Method for controlling blood glucose and artificial pancreatic system for performing same

Abstract

The present invention relates to a method of controlling glucose in a flexible structure dual hormone artificial pancreas, the method managing optional meal and / or exercise alerts by means of coordinated control behavior, the method comprising: measuring a plasma glucose signal; calculating an incremental plasma glucose measure (y); defining an incremental plasma glucose model; defining carbohydrate uptake as a carbohydrate content estimated depending on the patient; defining an expected postprandial incremental plasma glucose according to the administered insulin bolus; defining a corrected incremental plasma glucose, a corrected insulin infusion, and a corrected carbohydrate uptake; defining a virtual control behavior, which is allocated between a regulatory behavior and an anti-regulatory behavior; and calculating the control behavior with a 2-DOF feedback controller having a pre-filter having a nominal value s).

Description

Technical Field

[0001] This invention belongs to the field of glucose control technology. Specifically, this invention relates to more effectively controlling glucose levels through regulatory control by means of insulin administration and counter-regulatory control by means of glucagon and / or rescue carbohydrate administration.

[0002] One object of the present invention is to provide a glucose control method that can manage optional meal and exercise notifications by means of a flexible dual-hormone control system.

[0003] A second objective of this invention is to provide a flexible dual-hormone artificial pancreas for the coordinated administration of insulin, glucagon, and / or rescue carbohydrates. Background Technology

[0004] Type 1 diabetes (DMT1) is an autoimmune disorder that damages pancreatic beta cells, leading to the inability to secrete insulin. This hormone plays a crucial role in glucose homeostasis because it is responsible for lowering plasma glucose concentrations.

[0005] Therefore, people with DMT1 often have chronically high glucose levels (hyperglycemia), leading to serious long-term health problems such as cardiovascular disease, kidney disease, retinopathy, and neuropathy. This is why exogenous insulin is required.

[0006] Artificial pancreas (AP) systems have emerged as a technical treatment for DMT1, offering improved glycemic control compared to conventional therapies. However, daytime control remains challenging due to significant fluctuations in glucose levels caused by meal intake and exercise. Existing hybrid AP systems are based on mealtime boluses of carbohydrates suitable for compensating for glucose spikes caused by meals. In open-loop therapy, patients have been responsible for estimating carbohydrate intake and notifying the system (meal notification).

[0007] Therefore, some patients may not realize the usefulness of AP and thus abandon its continued use. Thus, the AP system must be able to perform acceptablely without the obligation to notify patients of meals. However, a fully automated system without meal notifications may not be suitable for all patients. Some patients with extensive experience in carbohydrate estimation may prefer to assume (at least in some cases) the responsibility for meal notifications to reinforce post-meal control.

[0008] However, meal notifications, as a feedforward behavior, can interact with feedback behaviors in an AP designed to operate fully automatically without such notifications, leading to excessive insulin administration and consequently hypoglycemia (abnormally low glucose levels). Therefore, AP systems must include mechanisms that allow users to provide meal notifications without increasing the risk of hypoglycemia.

[0009] Exercise is another major challenge to the performance of AP systems. While very high-intensity exercise events can lead to hyperglycemia (abnormally high glucose levels), light to moderate aerobic exercise, most commonly performed by non-athletes, lowers glucose levels. Existing hybrid systems require users to plan their exercise several hours beforehand to reduce insulin infusions.

[0010] To eliminate exercise-induced hypoglycemia, the AP system must have a counter-regulatory control behavior that raises glucose levels. Carbohydrate supplementation is the most practical option for treating mild exercise-induced hypoglycemia. However, individuals concerned about weight gain may prefer subcutaneous glucagon administration as a non-calorie treatment for hypoglycemia. In clinical trials, dual-hormone AP systems with automatic insulin and glucagon administration have been shown to effectively shorten the duration of exercise-induced hypoglycemia.

[0011] Individuals who are less active or have a low propensity for hypoglycemia may find the additional cost of a dual-hormone AP system for timely glucagon administration unreasonable. For these patients, self-administering a low-dose glucagon pen may be a safer and more appropriate option for relieving non-severe hypoglycemia.

[0012] Based on the above analysis, the following three features are recommended for use in AP systems: 1) Effectively compensate for meals without any prior notice; 2) Allow patients to provide meal notifications (if they wish to do so) without compromising the performance of the feedback controller; and 3) Provide counter-regulatory control behaviors to manage hypoglycemic remission, such as in unannounced exercise events.

[0013] Many methods have been proposed to eliminate meal notifications, but none of them have a mechanism to reduce the interaction between feedback behavior and the event in which the patient provides meal notifications.

[0014] As for the feedforward of unplanned exercise results, most glucagon-insulin systems lack a carbohydrate recommendation module, and vice versa.

[0015] A few strategies that can work with glucagon and carbohydrates to alleviate hypoglycemia require meal notification.

[0016] However, as mentioned above, no coordinated dual-hormone AP system has been found in the prior art that allows patients to provide meal notifications (if they wish to do so) without meal notifications and allows patients to decide on counter-regulatory behaviors between glucagon and carbohydrates. Summary of the Invention

[0017] The method of the present invention is a blood glucose control method for a flexible dual-hormone artificial pancreas. The method utilizes coordinated control behavior to handle optional meal notifications and optional exercise notifications.

[0018] The system coordinates regulatory control behavior (insulin) and counter-regulatory control behavior (glucagon and / or carbohydrates).

[0019] As described above, the method of the present invention manages unannounced meals, although users can provide meal notifications (if desired) through a non-interactive feedback-feedforward scheme that reduces the interaction between feedback control and feedforward control.

[0020] In addition to insulin, the system provides counter-regulatory control behaviors to address induced hypoglycemia, such as due to unannounced exercise events. Users can choose whether they wish to apply this counter-regulatory control, such as by automatically infusing glucagon via a pump, advising the patient to administer glucagon by pen, or advising on carbohydrate intake.

[0021] Therefore, the method of the present invention can manage mealtimes (unannounced or announced, depending on user preference) and unannounced movement events, providing parameters within the range recommended by clinical guidelines in computer simulation studies. The most significant change in performance corresponds to whether the patient decides to be notified of a mealtime. As expected, when a mealtime notification is provided, the controller improves the percentage of hyperglycemic time during operation without a mealtime notification, without noticeable deterioration in hypoglycemic-related parameters.

[0022] Furthermore, it must be considered that in computer simulation studies, there are no relevant differences in performance parameters between using glucagon or carbohydrates as counter-regulatory behaviors, because by design they theoretically provide the same control, with glucagon administration having the advantage of non-calorie supply and being able to be automated by means of a pump.

[0023] This invention allows for near-optimal postprandial glucose control with or without notification of meals or exercise. Optimal is defined as a control method that minimizes postprandial blood glucose spikes while simultaneously preventing glucose levels from falling below a predetermined threshold, i.e., a predefined hypoglycemic limit (e.g., 70 mg / dL), under nominal conditions.

[0024] The feedback controller provides feedback control behavior that achieves near-optimal postprandial glucose control when no meal notification is given. Therefore, optional meal notification is managed using a non-interacting feedback-feedforward scheme that suppresses insulin infusion after a pre-meal bolus following optional meal notification, preventing overdose and maintaining performance.

[0025] The proposed solution defines a novel control strategy designed to operate with and without food or exercise notifications, achieving near-optimal postprandial glucose control under nominal conditions. The strategy includes a preprandial bolus in insulin administration control, designed to achieve near-optimal postprandial glucose in the absence of meal notifications, to achieve optimized coordination between feedback control and feedforward behavior. Simultaneously, it allows for the calculation of whether counter-regulatory control behavior is needed, such as due to exercise events.

[0026] It should be noted that the optimal open-loop strategy consists of administering an insulin bolus before meals, i.e., pulse-feedforward control behavior, which feedback cannot produce because feedback control is not possible until the output is affected by a disturbance. This leads to the conclusion that the best performance achievable through feedback is obtained by administering an optimal bolus size when the output is affected by a disturbance.

[0027] Patient-reported food intake and administration of related bolus injections of CGM ( The expected postprandial variability is calculated on a meal and insulin model. This expected curve is used to calculate the correction term for the control behavior provided by the feedback controller, preventing the interaction between feedback and feedforward behavior that could lead to overdose of insulin. Therefore, the notified postprandial corrective feedback control behavior only adjusts for deviations from the expected trajectory in the postprandial period due to meal bolus injection.

[0028] Therefore, the method of the present invention is designed for near-optimal glycemic control with or without meal and / or exercise notifications, and is configured with a near-optimal feedback controller and a coordination strategy between feedback-feedforward behavior, the feedback controller being designed to operate without meal notifications, and the coordination strategy between feedback-feedforward behavior being used to adapt to meal boluses from optional notifications. Simultaneously, the method of the present invention allows for the calculation of counter-regulatory control behavior when necessary.

[0029] The method includes the step of measuring a plasma glucose signal (G(t)) using a continuous glucose monitor (CGM), represented by its Laplace transform (G(s)). In the following, any signal can be represented in either the time domain (t) or the Laplace domain (s), which are correlated via the Laplace transform. An incremental plasma glucose measurement (y) is introduced, which is calculated as:

[0030] in It is the application of baseline glucose infusion The resulting baseline glucose level.

[0031] Similarly, incremental insulin infusion is defined as:

[0032] in( () is total insulin infusion.

[0033] Furthermore, the definition of incremental plasma glucose (y) takes into account carbohydrate intake (d), insulin infusion (u), glucagon administration (v), and rescue carbohydrate administration (w), and is defined as follows:

[0034] in , and .

[0035] Carbohydrate administration (w) can be further processed using a quantifier to provide an amount that the patient can easily administer manually. Glucagon administration can also be quantified when the patient is advised to use a pen for simple manual administration. Alternatively, the quantified glucagon signal can be automatically administered using a pump, instead of administering a continuous glucagon signal.

[0036] In the context of this invention, It is a time-invariant linear transfer function that correlates insulin, glucagon, rescue carbohydrates, and carbohydrate intake from meals with glucose, and can be defined using various models known in the art, and is affected by... and A feasible constraint is required, namely, the order of the numerator must be less than or equal to the order of the denominator.

[0037] Preferably, the model With the help of the following structure definition:

[0038] in , , and It is a transfer function of the following form:

[0039] in , , and It is a parameter representing the delay. , , , , , , , , , , and These are the parameters obtained previously, and j is the index representing the patient.

[0040] Carbohydrate intake is defined as:

[0041] Where M This is the estimated carbohydrate content reported by the patient.

[0042] Similarly, the expected postprandial increase in plasma glucose (y ) is defined as:

[0043] in This refers to insulin boluses administered in conjunction with the patient's manual meal notification.

[0044] Preferably, insulin bolus injection The following calculation is performed using the feedforward controller C(s):

[0045] Similarly, the corrected incremental plasma glucose Corrected insulin infusion and corrected carbohydrate intake Defined as:

[0046] Using corrected insulin infusion ( Virtual control behavior It can be defined as:

[0047] Make

[0048] This newly acquired virtual control behavior was subsequently categorized into regulatory behavior ( ) and counter-regulatory behavior ( ):

[0049] Further categorize counter-regulatory behaviors into

[0050] in This indicates the counter-regulatory effect achieved through glucagon infusion, and This indicates that the counter-regulatory effect implemented by carbohydrate recommendations results in:

[0051] Regulatory and counter-regulatory behaviors are calculated as follows:

[0052] in It is a function (switching signal) that defines the change between regulatory and counter-regulatory behaviors, for example... or Filtering, such as a moving average filter with a specific sample window, and It is an adjustable factor, and among them It is the previously obtained controller gain, and Defined as:

[0053] in It is calculated as a 2-DOF feedback controller with a nominal pre-filter. like:

[0054] And among them It is a central linear controller designed for use in stable systems.

[0055] And it reduces interference (s), and in a preferred embodiment, can be calculated as control based on the disturbance observer.

[0056] Preferably, the central controller Defined as:

[0057] And the filter and These can be low-pass filters; more specifically, they can be defined as:

[0058] in , , and These are the parameters calculated previously, and It is relative to The nominal or target value of incremental glucose.

[0059] The method further includes the step of allocating counter-regulatory behavior between glucagon administration and rescue carbohydrate administration, in the form of:

[0060] in It is an adjustable parameter ranging from 0 to 1.

[0061] In some implementations, under certain conditions where one counter-regulatory behavior is more advantageous than another, This can vary over time. For example, when high insulin levels are involved (e.g., after a large insulin bolus), or when excessive glucagon has accumulated, it is preferable to administer rescue carbohydrates instead of glucagon to avoid side effects such as nausea, or according to user preference.

[0062] Similarly, the control behavior to be applied is calculated as

[0063] The method of the present invention may further include the step of correcting the regulatory and counter-regulatory behaviors to increase safety when the obtained control behavior is applied, such that:

[0064] in It is a function that defines the change between regulatory and counter-regulatory behaviors, as described above.

[0065] Furthermore, the method of the present invention may also include the step of: […]. Implement corrections to prevent glucagon administration from acting as a feedforward behavior.

[0066] In the case described, Divided into feedback function ( ) and feedforward behavior ( )like:

[0067] same, Calculated as:

[0068] As mentioned above, the controlled behavior of rescue carbohydrates can be discretized into a quantitative level of administering a specific dose. In this case, discretized control behavior ( It can be calculated as follows:

[0069] in Represents the nearest integer operator. It is the minimum threshold for activating the recommended cumulative carbohydrate intake. ,and:

[0070] in It is a specific time range The predicted glucose value within, and and These are the thresholds for predicting glucose and measuring glucose, respectively, for activating HypoFlag. The prediction can be calculated as follows:

[0071] in This is the discretized filtered glucose derivative, i.e., the following discretization:

[0072] in It is a previously determined time constant, and

[0073] Where BW is the user's weight (kg), It is the sampling period, and It's the clearance rate.

[0074] Preferably, the threshold Equal to 60 mg / dl, and threshold Equals 54 mg / dl. Furthermore, the time range... The time can range from 5 minutes to 300 minutes, with 60 minutes being the preferred value.

[0075] The same method can be applied to the administration of glucagon as a quantitative level of a specific dose. ) control behavior. In this case, specific control behavior ( The following calculations were performed:

[0076] in It is the minimum threshold for activating the recommended cumulative glucagon. ,and:

[0077] in This is the discretized filtered glucose derivative, i.e., the following discretization:

[0078] in It is a previously determined time constant, and:

[0079] Preferably, when a specific dose is used as a control behavior, the expected postprandial incremental plasma glucose It can be calculated by adding a term corresponding to the counter-regulatory behavior of the specific dose, such as:

[0080] in and It is an adjustable gain.

[0081] Preferably, and Designed so that the counter-regulatory behavior (v, w) equals the quantification level ( , To achieve the required maximum output () The increase is obtained as follows:

[0082] in Describes the inverse Laplace transform operator, and Due to the level of quantification ( , The required output change caused by ).

[0083] Preferably, the expected postprandial increase in plasma glucose can be converted to saturation. The result is:

[0084] in and These are the previously determined upper and lower threshold values.

[0085] Optionally, the above adjustable gain k is defined as .

[0086] As an alternative, the adjustable gain k is defined as:

[0087] in It is the inverse Laplace transform, and and These are adjustable parameters, among which This represents the expected size of the meal in the worst-case scenario, and This corresponds to the meal. The lower limit of tolerance for increased postprandial glucose.

[0088] Choose any location Defined as the size of a meal notified by the patient whenever an optional meal notification is given. ),and Defined as corresponding to this meal The incremental postprandial glucose tolerance limit is determined by keeping the calculated k value within an adjustable time window from the time of notification to eat.

[0089] In some implementation methods, the insulin bolus is administered ( ) is the scale of meals notified to the patient ( The group injection was calculated from... Define, where It is a feedforward controller, which can be defined as follows:

[0090] As an alternative, the feedforward controller The superbolus strategy can be followed, defined as:

[0091] in > 0 represents the time period during which the pump is shut off.

[0092] The present invention also relates to an artificial pancreas system incorporating the glucose control method as defined above, the system comprising: Pump, the pump being used to coordinate control behavior ( ) Delivers automated insulin infusion, and / or according to coordinated control behavior ( ) Delivers automated glucagon infusion; A continuous glucose monitor, wherein the continuous glucose monitor is used to detect plasma glucose signal (G(t)); and A first computing unit is configured to perform the steps of the method of the present invention.

[0093] As an alternative, the system may include a second pump for delivering an automated glucagon infusion.

[0094] In alternative implementations, the artificial pancreas system includes a display for informing the patient of controlled administration of recommended glucagon and / or rescue carbohydrates. In these cases, the system may also include a pen device for the patient to deliver a recommended dose of glucagon.

[0095] The present invention also relates to a computer program adapted to perform the steps of a defined method using a computing unit of an artificial pancreas system, and to a computer-readable storage medium containing the computer program. Attached Figure Description

[0096] To supplement the ongoing description and to aid in a better understanding of the features of the invention according to preferred practical embodiments, a set of figures has been added as part of this description, in which the following are depicted in an illustrative and non-limiting manner: Figure 1 A schematic diagram of the control strategy followed by the method of the present invention is shown.

[0097] Figure 2 A graph showing the percentage of time glucose is within a specific range.

[0098] Figure 3 The method according to the present invention is shown. Figure 2 The calculated control behavior based on the configuration. Detailed Implementation

[0099] To obtain the parameters, a patient-oriented control model based on an identification method is proposed, thereby generating personalized gains based on relevant clinical parameters readily available to the patient.

[0100] The interaction between glucose and insulin is a complex system that can be represented using high-dimensional nonlinear models. However, for control design purposes, simpler models with a small number of individual parameters are typically used.

[0101] The proposed AP was validated in a challenging scenario with meal and exercise events using a cohort of 10 adults in an academic version of the UVa / Padova simulator.

[0102] Figure 1 The proposed fully autonomous and coordinated artificial pancreas system is described. The core of the system comprises four basic components: asymmetric coordination control (A), a carbohydrate and glucagon suggestion module (B), a patient behavior adaptation module (C), and a central controller and pre-filter (D). Asymmetric coordination control allocates the output of the patient behavior adaptation module to insulin infusions that respectively lower and raise glucose levels. And counter-regulatory control behavior. As a counter-regulatory control behavior, the user can configure the controller to provide automatic glucagon infusion (…). ), and administer (e.g., with a pen) a quantified dose of glucagon ( Recommendations, or intake of carbohydrates ( The system provides suggestions, such as those derived from the carbohydrate and glucagon suggestion modules. The central controller is designed for operation without meal notifications, but the system can also manage notified meal times via a patient behavior adaptation module. This module also adapts the central controller's output when the user manually applies quantified counter-regulatory behaviors (such as carbohydrate intake or glucagon injection).

[0103] The system design assumes the following simplified relationship: incremental glucose Incremental insulin infusion rate (mg / dl) relative to baseline status (pmol / kg / min) relative to baseline infusion Between, glucagon infusion rate (mg / kg / min), and rescue carbohydrate application rate (mg / kg / min): (1) in Corresponding to oral carbohydrate intake (mg / kg / min), its scale Modeling with pulse signals, i.e. It has a time delay. , , and The linear transfer function is described as follows: (2)

[0104] superscript This represents the gain adapted to the patient. For the entire range of t, the control input is subject to saturation limiting. and , The dynamics in (2) assume three approximations to simplify model identification: 1) The two compartments represent the input ( , , or Absorption of ) 2) A compartment simulates the effect of this input on glucose, and 3) All models have the same delay of 15 minutes.

[0105] It should be noted that similar gain-based personalized models are routine for diabetes control purposes.

[0106] In the UVa / Padova release, the aforementioned models were identified for 10 virtual adults. Table I shows the control-oriented patient model parameters corresponding to the virtual adults identified in the UVa / Padova simulator.

[0107]

[0108] Table I.

[0109] The proposed control strategy should allow patients to provide meal notifications, leading to insulin bolus administration (if desired). However, doing so without notifying the feedback controller could result in insulin overdose. To avoid this, meal notifications are communicated to the central controller via a non-interactive feedback-feedforward protocol implemented in the patient behavior adaptation module. The variance of the expected output is defined as... (3) Based on insulin bolus The calculation is as follows:

[0110] in It indicates the carbohydrate content (g) of the meal, CIR is the carbohydrate-to-insulin ratio (g / U), and BW is body weight (kg). ∈ (0, 1) is the attenuation factor, and The term converts insulin units from U to pmol / kg. Subtracting (1) and (3) yields... (4) in , as well as The controller can now be designed based on model (4), which incorporates the patient-performed feedforward behavior. Note that ideally, where the patient provides notification about the exact size of the meal, resulting in the administration of an appropriate bolus of insulin, the feedback controller should be inactive as long as the output follows the expected path, thus avoiding unwanted interaction between feedback and feedforward behavior.

[0111] Having more manipulated inputs than the controlled output (such as in the case of MISO (Multiple Input Single Output) model (4)) can be used to improve performance under different operating conditions and manage input saturation, but it also brings difficulties to the design process. To simplify this task and take into account the nature of different control behaviors, model (4) is conveniently rewritten as follows: (5) in

[0112] Virtual control behavior, considered to include both regulatory and counter-regulatory behaviors, is defined as... (6) (7) (8) The latter two items represent the use of glucagon infusion ( or carbohydrate intake The counter-regulatory effects implemented are provided by the following

[0113] The counter-regulatory modality should be more aggressive to compensate for any drop in glucose levels that could have serious consequences for the patient. The outcome of the patient behavior adaptation module. The final control to be assigned. Defined as (9) in Defined below. The following coordination scheme is proposed. (10) (11) Among them, the controller gain increases in the inverse regulation mode. Note that this definition meets the following restrictions. For all t As applied by equation (6), unless saturation takes effect. Furthermore, to conform to (8) and thus ensure coordination of the two available counter-regulatory behaviors, the following allocation is chosen: , (12) in It can be adjusted in a time-varying manner to allow selection between glucagon or rescue carbohydrates.

[0114] The aforementioned coordination logic incorporates two mechanisms to minimize the overhead of counter-regulatory behavior. On one hand, the switching occurs on the filtered signal. This was performed on the signal. The signal was obtained by applying a moving average filter with a three-sample window to the signal. This is calculated to avoid unwanted activation of counter-regulatory modes due to noise inherent in the control behavior. Furthermore, the threshold for activating counter-regulatory modes is determined by an adjustable factor. Slight offset. This modification implies the existence of a small dead zone. No counter-regulatory behavior is performed within this small dead zone. The min and max functions are only used to ensure the feasibility of the obtained control behavior. Finally, through the definition above... , and The control signals that ensure closed-loop coordination can be calculated from (13) to (15), as follows: , , .

[0115] A positive insulin pulse, or insulin bolus, is the optimal insulin infusion that limits postprandial peak insulin levels while minimizing glucose deficiency at the same time. The following central controller can be derived from the equivalent model (5): (16) in

[0116] It has adjustable parameters , , , A strictly suitable filter. The control behavior (16) can be rewritten as: (17) This indicates that the output of the central controller is contributed by the following two items: one of which generates the self-tracking error. And another is due to the feedforward behavior provided. This leads to a correction term. Note that in some cases, such as during the postprandial period following a pre-meal bolus injection, the estimate may be overstated. This could lead to undesirable counter-regulatory control behavior. To avoid this, the two contributions are combined as follows: (18) Rescue carbohydrate recommendations must be provided as specific, quantified doses, not continuous rates. Therefore, continuous signals... It must be converted into a quantized pulse. Let... Given the sampling period, the recommended carbohydrate intake at sampling time k (expressed as...) The following formula is given. (19) in · Represents the most recent integer operator. It is a quantitative level and It is the minimum threshold for activating the recommended cumulative carbohydrate signal. The auxiliary variable is defined by the following formula (20) in It is a previously determined time constant that depends on the rate of carbohydrate accumulation, where the clearance rate Described as This is an alert indicating a risk of low glucose. It corresponds to the following conditions. (twenty one) in This is the actual reading from the continuous glucose monitor (CGM), and 1-hour expected glucose prediction is calculated using the following formula. (twenty two) in yes Discretization, i.e., using a time-constant low-pass filter. The glucose derivative of the filter.

[0117] The above quantification method will activate carbohydrate recommendations. Multiples of mg, as long as achieve Instead, the recommended carbohydrate intake will be based on cumulative signals. Subtract from the middle to compensate for the excess. In order to reduce the delay in the accumulation process, Set to below The consequence of quantization delay is that when continuous equivalents... When the value approaches 0, the controller may activate the suggestion too late. Establish the conditions in (19). To reduce the risk of these late arrival suggestions. Furthermore, Includes the forgetting factor, and when it is large enough to activate a suggestion but not meet the conditions. Set the time to zero.

[0118] Unlike rescue carbohydrate intake, glucagon infusion can be automated via a pump (a pump other than an insulin pump or a dual-chamber pump). The system can continuously administer glucagon and simultaneously provide quantified carbohydrate recommendations. However, some patients may prefer manual infusion of low-dose glucagon as a calorie-free alternative to recover from mild hypoglycemia without the need for an additional (or more complex) pump. Glucagon quantification strategies involve administering glucagon in a corresponding manner... , , , and replace , , , and To achieve an equivalent and equal approach to carbohydrate quantification schemes.

[0119] The quantified anti-regulatory behavior may differ from what the central controller calculates. To inform the central controller of this, the non-interacting feedback-feedforward scheme is modified by redefining the expected output change in (3) as follows: (twenty three) in and The increase in the maximum required output is defined when the anti-regulatory behavior is equivalent to the corresponding quantization level. Therefore, these parameters can be derived from the following...

[0120] in This represents the inverse Laplace transform operator, and Indicates by or The resulting desired output change. Furthermore, to prevent unexpected high or low glucose output, the signal... Saturation is as follows: (twenty four) It should be noted that the changes in the expected output in (24) are for practical considerations for implementation purposes; the central controller design is still based on the decomposition model (5) derived from (4).

[0121]

[0122] Table II.

[0123] Table II includes controller parameters. The gain k of the central controller is calculated according to the following formula for the object. j Personalization was performed: (25) in It is a safety factor set to 0.7. (Table II) , and The value was obtained through simulation. Since the inverse adjustment gain is the adjusted value... and factors Heuristically, these parameters were initially personalized using an optimization-based strategy. However, optimization-based adjustments were not feasible for clinical trials; therefore, population values ​​were also assessed by taking the median of the optimal values ​​for each parameter. No significant differences were found between using optimal adjustments and using population values ​​in the simulation. Therefore, for simplicity, population parameters were chosen in this study. Regarding the quantification-related parameters, carbohydrate quantification levels were set to the standard dose in commercial gels, i.e., 15 g. Glucagon was recommended to be quantified to 0.08 mg because this quantification level produced safe and effective control in recent clinical trials. The remaining quantification-related parameters are listed in Table II. , and and defining conditions The threshold is adjusted using simulation. As can be seen from Table II, and All are set to one-third of the corresponding quantification level.

[0124] As for the results obtained, the six configurations of the proposed controller—a hybrid system with glucagon, a hybrid system with carbohydrate recommendations, a no-meal notification system with glucagon, and a no-meal notification system with carbohydrate recommendations, wherein glucagon is considered as recommended or continuous infusion administration—were simulated against 10 virtual adults included in the extended version of the UVa / Padova simulator.

[0125] The scenario consisted of 14 days, with 3 meals per day, and randomized carbohydrate content and timing: 37.0 g [30.0, 45.0] g at 8:05 [7:45, 8:15] h (median [25th percentile, 75th percentile]), 53.0 g [36.0, 58.0] g at 13:05 [12:50, 13:15] h, and 75.5 g [53.0, 90.0] g at 19:50 [19:35, 20:10] h. In addition, 8 sessions of aerobic exercise were randomly scheduled every other day from day 1 at 19:17 [14:40, 22:10] h, lasting 55.0 minutes [50.0, 60.0] minutes and at an intensity of 48.5% [47.5, 53.0]% of maximum oxygen uptake. The effect of exercise on glucose was achieved by increasing insulin sensitivity. Finally, the controller configuration was tested against the following sources of variation implemented in the academic version of the simulator, including noise (built-in sensor model dexcom25): nominal values ​​of the meal absorption rate and carbohydrate bioavailability parameters of the meal absorption model varied per meal with uniform distributions of ±30% and ±10%, respectively; nominal values ​​of parameters describing insulin pharmacokinetics were modified per meal according to a uniform distribution of ±30%; diurnal variation of insulin sensitivity was represented by a sinusoidal variation over a 24-hour period, following a uniform ±30% random amplitude and random phase; and erroneous estimates of meal carbohydrate content were implemented. It should be noted that for controller configurations with recommendations for carbohydrates and glucagon, users follow the recommendations simultaneously (i.e., consuming carbohydrates or administering glucagon). Even more ideally, other studies in the literature have considered this assumption. The performance of the controller configuration was evaluated by the percentage of time within, below, or above the range, and by daily consumption of insulin, glucagon, and carbohydrates.

[0126] Figure 2 This graph shows the percentage of time that glucose levels were within a specific range. The boxes below and above represent the 25th and 75th percentiles, respectively, and the thick horizontal black line corresponds to the median. The crosses represent the mean.

[0127] The ends of the upper and lower whiskers represent the maximum and minimum values, respectively, not exceeding (at most) 1.5 times the interquartile range. Black dots extending beyond the whiskers correspond to outliers.

[0128] Notified and unnotified meals were considered. Furthermore, glucagon administration via pump (glucagon) and glucagon recommendations quantified via pen (glucagon recommendations) were evaluated. Carbohydrates were always quantified and patients were advised to consume them manually (carbohydrate recommendations).

[0129] like Figure 2 As shown, all controller configurations were able to manage unannounced meal and exercise events, with averages within their respective ranges. Hypoglycemic periods were rare in all configurations. Only subjects 3 and 6 had periods of CGM <54 mg / dL, although their time within this range was less than 0.70%. Six subjects had periods of CGM <70, but these were clinically acceptable. The patient with the longest hypoglycemic time (subject 3) had a CGM below 70 mg / dL in only 1.6% of cases. The reduction in hypoglycemic time was achieved without excessive application of counter-regulatory control behaviors.

[0130] Figure 3 The method according to the present invention is shown Figure 2 The configuration calculation control behavior.

[0131] In glucagon preparation, such as Figure 3 As shown, the required median daily dose is less than 1 mg, which is the threshold at which subjects frequently experience nausea. The required daily dose in the carbohydrate configuration is similar to that of other controllers evaluated in the literature in equivalent synthetic scenarios.

Claims

1. A method for controlling glucose in a flexible, dual-hormone artificial pancreas, the method being able to manage optional mealtime cues and optional exercise cues by means of coordinated control behaviors, the method comprising the steps of: - Plasma glucose signal (G(t)) is measured using a continuous glucose monitor (CGM); - Incremental plasma glucose measurement is calculated using the following formula: in This is the baseline glucose level; - Considering carbohydrate intake (d) and relative to baseline infusion The incremental insulin infusion (u), glucagon administration (v), and rescue carbohydrate administration (w) are used to define the model for incremental plasma glucose (y) as follows: in , and ; and among them It is a time-invariant linear transfer function that correlates carbohydrate intake (d), incremental insulin infusion (u), glucagon administration (v), and rescue carbohydrate administration (w) with incremental plasma glucose (y), and is subject to and The order of the numerator cannot exceed the order of the denominator; - Define carbohydrate intake as: in This is the estimated carbohydrate content reported by the patient. - Define the expected postprandial increase in plasma glucose for: in It is an insulin bolus administered in association with the patient's manual meal notification; - Define the corrected incremental plasma glucose Corrected insulin infusion and corrected carbohydrate intake for: - Define virtual control behavior for: Make: - Control behavior Divided into regulatory control behaviors and counter-regulatory control behavior : - The assigned counter-regulatory behavior is as follows in This represents the counter-regulatory effect implemented through glucagon infusion, and This represents the counter-regulatory effect implemented through carbohydrate intake, resulting in: - The regulatory and counter-regulatory behaviors are calculated as follows: in It is a function that defines the change between regulatory and counter-regulatory behaviors, and It is an adjustable factor, and among them It is the previously obtained controller gain, and Defined as: in Calculated as a pre-filter with nominal value The 2-DOF feedback controller is as follows: And among them It is designed for use in stabilizing systems and attenuate interference (s) central linear controller, - The counter-regulatory behavior is calculated in the following form: in It is an adjustable parameter; y - The control behavior is calculated as follows:

2. The method according to claim 1, wherein the model It was obtained as follows: in , , and It is a transfer function of the following form: in , , and It is a parameter representing the delay. , , , , , , , , , , and These are the parameters obtained previously, and j is the index representing the patient.

3. The method according to any one of claims 1 to 2, further comprising the following step: Corrective and counter-regulatory behaviors can be implemented in the following ways: in It is a function that defines the change between regulatory and counter-regulatory behaviors, and It is an adjustable factor.

4. The method according to any one of claims 1 to 3, wherein equal Or with a specific sample window The moving average filter.

5. The method according to any one of claims 1 to 4, further comprising the following step: right The following corrections will be implemented: Define feedback behavior ( ) and feedforward behavior ( )as follows: in It is relative to The nominal value of the incremental glucose.

6. The method according to any one of claims 1 to 5, wherein the controller Defined as: And the filter and Defined as: in , , and These are the parameters calculated previously, and r is relative to... The nominal value of the incremental glucose.

7. The method according to any one of claims 1 to 6, wherein, based on the condition that one counter-regulatory behavior is more advantageous than another, It is variable over time.

8. The method according to any one of claims 1 to 5, wherein the rescue carbohydrates are quantified as a level ( The specific dose is administered and calculated as follows: And among them It is the minimum threshold for activating the recommended cumulative carbohydrate intake. ,and: in It defines the time range. Internal The prediction and These are the predetermined thresholds for predicting glucose and measuring glucose, respectively; and where: in BW It is the user's weight (kg). It is the sampling period, and It's the clearance rate.

9. The method of claim 8, wherein for The prediction calculation is as follows: It is the discretization of the glucose derivative after filtering: in It is a previously determined time constant.

10. The method according to any one of claims 1 to 9, wherein glucagon is used as a quantification level ( The specific dose is administered and calculated as follows: And among them It is the minimum threshold for activating the recommended cumulative glucagon. ,and: in It defines the time range. Internal The prediction, and and These are the predetermined thresholds for predicting glucose and measuring glucose, respectively; and where: in BW It is the user's weight (kg). It is the sampling period and It's the clearance rate.

11. The method of claim 10, wherein for The prediction calculation is as follows: Discretization of the glucose derivative after filtering in It is a previously determined time constant.

12. The method according to any one of claims 8 to 11, wherein It is 60 mg / dl. The concentration was 54 mg / dl, and the time range was... It takes 60 minutes.

13. The method according to any one of claims 8 to 12, wherein the expected postprandial increment plasma glucose is calculated by adding a term corresponding to a specific dose of the counter-regulatory behavior. ,like: in and It is an adjustable gain.

14. The method of claim 13, wherein the adjustable gain and It was determined that when the counter-regulatory behavior (v, w) equals the quantification level ( , To achieve the required maximum output, increase The required gain is obtained as follows: in Represents the inverse Laplace transform operator, and It is determined by the level of quantification ( , The required output change caused by ).

15. The method according to any one of claims 1 to 14, wherein the expected postprandial increase in plasma glucose is... Saturation is as follows:

16. The method of claim 15, wherein 180 and It is 70.

17. The method according to any one of claims 6 to 16, wherein the controller Adjustable gain Defined as .

18. The method according to any one of claims 6 to 16, wherein the adjustable gain Defined as: in It is the inverse Laplace transform, and and These are adjustable parameters, among which This represents the expected size of the meal in the worst-case scenario, and It is in response to the meal The lower limit of glucose tolerance after an increased meal.

19. The method of claim 18, wherein Defined as the size of the meal notification given to the patient each time an optional meal notification is given. ),and Defined as a response to the meal The incremental postprandial glucose tolerance limit is determined by keeping the calculated k value within an adjustable time window from the time of notification to eat.

20. The method according to any one of claims 1 to 19, wherein the insulin bolus administered as a result of optional meal notification ( The meal scale reported by the patient ( ) is calculated as by Defined group note, where It is a feedforward controller, as given below:

21. The method according to any one of claims 1 to 19, wherein the administered insulin bolus ( The meal scale reported by the patient ( ) is calculated as by The defined supergroup betting, where It is a feedforward controller, as given below: in > 0 represents the time period during which the pump is shut off.

22. An artificial pancreas system for performing the blood glucose control method according to any one of claims 1 to 21, comprising: - Pump (3), which is based on coordinated control behavior ( ) deliver insulin, and / or according to coordinated control behavior ( Automated infusion of glucagon; - A continuous glucose monitor (2) for measuring plasma glucose signals (G(t)); and - A first computing unit (1) is configured to perform the steps of any one of claims 1 to 21.

23. The artificial pancreas system of claim 22, further comprising a second pump for automatically delivering glucagon.

24. The artificial pancreas system according to any one of claims 22 to 23, further comprising a display for notifying controlled administration of recommended glucagon and / or rescue carbohydrates.

25. The artificial pancreas system according to any one of claims 22 to 24, further comprising a pen device for use by a patient to deliver recommended glucagon-controlled behavior.

26. A computer program adapted to perform the steps of the method according to any one of claims 1 to 21 using a computing unit in an artificial pancreas system according to any one of claims 22 to 25.

27. A computer-readable storage medium comprising the computer program of claim 26.

Images