ES2334733B2 - ADAPTIVE CONTROL SYSTEM IN CLOSED LINK OF CONTINUOUS INSULIN INFUSION BY INVESTING A GLUCEMIA MODEL. - Google Patents

Abstract

Adaptive closed loop control system continuous insulin infusion by inversion of a model of blood glucose

It is a method of glycemic control in closed loop that aims to induce normoglycemia in a patient with diabetes The method is applicable through a system of insulin infusion intravenously or subcutaneously.

Use as the sole source of information the measurement of a continuous subcutaneous glucose sensor and employs a Mathematical model of blood glucose dynamics adapted to the patient. Originality consists in combining a control method by inversion of a blood glucose model with an adjustment method adaptive to control blood glucose deviations from intake. This method is applicable to other processes in the medical field, veterinary or industrial, being necessary to have a model mathematician of its dynamics and that the variables to be controlled are of nature continues.

Description

Adaptive closed loop control system continuous insulin infusion by inversion of a model of blood glucose

Technical sector

This invention is part of the sector of automatic control of biological systems and presents as main field of application type 1 diabetes in which a insulin treatment, however it can be applied in any another field of medicine that seeks to remedy, through the technology, a physiological dysfunction by dosing A drug to the patient. In the same way and with the same idea you can apply to the veterinary sector for dispensing drugs to animals. It can also be applied in the industrial sector to control in closed loop any continuous system in time which is available information of the variables to control.

         \ vskip1.000000 \ baselineskip
      

State of the art

Already since the 70s is trying create the Artificial Pancreas (PA) with different measuring equipment, of infusion and with diverse methods of glycemic control [6]. The The present invention falls within the field of glycemic control in closed loop

To date there are many jobs presented in this field and few who have taken the practice, mainly due to the unavailability of methods Effective continuous glucose measurement and difficulties for your post processing One of the problems presented by glycemic control methods is the complexity of the scenario of operation due largely to variations in the Physiology of patients and variations of actions Daily of the patient such as intakes and physical exercise. TO this variability must be added the problems of delays present in the control process, such as the delay associated with insulin absorption and the delay associated with the measurement subcutaneous glucose that is related to the concentration in blood that was a while ago (15-30 minutes). The most control methods work reactively to a error, so the delay of the control action (absorption of an insulin infusion) can cause risk situations, due to episodes of hypo and hyperglycemia, being the most compromising for life the first.

To date the scientific community has worked primarily with the following types of algorithms closed loop control: Proportional Integral and Derivative control (PID), predictive model control (MPC), fuzzy control and robust control, in its invariant and adaptive versions [3]. In the year 2001 appears a patent that protects a device from insulin dosage based on a PID method [5] and in 2006 Steil [15] conducts a clinical experiment with a PID regulator. For avoid hypoglycemia, a tactic used in drivers that depend on an error, is to raise the level of reference (ideal value of normoglycemia) what it has as consequence an increase of the average glycemia, moving away the control glycemic regarding the ideal medical action and compromising the Safety of hyperglycemia therapy.

According to the scientific community it is necessary propose an adaptive method [4] in the face of variations in patient parameters if you want to achieve results adjusted to each patient, we must also develop methods that "cancel" existing delays. For these reasons, authors focus their invention on the adaptive control method fed back by reversing a blood glucose model (or model of the glucoregulatory system), which runs the system of control.

Some papers have already been presented with adaptive glycemic control such as Fisher et al . [4] that in 1987 the need to use adaptive control systems was already raised; in 1988 Woodruff et al . [7] proposed a self-tuning adaptive control with a Kalman filter; in 1992 Jushász et al . [8] propose an adaptive control by reference model for a small number of measures; in 1994 Candas et al . [9] proposed an adaptable controller with a very simple model of the glucorregulatory system; in 2001 Bellazzi et al . [10] affirm that adaptive control by reference model (MIT Rule) and self-tuned systems are good solutions; 2003 Chee et al . [11] perform an adaptive PID control based on an adjustable table; In 2004 Hovorka [12] carried out a study of adaptive techniques with few measures and with continuous measurement, this study can be considered as the basis of new glycemic control strategies in closed loop.

Also some works have been presented concerning the inversion of a blood glucose model as Neatposarnvanit and Boston that in 2002 [13] used this technique to define estimates of insulinemia. Yes well no they introduced the concept of closed loop control for investment of the glycemic model and were not used as a system of automatic regulation

The present invention unites the novelty of control glycemic by inversion of a glycemic model with adaptation of its parameters by a reference model which in turn is a closed loop glycemic control by ideal inversion that does not consider delays in insulin absorption or measurement, in order to perform an automatic calculation of the doses of insulin infusion that would be applied directly to a person with diabetes

The authors of the present invention have presented at a national congress a preliminary proposal of glycemic control by invariant inversion of the Berger model Rodbar Guyton and Lehmann [1] [2]. In said proposal [14] the control performed is invariant, does not include adaptation, and introduces a basic topology that allows to build an adaptive controller in patients simulated exclusively with the Berger Rodbar model Guyton and Lehmann. The investment controller described is very simple and irrelevant in real patients because it lacks a model of the glucose sensor, takes into account exclusively the latest administration of insulin and use static estimators conditioned by the aforementioned model [1] [2].

The present invention has as a novelty the definition of a new topology that allows the method of investment can be applied to real patients as it takes into account the continuous glucose sensor and the infusion device insulin. Additionally, to estimate patient insulinemia considers the previous insulin administrations relevant in the acting moment The estimators have been modified and by both the estimation methods presented in this document They are also novelties of this invention. Another novelty regarding published work [14] is the concrete method of adaptation, according to the conceptual topology, which adapts the parameters of the method Inverse continuous infusion in an automatic way that we can denominate as adaptive reverse controller. In the publication [14] the possibility of the existence of the system is mentioned adaptive presenting a cost function. In this patent an alternative cost function is claimed with which they have obtained better results.

In the field of industrial property although there are adaptive control patents, none were found that applies adaptive control to glycemic control by Model inversion. Only one reference was found that protects a glycemic control device with a very method basic [5]. There are referenced other patents that have calculated from other non-automatic forms continuous infusions by pump insulin.

1.1. Detailed description of the invention

The basis of the adaptive control system of glycemic control presented in the invention is a control by inversion of the patient's glucorregulatory system model, also called the blood glucose model, which gets the dose, intravenous and subcutaneous, adequate insulin to induce normoglycemia The reverse adaptive method calculates insulinemia optimal (11) so that blood glucose variations ideally be null with respect to a reference blood glucose (19) independently of the metabolic state and the actions of the patient such as food intakes and physical exercise, considered as disturbing actions of the metabolic control system. Until the date and as far as the authors have been able to inquire, it has not been applied the concept of control by investment of a model of glycemia in glycemic control in closed loop, that is, Use the glucose measurement as a basis for calculating the following insulin infusion dose. The mechanism of adaptation of control parameters of the regulator that calculates insulinemia Optimal is done by an adaptation system based on the rule from MIT (14) [16].

         \ vskip1.000000 \ baselineskip
      

The steps of the adaptive method closed loop insulin dosing by inversion of a blood glucose model that runs the control system are the following for each moment of action:

Step 1:

Blood glucose measurement (6) of the patient, by route intravenous or subcutaneous, said level being dependent on the food intakes (4) of the patient and record of a history finite of previous insulin administrations (5) at patient.

Step 2:

From the previous doses of insulin (5), estimate the patient's insulinemia (7) by using a model of insulin absorption dynamics.

Step 3:

From the blood glucose measurement (6), of the doses previous insulin (5) and insulin estimation (7) are estimate the disturbances generated by the absorption of intakes of food (9), using a defined intake estimator (8) by equation 6.

Step 4:

From the blood glucose measurement (6), of the estimate of insulinemia (7) and estimate of absorption of intakes (9), is calculated, using a reverse regulator (10) defined by equation 5, optimal insulinemia (11). Bliss Optimal insulinemia would eliminate the variability of blood glucose patient due to the absorption of food intakes.

Step 5:

From the results of step 4 and, using an insulin dose calculator (12), defined by the equation 8, the dose, or bolus of insulin (5), necessary for induce normoglycemia in the patient, which will be different depending on either the administration of insulin subcutaneously or via intravenous

Step 6:

Using an adaptation system based on the MIT rule (14) to adapt the method to the patient, defined according to equation 13, and through a glycemic control loop reference ideal, which does not consider the measurement delays of blood glucose or absorption, the reference insulinemia is calculated (18) and  reference blood glucose (19).

Step 7:

The reference insulinemia (18) is compared with the insulinemia (7) of the patient, and the reference glycemia (19) with the blood glucose measurement (6) to obtain the errors of insulin deviation (22) and blood glucose (21) . With the errors, a cost function is defined, defined by equation 12, to obtain the temporary variations of the control parameters of the inverse regulator (10) and of the insulin dose calculator (12), chosen as degrees of freedom, which are the gain K IV (15) and K u (16) respectively, defined according to equation 14, which in turn are used to adjust the dose of insulin to be administered to a patient concrete, that is to say to adjust the control loop to the patient.

Step 8:

Validation of the dose, or bolus of insulin, which is you will administer to the patient using safety standards based on limiting insulin boluses between a minimum and a maximum, and cancel the insulin bolus when episodes occur hypoglycemic, according to equation 11. Once the insulin the patient would start the method again with the step one.

         \ vskip1.000000 \ baselineskip
      

The idea of the inverse regulator (10) arises when it is desired to obtain adequate insulin so that, under conditions of normoglycemia, consumption, external glucose contributions and Glucose transformations are compensated at all times. The difficulty of reversing a glycemic model lies in the no linearity of the model to be reversed and especially of the safety rules. The patient's blood glucose model is given by the following equation:

one

Where G is glycemia (6), G in the absorption of intakes, G NHGB liver balance, G out peripheral consumption, G kid renal excretion or glycosuria, V The volume of insulin dilution and t is the instant evaluation time.

         \ vskip1.000000 \ baselineskip
      

Under normoglycemia conditions there is no renal excretion, so to get the variations to be void, it must be met that:

100

The distribution ratio of the intake of food, between the metabolic capacity of the liver and consumption insulin dependent peripheral, we can consider it 50% in people with diabetes [17]. Independent peripheral consumption of insulin is provided by the liver, since it always has to exist regardless of intake. With these Considerations are that glycemic variations are:

2

Where G out ( I eq), is the peripheral consumption dependent on insulinemia; G out ( GI ) is the peripheral consumption independent of insulin and the rest of the parameters are detailed in [14].

         \ vskip1.000000 \ baselineskip
      

Clearing from the previous equation the insulinemia:

3

Where I IVp, I IVh are the partial insulinemias necessary to absorb the x proportions of intakes given by peripheral consumption and liver balance respectively.

         \ newpage
      

Clearing insulinemia from equation 4 partial and summed in a weighted way with an offset:

4

where: I IVo, is the optimal insulinemia to absorb the intake (4) adjusted to the patient (1), through the following degrees of freedom: K IV, is a performance gain and I _ IVb} that would adjust the minimum level of insulinemia. The degrees of freedom are chosen in such a way that the glycemic control is adjusted to each patient.

         \ vskip1.000000 \ baselineskip
      

Equation 5 defines the inverse regulator (10) which is used to obtain optimal insulinemia (11), which would eliminate the variability of the patient's blood glucose due to absorption of food intakes.

The most important disturbances on the Glycemic level are intakes and physical exercise. With the intakes glucose levels are increased and with exercise increase or decrease depending on insulin levels. At case of disturbances generated by food intakes, to solving the intake absorption flow of Equation 2 is necessary to "measure" intakes (4), which in a practical way and Efficient is such a difficult task as controlling diabetes, therefore a method is used that estimates the absorption of them. The intake estimator (8) performs a approximate calculation of food intake absorption (9) of actual intakes (4) without the need to resort to a measure external to the patient's method or intervention to Provide detailed information on intakes.

         \ vskip1.000000 \ baselineskip
      

The novelty of the invention is the realization of an intake absorption estimator based on a model of glycemia and conditioned by degrees of freedom that can fit a specific patient. This absorption estimator of intakes (8) obtains the estimate of the absorption of intakes (9) by the following equations:

5

Where \ hat {G} in is the estimate of absorption; hat IV is an estimate of the patient's insulinemia; \ hat {G} an estimate of glycemic activity and A and B degrees of freedom for adjusting the estimator to each patient.

         \ vskip1.000000 \ baselineskip
      

The direct result of the inverse regulator (10) is the desired plasma insulin concentration for a control optimal, it is directly applicable intravenously and therefore it is not directly applicable subcutaneously. Therefore, it has to raise an insulin bolus calculator (12) that converts the Optimal insulinemia data (11), under safety standards for safeguard the patient's physical integrity, in an infusion Effective insulin administration subcutaneously. The calculator insulin dose (12) per inversion calculates Approximate insulin infusions subcutaneously using only the information of optimal insulinemia (11), using the following equations:

6

The discretizations of the differential equations have been made by rectangular techniques; I IVo, is the optimal insulinemia to cancel glycemic variability; \ hat {I}, is the estimation of infusions; f (k) is the absorption of 1U of rapid insulin and b is a kinetic parameter of insulin; See [14] for more details.

From the above equations the dose is obtained or bolus of insulin to administer to the patient:

7

Where: qi are important weights, extracted from equations 7.

         \ vskip1.000000 \ baselineskip
      

Glycemic control provides a degree of freedom for adaptation to each patient, but also must be taken into history of absorption due to previous doses of insulin (5) and the form of administration, since the result is different if the infusion is subcutaneous or intravenous:

         \ vskip1.000000 \ baselineskip
      

- Subcutaneous route:

8

         \ vskip1.000000 \ baselineskip
      

- Intravenously:

9

Where: K u (16), is a patient adjustment gain or bolus sensitivity, and is a degree of freedom obtained for bolus control, Where: qi are important weights, extracted from Equations 7, I IVo is the optimal insulinemia proposed by the inverse regulator (10) and I IV is an estimate of the patient's insulinemia.

         \ vskip1.000000 \ baselineskip
      

For validation of the subcutaneous insulin bolus safety standards are used (12) based on limiting insulin boluses between a minimum and a maximum and in condition the insulin bolus when hypoglycemic episodes occur, thereby  It is a limitation of bolus insulin, according to the equations:

10

Where u b is a basal insulin level and u m a maximum level.

         \ vskip1.000000 \ baselineskip
      

Another novelty of the present invention is a specific method of adaptation, which we can call as adaptive reverse controller. The proposed method obtains the temporary variations of the control parameters of the inverse regulator (10) and of the insulin bolus calculator (12), chosen as degrees of freedom, which are the gain K IV (15) and K _ {u} (16) respectively, which in turn are used to adjust the dose of insulin to be administered to a specific patient, that is to say to adjust the control loop to the patient. The adaptation is done through a recursive equation obtained by an adaptation system based on the MIT Rule (14) [16].

The present invention provides the following developments to the adaptation method based on the MIT Rule:

-

Use an ideal glycemic control of closed loop reference that does not consider measurement delays of blood glucose or insulin absorption.

-

Use a compound cost function by two weighted addends, one that penalizes deviations from blood glucose, that is, the differences between the patient's blood glucose regarding the reference, and another that penalizes deviations in insulinemia, that is, of the patient's insulinemia Regarding the reference.

eleven

Where e G is the error of criminalization of the blood glucose deviation (21) measured with respect to the reference model, e i is the error of criminalization of the absorption delays for insulinemia deviations (22) of the infusions proposed by the dose calculator, and x G and x I weights.

The MIT rule states that you have to vary the parameters in time in the opposite direction to the gradient of The cost function:

12

After performing the partial derivatives of the variables of interest regarding the parameters and regrouping terms:

13

Where P I, F I and F G are functions resulting from the derivation.

         \ vskip1.000000 \ baselineskip
      

1.2. Brief description of the drawings

Figure 1: It is the topology of the adaptive control system in closed loop of continuous insulin infusion by inversion of a glycemic model, based on an adaptation system based on the MIT rule (14) by reference model. The adaptation system interacts with the patient control loop (13) to the patient through the degrees of freedom chosen for glycemic control, which are K IV (15) and K u (16), control parameters of the inverse regulator and insulin dose calculator respectively. In the patient's control loop (13) is the patient (1) with infusion system (23) and a glucose meter (3) with which the blood glucose value (6) and the systems in charge are obtained to determine the optimal dose: the inverse regulator (10) with which the optimal insulinemia (11) is obtained, the food intake absorption estimator (8) that proposes the absorption of food intakes (9) from the actual intakes (4), the patient insulimemia estimator (2) with which the patient's insulinemia value (7) is estimated due to previous insulin infusions (5) and the insulin dose calculator (12).

Figure 2: It is the ideal closed loop control topology in closed loop, which does not consider insulin absorption delays or glucose measurement delays. Said topology sets a reference glycemia model (19) obtained with a glycemia model (17) and reference insulinemia model (18) obtained through an ideal inverse regulator (20). With the data of the reference glycemia (19) and glycemia (6) of the patient and reference insulinemia (18) and insulinemia (7) of the patient, the deviations of glycemia (21) and insulinemia (22) are obtained with which automatically adjusts parameters K IV (15) and K u (16), control parameters of the inverse regulator and insulin dose calculator respectively.

         \ vskip1.000000 \ baselineskip
      

1.3. Detailed exposition of an embodiment of the invention

This section describes a possible detailed practical realization for a person with diabetes You can treat your disease with the proposed invention.

To begin to execute the method that is executed in the control system proposed in the present invention it is necessary to measure the blood glucose (6) of the patient, for example using a continuous glucose sensor. With the measurement of glycemia (6) and the previous doses of insulin (5) that have been administered in previous moments, the insulinemia (7) of the patient with an estimator of insulinemia (2) is estimated. The estimation of the patient's insulinemia is used: to estimate with a model of absorption of intakes (8), the absorption of food intakes (9) caused by the intakes (4) and calculate the bolus of insulin with the dose calculator insulin (12). With the data to estimate the absorption of intakes (9), insulinemia (7), and glycemia (6), the temporal variations of the parameters that adapt the control loop of the specific patient (13) are obtained. The adaptation is carried out by means of a reference model, which is an ideal control system that does not contemplate the delays of absorption of the boluses of insulin and nor in the measure of blood glucose of the patient. Comparing the ideal behavior given by the reference model and the real behavior of the patient, deviations of glycemia (21) and insulinemia (22) are obtained with which the parameters of the inverse regulator, K IV (15) are adjusted ), and of the dose calculator, K u (16), of the patient control loop (13). Finally, the effective value of the insulin dose is obtained with the insulin dose calculator (12).

1.4. Industrial application

This patent has its principal application as an autonomous system of subcutaneous insulin infusion for the treatment of diabetes, but it can also be used in intravenous administration of insulin by directly applying the regulator output without dose estimator or can even be applied in any other medical administration system controlled and autonomous medication in the medical field as per example intensive care units, or even in the field vet. The application in the industrial field is that of process controller with continuous variables such as the temperature, pressure, kinematic variables, forces, energy, power or ph to name a few.

This invention presents an interesting application in the field of diabetes treatment, in such a way which prevents patient intervention thus eliminating the component subjective that insulin therapies present. The method presented decides for the patient how much insulin has to be administer at any time, regardless of whether you perform a intake or is fasting In addition, the method is not only adaptable to changes and variations that the body may suffer human over time, but adapts to any patient. Therefore the universal application of the invention for the treatment of diabetes, which could be installed on any hardware system and applied in a way rapid to a patient, reducing the time that the endocrine you need to treat a patient's diabetes, thus reducing the health cost.

Bibliography

[1] ED Lehmann and T. Deutsch . A Physiological Model of Glucose Insulin Interaction. An. Intern. Conf. IEEE EMBS : 13: 5: 2274-2275. Apr. 1991.

[2] M. Berger and D. Rodbard . Computer Simulation of Plasma Insulin and Glucosa Dynamics alter Subcutaneous Insulin Injection. DiabetesCare : 12: 10. 1989 .

[3] DC Klonov. The artificial pancreas: how sweet engineering will solve bitter Problems. J. of Diabetes Science and Tecn : 1: 1: 72-81. 2007

[4] Fisher U., Schenk W., Salzsieder E. Albrecht G., Abel P. and Freyse EJ Does Physiological Blood Glucose Control Require an Adaptive Strategy ?. IEEE Trans. Biomed Eng. Vol. 34, pp. 575-582, 1987 .

[5] F. HOFFMANN-LA ROCHE AG. Device for the dosage of a glycemic regulatory hormone of a patient . Patent ES 2 264 447 T3 publication 1255578, 2001 .

[6] R. Hovorka . Continuous glucose monitoring and close-loop systems. Diabetes UK : 23: 1-12. 2005.

[7] EA Wooddruff , S. Gulaya and RB. Northrop . The Closed Loop Regulation of Blood Glucose in Diabetics. IEEE CH 2666-6 / 88 / 0000-0054- \ textdollar1.00, 54-56, 1988 .

[8] C. Juhász , ER Carson and Z. Benyó . Model Reference Adaptative Advisory System Based on Sparse Measurements for Insulin Adjustment in Diabetes Mellitus. IEEE, pp. 2261-2262, Feb. 1992.

[9] B. Candas and J. Radziuk . An Adaptive Plasma Glucose Controler Based on a Nonlienar Insulin / Glucose Model. IEEE Trans. On Biomed Engin , vol 41, 2, 1994 .

[10] R. Bellazzi , G. Nucci and C. Cobelli . The Subcutaneous Route to Insulin-Dependent Diabetes Therapy. IEEE engin. In medic and biol, 2001.

[11] F. Chee , TL Fernando , AV Savkin and VV Heeden . Expert PID Control System for Blood Glucose Control in Critically T1 Patients. IEEE Trans. On Biom. Eng. , Vol 7, 4, 2003 .

[12] R Hovorka , LJ Chassin , ME Wilinska , V Canonico , JA Akwi , MO Federici , MM Benedetti , I Hutzli , C Zaugg , H Kaufmann , M Both , T Vering , HC Schaller , L. Schaupp , M Bonenlez and TR Pieber Closing the Loop: The Adicol Experience. Diabetes Tech and Therap, vol. 6, 3: 307-318, 2004 .

[13] C. Neatpisamvanit and JR Boston . Estimation of Plasma Insulin from Plasma Glucose. IEEE Trans. On Biomedical Engineering. 2002 ; 49: 1253-1259.

[14] A. Rodríguez , E. Hernando , G. García , C. Pérez and EJ Gómez . Investment Algorithms for Control in Closed Loop of Type I Diabetes. XXIII Annual Congress of the Spanish Society of Biomedical Engineering (CASEIB 05). Madrid, pp. 99-102, 84-7402-325-4, November 2005.

[15] GM Steil , K. Rebrin , C. Darwin , F. Hariri and MF Saad . Feasibility of Automating Insulin Delyvery for the Treatment of Type I Diabetes. Diabetes 55: 3344-3350, 2006 .

[16] Landau, ID Reference Model A survey of Adaptive Techniques Theory and Applications. Automatic: 10: 353-379. Pergamon Press 1974

[17] R. Eaton , W. Spencer , D. Schade , B. Schafer and W. Corbett . Diabetics Glucose Control: Matchinng Plasma insulin Concentration to Dietary and Stress Hyperglycemia. Diabetes Care 1: 40-44, 1978 .

[18] Goodwin, SF Record, ME Salgado. System Desing Control. GC Prentice Hall , New Jersey 2001 .

Claims (7)

1. Closed loop adaptive control system of continuous insulin infusion by inversion of a model of blood glucose set to determine the optimal insulin dose to administer to the diabetic patient, subcutaneously or intravenously, in order to induce normoglycemia, which understands:

-

glucose meter (3) configured to obtain the patient's blood glucose value (6), by route intravenous or subcutaneous and means of recording a historical finite of previous insulin administrations (5) at patient,

-

insulinemia estimator (2) configured to estimate, from the previous doses of insulin (5), the insulinemia value (7) of the patient by using a model of insulin absorption dynamics,

-

intake intake estimator of food (8) configured to estimate the intake intake intake of food (9), from the blood glucose measurement (6), of the doses previous insulin (5) and insulin estimation (7),

-

a reverse regulator (10) configured to calculate insulinemia optimal (11), capable of eliminating blood glucose variability of patient, from the blood glucose measurement (6), of the estimate of insulinemia (7) and of the estimation of the absorption of intakes (9),

-

a insulin dose calculator (12) configured to calculate the insulin dose or bolus to induce normoglycemia, which will vary in function of whether the route of administration is intravenous or subcutaneous,

-

a adaptation system based on the MIT rule (14) configured to calculate the reference insulinemia (18) and glycemia of reference (19) for adapting the method to the patient concrete,

-

means for obtaining the temporary variations of the control parameters of the inverse regulator (10) and of the insulin dose calculator (12), chosen as degrees of freedom, which are the gain K IV (15) and K _ {u} (16) respectively, which in turn are used to adjust the dose of insulin to be administered to a particular patient, that is to say to adjust the control loop to the patient,

-

media of dose validation, or insulin bolus, to administer to patient based on a limitation of insulin boluses between a minimum and a maximum based on safety criteria and cancel the administration of the bolus of insulin in case of a hypoglycemic episode, and

-

infusion system (23) configured to administer insulin to the patient (1).

           \ vskip1.000000 \ baselineskip
        

2. Adaptive control system according to claim 1 characterized in that the food intake absorption estimator (8) is configured to estimate the absorption of food intake (9) by the following equation:

           \ vskip1.000000 \ baselineskip
        

14

           \ vskip1.000000 \ baselineskip
        

Where \ hat {G} in is the estimate of absorption; hat IV is an estimate of the patient's insulinemia; \ hat {G} an estimate of glycemic activity and A and B degrees of freedom for estimator adjustment.

           \ vskip1.000000 \ baselineskip
        

3. Adaptive control system according to claims 1 and 2 characterized inverse regulator (10) is configured to perform the calculation of optimal insulinemia (11) for the absorption of intakes (4) by the following equation:

           \ vskip1.000000 \ baselineskip
        

fifteen

           \ vskip1.000000 \ baselineskip
        

where: I IVo, is the optimal insulinemia to absorb the intake (4) adjusted to the patient (1), through the following degrees of freedom: K IV, is a performance gain and I _ IVb} that would adjust the minimum level of insulinemia.

           \ vskip1.000000 \ baselineskip
        

4. Adaptive control system according to claims 1 to 3 characterized in that the insulin dose calculator (12) is configured to: - in the case that the patient is administered intravenously, perform insulin dose calculation by the following equation:

           \ vskip1.000000 \ baselineskip
        

16

           \ vskip1.000000 \ baselineskip
        

Where: u IV are intravenous insulin doses, K u (16), is a gain of adjustment to the patient or bolus sensitivity and is a degree of freedom obtained for bolus control, I IVo} is the optimal insulinemia proposed by the inverse regulator (10) and I IV is an estimate of the patient's insulinemia, - in the case that the patient is administered subcutaneously, perform insulin dose calculation by the following equation:

           \ vskip1.000000 \ baselineskip
        

17

           \ vskip1.000000 \ baselineskip
        

Where: u SC are subcutaneous insulin doses; K {u} (16) is a gain adjustment to the patient or bolus sensitivity and a degree of freedom is obtained for controlling the bolus; qi are important weights, extracted from equations 7; I IVo is the optimal insulinemia proposed by the inverse regulator (10) and I IV is an estimate of the patient's insulinemia.

           \ vskip1.000000 \ baselineskip
        

5. Adaptive control system according to claims 1 to 4 characterized in that the reference insulinemia (18) and the reference glycemia (19) are calculated based on a closed loop glycemic control by ideal inversion, without delays in the absorption of insulin bolus or glucose measurement, that is, a direct intravenous control.

           \ vskip1.000000 \ baselineskip
        

6. Adaptive control system according to claims 1 to 5, characterized in that the following cost function is used to adjust the degrees of freedom of the patient's control loop (13) through the MIT rule:

           \ vskip1.000000 \ baselineskip
        

18

           \ vskip1.000000 \ baselineskip
        

Where e G is the error of penalization of blood glucose deviations (21) measured with respect to the reference model, e I is the error of penalty for insulinemia deviations (22) of the absorption delays of the infusions proposed by the dose calculator, and x G and x I weights and the parameters K IV (15) and K u (16) are adjusted according to the equation:

           \ vskip1.000000 \ baselineskip
        

19

           \ vskip1.000000 \ baselineskip
        

Where P I, F I and F G are functions resulting from the derivation; T S is the method execution period.

           \ vskip1.000000 \ baselineskip
        

7. System according to adaptive control claims 1 to 6 characterized in that for the validation of insulin bolus proposed by the insulin dose calculator (12), restrictions insulin infusion based limit of insulin boluses are obtained from one minimum and maximum, and condition the bolus of insulin when hypoglycemic episodes occur, according to the equations: twenty Where u IV are intravenous insulin doses; u SC are subcutaneous insulin doses; u b is a basal level of infusion; u m a maximum level and G the blood glucose measurement (6).