter self-conversion program, the relationship between the blood sugar level and the rate of change of blood sugar in a healthy person was derived as a linear relational expression. It goes without saying that higher-order relational expressions can be derived without limitation, and that various other design changes can be made without departing from the spirit of the present invention.

[Brief explanation of the drawing]

Fig. 1 is a blood sugar response characteristic curve diagram when a glucose load is applied to a large number of healthy subjects, Fig. 2 is a flowchart for creating a parameter self-conversion program for implementing the present invention, and Fig. 3 is a diagram for implementing the present invention. FIG. 4 is a block control circuit diagram of an artificial pancreas.
A characteristic curve diagram showing the blood sugar response characteristics and insulin injection amount characteristics when OHBA is injected. Figure 5 shows the blood sugar response characteristics and insulin when the artificial pancreas according to the present invention is applied to a pancreatectomized diabetic dog and epinephrine is injected. It is a characteristic curve diagram showing injection amount characteristics. 10...patient, 12...glucose sensor, 1
4...Controller, 16...Arithmetic circuit, 18...
...insulin infusion pump, 20...insulin reservoir.

Claims (1)

[Claims] 1. A glucose sensor that continuously measures blood sugar levels;
An artificial pancreas device comprising: an arithmetic circuit that calculates an insulin injection amount according to a measured blood sugar level; an insulin injector; and a controller that controls the insulin injector based on the arithmetic value calculated by the arithmetic circuit. The calculation circuit includes a calculation circuit that calculates the insulin injection amount based on a calculation formula, and a conversion circuit that self-converts the parameters of the calculation formula at a predetermined time point. The conversion circuit includes a circuit that detects a deviation value between the rate of change in blood sugar level determined by a relational expression that determines the blood sugar level and the measured blood sugar level, and when the deviation value reaches a predetermined value, An artificial pancreas device, characterized in that it is configured to self-convert the formula and calculate the amount of insulin to be injected in the calculation circuit based on the converted arithmetic expression. 2 The calculation circuit uses the following calculation formula IIR=Kp・BG(t) +Kd・ΔBG(t)+Kc However, Kp=F・D・θ・a (proportional operation coefficient) Kd=(a+F・D・b)・θ (differential action coefficient) Kc = F・D・θ・c (basal insulin secretion constant) [In the formula, IIR: Insulin injection amount (μU/min) BG (t): Blood sugar level at time t (mg/100ml) ) ΔBG (t): Rate of change in blood sugar level at time t (mg/100ml・min) F: Diffusion constant (dimensionless) D: Rate of insulin taken into tissue per minute (min ^-1 ) θ: Insulin Space [body weight x 15.7/100 (ml)] a, b, c: parameters that are constants unique to each individual and are self-converted] Claim 1 comprising calculating the insulin injection amount based on the following: Artificial pancreas device as described in Section. 3. The conversion circuit converts the target value of the rate of change in blood sugar [ΔBGp(t)] into: a) When blood sugar level is 165mg/100ml or more ΔBGp(t)=-1.5 b When blood sugar level is 55 to 165mg/100ml ΔBGp(t) =-0.0176・BG(t)+1.408 c When the blood sugar level is 55mg/100ml or less, ΔBGp(t)=0.44, and the following formula between the measured blood sugar level [ΔBG(t)] and ΔBG(t) -ΔBGp(t)|≧L When the relationship of L=0.3 to 0.5 holds true, the artificial body according to claim 2, which comprises self-converting the parameters (a, b, c) of the insulin injection amount calculation formula. pancreatic apparatus. 4 The conversion circuit is an indicator of blood sugar level dispersion (RI)
is the following formula: RI=√ ₂ /BG(t)×100≦R However, [In the formula, BGi: blood glucose level k before time i minutes: amplitude of hyperbolic tangent function ω: coefficient of fitting R: freely set value (for example, 1.0)] When the following is satisfied, the parameter (a) of the insulin injection amount calculation formula , b, c). 5. The artificial pancreas device according to claim 1, wherein the arithmetic circuit sets a refractory time until the next number of parameter conversions becomes possible after parameter conversion.