Closed-loop blood glucose control system and method

Abstract

To provide a closed-loop blood glucose control system and method.SOLUTION: A control device (30) for determining a recommendation value of a control parameter of an insulin infusion device (20).SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to the field of closed-loop blood glucose control systems for providing controlled insulin administration to a patient, also known as artificial pancreas. [Background technology]

[0002] An artificial pancreas is a system that automatically adjusts a diabetic's or user's insulin intake based on a history of measured blood glucose levels, dietary history, and insulin history.

[0003] In particular, the present invention relates to an insulin sensitivity factor (ISF) that can be used to determine appropriate insulin dosage based on a patient's body's sensitivity to insulin, thereby allowing for more precise adjustment of insulin dosage to maintain optimal glycemic control.

[0004] For example, it would be desirable to improve the performance of systems based on insulin sensitivity coefficients by improving the accuracy and reliability of physiological models. More accurate prediction of insulin sensitivity would allow for more accurate estimation of insulin requirements, thereby reducing the risk of both hyperglycemia and hypoglycemia. This improved prediction accuracy is critical for optimizing insulin therapy and ensuring overall improved diabetes management.

[0005] U.S. Patent Application Publication No. 20090054753A1 discloses insulin sensitivity as a general measure of the body's response to insulin administration. This coefficient can change as the patient's physiological state changes and can be useful in determining the patient's response to treatment. A patient's insulin sensitivity can be determined in a variety of ways, such as by input from a healthcare provider, by inference from other symptoms, or by determination from previous insulin administration and blood glucose measurement information. Insulin sensitivity can be used as a parameter to determine the next sampling time. [Prior art documents] [Patent documents]

[0006] [Patent Document 1] U.S. Patent Application Publication No. 20090054753A1 Summary of the Invention [Problem to be solved by the invention]

[0007] The aim of the present invention is to improve this situation.

[0008] SUMMARY OF THE INVENTION It is therefore an object of the present invention to at least partially solve the technical problems set out above. [Means for solving the problem]

[0009] The present invention therefore relates to a control device for determining recommended values ​​for control parameters of an insulin infusion device, the control device comprising: an acquisition unit configured to acquire user data, each data item of the user data having a timestamp, the user data being associated with a particular user, the user data comprising: the amount of insulin injected into a particular user; The amount of carbohydrates consumed by a particular user, and A plurality of physiological data values ​​for a particular user, including at least a measured blood glucose level. an acquisition unit including at least a recommendation unit configured to determine a recommendation based at least on an insulin sensitivity factor (ISF); Equipped with The ISF is a function of the measured blood glucose level of at least one of the physiological values.

[0010] Using an ISF that is a function of at least one measured blood glucose level allows the recommendation unit to more accurately determine recommendations that are tailored to the needs of a particular user, since the ISF may vary from one particular user and from one measured blood glucose level to another.

[0011] According to the present invention, the ISF represents the effect of a given insulin dose on a particular user's blood glucose level.

[0012] According to the present invention, the ISF represents the effect of a unit of insulin on a particular user's measured blood glucose level, where an insulin unit corresponds to the "bioequivalent amount" of 34.7 μg of pure crystalline insulin.

[0013] According to the present invention, the ISF being a function of the measured blood glucose level of at least one of the physiological numerical data means that the ISF changes depending on the measured blood glucose level, and therefore if the measured blood glucose level is not constant over time, the ISF will also change over time.

[0014] According to one embodiment, the control device further comprises an insulin delivery unit, such as a subcutaneous insulin delivery device, configured to deliver exogenous insulin, in particular continuous infusion insulin such as basal insulin and / or bolus insulin, into the subcutaneous tissue of the patient in response to an insulin delivery control signal.

[0015] According to one embodiment, the subcutaneous insulin delivery device is an insulin pump.

[0016] According to one embodiment, the recommendation corresponds to the amount of insulin to be infused at a future time step.

[0017] According to one embodiment, this future time step is within less than a few seconds of the most recent timestamp of the particular user's physiological value. This delay of less than a few seconds corresponds to the computation time required to determine the recommendation. The recommendation can be of any type, such as a bolus recommendation or a basal recommendation.

[0018] According to the present invention, the terms "injected" or "injection" should be understood as virtual injection when the training of the reinforcement learning algorithm is performed using a simulation in which a specific user is a virtual user.

[0019] According to one embodiment, the control device comprises a conversion unit configured to convert the recommended values ​​of insulin into control parameters of the fluid infusion device, which according to one embodiment are in the form of bolus and / or basal doses, allowing the infusion device to inject the recommended values ​​for a particular user.

[0020] According to the present invention, a measured blood glucose level is a blood glucose level measured for a particular user. The blood glucose level can be measured by any means, such as a continuous glucose monitor (CGM) or a background glucose monitor (BGM).

[0021] According to the present invention, the amount of carbohydrates ingested by a particular user corresponds to the amount of sugar ingested in, for example, a meal.

[0022] According to one embodiment, the ISF is a function of measured blood glucose levels having a timestamp within 1-3 hours before the current time, preferably 2 hours before the current time. This allows the recommendation unit to more accurately determine recommendations, as measured blood glucose levels with older timestamps may not accurately reflect the condition of a particular user as compared to measured blood glucose levels with more recent timestamps.

[0023] According to one embodiment, the ISF is a decreasing function of the measured blood glucose level of at least one of the plurality of physiological values.

[0024] Such a configuration allows the control device to accurately determine the recommended value, and thus assists in managing the blood glucose level of a particular user. Indeed, the ISF, which is a decreasing function of the measured blood glucose level of at least one of the physiological values, allows the recommendation unit to accurately determine the recommended value.

[0025] According to one embodiment, the ISF is a decreasing function of a measured blood glucose level of at least one of the plurality of physiological numerical data, and the measured blood glucose level of at least one of the plurality of physiological numerical data is greater than 100 mg / dl.

[0026] According to one embodiment, the ISF is a function of the measured blood glucose level of at least one of the physiological values, such as ISF=m×Gly+b, where: Gly is at least one measured blood glucose level among the plurality of physiological numerical data, for example, the most recently measured blood glucose level; m is a coefficient, b is a constant.

[0027] Such a configuration allows the control device to accurately determine the recommended value, and thus assists in managing the blood glucose level of a particular user. Indeed, the ISF, which is a decreasing function of the measured blood glucose level of at least one of the physiological values, allows the recommendation unit to accurately determine the recommended value.

[0028] According to one embodiment, m varies with Gly, e.g., 1≦m≦1.5 when Gly<100 mg / dL, −0.8≦m≦−0.6 when 100≦Gly<160 mg / dL, and −0.2≦m≦0 when Gly>160 mg / dL. This configuration allows the ISF to more accurately represent a particular user's physiological state. According to one embodiment, m varies with Gly by smoothly transitioning from 1 to −0.8 according to the aforementioned evolution of m.

[0029] According to one embodiment, b varies with Gly, e.g., −50≦b≦−30 when Gly<100 mg / dL, 150≦b≦170 when 100≦Gly<160 mg / dL, and 50≦b≦70 when Gly>160 mg / dL. This configuration allows the ISF to more accurately represent the physiological state of a particular user. According to one embodiment, b varies with Gly by smoothly transitioning from −50 to 170 according to the aforementioned gradual changes in b. According to another embodiment, the ISF is a piecewise affine function with an undefined number of intervals.

[0030] According to one embodiment, the ISF is a piecewise affine function of Gly, where m=m1 and b=b1 if Gly<100 mg / dl, m=m2 and b=b2 if 100≦Gly<160 mg / dl, and m=m3 and b=b3 if Gly>160 mg / dl. This configuration allows the ISF to more accurately represent the physiological state of a particular user. According to another embodiment, the ISF is a piecewise affine function with an undefined number of intervals. The piecewise affine function is a nonlinear exponential function, such as ISF(Gly)=a×ê((c×Gly)).

[0031] The control device 30 is configured to modify m and b, as well as m1, m2, m3, b1, b2, and b3, over time. This allows the control device 30 to more accurately adapt to the idiosyncrasies of a particular user. m and b, as well as m1, m2, m3, b1, b2, and b3, may be modified over time using any known method, such as the automated learning method detailed above applied to the ISF, etc.

[0032] According to one embodiment, the function in the piecewise affine function is the logarithmic function a*log(Gly)+b.

[0033] Note that the piecewise affine functions in the above example are linear and nonlinear functions. These piecewise affine functions can be a mixture of linear and nonlinear functions, e.g., linear for Gly<100 mg / dL and nonlinear for Gly≧100 mg / dL.

[0034] According to one embodiment, the recommendation unit is configured to determine the recommendation based at least on a predicted blood glucose value determined by calculating a physiological model of the glucose-insulin system corresponding to the particular user using at least a portion of the user data and the ISF.

[0035] Such a configuration allows for accurate determination of recommended values, thereby assisting in the management of blood glucose levels for a particular user. Indeed, the use of the ISF allows for specific adaptation of recommended values ​​to a particular user. Furthermore, the ISF, which is a function of the measured blood glucose level of at least one of the physiological values, allows the physiological model to closely replicate the physiological behavior of a particular user, thereby enabling the recommendation unit to accurately determine recommended values.

[0036] Furthermore, such a configuration allows for more accurate testing or fine-tuning of existing algorithms, for example using a simulator. Indeed, an improved physiological model more closely replicates the physiological behavior of a particular user, thus allowing for more accurate evaluation of existing algorithms.

[0037] According to one embodiment, the physiological model can be of any type, such as the Hovorka model, the Bergman minimal model, or the Dalla Man model, etc. According to a preferred embodiment, the physiological model is the Hovorka model or a model derived from the Hovorka model.

[0038] According to one embodiment, the recommendation unit may be configured to determine the recommendation based at least on an integral-differential method (PID) and an ISF, where the PID is based on at least one measured or predicted blood glucose value and at least one blood glucose target value, and where the ISF is used as a weighting factor.

[0039] According to one embodiment, the recommendation unit may be configured to determine a recommendation based at least on a neural network and the ISF. The neural network uses as inputs at least one amount of insulin injected for the specific user, at least one amount of carbohydrates ingested by the specific user, and at least one physiological value of the specific user, and outputs a raw recommendation, where the ISF is used as a weighting factor. The neural network can be of any type and is trained as follows: - Collect data. - Preprocess the data. - Design neural network architectures, e.g., regression models. - Define the input features. - Define the output features. - Train a neural network using preprocessed training data. - Validate the performance of the model, and - Evaluate the performance of the model using metrics such as mean absolute error (MAE) or mean squared error (MSE).

[0040] According to one embodiment, the recommendation unit is configured to determine the recommendation based at least on a predicted blood glucose level determined by calculating a physiological model of the glucose-insulin system corresponding to the particular user using an insulin action rate that is an increasing function of the measured blood glucose level of at least one of the plurality of physiological numerical data.

[0041] This configuration allows for a more accurate insulin on board (IOB) value to be obtained, and therefore allows for a more accurate determination of the recommended value by the recommended unit.

[0042] According to one embodiment, the measured blood glucose value of at least one of the plurality of physiological numerical data corresponds to the last measured blood glucose value of the plurality of physiological numerical data, or a plurality of blood glucose values ​​measured in the last few of the plurality of physiological numerical data, or an average value of a plurality of blood glucose values ​​measured in the last few of the plurality of physiological numerical data.

[0043] According to one embodiment, ISF is not a function of the total daily insulin dose, nor is it an average of multiple previous total daily insulin doses. The total daily insulin dose is the total amount of insulin infused for a particular user during the day. In fact, ISF is not a constant for blood glucose levels.

[0044] According to one embodiment, the ISF is a decreasing function of the measured blood glucose level of at least one of the physiological values ​​and is dependent on at least one past ISF.

[0045] Such a configuration allows for accurate determination of the recommended value, thereby reducing the risk to a particular user and helping the user manage their blood glucose levels. Indeed, an ISF that relies on at least one previous ISF allows for avoiding sudden changes in the determination of the recommended value.

[0046] According to the present invention, a past ISF is a previously calculated ISF, which may be calculated, for example, in the same manner as the current ISF, or may be an arbitrarily initialized past ISF.

[0047] According to one embodiment, the ISF is a decreasing function of the measured blood glucose level of at least one of the plurality of physiological values ​​and is based on an average value of a plurality of past ISFs.

[0048] Such a configuration makes it possible to further smooth out variations in the determination of the recommendation value, and therefore to reduce the risk to a particular user.

[0049] According to one embodiment, the ISF is based on an auto-learned glycemic factor, e.g., ISF=avgISF*Gly*glyF, where: avgISF is the average value of the past ISF, glyF is the glycemic index.

[0050] According to one embodiment, avgISF is the average value of all ISF calculated over a one week period.

[0051] According to one embodiment, the ISF is a decreasing function of at least the measured blood glucose level among the plurality of physiological numerical data, and is based on an automatically learned ISF.

[0052] According to one embodiment, the ISF relies on an auto-learned glycemic factor, e.g., ISF=ALISF*Gly*glyF, where: ALISF is an auto-learned ISF.

[0053] Such a configuration allows for accurate determination of recommendations and therefore tailors them to a particular user over time, thereby helping the particular user manage their blood glucose levels, which is particularly useful when a particular user's ISF is highly variable in response to measured blood glucose levels.

[0054] Such a configuration allows for accurate determination of recommendations, which can then be adapted to a particular user over time to help manage that user's blood glucose levels. Indeed, a particular user's ISF may change due to multiple factors, such as conditions unrelated to diabetes.

[0055] According to the present invention, the term "automatically learned ISF" refers to an ISF determined using an automatic learning method. The automatic learning method can be of any type, such as a process by which a system, typically a computer program or artificial intelligence, autonomously acquires knowledge, skills, or behaviors without explicit programming or direct human intervention. Automatic learning systems often rely on algorithms, data, and feedback mechanisms to improve their performance over time and adapt to changing situations or tasks through self-adjustment and refinement.

[0056] According to one embodiment, the auto-learned ISF is determined by grouping user data based at least on their timestamps by creating a plurality of time segments, each time segment having an ISF, creating a segment correction for each time segment based at least on a past ISF and at least one physiological value from the plurality of physiological values ​​during a time segment from the plurality of time segments, and creating an auto-learned ISF for a predetermined future time period based at least on a segment correction from the segment corrections and the ISF for the predetermined time period. The use of segments allows for accurate determination of the auto-learned ISF because each segment correction can be created in a different way, for example, for each time segment or depending on the particular subject of interest of the time segment.

[0057] According to one embodiment, the ISF relies on an auto-learned glycemic factor.

[0058] According to one embodiment, the ISF is based on an auto-learned glycemic factor such as ISF=pISF*Gly*ALglyF, where: pISF is a past ISF, avgISF is the average of the past ISFs, for example the average of the last five ISFs in the most recent time period, ALglyF is the auto-learned glycemic index.

[0059] Such a configuration allows for accurate determination of recommendations, which can then be adapted to a particular user over time to help manage that user's blood glucose levels, which is particularly useful when a particular user's ISF is highly variable in response to measured blood glucose levels.

[0060] According to one embodiment, pISF may be replaced by avgISF, in effect ISF=avgISF*Gly*ALglyF, which allows for a further smoothing of variations in the determination of the recommendation value and thus a reduction in the risk for a particular user.

[0061] According to one embodiment, ALglyF is a combination of coefficients and constant terms.

[0062] According to the present invention, the term "auto-learned glycemic coefficient" refers to a glycemic coefficient determined using an auto-learning method. The auto-learning method can be of any type, such as a process by which a system, typically a computer program or artificial intelligence, autonomously acquires knowledge, skills, or behaviors without explicit programming or direct human intervention. Auto-learning systems often rely on algorithms, data, and feedback mechanisms to improve their performance over time and adapt to changing situations or tasks through self-regulation and refinement.

[0063] According to one embodiment, the automatically learned glycemic factor is determined by grouping user data based at least on their timestamps by creating a plurality of time segments, each time segment having a glycemic factor; creating a segment correction value for each time segment based at least on at least one past glycemic factor and at least one physiological value among the plurality of physiological values ​​during a time segment among the plurality of time segments; and creating an automatically learned glycemic factor for a predetermined time period in the future based at least on a segment correction value among the segment correction values ​​and the glycemic factor for the predetermined time period. The use of segments allows the automatically learned glycemic factor to be accurately determined, since each segment correction value can be created in a different way, for example, for each time segment or depending on the particular interest of the time segment.

[0064] According to one embodiment, the ISF is a decreasing function of the measured blood glucose level of at least one of the plurality of physiological values ​​and is based on an auto-learned ISF and an auto-learned blood glucose coefficient.

[0065] According to one embodiment, this ISF is based on an auto-learned ISF and an auto-learned glycemic factor, for example, ISF=ALISF*Gly*ALglyF.

[0066] Such a configuration allows for accurate determination of recommended values ​​and therefore helps manage a particular user's blood glucose levels by adapting to the particular user over time. Indeed, both a particular user's ISF and glyF may vary due to multiple factors, such as, for example, diseases unrelated to diabetes.

[0067] According to one embodiment, the recommendation unit is configured to determine the recommendation value based at least on the ISF and the correction factor.

[0068] According to one embodiment, this correction factor is a factor applied to the ISF, which allows the recommendation unit to take into account the specific circumstances of a particular user, such as diseases unrelated to diabetes.

[0069] According to one embodiment, the recommendation unit determines the recommendation value by considering the ISF multiplied by a correction factor.

[0070] Furthermore, the present invention relates to a method for determining recommended values ​​of control parameters of an insulin infusion device, implemented by a control device as described above, the method comprising: obtaining user data; determining a recommended value based at least on the ISF; Including, The ISF is a function of the measured blood glucose level of at least one of the physiological values.

[0071] The embodiments, technical effects, and definitions disclosed herein with respect to this control device also apply to the method described herein. The method includes steps that fully utilize the functions and features of the control device described herein. Thus, all embodiments, technical effects, and definitions related to the device, including but not limited to the change in ISF over time and how it is calculated, apply to this method as well. This ensures a comprehensive and unified understanding of both the control device and method aspects of the present invention, thereby facilitating the implementation and use of the disclosed technology across a variety of applications.

[0072] Furthermore, the invention relates to a computer program comprising instructions for causing a control device as described above to carry out the steps of the method as described above.

[0073] Embodiments of the invention will now be described with reference to the drawings, which will be briefly described below. [Brief explanation of the drawings]

[0074] [Figure 1] FIG. 1 illustrates a method for validating a control algorithm according to one embodiment of the present invention. [Figure 2] FIG. 2 illustrates a control device according to one embodiment of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0075] FIG. 1 illustrates a control device 30 for determining recommended values ​​for control parameters of an insulin infusion device 20. The recommended values ​​correspond to the amount of insulin to be infused at a future time step. This future time step is within a few seconds of the most recent timestamp of a specific user's physiological value. This delay of less than a few seconds corresponds to the computation time required to determine the recommended values. The recommended values ​​can be of any type, such as a bolus recommendation or a basal recommendation. The terms "infused" or "infusion" should be understood to refer to a virtual infusion when the reinforcement learning algorithm is trained using a simulation in which the specific user is a virtual user. Such simulations can be useful for validating or improving the algorithm and thus reducing risk to the specific user. In other cases, the terms "infused" or "infusion" should be understood to refer to a normal infusion.

[0076] The control device 30 includes an acquisition unit 32 configured to acquire user data, each of which has a timestamp and is associated with a specific user. The user data includes an amount of insulin injected for the specific user, an amount of carbohydrates ingested by the specific user, and multiple physiological numerical data of the specific user, the multiple physiological numerical data of the specific user including at least a measured blood glucose level.

[0077] The measured blood glucose level is the blood glucose level measured for a particular user. The blood glucose level can be measured by any means, such as a continuous glucose monitor (CGM) 12 or a blood glucose monitor (BGM). Preferably, the blood glucose level is measured using a CGM 12. The amount of carbohydrates ingested by a particular user corresponds to the amount of sugar ingested in a meal, for example.

[0078] Furthermore, the control device 30 comprises a recommendation unit 34. The recommendation unit 34 is configured to determine a recommendation based at least on an insulin sensitivity factor (ISF). The ISF represents the effect of a given insulin dose on a specific user's blood glucose level. More precisely, the ISF represents the effect of an insulin unit on a specific user's measured blood glucose level. An insulin unit corresponds to the "bioequivalent amount" of 34.7 μg of pure crystalline insulin.

[0079] According to one embodiment, the recommendation unit 34 is configured to determine the recommendation based at least on a predicted blood glucose level determined by calculating a physiological model of the glucose-insulin system corresponding to the specific user using at least a portion of the user data and the ISF. Such a configuration enables the recommendation unit 34 to accurately determine the recommendation, thereby assisting the specific user in managing their blood glucose levels. Indeed, the use of the ISF allows the control device 30 to specifically tailor the recommendation to the specific user. Furthermore, the ISF, which is a function of the measured blood glucose level of at least one of the physiological values, enables the physiological model to closely replicate the physiological behavior of the specific user, thereby enabling the recommendation unit to accurately determine the recommendation. Furthermore, such a configuration allows for more accurate testing or fine-tuning of existing algorithms, for example, using a simulator. Indeed, an improved physiological model more closely replicates the physiological behavior of the specific user, thereby enabling more accurate evaluation of existing algorithms. The physiological model can be of any type, such as the Hovorka model, the Bergman minimal model, or the Dalla Man model, etc. Preferably, the physiological model is the Hovorka model or a model derived from the Hovorka model.

[0080] Also, according to one embodiment, the recommendation unit 34 may be configured to determine the recommendation based at least on an integral-differential method (PID) and an ISF, where the PID is based on at least one measured or predicted blood glucose level and at least one blood glucose target level, and the ISF is used as a weighting factor.

[0081] According to one embodiment, the recommendation unit 34 may be configured to determine a recommendation based at least on a neural network and the ISF. The neural network uses as inputs at least one amount of insulin injected for the specific user, at least one amount of carbohydrates ingested by the specific user, and at least one physiological value of the specific user, and outputs a raw recommendation, where the ISF is used as a weighting factor. The neural network can be of any type and is trained as follows: - Collect data. - Preprocess the data. - Design neural network architectures, e.g., regression models. - Define the input features. - Define the output features. - Train a neural network using preprocessed training data. - Validate the performance of the model, and - Evaluate the performance of the model using metrics such as mean absolute error (MAE) or mean squared error (MSE).

[0082] The recommendation unit 34 is configured to determine the recommendation based at least on a predicted blood glucose level determined by calculating a physiological model of the glucose-insulin system corresponding to the specific user using an insulin action rate that is an increasing function of the measured blood glucose level of at least one of the plurality of physiological numerical values. Such a configuration makes it possible to obtain a more accurate Insulin in Body (IOB) value, and therefore to more accurately determine the recommendation by the recommendation unit 34. The measured blood glucose level of at least one of the plurality of physiological numerical values ​​corresponds to the last measured blood glucose level of the plurality of physiological numerical values, or a plurality of blood glucose levels measured in the last few of the plurality of physiological numerical values, or an average value of a plurality of blood glucose levels measured in the last few of the plurality of physiological numerical values.

[0083] It should be understood that in the present invention, ISF is not a function of the total daily insulin dose, nor is it an average of multiple past total daily insulin doses. The total daily insulin dose is the total amount of insulin infused for a particular user during the day. In fact, ISF is not a constant for blood glucose levels.

[0084] The ISF is a function of at least one measured blood glucose level among the physiological numerical data. Using an ISF that is a function of at least one measured blood glucose level allows the recommendation unit 34 to more accurately determine a recommendation tailored to the needs of a specific user. This is because the ISF may vary for each specific user and for each measured blood glucose level. The fact that the ISF is a function of at least one measured blood glucose level among the physiological numerical data means that the ISF varies depending on the measured blood glucose level. Therefore, if the measured blood glucose level is not constant over time, the ISF also varies over time. The ISF is a function of a measured blood glucose level that has a timestamp within one to three hours before the current time, preferably two hours before the current time. This configuration allows the recommendation unit 34 to more accurately determine a recommendation because a measured blood glucose level with an older timestamp does not accurately reflect the state of the specific user as compared to a measured blood glucose level with a more recent timestamp.

[0085] The ISF is a decreasing function of the measured blood glucose level of at least one of the plurality of physiological numerical values, where the measured blood glucose level of at least one of the plurality of physiological numerical values ​​is greater than 100 mg / dL. Such a configuration enables the control device 30 to accurately determine a recommended value, and thus, such a configuration helps manage the blood glucose level of a particular user. Indeed, the ISF, which is a decreasing function of the measured blood glucose level of at least one of the plurality of physiological numerical values, enables the recommendation unit 34 to accurately determine a recommended value.

[0086] The ISF can be calculated as a function of the measured blood glucose level of at least one of the physiological values ​​as follows: ISF = m × Gly + b Here, Gly is the measured blood glucose level of at least one of the plurality of physiological numerical data, for example, the blood glucose level measured most recently. m is a coefficient, and b is a constant.

[0087] Such a configuration allows the control device 30 to accurately determine the recommended value, and thus assists in managing the blood glucose level of a particular user. Indeed, the ISF, which is a decreasing function of the measured blood glucose level of at least one of the physiological values, allows the recommendation unit 34 to accurately determine the recommended value.

[0088] According to one embodiment, the ISF is a piecewise affine function of Gly, where m=m1 and b=b1 if Gly<100 mg / dl, m=m2 and b=b2 if 100≦Gly<160 mg / dl, and m=m3 and b=b3 if Gly>160 mg / dl. This configuration allows the ISF to more accurately represent a specific user's physiological state. Therefore, b varies depending on Gly, e.g., −50≦b1≦−30 if Gly<100 mg / dl, 150≦b2≦170 if 100≦Gly<160 mg / dl, and 50≦b3≦70 if Gly>160 mg / dl. This configuration allows the ISF to more accurately represent a specific user's physiological state. According to one embodiment, b varies with Gly by smoothly transitioning from −50 to 170 in accordance with the aforementioned gradual change in b. According to another embodiment, the ISF is a piecewise affine function with an undefined number of intervals. Furthermore, m varies with Gly, e.g., 1≦m1≦1.5 when Gly<100 mg / dL, −0.8≦m2≦−0.6 when 100≦Gly<160 mg / dL, and −0.2≦m3≦0 when Gly>160 mg / dL. This configuration allows the ISF to more accurately represent the physiological state of a particular user. According to one embodiment, m varies with Gly by smoothly transitioning from 1 to −0.8 in accordance with the aforementioned gradual change in m.

[0089] According to another embodiment, the ISF is a piecewise affine function with an undefined number of intervals, where the function is a nonlinear exponential function, such as ISF(Gly)=a×ê((c×Gly)).

[0090] According to one embodiment, the function in this piecewise affine function is the logarithmic function a*log(Gly)+b.

[0091] Note that the piecewise affine functions in the above example are linear and nonlinear functions. These piecewise affine functions can be a mixture of linear and nonlinear functions, e.g., linear for Gly<100 mg / dL and nonlinear for Gly≧100 mg / dL.

[0092] The control device 30 is configured to modify m and b, as well as m1, m2, m3, b1, b2, and b3, over time. This allows the control device 30 to more accurately adapt to the idiosyncrasies of a particular user. m and b, as well as m1, m2, m3, b1, b2, and b3, may be modified over time using any known method, such as the automated learning method detailed above applied to the ISF, etc.

[0093] Preferably, the ISF relies on at least one previous ISF. This configuration allows the recommendation unit 34 to accurately determine the recommended value, thereby reducing the risk to a particular user and helping the user manage their blood glucose level. In fact, an ISF that relies on at least one previous ISF can avoid sudden changes in the determination of the recommended value. A previous ISF is an ISF that has been previously calculated. Such a previous ISF can be calculated, for example, in the same way as the current ISF, or can be a previous ISF that has been arbitrarily initialized, for example, by a medical professional.

[0094] For this reason, the ISF is based on an average of multiple past ISFs, which allows the recommendation unit 34 to further smooth out variations in the recommendation determination, thereby reducing risk for a particular user.

[0095] According to one embodiment, the ISF is based on an auto-learned glycemic factor, e.g., ISF=avgISF*Gly*glyF, where: avgISF is the average of all ISFs calculated over the week, glyF is the glycemic index.

[0096] According to one embodiment, the ISF relies on an auto-learned glycemic factor, e.g., ISF=ALISF*Gly*glyF, where: ALISF is an auto-learned ISF.

[0097] This configuration allows the recommendation unit 34 to accurately determine the recommended values, which can then be adapted to the specific user over time, thereby helping the specific user manage their blood glucose levels. This configuration is particularly useful when the specific user's ISF is highly variable depending on the measured blood glucose levels. This configuration also allows the recommendation unit 34 to accurately determine the recommended values, which can then be adapted to the specific user over time, thereby helping the specific user manage their blood glucose levels. In fact, the ISF of a specific user may vary due to multiple factors, such as diseases unrelated to diabetes. According to the present invention, the term diabetes refers to a condition dependent on exogenous insulin.

[0098] The term "auto-learned ISF" refers to an ISF determined using an auto-learning method, which can be of any type, such as a process by which a system, typically a computer program or artificial intelligence, autonomously acquires knowledge, skills, or behaviors without explicit programming or direct human intervention. Auto-learning systems often rely on algorithms, data, and feedback mechanisms to improve their performance over time and adapt to changing situations or tasks through self-adjustment and refinement.

[0099] According to one embodiment, the auto-learned ISF is determined by grouping user data based at least on their timestamps by creating a plurality of time segments, each time segment having an ISF, creating a segment correction for each time segment based at least on a past ISF and at least one physiological value from the plurality of physiological values ​​during a time segment from the plurality of time segments, and creating an auto-learned ISF for a predetermined future time period based at least on a segment correction from the segment corrections and the ISF for the predetermined time period. The use of segments allows for accurate determination of the auto-learned ISF because each segment correction can be created in a different way, for example, for each time segment or depending on the particular subject of interest of the time segment.

[0100] Additionally, the ISF may be based on an auto-learned glycemic factor such as ISF=pISF*Gly*ALglyF, where: pISF is a past ISF, avgISF is the average of the past ISFs, for example the average of the last five ISFs in the most recent time period, ALglyF is the auto-learned glycemic index.

[0101] This configuration allows the recommendation unit to accurately determine the recommended value, thus helping the specific user manage their blood glucose levels by adapting to the specific user over time. This is particularly useful when the specific user's ISF is highly variable depending on the measured blood glucose levels. pISF may be replaced by avgISF. In fact, ISF=avgISF*Gly*ALglyF allows for further smoothing of variations in the determination of the recommended value, thus reducing the risk for the specific user.

[0102] According to the present invention, the term "auto-learned glycemic coefficient" refers to a glycemic coefficient determined using an auto-learning method. The auto-learning method can be of any type, such as a process by which a system, typically a computer program or artificial intelligence, autonomously acquires knowledge, skills, or behaviors without explicit programming or direct human intervention. Auto-learning systems often rely on algorithms, data, and feedback mechanisms to improve their performance over time and adapt to changing situations or tasks through self-regulation and refinement.

[0103] According to one embodiment, the automatically learned glycemic factor is determined by grouping user data based at least on their timestamps by creating a plurality of time segments, each time segment having a glycemic factor; creating a segment correction value for each time segment based at least on at least one past glycemic factor and at least one physiological numerical value of the plurality of physiological numerical values ​​during a time segment of the plurality of time segments; and creating an automatically learned glycemic factor for a predetermined time period in the future based at least on a segment correction value of the segment correction values ​​and the glycemic factor for the predetermined time period. The use of segments allows the automatically learned glycemic factor to be accurately determined, since each segment correction value can be created in a different way, for example, for each time segment or depending on the particular interest of the time segment.

[0104] According to one embodiment, the ISF is a decreasing function of at least the measured blood glucose level among the physiological values ​​and relies on an automatically learned ISF and an automatically learned blood glucose factor, thus combining the above-mentioned embodiments. The ISF relies on an automatically learned ISF and an automatically learned blood glucose factor, e.g., ISF = ALISF * Gly * ALglyF. This configuration allows for accurate determination of recommendations and thus helps manage a specific user's blood glucose levels by adapting to the specific user over time. In practice, both the ISF and glyF of a specific user may change due to multiple factors, such as diseases unrelated to diabetes.

[0105] The recommendation unit 34 is configured to determine the recommended value based at least on the ISF and a correction factor. The correction factor is a factor applied to the ISF that allows the recommendation unit 34 to take into account the specific circumstances of a specific user, such as diseases unrelated to diabetes. The recommendation unit 34 determines the recommended value by considering the ISF multiplied by the correction factor.

[0106] Additionally, the control device 30 includes an insulin delivery unit 20. A preferred insulin delivery unit 20 is a subcutaneous insulin delivery device configured to deliver exogenous insulin into the patient's subcutaneous tissue in response to an insulin delivery control signal, such as an insulin pump. In particular, the insulin delivery unit 20 delivers continuous infusion insulin and / or bolus insulin.

[0107] According to one embodiment, the control device 30 comprises a conversion unit, not shown in the drawings, configured to convert the recommended values ​​of insulin into control parameters of the fluid infusion device 20, which may take the form of bolus and / or basal doses, allowing the infusion device to inject the recommended values ​​for a particular user.

[0108] Additionally, as shown in Figure 2, the present invention relates to a method for determining recommended values ​​for control parameters of an insulin infusion device 20, as described above. The embodiments, technical effects, and definitions disclosed herein in connection with the control device 30 also apply to the method described herein. The method includes steps that fully utilize the functions and features of the control device 30 described herein. Accordingly, all embodiments, technical effects, and definitions related to the device, including but not limited to, changes in ISF over time and methods for calculating the same, apply to the method as well. This ensures a comprehensive and unified understanding of both the control device 30 and method aspects of the present invention, thereby facilitating implementation and utilization of the disclosed technology across a variety of applications.

[0109] This method is - a step 50 of obtaining user data; - step 58 of determining the recommended value based at least on the ISF; Including, The ISF is a function of the measured blood glucose level of at least one of the physiological values.

[0110] According to one embodiment, where SF is based on an automatically learned glycemic factor, e.g., ISF=ALISF*Gly*glyF, the method comprises: - a step 52 of grouping user data based at least on their timestamps by creating a plurality of time segments, each time segment having an ISF; - generating 54 a segment correction for each time segment based at least on a past ISF and at least one physiological value of the plurality of physiological values ​​during a time segment of the plurality of time segments; - generating an auto-learned ISF for a future predetermined period based at least on a segment correction value among the segment correction values ​​and an ISF for the predetermined period; Includes:

[0111] The use of segments allows for accurate determination of the automatically learned ISF, since each segment correction can be generated differently, for example, for each time segment or depending on the particular interest of the time segment.

[0112] Furthermore, the invention relates to a computer program comprising instructions for causing the control device 30 described above to carry out the steps of the method described above.

[0113] Furthermore, the present invention relates to a control system 10 comprising an insulin infusion device 20, a CGM 12, and a control device 30 comprising a recommendation unit 34 and an acquisition unit 32.

[0114] While illustrative embodiments of the present invention have been described, those skilled in the art will recognize that various changes, omissions, and / or additions may be made therein without departing from the spirit and scope of the invention, and that equivalents may be substituted for elements of the present invention. In addition, many modifications may be made to adapt a particular situation or material to the teachings of the invention without departing from its scope. Therefore, the invention is not limited to the particular embodiments disclosed for carrying out the invention, but is intended to include all embodiments falling within the scope of the appended claims. Further, unless otherwise specified, any use of terms such as "first," "second," etc., does not imply any order or importance, but rather is used to distinguish one element from another. [Explanation of symbols]

[0115] 10. Control System 12 Continuous glucose monitor, CGM 20 Insulin infusion devices, insulin delivery units, fluid infusion devices 30 Control Device 32 Acquisition Units 34 Recommended Units

Claims

1. A control device (30) for determining recommended values ​​of control parameters of an insulin infusion device (20), comprising: An acquisition unit (32) configured to acquire user data, each datum of the user data having a timestamp, the user data being associated with a particular user, the user data comprising: the amount of insulin injected into said particular user; the amount of carbohydrates consumed by said particular user; and a plurality of physiological numerical data of the particular user, including at least a measured blood glucose level; an acquisition unit (32) including at least a recommendation unit (34) configured to determine said recommendation based at least on an insulin sensitivity factor (ISF); Equipped with The control device (30) wherein the ISF is a function of a measured blood glucose level of at least one of the plurality of physiological values.

2. A control device (30) for determining recommended values ​​of control parameters of an insulin infusion device (20) as described in claim 1, wherein the ISF is a decreasing function of the measured blood glucose level of at least one of the plurality of physiological numerical data.

3. The ISF is a function of the measured blood glucose level of at least one of the plurality of physiological values, such as ISF=m×Gly+b, where: Gly is the at least one measured blood glucose level among the plurality of physiological numerical data, for example, the most recently measured blood glucose level, m is a coefficient, 2. The control device (30) for determining recommended values ​​of control parameters of an insulin infusion device (20) according to claim 1, wherein b is a constant.

4. A control device (30) for determining recommended values ​​of control parameters of an insulin infusion device (20) as described in any one of claims 1 to 3, wherein the recommendation unit (34) is configured to determine the recommended values ​​based at least on a predicted blood glucose level determined by calculating a physiological model of the glucose-insulin system corresponding to the particular user using at least a portion of the user data and the ISF.

5. 5. A control device (30) for determining recommended values ​​of control parameters of an insulin infusion device (20) as described in claim 4, wherein the recommendation unit (34) is configured to determine the recommended values ​​based at least on a predicted blood glucose level determined by calculating a physiological model of the glucose-insulin system corresponding to the specific user using an insulin action rate that is an increasing function of the measured blood glucose level of at least one of the plurality of physiological numerical data.

6. A control device (30) for determining recommended values ​​of control parameters of an insulin infusion device (20) as described in any one of claims 1 to 5, wherein the ISF is a decreasing function of the measured blood glucose level of at least one of the plurality of physiological numerical data and is based on at least one past ISF.

7. A control device (30) for determining recommended values ​​of control parameters of an insulin infusion device (20) as described in any one of claims 1 to 6, wherein the ISF is a decreasing function of the measured blood glucose level of at least one of the plurality of physiological numerical data and is based on an average value of a plurality of past ISFs.

8. A control device (30) for determining recommended values ​​of control parameters of an insulin infusion device (20) as described in any one of claims 1 to 7, wherein the ISF is a decreasing function of the measured blood glucose level of at least one of the plurality of physiological numerical data and is based on an automatically learned ISF.

9. A control device (30) for determining recommended values ​​of control parameters of an insulin infusion device (20) according to any one of claims 1 to 8, wherein the ISF relies on an automatically learned glycemic coefficient.

10. A control device (30) for determining recommended values ​​of control parameters of an insulin infusion device (20) described in any one of claims 1 to 9, wherein the ISF is a decreasing function of the measured blood glucose level of at least one of the plurality of physiological numerical data and is based on an automatically learned ISF and an automatically learned blood glucose coefficient.

11. A control device (30) for determining a recommended value of a control parameter of an insulin infusion device (20) as described in any one of claims 1 to 10, wherein the recommendation unit (34) is configured to determine the recommended value based at least on the ISF and a correction factor.

12. 10. A method for determining recommended values ​​of control parameters of an insulin infusion device (20) implemented by the control device (30) of claim 1, comprising: obtaining said user data (50); determining (52) the recommended value based at least on the ISF; Including, The method, wherein the ISF is a function of a measured blood glucose level of at least one of the plurality of physiological values.

13. A computer program comprising instructions for causing the control device (20) of claim 1 to perform the steps of the method of claim 12.

Images