Treatment area evaluator

Abstract

Systems and methods are provided for identifying a treatment zone in which a particular type of glycemic dysfunction exists that can be addressed by a change in a behavior and / or treatment parameter strategy. The systems and methods described herein evaluate large historical datasets to: identify one or more treatment zones in which glycemic dysfunction is most easily addressed; quantify the impact of a variety of different treatment adjustments on glycemia according to adjustments to historical doses or parameters of an expected dosing strategy to determine the best way to improve; and / or determine a patient dosing strategy to provide a treatment recommendation that is tailored to a patient's preferred behaviorally dosing strategy.

Description

[0001] RELATED APPLICATIONS INCORPORATED BY REFERENCE

[0002] This application claims priority to U.S. Provisional Patent Application 62 / 950,029, filed December 18, 2019, entitled “THERAPEUTIC ZONE ASSESSOR,” the entire contents of which are incorporated herein by reference and hereby expressly made a part of this specification.

[0003] BACKGROUND

[0004] In recent years, the availability and reliability of glucose time series data has improved with the increased adoption of CGM (continuous glucose monitoring) and related devices.

[0005] Identifying multiple patterns in large historical datasets requires a certain degree of sophistication, at least in part due to the overlapping nature of these patterns which cannot be resolved by human assessment. Even with pattern analysis tools, physicians cannot reliably determine from data review which aspect of diabetes treatment is most suitable for each particular patient’s situation.

[0006] Due to the numerous variables and factors involved in diabetes management, current approaches for identifying patterns and making recommendations lack reliability and ease of use, taking into account, for example, a combination of data trustworthiness, treatment availability for specific aspects of diabetes risk, action improvement, patient preferred diabetes management strategy, and associated risks and therapies.

[0007] Typically, a clinician reviews a patient’s CGM curve and corresponding insulin delivery pattern over a period of time, such as 14 days, or views the data in a more unified format, for example, a graph showing data for each day of the 14 days superimposed on a 24-hour timeline to visually highlight prominent pattern areas during specific times of the day.

[0008] However, visualization of the data does not readily highlight many important factors, risks, and potential outcomes needed for effective treatment optimization. Furthermore, the trustworthiness of the data by visual inspection is unknown or unclear.

[0009] In view of the above problems, various aspects and embodiments of the present disclosure are presented.

[0010] ABSTRACT

[0011] Systems and methods are provided for identifying a therapeutic zone in which a particular type of glycemic dysfunction exists, which can be addressed by changes in strategy for behavior and / or treatment parameters.

[0012] The systems and methods described herein evaluate large historical data sets to: identify one or more treatment zones of the most easily addressed glycemic dysfunction; quantify the impact of a plurality of different treatment adjustments on glycemia according to parameters of adjustment to historical dosing or expected dosing strategies to determine the best way to improve; and / or determine a patient dosing strategy to provide a treatment recommendation that is tailored to a patient's preferred behavioral dosing strategy.

[0013] In one implementation, a method includes deriving at least one single-symptom-specific risk curve using a glycemic risk analyzer; determining at least one treatment-related zone associated with the at least one single-symptom-specific risk curve using a treatment zone evaluator; determining an importance value for the at least one treatment-related zone using a zone importance quantifier; and outputting information based on the importance value.

[0014] Implementations can include some or all of the following features. The method further includes receiving glucose data, wherein the at least one single-symptom-specific risk curve is derived using the glucose data. The glucose data includes at least one of CGM (continuous glucose monitoring) readings, trustworthiness readings assigned to CGM values, self-monitored blood glucose readings, or retrospective calibrated or corrected CGM readings. The glucose data includes a time period of at least one week. The at least one single-symptom-specific risk curve describes a hypoglycemic risk or a hyperglycemic risk as a function of time of day using the glucose data. Deriving the at least one single-symptom-specific risk curve includes evaluating a steepness (first and second derivatives of the curve), a frequency, a severity, a curvature, a mean value in the curve over 24 hours, or a variability of the curve (mean and standard deviation). The at least one single-symptom-specific risk curve is based on a CGM signal indicating a glycemic dysfunction in a selected time period, indicating a cyclic time window characterized by a predetermined severity and frequency of hypoglycemia or hyperglycemia in the selected time period. The at least one single-symptom-specific risk curve represents at least one of hypoglycemia separated from hyperglycemia or at least one of hyperglycemia separated from hypoglycemia. Determining the at least one treatment-related zone includes identifying the at least one treatment-related zone from the at least one single-symptom-specific risk curve. The at least one treatment-related zone is an interval over 24 hours of a day in which the patient's BG data indicates that at least one of the patient's basal rate or dose or bolus strategy is systematically not optimal. The at least one treatment-related zone is identified and associated with the at least one risk curve and includes at least one interval of the day in which one or more single-symptom-specific risk curves indicate a potential glycemic dysfunction.

[0015] Implementations can also include some or all of the following features. The method further includes determining a time period of the at least one single specific symptom risk curve that can be mitigated by adjusting a parameter or timing of insulin therapy. Determining the at least one therapy-related region is based on predicting at least one of a candidate behavior change or a therapy change to reduce a single symptom glucose risk without subsequently exacerbating another symptom. Determining the importance value of the at least one therapy-related region includes prioritizing regions that are of significant therapeutic interest or resolvable. Determining the importance value of the at least one therapy-related region includes assessing a size of the risk. Determining the importance value of the at least one therapy-related region includes considering at least one of a time of day or a proximity of one risk curve to another risk curve. The importance value is a peak value of the at least one single specific symptom risk curve. Deriving the at least one single specific symptom risk curve is based on data that is above a certain level of confidence. The output information includes at least one of outputting a number, an alphanumeric, or a graphical information. The output information includes outputting at least one of a risk curve or a therapy-related region to a connected insulin pump or insulin pen, or to a bolus calculator. The output information includes outputting a graphical representation of at least one of a risk curve or a therapy-related region or a relative importance of at least one of a risk curve or a therapy-related region.

[0016] In one implementation, a system includes a glucose risk analyzer configured to derive at least one single specific symptom risk curve, a therapy region evaluator configured to determine at least one therapy-related region associated with the at least one single specific symptom risk curve, a region importance quantifier configured to determine an importance value of the at least one therapy-related region, and a therapy region report generator configured to output information based on the importance value.

[0017] Implementations can include some, or all, of the following features. The blood glucose risk analyzer is further configured to receive glucose data, wherein the at least one single specific symptom risk curve is derived using the glucose data. The glucose data includes at least one of CGM (continuous glucose monitoring) readings, trustworthiness readings assigned to CGM values, self-monitoring blood glucose readings, or retrospective calibrated or corrected CGM readings. The glucose data includes a time period in at least one week. The at least one single specific symptom risk curve describes hypoglycemic risk or hyperglycemic risk as a function of time of day using the glucose data. Deriving the at least one single specific symptom risk curve includes evaluating steepness (first and second derivatives of the curve), frequency, severity, curvature, mean values in the curve over 24 hours, or variability of the curve (mean and standard deviation). The at least one single specific symptom risk curve is indicative of glycemic dysfunction based on CGM signals over a selected time period, indicative of recurrent time windows characterized by a predetermined severity and frequency of hypoglycemia or hyperglycemia over the selected time period. The at least one single specific symptom risk curve represents at least one of hypoglycemia separated from hyperglycemia or at least one of hyperglycemia separated from hypoglycemia. Determining the at least one treatment-related region includes identifying the at least one treatment-related region from the at least one single specific symptom risk curve. The at least one treatment-related region is an interval in 24 hours of a day in which the patient's BG data indicates that at least one of the patient's insulin basal rate or dose or bolus strategy is systematically not optimal. The at least one treatment-related region is identified and associated with the at least one risk curve and includes at least one interval in a day in which one or more single specific symptom risk curves indicate potential glycemic dysfunction.

[0018] Implementations can also include some or all of the following features. The therapy area evaluator is further configured to identify time periods of the at least one single specific symptom risk profile that can be mitigated by adjusting a parameter or timing of insulin therapy. Determining the at least one therapy-related area is based on predicting at least one of a candidate behavior change or a therapy change to reduce a single symptom glucose risk without subsequently exacerbating another symptom. Determining the at least one therapy-related area includes prioritizing areas of significant therapy interest or resolution. Determining the at least one therapy-related area includes evaluating a size of a risk. Determining the at least one therapy-related area includes considering at least one of a time of day or a proximity of one risk profile to another risk profile. The importance value is a peak value of the at least one single specific symptom risk profile. Deriving the at least one single specific symptom risk profile is based on data above a certain level of confidence. The output information includes at least one of outputting a number, an alphanumeric, or a graphical information. The output information includes outputting at least one of a risk profile or a therapy-related area to a connected insulin pump or insulin pen, or to a bolus calculator. The output information includes outputting a graphical representation of at least one of a risk profile or a therapy-related area or a relative importance of at least one of a risk profile or a therapy-related area.

[0019] In one implementation, a system includes at least one processor; a non-transitory computer readable medium comprising: instructions that, when executed by the at least one processor, cause the system to: derive at least one single specific symptom risk profile; determine at least one therapy-related area related to the at least one single specific symptom risk profile; determine an importance value of the at least one therapy-related area; and output information according to the importance value.

[0020] Implementations can include some, or all, of the following features. The system also includes instructions that, when executed by the at least one processor, cause the system to receive glucose data, wherein the at least one single symptom-specific risk curve is derived using the glucose data. The glucose data includes at least one of CGM (continuous glucose monitoring) readings, trustworthiness readings assigned to CGM values, self-monitoring blood glucose readings, or retrospective calibrated or corrected CGM readings. The glucose data includes a time period in at least one week. The at least one single symptom-specific risk curve describes hypoglycemic risk or hyperglycemic risk as a function of time of day using the glucose data. Deriving the at least one single symptom-specific risk curve includes evaluating steepness (first and second derivatives of the curve), frequency, severity, curvature, mean values in the curve over 24 hours, or variability of the curve (mean and standard deviation). The at least one single symptom-specific risk curve is based on CGM signals indicating glycemic dysfunction over a selected time period, indicating a recurrent time window characterized by a predetermined severity and frequency of hypoglycemia or hyperglycemia over the selected time period. The at least one single symptom-specific risk curve represents at least one of hypoglycemia separated from hyperglycemia or at least one of hyperglycemia separated from hypoglycemia. Determining the at least one therapy-related zone includes identifying the at least one therapy-related zone from the at least one single symptom-specific risk curve. The at least one therapy-related zone is an interval over 24 hours in a day in which the patient’s BG data indicates that at least one of the patient’s insulin basal rate or dose or bolus strategy is systematically not optimal. The at least one therapy-related zone is identified and associated with the at least one risk curve and includes at least one interval in a day in which one or more single symptom-specific risk curves indicate potential glycemic dysfunction.

[0021] Implementations can also include some or all of the following features. The system further includes instructions that, when executed by the at least one processor, cause the system to identify a time period of at least one single specific symptom risk profile that can be mitigated by adjusting a parameter or timing of insulin therapy. Determining the at least one therapy-related region is based on predicting at least one of a candidate behavioral change or a therapy change to reduce a single symptom glucose risk without subsequently exacerbating another symptom. Determining the at least one therapy-related region includes prioritizing regions that are of high therapeutic significance or resolvable. Determining the at least one therapy-related region includes assessing a size of a risk. Determining the at least one therapy-related region includes considering at least one of a time of day or a proximity of one risk profile to another risk profile. The importance value is a peak of the at least one single specific symptom risk profile. Deriving the at least one single specific symptom risk profile is based on data that is above a certain level of confidence. The output information includes at least one of outputting a number, an alphanumeric, or a graphical information. The output information includes outputting at least one of a behavioral change or a therapy change to a therapy time region to reduce a single symptom in a time window. The output information includes outputting information to a connected insulin pump or insulin pen, or to a bolus calculator. The output information includes outputting a graphical representation of at least one of a risk profile or a therapy-related region or a relative importance of at least one of a risk profile or a therapy-related region.

[0022] In one implementation, a method includes receiving glucose and insulin data; identifying therapy improvement opportunities using the glucose and insulin data; determining candidate changes to insulin therapy; evaluating improvements in therapy risk based on the candidate changes; quantifying improvements of the candidate changes; and outputting at least one of the candidate changes based on the improvements.

[0023] Implementations can include some or all of the following features. Glucose and insulin data is received from at least one of a patient or a connected system or device. Identifying a therapy improvement opportunity includes receiving a user selection of at least one of a meal time, a time of day, or a parameter setting. The parameter setting is a carbohydrate ratio. The candidate change to insulin therapy includes a percentage increase or decrease to a bolus therapy or a basal therapy. The candidate change to the insulin therapy includes a change to an insulin delivery parameter related to the bolus therapy or the basal therapy. The candidate change is according to a carbohydrate ratio, a correction factor, a basal rate, or a graph. The candidate change includes basal sensitivity. The candidate change includes a percentage change in a basal dose or a bolus dose in a therapy zone. Quantifying an improvement in the candidate change includes comparing risk profile values. Outputting at least one of the candidate changes based on the improvement includes outputting a candidate change that provides an optimized risk profile. Outputting at least one of the candidate changes includes providing an output in a form of a graph to a user interface or a connected device, where the graph illustrates at least one of the candidate change or the optimized risk output. The connected device includes a bolus calculator. A natural language processor outputs to describe the candidate change and the optimized risk outcome. The output identifies an optimized therapy zone or set of zones.

[0024] In one implementation, a system includes a therapy improvement identifier configured to evaluate a patient's reconciled glucose and insulin data to identify a zone in a patient's diabetes management routine in which to make a therapy optimization, and to generate a therapy improvement; a relative insulin optimizer configured to propose a change to therapy, evaluate an impact of the change, and quantify an improvement associated with the change; and a relative insulin optimizer report generator to provide an output.

[0025] Implementations can include some or all of the following features. The relative insulin optimizer includes a change proposer configured to propose a change to insulin therapy, an impact evaluator configured to evaluate an impact of a candidate therapy change by estimating an impact on a risk profile of historical glucose values, and an improvement quantifier configured to quantify an improvement in the candidate therapy change. The change proposer is further configured to propose the change as a percentage change to at least one of a basal or a bolus in a time window. The improvement quantifier is configured to quantify the improvement in the candidate therapy change based on a percentage improvement or change in a glycemic outcome metric. The relative insulin optimizer report generator is configured to output the candidate therapy change to a user. The user is one of a clinician, a patient, or a connected device or system. The therapy improvement identifier includes a user selection of a therapy to be optimized or a time of day. The user is a patient or a clinician. The therapy improvement is identified by an algorithm.

[0026] In one embodiment, a system comprises: at least one processor; and a non-transitory computer-readable medium comprising: instructions that, when executed by the at least one processor, cause the system to: receive glucose and insulin data; identify a therapy improvement opportunity using the glucose and insulin data; determine a candidate change to an insulin therapy; evaluate an improvement in therapy risk according to the candidate change; quantify the improvement in the candidate change; and output at least one candidate change based on the improvement.

[0027] Implementations can include some or all of the following features. The glucose and insulin data is received from at least one of a patient or a connected system or device. Identifying a therapy improvement opportunity includes receiving a user selection of at least one of a meal time, a time of day, or a parameter setting. The parameter setting is a carbohydrate ratio. The candidate change to an insulin therapy includes a percentage increase or decrease to a bolus therapy or a basal therapy. The candidate change to the insulin therapy includes a change to an insulin delivery parameter related to the bolus therapy or the basal therapy. The candidate change is according to a carbohydrate ratio, a correction factor, a basal rate, or a graph. The candidate change includes basal sensitivity. The candidate change includes a percentage change in a basal dose or a bolus dose in a therapy zone. Quantifying the improvement in the candidate change includes comparing risk profile values. Outputting at least one of the candidate changes based on the improvement includes outputting a candidate change that provides an optimized risk profile. Outputting at least one of the candidate changes includes providing an output in the form of a graph to a user interface or a connected device, wherein the graph illustrates at least one of the candidate change or the optimized risk output. The connected device includes a bolus calculator. A natural language processor outputs to describe the candidate change and the optimized risk result. The output identifies an optimized therapy zone or set of zones.

[0028] In one embodiment, a method comprises: receiving at least one of glucose data, insulin data, or other diabetes-related data for a patient; identifying a therapy improvement opportunity using the at least one of glucose data, insulin data, or other diabetes-related data; determining an insulin dosing strategy for the patient; scoring the insulin dosing strategy for patient adherence; optimizing the insulin dosing strategy; and providing an output to a user including optimized insulin strategy parameters.

[0029] Implementations can include some or all of the following features. The other diabetes-related data includes at least one of meal information, specific meals, meal times, meal sizes, carbohydrate estimates, ingredient information, or exercise information. The glucose data, insulin data, or other diabetes-related data are received from at least one of the patient or a connected system or device. Identifying the therapy improvement opportunity includes receiving a selection by the user of at least one of a meal time, a time of day, or a parameter setting. The parameter setting is a carbohydrate ratio. The insulin dosing strategy includes a diabetes management or insulin strategy implemented by the patient in practice, as determined from at least one of the patient's glucose data, insulin data, or other diabetes-related data. Optimizing the insulin dosing strategy includes determining whether the patient adheres to a known insulin strategy and analyzing the impact of a percentage change in a determined insulin strategy parameter. The user is at least one of a clinician, the patient, or a connected device or system. The natural language processor outputs to describe a candidate change and a risk outcome of the optimization. Providing the output includes outputting to a user interface or a connected device in the form of a chart illustrating an optimized insulin strategy parameter. The connected device includes a bolus calculator.

[0030] In one implementation, a system includes a therapy improvement identifier configured to evaluate a patient's reconciled glucose and insulin data to identify areas of ongoing therapy optimization in the patient's diabetes management routine and generate a therapy improvement, an insulin strategy optimizer configured to determine whether the patient adheres to a known insulin strategy and analyze the impact of a percentage change in a determined insulin strategy parameter, and a therapy identification optimization report generator providing an output.

[0031] Implementations can include some or all of the following features. The insulin strategy optimizer includes: an insulin strategy identifier configured to identify a diabetes management or insulin strategy implemented in practice by a patient as determined from reconciled glucose and insulin data; a compliance scorer configured to quantify a patient's compliance with the identified insulin strategy; and the insulin strategy within the identified behavior is optimized by the insulin strategy optimizer configured to optimize the identified insulin strategy. The diabetes data includes insulin data and meal data. The insulin strategy identifier is configured to identify a dosing pattern and describe the identified patient's insulin strategy based thereon. The insulin strategy is a behavioral methodology applied by the patient in diabetes management including at least one of use of an insulin pump, multiple daily injections, or a type 2 therapy. The compliance scorer is configured to generate a score computed for a degree of compliance of the patient with the identified insulin strategy. The insulin strategy within the identified behavior is optimized by the insulin strategy optimizer configured to iteratively propose a percentage change to a strategy parameter in a selected therapy region or group of regions. The output includes optimized insulin strategy parameters. The therapy identification optimization report generator is configured to output a candidate therapy change to a user. The user is at least one of a clinician, a patient, or a connected device or system. The natural language processor outputs to describe the candidate change and optimized risk outcomes.

[0032] In one implementation, a system includes: at least one processor; a non-transitory computer-readable medium comprising: instructions that, when executed by the at least one processor, cause the system to: receive at least one of glucose data, insulin data, or other-diabetes-related data of a patient; identify a therapy improvement opportunity using at least one of the glucose data, the insulin data, or the other-diabetes-related data; determine an insulin dosing strategy of the patient; score the insulin dosing strategy for patient compliance; optimize the insulin dosing strategy; and provide an output including optimized insulin strategy parameters to a user.

[0033] Implementations can include some or all of the following features. The other diabetes-related data includes at least one of meal information, specific meals, meal times, meal sizes, carbohydrate estimates, ingredient information, or exercise information. The glucose data, insulin data, or other diabetes-related data are received from at least one of the patient or a connected system or device. Identifying the therapy improvement opportunity includes receiving a selection by the user of at least one of a meal time, a time of day, or a parameter setting. The parameter setting is a carbohydrate ratio. The insulin dosing strategy includes a diabetes management or insulin strategy implemented by the patient in practice, as determined from at least one of glucose data, insulin data, or other diabetes-related data of the patient. Optimizing the insulin dosing strategy includes determining whether the patient is following a known insulin strategy and analyzing an impact of a percentage change in determined insulin strategy parameters. The user is at least one of a clinician, the patient, or a connected device or system. The natural language processor outputs to describe a candidate change and a risk outcome of the optimization. Providing the output includes outputting to a user interface or a connected device in a form of a chart illustrating optimized insulin strategy parameters. The connected device includes a bolus calculator.

[0034] The summary is provided to present a simplified summary of some of the concepts in a form that is brief enough to introduce a small number of concepts, which are further described in the detailed description below. The summary is not intended to identify key or essential features of the claimed subject matter, nor is it intended for use in limiting the scope of the claimed subject matter. BRIEF DESCRIPTION OF DRAWINGS

[0035] The foregoing summary, as well as the following detailed description of the exemplary implementations, will be better understood when read in conjunction with the appended drawings. For the purpose of illustrating the implementations, exemplary constructions of the implementations are shown in the drawings; however, the implementations are not limited to the specific methods and means disclosed. In the drawings:

[0036] FIG. 1 is a high level functional block diagram of an implementation of the present invention;

[0037] FIG. 2 is a system diagram of an implementation of a therapy zone identifier;

[0038] FIG. 3 is a flow diagram of a method of identifying a therapy zone with potential for improved glycemic outcomes;

[0039] FIG. 4 is a chart illustrating one example of superimposing multi-day CGM data over a 24 hour time window from one type 1 diabetes patient;

[0040] FIG. 5 is a graph illustrating one example of a single specific symptom risk curve based on a glycemic risk quantifier;

[0041] FIG. 6 is a chart illustrating one example of a recognized treatment zone derived from a single specific symptom risk curve;

[0042] FIG. 7 is a system diagram implementing a relative insulin optimizer;

[0043] FIG. 8 is a flowchart of a method for selecting, evaluating, and affecting candidate insulin therapy changes for a patient;

[0044] FIG. 9A and FIG. 9B are charts showing glucose and insulin data, respectively, for a patient over a period of time;

[0045] FIG. 10 is a visualization chart showing CGM data overlaid over a 24-hour period per day;

[0046] FIG. 11 is a chart showing the reliability of CGM and insulin data for the data shown in FIG. 9A , FIG. 9B and FIG. 10 .

[0047] FIG. 12 is a chart showing a risk curve for a patient over a time window in some embodiments;

[0048] FIG. 13A and FIG. 13B are charts showing operable and inoperable risk, respectively, as a function of time of day;

[0049] FIG. 14 is a chart of one embodiment of the present invention showing a blood glucose risk curve and corresponding treatment zones determined from the risk curves described with reference to FIG. 12 , FIG. 13A and FIG. 13B .

[0050] FIG. 15 is a system diagram implementing an insulin strategy optimizer;

[0051] FIG. 16 is a flowchart of a method for identifying, scoring, and optimizing a patient’s insulin strategy;

[0052] FIG. 17 is a chart showing CGM data for a patient over a 10-day period;

[0053] FIG. 18 is a chart illustrating a risk curve function output from an analysis of the 10-day CGM data shown in FIG. 17 ;

[0054] FIG. 19 is a graph illustrating a comparison of risk curves from historical data and from a playback simulation;

[0055] FIG. 20 illustrates a risk curve from a playback;

[0056] FIG. 21 is a graph showing two hyperglycemic risk curves;

[0057] FIG. 22 is a graph showing two corresponding hyperglycemic treatment zones;

[0058] FIG. 23 is a graph showing optimized bolus and basal parameters;

[0059] FIG. 24 and FIG. 25 shows a graph showing an exemplary risk curve; and

[0060] FIG. 26 shows an exemplary computing environment in which exemplary embodiments and aspects can be implemented. DETAILED DESCRIPTION

[0061] The claimed subject matter is described with reference to the drawings, wherein like reference numerals are used to refer to like elements throughout. In the following description, for purposes of explanation, numerous specific details are set forth in order to provide a thorough understanding of the claimed subject matter. It can be apparent, however, that the claimed subject matter can be practiced without these specific details. In other instances, structures and devices are shown in block diagram form in order to facilitate describing the claimed subject matter.

[0062] FIG. 1 is a high level functional block diagram 100 of an embodiment of the present invention. A processor 130 is in communication with an insulin device 110 and a glucose monitor 120. The insulin device 110 and glucose monitor 120 are in communication with a patient 140 to deliver insulin and monitor glucose levels of the patient 140, respectively. The processor 130 is configured to perform the calculations and other operations and functions described further herein. The insulin device 110 and glucose monitor 120 can be implemented within a single device or across multiple devices, either as separate devices or as a single device. The processor 130 can be implemented locally to the insulin device 110, glucose monitor 120, or as a standalone device (or any combination of two or more of the insulin device 110, glucose monitor 120, or standalone device). The processor 130 or a portion of the system can be disposed remotely, e.g., within a server or cloud-based system.

[0063] Examples of insulin devices, such as insulin device 110, include insulin syringes, external pumps, and patch pumps that deliver insulin, typically into subcutaneous tissue. Insulin device 110 also includes devices that deliver insulin in different ways, e.g., insulin inhalers, insulin jet injectors, intravenous infusion pumps, and implantable insulin pumps. In some embodiments, a patient will use more than two insulin delivery devices in combination, e.g., a syringe to inject long-acting insulin and an inhaler to use before meals. In other embodiments, these devices can deliver other drugs that help control glucose levels, e.g., glucagon, pramlintide, or glucagon-like peptide-1 (GLP-1).

[0064] Examples of glucose monitors, such as glucose monitor 120, include continuous glucose monitors that record glucose values at prescribed intervals, e.g., 1, 5, or 10 minutes, etc. These continuous glucose monitors, for example, can use electrochemical or optical sensors that are inserted transcutaneously, implanted completely, or measured non-invasively of tissue. Examples of glucose monitors, such as glucose monitor 120, also include devices that periodically extract blood or other fluid to measure glucose, e.g., intravenous blood glucose monitors, micro- perfusion sampling, or periodic finger pricking. In some embodiments, glucose readings are provided in near real-time. In other embodiments, glucose readings determined by a glucose monitor can be stored on the glucose monitor itself for later retrieval.

[0065] Insulin device 110, glucose monitor 120, and processor 130 can be implemented using various computing devices, such as smartphones, desktop computers, laptop computers, and tablet computers. Other types of computing devices can be supported. Suitable computing devices are illustrated in FIG. 26 as computing device 2600 and cloud-based applications.

[0066] Insulin device 110, glucose monitor 120, and processor 130 can communicate over a network. The network can be various types of networks, including public switched telephone networks (PSTNs), cellular telephone networks, and packet-switched networks (e.g., the Internet). Although only one insulin device 110, one glucose monitor 120, and one processor 130 are illustrated in FIG. 1 there is no limit to the number of insulin devices, glucose monitors, and processors that can be supported. Activity monitor 150 and / or smartphone 160 can also be used to collect or gather meal and / or activity data from and about patient 140 and provide the meal and / or activity data to processor 130.

[0067] The processor 130 can execute an operating system and one or more applications. The operating system can control the applications executed by the insulin device 110 and / or the glucose monitor 120 and control the manner in which the applications interact with the one or more sensors, or other ancillary means of the insulin device 110 and / or the glucose monitor 120.

[0068] The processor 130, in some embodiments, receives data from the insulin device 110 and the glucose monitor 120, as well as from the patient 140, and can be configured and / or used to perform one or more of the calculations, operations, and / or functions described further herein.

[0069] FIG. 2 is a system diagram of an embodiment of a therapy zone identifier 210, which provides an overall framework for the therapy zone identification system described herein. As shown, the system uses blood glucose data 205, which can be any diabetes data associated with a host such as a human, and can include CGM data only, BG (blood glucose) data, or other glucose or diabetes related data, depending on the embodiment.

[0070] The therapy zone identifier 210 identifies intervals in a 24 hour day in which the patient’s BG data indicates that the patient’s medication dosing / strategy needs improvement, e.g., the patient’s basal rate / dose of insulin and / or bolus strategy is not optimal systemically, or puts the patient at risk for diabetes complications. In embodiments, the therapy zone identifier 210 uses only blood glucose data (e.g., does not use insulin data), e.g., data from the glucose monitor 120.

[0071] The blood glucose risk analyzer 220 quantifies changes in low blood glucose, high blood glucose, or both (variable risk) average over multiple days by quantifying the risk as a pattern.

[0072] The therapy zone evaluator 230 evaluates the risk curve and evaluates more than one time window in which dosing can be adjusted to systemically improve blood glucose outcomes related to more than one risk curve feature. The time of day is often an important factor to consider.

[0073] The zone importance quantifier 240 quantifies the relative importance of therapy zones.

[0074] The report generator 250 outputs 260 in the form of numeric, alphanumeric, and / or graphical information based on the quantification of therapy related zone importance.

[0075] FIG. 3is a flowchart of a method 300 of identifying a treatment area with potential to improve glycemic outcomes; the method 300 of identifying a treatment area with potential to improve glycemic outcomes includes evaluating a large historical data set to identify a treatment area with the fastest matching glycemic dysfunction.

[0076] In step 310, glucose data is received (e.g., from the glucose monitor 120, the patient 140, the activity monitor 150, and / or the smartphone 160, in some embodiments). These data typically include glucose level measurements, such as: CGM readings, trustworthiness readings assigned to CGM values, self-monitoring blood glucose readings (blood glucose meter), retrospective calibrated or corrected CGM readings, etc. The glucose data typically includes a time period selected from at least one week; however, larger data sets can provide longer term patterns. The glucose data provides a retrospective analysis of CGM as a function of time in a day (e.g., over 24 hours in a day). FIG. 4 is a chart 400 illustrating one example of multiple days of CGM data superimposed over a 24-hour time window from one type 1 diabetic patient;

[0077] In step 320, one or more single specific symptom risk curves are derived. More specifically, glycemic dysfunction single symptom curves are evaluated, using the glucose data received from step 310, to describe hypoglycemic risk or hyperglycemic risk as a function of time in a day. The risk curves can be derived using any known glycemic risk quantifier, such as described in US 2018 / 0020988 entitled "Methods, Systems, and Computer-Readable Media for Assessing Actionable Glycemic Risk" by inventor Stephen D. Patek, which is incorporated by reference herein in its entirety. For example, the glycemic risk quantifier can evaluate steepness (first and second derivatives of the curve), frequency, severity, curvature, average in the curve over 24 hours, variability of the curve (average and standard deviation), etc., all of which can be meaningful in identifying areas. The glycemic risk quantifier includes any risk-related score or value, such as the low blood glucose index (LBGI) and the high blood glucose index (HBGI). The glycemic risk quantifier can quantify changes in hypoglycemia, hyperglycemia, and / or both in the historical data. In some embodiments, the method resolves various types of risk at the same time in a day or corresponding to adjacent times in a day experienced all hyperglycemia and hypoglycemia in different days in the historical record.

[0078] In some embodiments, the one or more single specific symptom risk curves are indicative of glycemic dysfunction based on CGM signals over a selected time period, the analysis based on the one or more single specific symptom signals is indicative of a circulatory time window (cycling over the same time period within each 24 hour period), the time window characterized by a predetermined severity and frequency of the single symptom (e.g., hypoglycemia or hyperglycemia) over the selected time period.

[0079] In some embodiments, the time of day is identified as a consistent pattern in hypoglycemia, hyperglycemia, and / or a combination of the two.

[0080] In some embodiments, the single specific symptom risk curve is representative of a) at least one of hypoglycemia separated from hyperglycemia or b) at least one of hyperglycemia separated from hypoglycemia. In an exemplary embodiment, the single symptom is hypoglycemia, the quantifier can evaluate a risk index value below a threshold and a duration greater than a threshold, e.g., glucose average less than or equal to 70 mg / dl for 30 minutes. In an exemplary embodiment, the single symptom is hyperglycemia, the quantifier can evaluate a risk index value above a threshold and a duration greater than a threshold, e.g., glucose average greater than or equal to 180 mg / dl for 2 hours.

[0081] Other single specific symptom risk curves can be defined and evaluated, e.g., a combination of health data including diet and exercise as well as cognitive symptoms with an Alc greater than a threshold, which can be measured using techniques known in the art.

[0082] Preferably, the single specific symptom signal is determined over a single time period within the 24 hours of a day to avoid mixing glycemic dysfunction. However, other forms of segmentation can also be evaluated, e.g., weekdays versus weekends. Generally, the single specific symptom signal refers to a signal during a particular time period in which there is no concurrent single specific symptom signal of a different type of predetermined severity or frequency. For example, in which a single hypoglycemic risk curve can not have concurrent hyperglycemia above a value / duration threshold. FIG. 5 is a chart 500 illustrating one example of a single specific symptom risk curve 510 based on a risk quantifier 520 provided by a glycemic risk quantifier, derived from FIG. 4 glucose data for a patient as shown. The risk curve 510 is identified based on the risk magnitude shown by the risk quantifier 520.

[0083] In step 330, one or more treatment-related regions associated with the one or more single specific symptom risk curves are evaluated (e.g., determined). In other words, the treatment regions are identified from the risk curves.

[0084] Generally, in the scope of insulin therapy, the treatment zone is a time interval in the 24 hours for which the patient's BG data indicates (e.g., is indicative of) that the patient's insulin basal rate / dose and / or bolus strategy is not optimal systemically. In other scopes, with respect to other treatment methods such as glucagon, type 2 drugs, or even food, there can also be corresponding evaluated treatment zones as can be appreciated by those skilled in the art.

[0085] In some embodiments, the treatment zone is identified that is associated with risk curves that are considered to be a time interval in the day that is indicative of potential glycemic dysfunction for one or more single specific symptom risk curves. In some cases, the identified treatment zone is associated with the presence of only one symptom, e.g., a single time period in the day that is at risk for hypoglycemia (or conversely hyperglycemia risk). In such cases, the treatment zone can be identified as a time window that is distinct from (e.g., expanded in, separated from, adjacent to, and / or superimposed on) the associated glycemic risk curve as a time window in which a root cause behavior / treatment change can be implemented to mitigate the glycemia, e.g., to temper the corresponding hypoglycemia risk curve (or conversely, the hyperglycemia risk curve) by systemically reducing insulin for the interval time period in the day (the treatment zone).

[0086] In other cases, the treatment zone evaluator 230 can associate more than one treatment-related single-symptom risk curve with more than one treatment zone. For example, a patient's risk curve can indicate (1) a risk of hyperglycemia occurring within a particular time interval of the day (i.e., a hyperglycemia risk curve), and a period of hypoglycemia risk (i.e., a subsequent related hypoglycemia risk curve) or a period of unresolved risk (i.e., historical data indicates a subsequent related interval of both hyperglycemia and hypoglycemia) in a subsequent time interval, and the corresponding determined treatment zone (or zones) would represent a period (or periods) of the day during which multiple corresponding risk curves can be mitigated by adjusting parameters or timing of insulin therapy (e.g., by more effectively addressing the initial source of hyperglycemia to avoid conditions that can promote hypoglycemia in a later time interval). In such cases, the related multiple risk curves can be contiguous, or can be separated by a period of time (an interval of the day) that represents no glycemic risk. By this means, adjacent or subsequent zones can be combined by virtue of their relatedness and addressed by treatment adjustments in a single treatment zone. For example, when hypoglycemia patterns consistently follow hyperglycemia, the risk curves can be addressed by treatment in relation to a single treatment zone, sometimes referred to in some embodiments as a single treatment zone. The prior art fails to consider the possibility of addressing multiple related glycemic risks by treatment adjustments within one or more composite treatment zones. In contrast, the systems and methods described herein extract treatment approaches by relating risk to treatment zones.

[0087] FIG. 6 is a chart illustrating one example of an identified treatment zone derived from a single specific symptom risk curve 600 shown in FIG. 5 glucose data from the patient CGM data shown in FIG. 4

[0088] In some embodiments, the evaluation of the risk curve that determines the treatment zone is not simply a threshold, for example, but can be a steepness (first and second derivatives of the curve), frequency, severity, curvature, average value of the curve over 24 hours, variability of the curve (average and standard deviation), etc., all of which can be meaningful in identifying a zone.

[0089] In some embodiments, a treatment zone is evaluated by predicting candidate behaviors and / or treatment changes to reduce single-symptom glycemic risk without increasing subsequent risk of another symptom. In other words, a treatment zone can be considered a zone in which treatment is made without risk of negatively affecting symptoms in adjacent time windows. In some cases, a feature of a treatment zone can be minimal symptom-free, mixed-symptom, or different single-symptom signaling.

[0090] ​In all cases, each determined treatment region is a time interval in the day that overlaps with and can include a portion of the time window of the associated risk profile. The one or more treatment regions represent the underlying treatment cause of the adjacent risk profile and can be assessed by looking at the time window immediately preceding and can overlap with the risk profile(s). In summary, the identified treatment regions define a time window (i.e., an interval in the day) on which the resolvable portion of the glycemic risk profile can be mitigated, as described in more detail herein.

[0091] In step 340, the importance of the treatment-related regions is quantified. In some embodiments, an importance value of the treatment-related regions is determined. Thus, the importance of the one or more treatment-related regions is quantified by a region importance quantifier. The region importance quantifier prioritizes regions that are more important or resolvable in terms of treatment. In some embodiments, the quantifier assesses the size of the risk and can also take into account the time of day and / or the proximity of one risk profile to another (e.g., in some cases, two different risk profiles are correlated with each other in that one affects the other due to not being independent of each other). The quantifier can use the mathematical function of the risk profile from step 320, e.g., the peak of the glycemic risk profile, where in an exemplary embodiment, the peak can be equated to the importance of the treatment region.

[0092] In embodiments where only glucose data is available, the quantifier can derive from the glycemic risk. However, if additional data is available, e.g., insulin, meals, and exercise, the quantifier can also take these into account in the assessment. In some embodiments, a priority can be assigned to the treatment regions based on a preference for solving the problem. For example, the system or user can assign importance between the bolus problem and the basal problem, or vice versa. In some embodiments, the region importance is informed by other factors, e.g., the current basal-bolus insulin ratio, e.g., 50% of basal is nearly too high. In some embodiments, as described in more detail elsewhere herein, insulin data is available and / or an insulin strategy is determined, which can be used to differentiate the priority of the treatment-related regions.

[0093] In some embodiments, as described in more detail elsewhere herein (e.g., using a playback analysis), the quantifier assesses the opportunity to reduce the risk in each of more than one region based on a candidate change to treatment, where the region importance quantifier prioritizes treatment regions based on the opportunity to reduce the risk profile of one region compared to another region.

[0094] In some embodiments, the systems and methods described herein can determine and apply an analysis of the trustworthiness of data prior to determining one or more data processing steps described herein. For example, the risk profile derived by the blood glucose risk analyzer is based on data above a certain trustworthiness level. In other words, trustworthiness enters the objective function of the risk profile. An example of trustworthiness analysis and application is described in U.S. Application No. 17 / 096,785, entitled “Joint State Estimation Prediction Evaluating Discrepancies of Predictive Data from Corresponding Received Data,” filed November 12, 2020, by inventor Stephen D. Patek, which is incorporated by reference herein in its entirety.

[0095] In step 350, numerical, alphanumeric, and / or graphical information is outputted based on the quantified importance of the therapy-related regions. Thus, the therapy region report generator outputs numerical, alphanumeric, and / or graphical information based on the quantified importance of the therapy-related regions. In one embodiment, the visualized relevant regions are simply outputted onto the report. In other embodiments, the behavior and / or therapy changes to therapy region times are identified and outputted to reduce one of the symptoms in the identified circadian time window.

[0096] In one example, when a hyperglycemic risk profile occurs at noon, e.g., the risk profile value is above the hyperglycemic risk threshold for the noon time period and there is no hypoglycemia around it, a time window from immediately (and possibly overlapping) at noon to 4pm can be evaluated. In this case, the time of day can be selectively inputted and a lookup table, decision tree, etc. can be used to analyze possible root causes, indicating two reasons for the hyperglycemic risk: 1) an under bolus at lunch time (e.g., possibly no bolus parameter, the patient can have underestimated the carbohydrate intake, or the patient can have not taken a bolus at all, etc.); or 2) an under basal. Given the two possible root causes in this example, the report can describe the two possible therapy problems to address 1) or 2).

[0097] Consider another example, where a hypoglycemic risk is determined before noon each day and the relevant therapy regions are evaluated. Three possible outputs of root causes: 1) an over aggressive morning basal; 2) an over bolus; and / or 3) a bolus too late (carbohydrate intake too late).

[0098] In the example of a consistently low blood glucose following a high blood glucose, a single therapy zone with a causal link to the zone can be output. The report generator can flag a time of day where adjustments can be made, e.g., high blood glucose following low blood glucose, recommending a change in basal at a time of day or a change in bolus parameters affecting another time of day. Prior recommendations can be set based on user preferences / settings, quantifiers, or other insights such as described in more detail elsewhere herein (e.g., candidate changes or relative improvement in the patient’s insulin strategy). In a combination scenario, where a single zone represents a low blood glucose zone and a high blood glucose zone connected in terms of therapy, multiple combinations can be analyzed according to their impact on the blood glucose risk curve.

[0099] The output can be in the form of a report of a user interface, can be sent to a connected insulin pump or insulin pen, or can be input to a bolus calculator. For example, a therapy zone associated with a time of day where a change in the carbohydrate ratio of a meal bolus has been identified (and possibly confirmed / verified by the user), the bolus calculator can be automatically programmed with the new carbohydrate ratio. Other suggestions can be more behavioral in nature, e.g., suggesting a bolus at an earlier time (e.g., before a meal).

[0100] The report generator can generate and output a graph of the risk curve, therapy-related zones, and / or their relative importance for review. Additionally or alternatively, a natural language generator or text generator can be used to convey the risk curve, therapy-related zones, and / or their relative importance.

[0101] FIG. 7 is a system diagram of embodiments of the relative insulin optimizer 710, which provides a general framework for insulin optimization in the systems and methods described herein.

[0102] Glucose and insulin data are received (e.g., from the glucose monitor 120, the patient 140, the activity monitor 150, and / or the smart phone 160 in some embodiments) and collated as needed according to their source 705.

[0103] The therapy improvement identifier 707 evaluates the collated glucose and insulin data 705 to identify therapy optimization zones in the patient’s diabetes management program. In some embodiments, the identifier 707 can include a user selection from a clinician or patient (e.g., where the user identifies a particular therapy or time of day to optimize). The user can select a particular meal time (e.g., lunch), a particular time of day (e.g., upon waking in the morning), a particular setting (e.g., carbohydrate ratio), etc. Any parameter or behavior affecting insulin therapy can be selected. In some embodiments, the therapy improvement 709 is identified by an algorithm, e.g., with respect to FIG. 2The described therapy area identifier 210; however, as can be appreciated by those skilled in the art, other algorithms for identifying areas in need of improvement can also be used.

[0104] Based on the determined therapy improvement 709, the relative insulin optimizer 710 proposes candidate changes to therapy, evaluates the impact of these changes, and quantifies the improvement associated with these changes.

[0105] The change proposer 720 proposes candidate changes to insulin therapy, e.g., percentage changes to basal and / or bolus in a particular time window (e.g., an area or group of areas).

[0106] The impact evaluator 730 evaluates the effect of the candidate therapy changes by evaluating the impact on the risk profile of historical glucose values.

[0107] The improvement quantifier 740 quantifies the improvement of the candidate therapy changes, e.g., based on the percentage improvement / change in the glycemic outcome metric.

[0108] The relative insulin optimizer report generator 750 provides output 760, e.g., outputting the candidate therapy changes to a user (clinician, patient, or connected device / system).

[0109] FIG. 8 is a flowchart of a method 800 of selecting, evaluating, and affecting candidate insulin therapy changes for a patient. The method receives glucose and insulin data from a patient and / or connected devices to run optimization algorithms to improve insulin therapy based on quantified improvements associated with candidate changes to insulin therapy. In one example, using glucose and insulin delivery data, using a replay prediction function, the impact of percentage changes in basal and / or bolus insulin in determined therapy areas is determined.

[0110] In step 810, glucose and insulin data is received from a patient and / or connected system / device (e.g., in some embodiments, from the glucose monitor 120, the patient 140, the activity monitor 150, and / or the smartphone 160). FIG. 9A and FIG. 9B is a graph 900, 950 showing glucose and insulin data, respectively, for a patient over about 54 days. FIG. 10 is a chart 1000 showing a visualization of all CGM data overlaid for each day of 24 hours. FIG. 11 is a graph 1100 showing the reliability of CGM and insulin data as a function of changes in time data for a day shown in the immediately above graph.

[0111] In step 820, a therapy improvement opportunity is identified as described in greater detail elsewhere herein. In some embodiments, therapy improvement opportunity identification can include user selection from a clinician or patient (e.g., where the user identifies a particular therapy or time of day to optimize). The user can select a particular meal time (e.g., lunch), a particular time of day (e.g., upon waking in the morning), a particular environment (e.g., carbohydrate ratio), or the like. Any parameter or behavior that affects insulin therapy can be selected. In some embodiments, the improvement is identified by an algorithm, such as the therapy zone identifier described in greater detail elsewhere herein, however, as will be appreciated by those skilled in the art, other algorithms for identifying areas in need of improvement are possible.

[0112] FIG. 12 is a graph 1200 illustrating a patient's risk curve over a window of time in some embodiments. The area above the zero line 1210 represents hyperglycemia, while the area below the zero line 1220 represents hypoglycemia as a function of time of day.

[0113] FIG. 13A and 13B are graphs 1300, 1350 illustrating operable and inoperable risk as a function of time of day, respectively. At about 13 hours (approximately 1 PM, after lunch), there is a peak in operable hyperglycemic risk, while at the same time of day, the level of inoperable risk is lower.

[0114] FIG. 14 is a graph 1400 of an embodiment of the present invention illustrating a blood glucose risk curve and corresponding therapy zone determined from the risk curves described with reference to FIG. 12 , 13A and 13B. Curve 1410 represents the risk curve and curve 1420 represents the corresponding therapy zone. In this example, the operable hyperglycemic risk from noon to 2 PM, indicating that from 10 AM to noon would be a candidate window for modification of insulin therapy.

[0115] In step 830, a candidate change to insulin therapy is proposed (e.g., determined) for the therapy zone. The candidate change to insulin therapy can include a percentage increase or decrease to a bolus or basal therapy and / or can include a change to an insulin delivery parameter related to a bolus or basal therapy.

[0116] Once the treatment zone is determined, candidate changes to therapy can be directly proposed based on, for example, the carbohydrate ratio therein, correction factors, basal rates, and / or profiles. Those skilled in the art will appreciate that any parameter used in diabetes management that impacts diabetes outcomes can be a candidate for change. The parameters can be specific to an insulin pump, bolus calculator, or any value related to insulin therapy, whether in type 1 or type 2 or single or multiple daily injection therapy, insulin pen therapy, insulin pump therapy, artificial pancreas therapy, beta cell therapy, and / or any aspect thereof.

[0117] In some implementations, basal sensitivity (percent change basal dose and / or basal rate) is a candidate change. In some embodiments, candidate changes can include a percent change to the treatment zone basal dose or bolus dose. Compound changes and / or combination changes can also be proposed, for example, two-factor sensitivity (e.g., differential recommendations for basal versus bolus). In some embodiments, time of day is also a factor that can be proposed for basal rate.

[0118] In step 840, an improvement in overall therapy risk based on the candidate change is evaluated (e.g., determined). In some embodiments, a playback prediction function, as described in U.S. Application No. 17 / 096,785, entitled “Joint State Estimation Prediction Evaluating Predictions Versus Differences in Corresponding Received Data,” by inventor Stephen D. Patek, filed November 12, 2020, which is incorporated herein by reference in its entirety, estimates the impact of the candidate change on historical glucose in the treatment zone, and runs a risk analysis function to determine a new risk profile based on the candidate change. For example, a percent change (5%, 10%, etc.) in historical bolus and / or basal rates is played back over the treatment zone, and the resulting risk profile is reevaluated.

[0119] In step 850, the relative improvement of the candidate change can be quantified. In one embodiment, the risk profile values are compared to quantify the improvement.

[0120] In step 860, the candidate change to insulin therapy that provides the optimized risk profile is output. The output can be in the form of a graph illustrating the candidate change and / or the optimized risk, output to a user interface or connected device, for example, a clinician report or connected bolus calculator. Additionally or alternatively, the output can be provided by a natural language processor to describe the candidate change and the optimized risk outcome. The output can identify the treatment zone or set of zones that are optimized, and in some embodiments, a text generator can be used to communicate the results.

[0121] Although this article has described and illustrated diabetes data associated with type 1 diabetes, the systems and methods used are applicable to type 2 diabetes in this article, for example, for basal titration acceleration to identify and assess the risk of more or less invasive type 2 injections (depending on drug type, drug dose, and / or injection time).

[0122] FIG. 15 This is a system diagram of an implementation of the insulin strategy optimizer 1510, which provides a general framework for insulin optimization in the systems and methods described herein.

[0123] Receive blood glucose and insulin data (e.g., from glucose monitor 120, patient 140, activity monitor 150, and / or smartphone 160) and organize them as needed into organized glucose and insulin data 1505, depending on their source.

[0124] Treatment improvement identifier 1507 evaluates the checked glucose and insulin data 1505 to identify areas for treatment optimization in the patient's diabetes management program. In some embodiments, identifier 1507 may be a user selection from a clinician or patient (e.g., where the user identifies a specific therapy or time of day to optimize). The user may select, for example, a specific meal time (e.g., lunch), a specific time of day (e.g., upon waking in the morning), a specific setting (e.g., carbohydrate ratio), etc. Any parameter or behavior affecting insulin therapy may be selected. In some embodiments, treatment improvement 1509 is identified by an algorithm, e.g., regarding... FIG. 2 The treatment area identifier 210 is described; however, as those skilled in the art will understand, other algorithms for identifying areas that need improvement are also possible.

[0125] Using records of glucose, insulin, and other data, the insulin strategy optimizer 1510 determines: (1) whether the patient adheres to a known insulin strategy (e.g., pre-meal bolus "functional insulin therapy" based on carbohydrate counts) and (2) the impact of percentage changes in determined insulin strategy parameters (e.g., carbohydrate ratio, correction factor, basal rate / dose, etc.). Compared to existing insulin optimizers that make assumptions about a patient's insulin strategy and / or require changes to the patient's behavioral insulin strategy, the system and method described herein identify a patient's current insulin dosing strategy from a multitude of different strategies to apply optimized insulin therapy to the patient's chosen behavioral insulin strategy, matching their actual reality-world strategy. Optimization focuses on how patients perceive their insulin therapy (i.e., strategy); allowing for more customized user recommendations, thereby improving adherence and improvement.

[0126] FIG. 16is a flowchart of a method 1600 of identifying, scoring, and optimizing a patient's insulin strategy. The method receives glucose, insulin, and other-diabetes related data, such as meal or exercise data, from the patient and / or connected devices. From the patient data, behavioral patterns can be identified to determine how the patient prefers to manage their diabetes. With an understanding of the patient's insulin dosing strategy, as described in greater detail herein, changes in the amount and / or timing of strategy-specific parameters can be evaluated and recommended.

[0127] In step 1610, glucose, insulin, and / or other-diabetes related data is received (e.g., from the glucose monitor 120, the patient 140, the activity monitor 150, and / or the smartphone 160, in some embodiments). The other-diabetes related data can include meal information, such as the specificity and / or timing of meals, the general or specific size of meals, carbohydrate estimates, ingredient information, etc. Additionally or alternatively, exercise information can be provided to include, for example, the type, duration, intensity, heart rate, calories burned, etc. of the exercise. The diabetes related data can be from connected devices and / or self-reported. As understood by one of skill in the art, this data can be reconciled with the glucose and insulin data.

[0128] In step 1620, a therapy improvement opportunity is identified, as described in greater detail herein. In some embodiments, the therapy improvement opportunity identification can include a user selection from a clinician or patient (e.g., where the user identifies a particular therapy or time of day to optimize). The user can select a particular meal time (e.g., lunch), a particular time of day (e.g., upon waking in the morning), a particular setting (e.g., carbohydrate ratio), etc. Any parameter or behavior that affects insulin therapy can be selected. In some embodiments, the improvement is identified by an algorithm, such as by the therapy area identifier 210, however, as can be appreciated by one of skill in the art, other algorithms for identifying areas in need of improvement are possible.

[0129] In step 1630, the patient’s insulin dosing strategy is determined. The insulin strategy identifier 1520 identifies the diabetes management / insulin strategy that the patient has implemented in practice as determined from the diabetes data (insulin data and meal data). The strategy identifier 1520 can include a series of questions for the patient or selections to be made by the patient. In some embodiments, the identifier 1520 identifies the dosing pattern and bases the patient’s insulin strategy on this. The identifier 1520 can use replay prediction functions (e.g., described in U.S. Application No. 17 / 096,785, entitled “Joint State Estimation Prediction to Evaluate Differences Between Predicted and Corresponding Received Data,” by inventor Stephen D. Patek, filed November 12, 2020, which is incorporated by reference herein in its entirety), among other data related to dosing to attempt to reproduce the patient’s historical insulin decisions in context.

[0130] By “insulin strategy” is meant the behavioral methodology that the patient applies in diabetes management, including the type of insulin pump use, multiple daily injections, and type 2 therapy, as can be appreciated by one of skill in the art.

[0131] As one example, the system and method identify the patient as a pump user and can further identify the type of pump use selected from: open loop (assess basal and / or bolus / timed), semi-closed loop, and closed loop (which can be further divided into, for example, artificial pancreas algorithm type A and artificial pancreas algorithm type B). Other insulin pump strategies can be appreciated by one of skill in the art, including programmable basal and bolus settings in terms of time and amount, and specific program or provider-recommended basal-bolus combination therapy.

[0132] As another example, the system and method identify whether the patient takes boluses, and if so, what behavioral strategy is associated with their routine bolus pattern. One bolus pattern used by some patients includes a fixed time bolus strategy throughout the day (i.e., bolus at a specific time of day), mealtime bolus, carbohydrate-counted bolus (e.g., where the patient regularly takes in different amounts of carbohydrates at most meals), non-carbohydrate-counted bolus (e.g., where the patient assesses (S / M / L)), pre-meal bolus (bolus first then titrate food), micro-bolus (e.g., average more than x number of boluses per day (where x is greater than 5, 6, 7, or more)), and the like, as can be appreciated by one of skill in the art.

[0133] Other examples include a sliding bolus, where the system and method identify whether the patient’s bolus in response to BG is above a certain range. Other concepts for insulin management can be considered as can be appreciated by others of skill in the art.

[0134] In step 1640, the strategy optionally scores the patient's adherence to the strategy. The insulin strategy scorer (adherence scorer) 1530 quantifies the patient's adherence to the determined insulin strategy, the strictness with which the patient adheres to the determined insulin strategy. A score is calculated based on the degree of adherence to the strategy by the patient, which can act as a gatekeeper to determine whether or not or how to proceed to the next step. For example, if the adherence score is above a given threshold level, the process proceeds to the next step, otherwise additional analysis, patient queries or feedback to a different algorithm (e.g., the FIG. 8 ) can be performed.

[0135] In step 1650, the determined strategy is executed for optimization. In other words, the optimization is confined to the range of diabetes management solutions preferred by the patient without requiring behavior modification. While not wishing to be bound by theory, by optimizing the insulin therapy confirmed by the patient behavior, a more efficient and effective therapy optimization can be achieved.

[0136] In general, the insulin strategy adapter (insulin strategy optimizer within the identify behavior 1540) adjusts the insulin optimization / recommendation based on the selected insulin strategy and / or based on the score associated with the insulin strategy. In some embodiments, the optimization iteratively proposes percentage changes in the strategy parameters in the selected therapy region or group of regions until a certain improvement and / or improvement through repeated changes is not visible in the quantification, the improvement can be quantified in a feedback loop fashion before.

[0137] In some embodiments, the optimization can be performed using a replay prediction function (described in U.S. Application No. 17 / 096,785, entitled "Joint State Estimation Prediction to Evaluate Differences in Predicted Data Versus Corresponding Received Data," by inventor Stephen D. Patek, filed November 12, 2020, which is incorporated by reference herein in its entirety.) to estimate the impact on historical BG. In these embodiments, the risk analysis function can be rerun on each iteration optimization and the percentage improvement / change in the BG outcome metric assessed until a certain criteria is met.

[0138] In 1660, the optimized insulin strategy parameters are output to a user interface or connected device, e.g., through a therapy identification optimization report generator 1550. In an embodiment, the therapy identification optimization report generator 1550 provides an output 1560, e.g., to a user (clinician, patient, or connected device / system) outputting candidate therapy changes. The output 1560 can be in the form of a chart illustrating the optimized insulin strategy parameters. In some embodiments, one or more parameters can be output to a user interface or connected device, e.g., a clinician report or connected bolus calculator. Additionally or alternatively, the output can be provided by a natural language processor to describe the candidate changes and optimized risk outcomes. The output can identify the time window that was optimized, and in some embodiments, a text generator can be used to communicate the results.

[0139] An example of the insulin strategy optimizer 1510 is now described on a dataset from a patient with type 1 diabetes. In this example, historical CGM data is analyzed to identify nocturnal hyperglycemia as a solvable risk profile. FIG. 17 is a chart 1700 showing the patient’s 10 days of CGM data. Despite the presence of an abnormal day with morning hypoglycemia, and a significant rebound in the day (indicated by curve 1710), a systematic pattern of nocturnal hypoglycemia can be seen in most of the trace (indicated by curve 1720). This is a combined example where a replayer is constructed, a therapy zone is identified, a strategy adjustment is made (e.g., a percentage change in strategy parameters (or parameter sweep)), and then replayed to find the best glycemic outcome.

[0140] FIG. 18 is a chart 1800 illustrating the risk profile function output from the analysis of the 10 days of CGM data shown in FIG. 17 is a chart 1800 illustrating the risk profile function output from the analysis of the 10 days of CGM data shown in

[0141] After analyzing the coexisting insulin data (provided by basal and bolus data) and meal data (provided by recognized carbohydrate data), it is determined that the patient is using functional insulin therapy (basal-bolus therapy), with occasional bolus recognized carbohydrates. From the data, the nominal parameters can be determined, including the existing basal prescription and current correction factor, and the carbohydrate ratio programmed into the bolus calculator.

[0142] In this example, the insulin strategy optimizer utilizes a playback simulation (described in U.S. Application No. 17 / 096,785, entitled “Joint State Estimation Prediction to Evaluate Differences Between Predicted Data and Corresponding Received Data,” filed November 12, 2020, by inventor Stephen D. Patek, the entirety of which is incorporated by reference herein). The playback simulation data collects the patient’s history of recognized carbohydrates. Boluses are simulated only at recognized carbohydrates. Inconsistencies between the playback simulation and the historical data are expected because boluses in the history are not simulated. The simulated doses are calculated strictly according to the current prescription of simulated BG, IOB (insulin on board), and recognized carbohydrates, as well as the current prescription of carbohydrate ratios and correction factors. This can be referred to as compliant functional insulin therapy.

[0143] FIG. 19 is a graph 1900, 1950 comparing the risk curves from the historical data and from the playback simulation. Thus, for this example, one would expect to see a close match between the simulated CGM trace and the historical CGM if the patient (1) ate exactly according to the carbohydrate notifications and (2) took boluses only at meal times. In this example, the historical boluses can have been delayed, and the estimates of on-board carbohydrates and insulin can not have been accurate. The patient compliance can optionally be scored here to determine compliance with functional insulin therapy.

[0144] Next, the playback simulation is run on the simulated CGM trace to set a large number of candidate parameters, with the basal insulin doses used being reduced. The parameters that minimize the patient’s (playback) exposure to both low and high blood sugar risks are stored for future reference. The risk curves and therapy regions are re-run. FIG. 20 is shown a risk curve 2000 from the playback.

[0145] As one example, FIG. 21 is a graph 2100 showing 2 high blood sugar risk curves 2120. Similar to the case described, these regions are smaller because the higher basal doses result in less time in the high blood sugar range. As another example, FIG. 22 is a graph 2200 showing two corresponding high blood sugar therapy regions 2220.

[0146] The carbohydrate ratios and correction factors can be further optimized. The playback simulation is run to set a large number of candidate parameters, with the basal insulin doses used being increased. The parameters that minimize the patient’s (playback) exposure to both low and high blood sugar risks are stored for future reference.

[0147] After running and quantifying various optimizations, the best combination of parameters is selected. FIG. 23is a chart 2300 showing optimized bolus and basal parameters.

[0148] FIG. 24 and FIG. 25 A chart is illustrated showing example risk curves 2400, 2500, 2550. The risk curves show that by applying optimized parameters, the patient can significantly reduce their glycemic risk, and thus, as described in greater detail herein, a recommendation for the optimized parameters can be output.

[0149] In another example, where a patient prefers a one-day-midday insulin bolus strategy, the system and method can receive the CGM as well as the insulin amount and time, and identify the treatment opportunity as an afternoon hypoglycemia in one time window (e.g., time x to time y). The insulin strategy is identified as a one-day-midday fixed time, based on a pattern bolus identified in a regular time bolus pattern in a day. The insulin strategy scorer determines a correlation of 85% to the one-day-midday fixed time. The insulin strategy adjustment runs a percentage change in the time and amount of the insulin bolus repeatedly and suggests a bolus 30 minutes before lunch and / or use 10% more insulin at the regular midday bolus. The report to the patient indicates that the 30 minutes time change and / or the 10% increase in the midday fixed bolus will reduce hypoglycemia by 20%, and the combination of both will reduce hypoglycemia by 25%.

[0150] In yet another example, a patient who prefers a mealtime bolus without carbohydrate calculation can use small, medium, large (S / M / L) meal estimates instead of three typical bolus amounts. In this example, the data inputs include the CGM, the amount of insulin, the time, and the meal time. The treatment opportunity is identified by the patient requesting a "bolus check" from the user interface. The determined insulin strategy is a mealtime bolus strategy with three typical doses, indicating S / M / L meal estimates. The insulin strategy scorer shows a correlation to the typical carbohydrate estimator used to estimate S / M / L. The insulin strategy adjustment suggests increasing the medium dose by 10% and / or decreasing the large dose by 10%. The output report to the patient of the medium bolus dose increase by 10% and the large bolus dose decrease by 10% will result in a reduction in hypoglycemia and hyperglycemia, where the combination of both will result in a specific percentage amount reduction (x%, where x is a determined, calculated, or estimated number) in hypoglycemia and a specific percentage amount reduction (x%) in hyperglycemia.

[0151] FIG. 26 An example computing environment in which example embodiments and aspects can be implemented is shown. The computing device environment is only one example of a suitable computing environment and is not intended to suggest any limitation as to the scope of use or functionality.

[0152] Many other general purpose or special purpose computing device environments or configurations can be used. Examples of well-known computing devices, environments, and / or configurations that can be suitable for use include, but are not limited to, personal computers, server computers, handheld or laptop devices, multiprocessor systems, microprocessor-based systems, network personal computers (PCs), minicomputers, mainframe computers, embedded systems, distributed computing environments that include any of the above systems or devices, and the like.

[0153] Computer-executable instructions, such as program modules, being executed by a computer can be used. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform particular tasks or implement particular abstract data types. Distributed computing environments can be used where tasks are performed by remote processing devices that are linked through a communications network or other data transmission medium. In a distributed computing environment, program modules and other data can be located in local and remote computer storage media including memory storage devices.

[0154] With reference to FIG. 26 , an exemplary system for implementing the aspects described herein includes a computing device, such as the computing device 2600. In its most basic configuration, computing device 2600 typically includes at least one processing unit 2602 and memory 2604. Depending on the exact configuration and type of computing device, memory 2604 can be volatile (such as random access memory (RAM)), non-volatile (such as read-only memory (ROM), flash memory, etc.), or some combination of the two. This most basic configuration is illustrated in FIG. 26 by dashed line 2606.

[0155] Computing device 2600 can have additional features / functionality. For example, computing device 2600 can include additional storage (removable and / or non-removable) including, but not limited to, magnetic or optical disks or tape. Such additional storage is illustrated in FIG. 26 by removable storage 2608 and non-removable storage 2610.

[0156] Computing device 2600 typically includes a variety of computer-readable media. Computer-readable media can be any available media that can be accessed by device 2600 and includes both volatile and nonvolatile media, removable and non-removable media.

[0157] Computer storage media includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storage of information such as computer readable instructions, data structures, program modules or other data. Memory 2604, removable storage 2608, and non-removable storage 2610 are all computer storage media examples. Computer storage media includes, but is not limited to, RAM, ROM, electrically erasable program read-only memory (EEPROM), flash memory or other memory technology, CD-ROM, digital versatile disks (DVD) or other optical storage, magnetic cassettes, magnetic tapes, magnetic disk storage or other magnetic storage devices, or any other medium which can be used to store the desired information and which can be accessed by computing device 2600. Any such computer storage media can be part of computing device 2600.

[0158] Computing device 2600 can contain communication connection(s) 2612 that allow the device to communicate with other devices. Computing device 2600 can also have input device(s) 2614 such as keyboard, mouse, pen, voice input device, touch input device, etc. Output device(s) 2616 such as a display, speakers, a printer, etc. can also be included. All of these devices are well known in the art and need not be discussed at length here.

[0159] In one embodiment, a method includes deriving at least one single specific symptom risk curve using a blood glucose risk analyzer, determining at least one treatment related region associated with the at least one single specific symptom risk curve using a treatment region evaluator, determining an importance value for the at least one treatment related region using a region importance quantifier, and outputting information based on the importance value.

[0160] Implementations can include some or all of the following features. The method further includes receiving glucose data, wherein the at least one single symptom-specific risk curve is derived using the glucose data. The glucose data includes at least one of CGM (continuous glucose monitoring) readings, trustworthiness readings assigned to CGM values, self-monitoring blood glucose readings, or retrospective calibrated or corrected CGM readings. The glucose data includes a time period in at least one week. The at least one single symptom-specific risk curve describes hypoglycemic risk or hyperglycemic risk as a function of time of day using the glucose data. Deriving the at least one single symptom-specific risk curve includes evaluating steepness (first and second derivatives of the curve), frequency, severity, curvature, mean values in the curve over 24 hours, or variability of the curve (mean and standard deviation). The at least one single symptom-specific risk curve is based on CGM signals indicating glycemic dysfunction over a selected time period, indicating recurrent time windows characterized by a predetermined severity and frequency of hypoglycemia or hyperglycemia over the selected time period. The at least one single symptom-specific risk curve represents at least one of hypoglycemia separated from hyperglycemia or at least one of hyperglycemia separated from hypoglycemia. Determining the at least one therapy-related zone includes identifying the at least one therapy-related zone from the at least one single symptom-specific risk curve. The at least one therapy-related zone is an interval in 24 hours of a day in which the patient’s BG data indicates that at least one of the patient’s insulin basal rate or dose or bolus strategy is systematically not optimal. The at least one therapy-related zone is identified and associated with the at least one risk curve and includes at least one interval in the day in which one or more single symptom-specific risk curves indicate potential glycemic dysfunction.

[0161] Implementations can also include some or all of the following features. The method further includes determining a time period of the at least one single specific symptom risk curve that can be mitigated by adjusting a parameter or timing of insulin therapy. Determining the at least one therapy-related region is based on predicting at least one of a candidate behavior change or a therapy change to reduce a single symptom glucose risk without subsequently exacerbating another symptom. Determining the importance value of the at least one therapy-related region includes prioritizing regions that are of significant therapeutic interest or resolvable. Determining the importance value of the at least one therapy-related region includes assessing a size of the risk. Determining the importance value of the at least one therapy-related region includes considering at least one of a time of day or a proximity of one risk curve to another risk curve. The importance value is a peak value of the at least one single specific symptom risk curve. Deriving the at least one single specific symptom risk curve is based on data that is above a certain level of confidence. The output information includes at least one of outputting a number, an alphanumeric, or a graphical information. The output information includes outputting at least one of a risk curve or a therapy-related region to a connected insulin pump or insulin pen, or to a bolus calculator. The output information includes outputting a graphical representation of at least one of a risk curve or a therapy-related region or a relative importance of at least one of a risk curve or a therapy-related region.

[0162] In one implementation, a system includes a glucose risk analyzer configured to derive at least one single specific symptom risk curve, a therapy region evaluator configured to determine at least one therapy-related region associated with the at least one single specific symptom risk curve, a region importance quantifier configured to determine an importance value of the at least one therapy-related region, and a therapy region report generator configured to output information based on the importance value.

[0163] Implementations can include some or all of the following features. The blood glucose risk analyzer is further configured to receive glucose data, wherein the at least one single specific symptom risk curve is derived using the glucose data. The glucose data includes at least one of CGM (continuous glucose monitoring) readings, trustworthiness readings assigned to CGM values, self-monitoring blood glucose readings, or retrospective calibrated or corrected CGM readings. The glucose data includes a time period in at least one week. The at least one single specific symptom risk curve describes hypoglycemic risk or hyperglycemic risk as a function of time of day using the glucose data. Deriving the at least one single specific symptom risk curve includes evaluating steepness (first and second derivatives of the curve), frequency, severity, curvature, mean values in the curve over 24 hours, or variability of the curve (mean and standard deviation). The at least one single specific symptom risk curve is indicative of glycemic dysfunction based on CGM signals over a selected time period, indicative of recurrent time windows characterized by a predetermined severity and frequency of hypoglycemia or hyperglycemia over the selected time period. The at least one single specific symptom risk curve represents at least one of hypoglycemia separated from hyperglycemia or at least one of hyperglycemia separated from hypoglycemia. Determining the at least one treatment-related region includes identifying the at least one treatment-related region from the at least one single specific symptom risk curve. The at least one treatment-related region is an interval in 24 hours of a day in which the patient’s BG data indicates that at least one of the patient’s insulin basal rate or dose or bolus strategy is systematically not optimal. The at least one treatment-related region is identified and associated with the at least one risk curve and includes at least one interval in the day in which one or more single specific symptom risk curves indicate potential glycemic dysfunction.

[0164] Implementations can also include some or all of the following features. The therapy area evaluator is further configured to identify time periods of the at least one single specific symptom risk profile that can be mitigated by adjusting a parameter or timing of insulin therapy. Determining the at least one therapy-related area is based on predicting at least one of a candidate behavior change or a therapy change to reduce a single symptom glucose risk without subsequently exacerbating another symptom. Determining the at least one therapy-related area includes prioritizing areas of significant therapy interest or resolution. Determining the at least one therapy-related area includes evaluating a size of a risk. Determining the at least one therapy-related area includes considering at least one of a time of day or a proximity of one risk profile to another risk profile. The importance value is a peak value of the at least one single specific symptom risk profile. Deriving the at least one single specific symptom risk profile is based on data above a certain level of confidence. The output information includes at least one of outputting a number, an alphanumeric, or a graphical information. The output information includes outputting at least one of a risk profile or a therapy-related area to a connected insulin pump or insulin pen, or to a bolus calculator. The output information includes outputting a graphical representation of at least one of a risk profile or a therapy-related area or a relative importance of at least one of a risk profile or a therapy-related area.

[0165] In one implementation, a system includes at least one processor; a non-transitory computer readable medium comprising: instructions that, when executed by the at least one processor, cause the system to: derive at least one single specific symptom risk profile; determine at least one therapy-related area related to the at least one single specific symptom risk profile; determine an importance value of the at least one therapy-related area; and output information according to the importance value.

[0166] Implementations can include some, or all, of the following features. The system also includes instructions that, when executed by the at least one processor, cause the system to receive glucose data, wherein the at least one single symptom-specific risk curve is derived using the glucose data. The glucose data includes at least one of CGM (continuous glucose monitoring) readings, trustworthiness readings assigned to CGM values, self-monitoring blood glucose readings, or retrospective calibrated or corrected CGM readings. The glucose data includes a time period in at least one week. The at least one single symptom-specific risk curve describes hypoglycemic risk or hyperglycemic risk as a function of time of day using the glucose data. Deriving the at least one single symptom-specific risk curve includes evaluating steepness (first and second derivatives of the curve), frequency, severity, curvature, mean values in the curve over 24 hours, or variability of the curve (mean and standard deviation). The at least one single symptom-specific risk curve is based on CGM signals indicating glycemic dysfunction over a selected time period, indicating a recurrent time window characterized by a predetermined severity and frequency of hypoglycemia or hyperglycemia over the selected time period. The at least one single symptom-specific risk curve represents at least one of hypoglycemia separated from hyperglycemia or at least one of hyperglycemia separated from hypoglycemia. Determining the at least one treatment-related zone includes identifying the at least one treatment-related zone from the at least one single symptom-specific risk curve. The at least one treatment-related zone is an interval over 24 hours in a day in which the patient’s BG data indicates that at least one of the patient’s insulin basal rate or dose or bolus strategy is systematically not optimal. The at least one treatment-related zone is identified and associated with the at least one risk curve and includes at least one interval in a day in which one or more single symptom-specific risk curves indicate potential glycemic dysfunction.

[0167] Implementations can also include some or all of the following features. The system further includes instructions that, when executed by the at least one processor, cause the system to identify a time period of at least one single specific symptom risk profile that can be mitigated by adjusting a parameter or timing of insulin therapy. Determining the at least one therapy-related region is based on predicting at least one of a candidate behavioral change or a therapy change to reduce a single symptom glucose risk without subsequently exacerbating another symptom. Determining the at least one therapy-related region includes prioritizing regions that are of high therapeutic significance or resolvable. Determining the at least one therapy-related region includes assessing a size of a risk. Determining the at least one therapy-related region includes considering at least one of a time of day or a proximity of one risk profile to another risk profile. The importance value is a peak of the at least one single specific symptom risk profile. Deriving the at least one single specific symptom risk profile is based on data that is above a certain level of confidence. The output information includes at least one of outputting a number, an alphanumeric, or a graphical information. The output information includes outputting at least one of a behavioral change or a therapy change to a therapy time region to reduce a single symptom in a time window. The output information includes outputting information to a connected insulin pump or insulin pen, or to a bolus calculator. The output information includes outputting a graphical representation of at least one of a risk profile or a therapy-related region or a relative importance of at least one of a risk profile or a therapy-related region.

[0168] In one implementation, a method includes receiving glucose and insulin data; identifying therapy improvement opportunities using the glucose and insulin data; determining candidate changes to insulin therapy; evaluating improvements in therapy risk based on the candidate changes; quantifying improvements of the candidate changes; and outputting at least one of the candidate changes based on the improvements.

[0169] Implementations can include some or all of the following features. Glucose and insulin data are received from at least one of a patient or a connected system or device. Identifying a therapy improvement opportunity includes receiving a user selection of at least one of a meal time, a time of day, or a parameter setting. The parameter setting is a carbohydrate ratio. The candidate change to insulin therapy includes a percentage increase or decrease to a bolus therapy or a basal therapy. The candidate change to the insulin therapy includes a change to an insulin delivery parameter related to the bolus therapy or the basal therapy. The candidate change is according to a carbohydrate ratio, a correction factor, a basal rate, or a graph. The candidate change includes basal sensitivity. The candidate change includes a percentage change in a basal dose or a bolus dose in a therapy zone. Quantifying an improvement in the candidate change includes comparing risk profile values. Outputting at least one of the candidate changes based on the improvement includes outputting a candidate change that provides an optimized risk profile. Outputting at least one of the candidate changes includes providing an output in a form of a graph to a user interface or a connected device, where the graph illustrates at least one of the candidate change or the optimized risk output. The connected device includes a bolus calculator. A natural language processor outputs to describe the candidate change and the optimized risk result. The output identifies an optimized therapy zone or set of zones.

[0170] In an implementation, a system includes a therapy improvement identifier configured to evaluate a patient's reconciled glucose and insulin data to identify a zone in a patient's diabetes management routine in which to make a therapy optimization, and to generate a therapy improvement; a relative insulin optimizer configured to propose a change to therapy, evaluate an impact of the change, and quantify an improvement associated with the change; and a relative insulin optimizer report generator to provide an output.

[0171] Implementations can include some or all of the following features. The relative insulin optimizer includes a change proposer configured to propose a change to insulin therapy, an impact evaluator configured to evaluate an impact of a candidate therapy change by estimating an impact on a risk profile of historical glucose values, and an improvement quantifier configured to quantify an improvement in the candidate therapy change. The change proposer is further configured to propose the change as a percentage change to at least one of a basal or a bolus in a time window. The improvement quantifier is configured to quantify the improvement in the candidate therapy change based on a percentage improvement or change in a glycemic outcome metric. The relative insulin optimizer report generator is configured to output the candidate therapy change to a user. The user is one of a clinician, a patient, or a connected device or system. The therapy improvement identifier includes a user selection of a therapy to be optimized or a time of day. The user is a patient or a clinician. The therapy improvement is identified by an algorithm.

[0172] In one embodiment, a system includes: at least one processor; and a non-transitory computer-readable medium comprising: instructions that, when executed by the at least one processor, cause the system to: receive glucose and insulin data; identify therapy improvement opportunities using the glucose and insulin data; determine candidate changes to insulin therapy; evaluate improvement in therapy risk according to the candidate changes; quantify improvement in the candidate changes; and output at least one candidate change based on the improvement.

[0173] Implementations can include some or all of the following features. The glucose and insulin data is received from at least one of a patient or a connected system or device. Identifying therapy improvement opportunities includes receiving a user selection of at least one of a meal time, a time of day, or a parameter setting. The parameter setting is a carbohydrate ratio. The candidate changes to insulin therapy include a percentage increase or decrease to a bolus therapy or a basal therapy. The candidate changes to the insulin therapy include a change to an insulin delivery parameter related to the bolus therapy or the basal therapy. The candidate changes are according to a carbohydrate ratio, a correction factor, a basal rate, or a graph. The candidate changes include basal sensitivity. The candidate changes include a percentage change in a basal dose or a bolus dose in a therapy zone. Quantifying improvement in the candidate changes includes comparing risk profile values. Outputting at least one of the candidate changes based on the improvement includes outputting a candidate change that provides an optimized risk profile. Outputting at least one of the candidate changes includes providing an output in the form of a graph to a user interface or a connected device, wherein the graph illustrates at least one of the candidate changes or the optimized risk output. The connected device includes a bolus calculator. A natural language processor outputs to describe the candidate changes and the optimized risk results. The output identifies an optimized therapy zone or set of zones.

[0174] In one embodiment, a method includes: receiving at least one of glucose data, insulin data, or other diabetes-related data for a patient; identifying therapy improvement opportunities using the at least one of glucose data, insulin data, or other diabetes-related data; determining an insulin dosing strategy for the patient; scoring the insulin dosing strategy for patient adherence; optimizing the insulin dosing strategy; and providing an output to a user including optimized insulin strategy parameters.

[0175] Implementations can include some or all of the following features. The other diabetes-related data includes at least one of meal information, specific meals, meal times, meal sizes, carbohydrate estimates, ingredient information, or exercise information. The glucose data, insulin data, or other diabetes-related data are received from at least one of the patient or a connected system or device. Identifying the therapy improvement opportunity includes receiving a selection by the user of at least one of a meal time, a time of day, or a parameter setting. The parameter setting is a carbohydrate ratio. The insulin dosing strategy includes a diabetes management or insulin strategy implemented by the patient in practice, as determined from at least one of the patient's glucose data, insulin data, or other diabetes-related data. Optimizing the insulin dosing strategy includes determining whether the patient adheres to a known insulin strategy and analyzing the impact of a percentage change in a determined insulin strategy parameter. The user is at least one of a clinician, the patient, or a connected device or system. The natural language processor outputs to describe a candidate change and a risk outcome of the optimization. Providing the output includes outputting to a user interface or a connected device in the form of a chart illustrating an optimized insulin strategy parameter. The connected device includes a bolus calculator.

[0176] In one implementation, a system includes a therapy improvement identifier configured to evaluate a patient's reconciled glucose and insulin data to identify areas of ongoing therapy optimization in the patient's diabetes management routine and generate a therapy improvement, an insulin strategy optimizer configured to determine whether the patient adheres to a known insulin strategy and analyze the impact of a percentage change in a determined insulin strategy parameter, and a therapy identification optimization report generator providing an output.

[0177] Implementations can include some or all of the following features. The insulin strategy optimizer includes: an insulin strategy identifier configured to identify a diabetes management or insulin strategy implemented in practice by a patient as determined from reconciled glucose and insulin data; a compliance scorer configured to quantify a patient's compliance with the identified insulin strategy; and the insulin strategy within the identified behavior is optimized by the insulin strategy optimizer configured to optimize the identified insulin strategy. The diabetes data includes insulin data and meal data. The insulin strategy identifier is configured to identify a dosing pattern and describe the identified patient's insulin strategy based thereon. The insulin strategy is a behavioral methodology applied by the patient in diabetes management including at least one of use of an insulin pump, multiple daily injections, or a type 2 therapy. The compliance scorer is configured to generate a score computed for a degree of compliance of the patient with the identified insulin strategy. The insulin strategy within the identified behavior is optimized by the insulin strategy optimizer configured to iteratively propose a percentage change to a strategy parameter in a selected therapy region or group of regions. The output includes optimized insulin strategy parameters. The therapy identification optimization report generator is configured to output a candidate therapy change to a user. The user is at least one of a clinician, a patient, or a connected device or system. The natural language processor outputs to describe the candidate change and optimized risk outcomes.

[0178] In one implementation, a system includes: at least one processor; a non-transitory computer-readable medium comprising: instructions that, when executed by the at least one processor, cause the system to: receive at least one of glucose data, insulin data, or other-diabetes-related data of a patient; identify a therapy improvement opportunity using at least one of the glucose data, the insulin data, or the other-diabetes-related data; determine an insulin dosing strategy of the patient; score the insulin dosing strategy for patient compliance; optimize the insulin dosing strategy; and provide an output including optimized insulin strategy parameters to a user.

[0179] Implementations can include some, or all, of the following features. The other diabetes-related data includes at least one of meal information, specific meals, meal times, meal sizes, carbohydrate estimates, ingredient information, or exercise information. The glucose data, insulin data, or other diabetes-related data are received from at least one of the patient or a connected system or device. Identifying the therapy improvement opportunity includes receiving a selection by the user of at least one of a meal time, a time of day, or a parameter setting. The parameter setting is a carbohydrate ratio. The insulin dosing strategy includes a diabetes management or insulin strategy implemented in practice by the patient as determined from at least one of the glucose data, the insulin data, or the other diabetes-related data of the patient. Optimizing the insulin dosing strategy includes determining whether the patient is following a known insulin strategy and analyzing an impact of a percentage change in determined insulin strategy parameters. The user is at least one of a clinician, the patient, or a connected device or system. The natural language processor outputs to describe a candidate change and a risk outcome of the optimization. Providing the output includes outputting to a user interface or a connected device in a form of a chart illustrating optimized insulin strategy parameters. The connected device includes a bolus calculator.

[0180] It should be appreciated that various technologies described herein can be implemented in connection with hardware components or software components, or combinations of both. Examples of exemplary hardware components that can be used include Field-Programmable Gate Arrays (FPGAs), Application-Specific Integrated Circuits (ASICs), Application-Specific Standard Products (ASSPs), System-on-a-Chip (SOCs), Complex Programmable Logic Devices (CPLDs), etc. The methods and apparatus of the presently disclosed subject matter, or certain aspects or portions thereof, can take the form of program code (i.e., instructions) embodied in tangible media, such as floppy diskettes, CD-ROMs, hard drives, or any other machine-readable storage medium wherein, when the program code is loaded into and executed by a machine, such as a computer, the machine becomes an apparatus for practicing the presently disclosed subject matter.

[0181] Although exemplary implementations can refer to utilizing the presently disclosed subject matter in the context of one or more stand-alone computer systems, the subject matter is not so limited, but rather can be implemented in connection with any computing environment, such as a network or distributed computing environment. Still further, aspects of the presently disclosed subject matter can be implemented in or across a plurality of processing chips or devices, and storage can similarly be effected across a plurality of devices. Such devices can include personal computers, network servers, and handheld devices, for example.

[0182] Although the subject matter has been described in language specific to structural features and / or methodological acts, it is to be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or acts described above. Rather, the specific features and acts described above are disclosed as example forms of implementing the claims.

Claims

1. A system comprising: a glycemic risk analyzer configured to derive at least one single specific symptom risk curve; a therapy zone evaluator configured to determine at least one therapy-related zone associated with at least one of the single specific symptom risk curves, the at least one therapy-related zone being a window of time in which insulin therapy is suboptimal such that insulin therapy can be adjusted to provide systemic improvement in glycemic outcomes; a zone importance quantifier configured to determine an importance value for at least one of the therapy-related zones; and a therapy zone report generator configured to output information based on the importance value to adjust insulin therapy delivered by an insulin therapy delivery device. the glycemic risk analyzer is further configured to receive glucose data, wherein the at least one single specific symptom risk curve is derived using the glucose data.

2. The system of claim 1, wherein, the glucose data comprises at least one of continuous glucose monitor (CGM) readings, confidence readings assigned to CGM values, self-monitored blood glucose readings, or retrospectively calibrated or corrected CGM readings.

3. The system of claim 2, wherein, the glucose data comprises a time period of at least one week.

4. The system of claim 2, wherein, the at least one single specific symptom risk curve describes hypoglycemic risk or hyperglycemic risk as a function of time of day using the glucose data.

5. The system of claim 2, wherein, deriving the at least one single specific symptom risk curve comprises evaluating at least one of steepness, frequency, severity, curvature, average in the curve over 24 hours, or variability of the curve.

6. The system of claim 1 or 2, wherein, the at least one single specific symptom risk curve is based on CGM signals indicating glycemic dysfunction over a selected time period, indicating a recurring time window characterized by a predetermined severity and frequency of hypoglycemia or hyperglycemia over the selected time period.

7. The system of claim 1 or 2, wherein, the at least one single specific symptom risk curve represents at least one of hypoglycemia separated from hyperglycemia or hyperglycemia separated from hypoglycemia.

8. The system of claim 1 or 2, wherein, determining the at least one therapy-related zone comprises identifying the at least one therapy-related zone from the at least one single specific symptom risk curve.

9. The system of claim 1 or 2, wherein, the at least one therapy-related zone is an interval in a 24-hour day in which patient BG data indicates that at least one of insulin basal rates or doses or bolus strategies are systematically out of optimal condition.

10. The system of claim 1 or 2, wherein, the at least one therapy-related zone is identified and associated with the at least one risk curve and comprises at least one interval in a day in which one or more of the single specific symptom risk curves indicate potential glycemic dysfunction.

11. The system of claim 1 or 2, wherein, the therapy zone evaluator is further configured to identify time periods in which the at least one single specific symptom risk curve can be mitigated by adjusting parameters or timing of insulin therapy.

12. The system of claim 1 or 2, wherein, determining the at least one therapy-related zone is based on predicting at least one of candidate behavioral changes or therapy changes to reduce single symptom glycemic risk without subsequently exacerbating another symptom.

13. The system of claim 1 or 2, wherein, determining the importance value for the at least one therapy-related zone comprises prioritizing zones that are more significant or addressable for therapy.

14. The system of claim 1 or 2, wherein, determining the importance value for the at least one therapy-related zone comprises evaluating a magnitude of risk.

15. The system of claim 1 or 2, wherein, ​ 16. The system of claim 1 or 2, wherein, Determining the importance value for at least one of the therapy-related zones includes considering at least one of time of day or proximity of one risk profile to another risk profile.

17. The system of claim 1 or 2, wherein, The importance value is a peak value of at least one of the single specific symptom risk profiles.

18. The system of claim 1 or 2, wherein, Deriving at least one of the single specific symptom risk profiles is based on data above a certain level of confidence.

19. The system of claim 1 or 2, wherein, The output information includes outputting at least one of numerical, alphanumeric, or graphical information.

20. The system of claim 1 or 2, wherein, The output information includes outputting at least one of a behavior change or a therapy change to a therapy time zone to reduce a time window of a single symptom.

21. The system of claim 1 or 2, wherein, The output information includes outputting the information to a connected insulin pump or insulin pen, or to a bolus calculator.

22. The system of claim 1 or 2, wherein, The output information includes outputting a graphical representation of at least one of the risk profiles or the therapy-related zones or relative importance of at least one of the risk profiles or the therapy-related zones.

23. A system comprising: at least one processor; and a non-transitory computer readable medium comprising instructions that, when executed by the at least one processor, cause the system to: derive at least one single specific symptom risk profile; determine at least one therapy-related zone associated with the at least one single specific symptom risk profile, the at least one therapy-related zone being a time window in which insulin therapy is suboptimal to enable insulin therapy to be adjusted to provide systematic improved glycemic outcomes; determine an importance value for at least one of the therapy-related zones; and output information based on the importance value to adjust insulin therapy delivered by an insulin therapy delivery device. Deriving at least one of the single specific symptom risk profiles using the glucose data.

24. The system of claim 23, further comprising instructions that, when executed by the at least one processor, cause the system to receive glucose data, wherein, The glucose data includes at least one of continuous glucose monitoring (CGM) readings, confidence readings assigned to CGM values, self-monitoring blood glucose readings, or retrospectively calibrated or corrected CGM readings.

25. The system of claim 24, wherein, The glucose data includes a time period of at least one week.

26. The system of claim 24, wherein, At least one of the single specific symptom risk profiles describes hypoglycemic risk or hyperglycemic risk as a function of time of day using the glucose data.

27. The system of claim 24, wherein, Deriving at least one of the single specific symptom risk profiles includes evaluating at least one of steepness, frequency, severity, curvature, average in the curve over 24 hours, or variability of the curve.

28. The system of claim 23 or 24, wherein, At least one of the single specific symptom risk profiles is based on CGM signals indicating glycemic dysfunction over a selected time period, indicating a recurring time window characterized by a predetermined severity and frequency of hypoglycemia or hyperglycemia over the selected time period.

29. The system of claim 23 or 24, wherein, At least one of the single specific symptom risk profiles represents at least one of hypoglycemia separated from hyperglycemia or hyperglycemia separated from hypoglycemia.

30. The system of claim 23 or 24, wherein, Determining at least one of the therapy-related zones includes identifying at least one of the therapy-related zones from at least one of the single specific symptom risk profiles.

31. The system of claim 23 or 24, wherein, At least one of the therapy-related zones is an interval in a 24-hour day in which the patient's BG data indicates that at least one of insulin basal rates or doses or bolus strategies are systematically out of optimal.

32. The system of claim 23 or 24, wherein, ​ 33. The system of claim 23 or 24, wherein, At least one of the therapy-related zones is identified and associated with at least one of the risk profiles, and includes at least one interval of the day in which one or more of the single specific symptom risk profiles indicate a potential blood glucose dysfunction.

34. The system of claim 23 or 24, further comprising instructions which, when executed by the at least one processor, cause the system to identify a time period in which at least one of the single specific symptom risk profiles can be mitigated by adjusting a parameter or timing of insulin therapy.

35. The system of claim 23 or 24, wherein, Determining at least one of the therapy-related zones is based on predicting at least one of a candidate behavioral change or a therapy change to reduce a single symptom blood glucose risk without subsequently exacerbating another symptom.

36. The system of claim 23 or 24, wherein, Determining an importance value for at least one of the therapy-related zones includes prioritizing zones of greater or resolvable therapy significance.

37. The system of claim 23 or 24, wherein, Determining an importance value for at least one of the therapy-related zones includes assessing a size of a risk.

38. The system of claim 23 or 24, wherein, Determining an importance value for at least one of the therapy-related zones includes considering at least one of a time of day or a proximity of one risk profile to another risk profile.

39. The system of claim 23 or 24, wherein, The importance value is a peak value of at least one of the single specific symptom risk profiles.

40. The system of claim 23 or 24, wherein, Deriving at least one of the single specific symptom risk profiles is based on data above a certain level of confidence.

41. The system of claim 23 or 24, wherein, Outputting information includes outputting at least one of a number, alphanumeric, or graphical information.

42. The system of claim 23 or 24, wherein, Outputting information includes outputting at least one of a behavioral change or a therapy change to a therapy timing zone to reduce a single symptom in a time window.

43. The system of claim 23 or 24, wherein, Outputting information includes outputting the information to a connected insulin pump or insulin pen, or to a bolus calculator.

44. The system of claim 23 or 24, wherein, Outputting information includes outputting a graphical representation of at least one of the risk profiles or the therapy-related zones or a relative importance of at least one of the risk profiles or the therapy-related zones.

Images