System and method for regimen compliance evaluation

Abstract

A diabetes management system adapted to determine compliance of a subject in treatment according to a basal insulin regimen, the system adapted to receive regimen data, blood glucose data, and insulin injection data. If one or more insulin injections are not received in accordance with the specified regimen and thus missing, the system is adapted to calculate a predicted dose-responsive blood glucose level for each missing injection. By comparing the received blood glucose data corresponding to missing insulin injections, it may be determined whether the subject is compliant with the basal insulin regimen for a given confidence interval.

Description

Technical Field

[0001] The present disclosure generally relates to systems and methods for assisting patients and healthcare practitioners in managing insulin therapy for diabetic patients. In one specific aspect, the present invention relates to systems and methods suitable for use in a diabetes management system that provides an optimized personalized basal insulin titration regimen. Background Art

[0002] Diabetes mellitus (DM) is impaired insulin secretion and varying degrees of peripheral insulin resistance, resulting in hyperglycemia. Type 2 diabetes is characterized by a progressive disruption of normal physiological insulin secretion. In healthy individuals, basal insulin secretion occurs continuously from pancreatic β cells, thereby maintaining stable glucose levels for a long time between meals. In healthy individuals, there is also a mealtime secretion, in which insulin is rapidly released in response to a meal with an initial first-phase peak, followed by a prolonged insulin secretion that returns to basal levels after 2-3 hours. Many years of poorly controlled hyperglycemia can lead to a variety of health complications. Diabetes is one of the leading causes of premature morbidity and mortality worldwide.

[0003] Effective control of blood / plasma glucose can prevent or delay many of these complications, but once they occur they may be irreversible. Therefore, striving to achieve good glycemic control to prevent diabetic complications is a major goal in the treatment of both type 1 and type 2 diabetes. Smart titrators with adjustable step length and physiological parameter estimation and predetermined fasting glucose target values ​​have been developed for managing insulin medication regimens.

[0004] There are many non-insulin treatment options for diabetes, however, the most stable response is generally achieved with insulin as the disease progresses. In particular, because diabetes is associated with progressive β-cell loss, many patients, especially those with long-standing disease, eventually need to be switched to insulin because the degree of hyperglycemia (e.g., HbA1c ≥ 8.5%) makes it unlikely that other agents will provide adequate benefit.

[0005] The ideal insulin regimen aims to mimic the physiological curve of insulin secretion as closely as possible. The insulin curve has two main components: continuous basal secretion and a mealtime surge after a meal. Basal secretion controls nocturnal and fasting blood sugars, while the mealtime surge controls postprandial hyperglycemia.

[0006] Based on the time of onset and the duration of their action, injectable preparations can be broadly divided into basal insulin (long-acting analogs [e.g., insulin detemir and insulin glargine] and ultra-long-acting analogs [e.g., insulin degludec (intended for once-daily dosing) and insulin icodec (intended for once-weekly dosing)]) and intermediate-acting insulins [e.g., isophane insulin] and prandial insulins (rapid-acting analogs [e.g., insulin aspart, insulin glulisine, and insulin lispro]). Premixed insulin preparations contain both basal and prandial insulin components.

[0007] For patients with type 2 diabetes, basal insulin is often the sole (initial) insulin therapy, whereas for patients with type 1 diabetes, basal insulin may be used in combination with a rapid-acting insulin taken before meals.

[0008] Typically, to determine the optimal basal insulin dose for a given patient, the patient is titrated from an initial recommended safe low dose of basal insulin (usually 10 U / day), and then the dose is increased until fasting plasma glucose (FPG) is within the target range, typically 80-130 mg / dL for patients with type 2 diabetes at a dose of typically 40-70 U / day. Alternatively, fasting blood glucose (FBG) values ​​are used. When approaching the target, dose adjustments should be more cautious and less frequent, and downward titration is recommended if any hypoglycemia occurs.

[0009] However, significant barriers exist not only in initiating insulin therapy but also in optimizing doses and intensifying treatment regimens, all of which are necessary steps to tailor treatment to individual needs and maintain glycemic control.

[0010] One challenge to effective titration is treatment adherence: failure to initiate, optimize, and intensify basal insulin therapy is driven by clinical inertia, which results in poor treatment adherence. This often involves forgetfulness, a perceived need to take medication, fear of hypoglycemia, and a lack of confidence or uncertainty about insulin titration.

[0011] Connected injection devices provide patients and clinicians with insights into treatment adherence and actionable treatment changes. Some decision support tools, such as the insulin titration support app, use input from connected devices to provide reliable and safe guidance. The integrity of the data from connected devices is critical because algorithms use this data to calculate safe and effective dosing recommendations for patients. Accordingly, it is also critical to understand whether the recommended dose was administered.

[0012] The nature of injection data is that they are sparse, and they are intended to represent whether a patient adheres to a treatment regimen. Therefore, "missing" data points represent non-adherence and are therefore important inputs to the algorithm.

[0013] In the case of bolus calculations, the problem of missing dose data has been addressed, for example in WO2021 / 172628, where it is disclosed that an assessment of blood glucose values ​​can indicate whether a given recommended dose was actually administered, or whether a recommended dose was not administered because it was not needed, such as because an expected meal was skipped. The assessment of blood glucose values ​​is based on overall considerations regarding the set blood glucose target range.

[0014] WO 2022 / 117713 discloses a data collection device for use in conjunction with an insulin infusion pump. The device is suitable for replacing missing data based on blood glucose data using predictive learning. The missing data may be data about non-insulin drugs such as painkillers, allergy medicines, or cold medicines.

[0015] WO 2020 / 043922 discloses a diabetes management system suitable for determining the compliance of a subject in treatment according to a basal insulin regimen. The system is suitable for receiving: regimen data listing the current dose size and the prescribed injection cycle, and multiple fasting blood glucose (FBG) measurements of the subject over a time course, thereby establishing an FBG history (FBGH), and multiple insulin injection data sets during the entire or partial time course, thereby establishing an insulin injection history (IH). The system includes one or more processors and a memory, the memory containing instructions that, when executed by one or more processors, perform data optimization based on FBGH and IH by processing missing data, wherein one or more time gaps in the subject data are interpolated by resampling the subject data at predefined time intervals.

[0016] In view of the above discussed problems of treatment compliance and hence of missing dose data during basal insulin titration, it is an object of the present invention to provide methods and systems that allow for more efficient and safer insulin titration even in the presence of missing dose data. Summary of the invention

[0017] In the disclosure of the present invention, embodiments and aspects will be described that achieve one or more of the above-mentioned objects or that will achieve objects that will be apparent from the following disclosure and description of exemplary embodiments.

[0018] In general, the present invention is based on the recognition that "missing" insulin dosage data points may occur due to technical problems, such as loss of communication with the injection device or the patient not properly installing a connected additional device on their injection device. Accordingly, it would be beneficial for a given titration regimen if non-compliance could be distinguished from missing data points.

[0019] Accordingly, in a first aspect of the invention, there is provided a diabetes management system adapted to determine the compliance of a subject in treatment according to a basal insulin regimen. The system is adapted to receive: regimen data listing the current dose size and the prescribed injection period; a plurality of fasting blood glucose (FBG) measurements taken of the subject over a time course, thereby establishing an FBG history (FBGH), wherein each corresponding glucose measurement in the plurality of glucose measurements comprises (i) an FBG value, and (ii) a corresponding FBG timestamp; and a plurality of insulin injection data sets during the whole or part of the time course, thereby establishing an insulin injection history (IH), wherein each injection data set comprises (i) an injection amount, and (ii) an injection timestamp indicating when the injection occurred during the time course. The system comprises one or more processors and a memory, wherein the memory comprises instructions which, when executed by the one or more processors, perform a method comprising the following steps: determining whether one or more insulin injection data sets have not been received (recorded) in accordance with a prescribed injection regimen over a period of time and are therefore missing. If the insulin injection data set is missing, the method includes the steps of calculating an expected dose response FBG value for each missing injection based on the received FBGH and IH and using a dose response algorithm, and determining, for a given confidence interval, whether the received FBG value corresponding to the missing insulin injection corresponds to the calculated dose response FBG value. If the received FBG value corresponds to the calculated dose response FBG value, the system determines that the subject is complying with the basal insulin regimen, or if the received FBG value does not correspond to the calculated dose response FBG value, the system determines that the subject is not complying with the basal insulin regimen. The FBG value can be derived from the received CGM data.

[0020] In this way, knowledge of FBGH and IH can be used to determine whether the patient actually received the unrecorded (ie, missing) injections according to the titration regimen.

[0021] In an exemplary embodiment, the predicted dose response FBG value is based on the received FBGH and IH and protocol dose data, ie, for unrecorded injections, the calculation is based on protocol data listing the times and dose sizes for the unrecorded injections.

[0022] In an exemplary embodiment, the diabetes management system is adapted to further provide insulin dosage recommendations for the subject, the method comprising the additional steps of receiving a dosage guidance request (DGR), and determining whether the subject has complied with the regimen in terms of insulin injections within a predetermined amount of time before receiving the DGR. If the subject complies (with or without missing dosage data), the system provides an updated dosage recommendation based on the received FBGH and IH, or if the subject does not comply, the system maintains the current dosage recommendation. The predetermined amount of time before receiving the DGR may be the time since the last DGR was raised.

[0023] In a second aspect of the invention, a method for determining the compliance of a subject in treatment according to a basal insulin regimen is provided. The method comprises the following steps: obtaining regimen data, the regimen data listing the current dose size and the prescribed injection period; obtaining multiple fasting blood glucose (FBG) measurements of the subject over a time course, thereby establishing a FBG history (FBGH), each corresponding glucose measurement in the multiple glucose measurements comprising (i) an FBG value, and (ii) a corresponding FBG timestamp; and obtaining multiple insulin injection data sets during the entire or part of the time course, thereby establishing an insulin injection history (IH), each injection data set comprising (i) an injection amount, and (ii) an injection timestamp, which indicates when the injection occurred during the time course. The method comprises the following further steps: determining whether one or more insulin injection data sets were not received according to the prescribed injection regimen over a period of time and are therefore missing. If the insulin injection data set is missing, the method includes the further steps of calculating an expected dose response FBG value for each missing injection based on the received FBGH and IH and using a dose response algorithm; determining, for a given confidence interval, whether the received FBG value corresponding to the missing insulin injection corresponds to the calculated dose response FBG value. If the received FBG value corresponds to the calculated dose response FBG value, the method determines that the subject is compliant with the basal insulin regimen, or if the received FBG value does not correspond to the calculated dose response FBG value, the method determines that the subject is not compliant with the basal insulin regimen. The FBG value can be derived from the received CGM data.

[0024] The predicted dose-response FBG value may be based on the received FBGH and IH and protocol dose data.

[0025] The method may be further adapted to provide an insulin dosage recommendation for the subject, the method comprising the additional steps of receiving a dosage guidance request (DGR), determining whether the subject has been compliant with the regimen for insulin injections within a predetermined amount of time prior to receiving the DGR, providing an updated dosage recommendation based on the received FBGH and IH if the subject is compliant, or maintaining the current dosage recommendation if the subject is not compliant. The predetermined amount of time prior to receiving the DGR may be the time since the last DGR was issued. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] The embodiments of the present invention will be described below with reference to the accompanying drawings, in which

[0027] Figure 1 FBG (dark grey) and dose data received from the patient (light grey) acquired over a time course are shown for the compliance case.

[0028] Figure 2 Shown for Figure 1 The compliance of the FBG distribution with respect to the confidence interval of the calculated response,

[0029] Figure 3 FBG (dark grey) and dose data received from the patient (light grey) acquired over the time course for the non-compliant case are shown.

[0030] Figure 4 Shown for Figure 3 The FBG distribution of the non-compliant case with respect to the confidence interval of the calculated response, and

[0031] Figure 5 An asymmetric weighting function is shown. DETAILED DESCRIPTION

[0032] In general, a diabetes dosage guidance system is provided that helps diabetic patients by generating recommended insulin dosages. In such a system, a given algorithm is used to generate recommended insulin dosages and treatment recommendations for diabetic patients based on blood glucose and insulin administration history.

[0033] Essentially, such a system comprises a backend engine ("engine") used in conjunction with an interaction system in the form of a client and an operating system, which backend engine is the primary aspect of the present invention.

[0034] From the engine's perspective, the client is the software component that requests dosing guidance. The client collects the necessary data (e.g., CGM data, insulin dosage data, patient parameters) and requests dosing guidance from the engine. The client then receives a response from the engine.

[0035] On a small local scale, the engine may be run directly as an app on a given user's smartphone, thus being a self-contained application comprising both the client and the engine. Alternatively, the system setup may be designed to be implemented as a backend engine suitable for use as part of a large-scale cloud-based diabetes management system. Such a cloud-based system would keep the engine always up to date (compared to an app-based system that runs entirely on, for example, a patient's smartphone), would allow for the implementation of advanced methods such as machine learning and artificial intelligence, and would allow the data to be used in conjunction with other services in a larger "digital health" setting. Ideally, such a cloud-based system would handle a large number of patient requests for dosage recommendations.

[0036] Although a "full" engine may be designed to take care of all computational aspects, it may be desirable to separate the engine into a local version and a cloud version to enable the near-patient side of the day-to-day portion of the dose guidance system to run independently without relying on cloud computing. For example, when a user issues a request for dose guidance through a client application, the request is transmitted to the cloud engine, which returns a dose recommendation. If cloud access is not available, the client application will run the dose recommendation calculation using the current local algorithm. Depending on the user's application settings, the user may or may not be informed.

[0037] In order to provide safe and effective dosing guidance recommendations, it is critical that the data on which the recommendations are based are as complete as possible.

[0038] Accordingly, in a first aspect, the present invention provides a diabetes management system adapted to determine a subject's compliance with treatment according to a basal insulin regimen.

[0039] More specifically, the present invention aims to distinguish between periods of non-compliance and periods of missing data in a time series of paired injection and glucose data. This is achieved by using periods of known compliance (i.e., available data points) to construct a dose-response model and by observing the corresponding glucose data to calculate the probability that a new missing injection data point is due to non-compliance or due to a genuine missing injection data point.

[0040] The advantage of this solution is that it enables more stable and accurate dosing guidance in settings where insulin pens are connected. By calculating the probability of adherence during the period of missing data from the connected pen, dosing guidance that would otherwise be delayed can be provided immediately upon restoration of connectivity.

[0041] More specifically, an exemplary diabetes management system is provided that is adapted to determine a subject's compliance in treatment according to a basal insulin regimen. The system is adapted to receive (a) regimen data that lists a current dose size and a prescribed injection period, (b) a plurality of fasting blood glucose (FBG) measurements taken of the subject over a time course, thereby establishing an FBG history (FBGH), wherein each respective glucose measurement of the plurality of glucose measurements comprises (i) an FBG value, and (ii) a respective FBG timestamp, and (c) a plurality of insulin injection data sets during the entire or partial time course, thereby establishing an insulin injection history (IH), wherein each injection data set comprises (i) an injection amount, and (ii) an injection timestamp indicating when the injection occurred over the time course.

[0042] The system includes one or more processors and a memory storing instructions that, when executed by the one or more processors, perform a method comprising the steps of: (A) determining that one or more insulin injection data sets were not received according to a prescribed injection regimen over a period of time and are therefore missing, and (B) calculating an expected dose response FBG value for each missed injection based on the received FBGH and IH and using a dose response algorithm.

[0043] The method comprises the further steps of: (C) determining, for a given confidence interval, whether the received FBG value corresponding to the missed insulin injection corresponds to the calculated dose-response FBG value, and (i) determining that the subject is complying with the basal insulin regimen if the received FBG value corresponds to the calculated dose-response FBG value, or (ii) determining that the subject is not complying with the basal insulin regimen if the received FBG value does not correspond to the calculated dose-response FBG value.

[0044] In an exemplary embodiment, the patient provides blood glucose data using a continuous glucose monitor (CGM), determines a fasting blood glucose (FBG) value based on the blood glucose data, and uses a connected drug delivery device, such as a pen device with an attached device, to record injections in a smartphone application that provides daily insulin titration guidance. If the fasting blood glucose is above the range, the application titrates up 4 units; if the fasting blood glucose is below the range, it titrates down 4 units; otherwise, no change is recommended. This is done every three days.

[0045] Example 1: Compliance

[0046] A patient has been compliant for the first 29 days of the period (as shown by the recorded data) and is now taking 40 units (IU) of insulin:

[0047] - No injections were recorded on days 33-37 (see Figure 1 )

[0048] - Day 38, patient requests dose recommendation

[0049] - Dose-response algorithm (see below) Assuming that 44 units were administered on days 33-37, a dose-response was created for the missing days based on data from days 1-37 (see Figure 2 )

[0050] - The evaluation algorithm (see below) determined that the mean of the fasting blood glucose measurements during days 33-37 fell within the 95% confidence interval for the response

[0051] - Based on the above information, the assessment algorithm determines that the patient has adhered to 44 units of insulin per day and continues to titrate from day 32 by increasing the recommended dose by 4 units

[0052] The above is summarized in Table 1 below.

[0053] Table 1: Daily FBG values ​​in compliant patients

[0054]

[0055] Example 2: Non-compliance

[0056] A patient has been compliant for the first 29 days of the period (as shown by the recorded data) and is now taking 40 units of insulin:

[0057] - No injections were recorded on days 33-37 (see Figure 3 )

[0058] - Day 38, patient requests dose recommendation

[0059] -Dose-response algorithm assumes that a dose of 44 units is administered on days 33-37 and the dose response is calculated based on data from days 1-37 (see Figure 2 )

[0060] - The evaluation algorithm determines that the mean of the fasting blood glucose measurements during days 33-37 falls outside the 95% confidence interval for this response

[0061] - Based on the above information, the assessment algorithm determines that the patient is non-adherent and recommends no dose change and to continue titration from Day 32

[0062] Table 2: Daily FBG values ​​in non-adherent patients

[0063]

[0064]

[0065] In Examples 1 and 2, all five injections were missed for a period of five injections according to the prescribed regimen. This period may be shorter or longer because the size of one or more of the injection doses may be known.

[0066] In the above embodiment, the dose response algorithm is based on a global linear regression model, assuming a linear relationship between glucose measurements, i.e.

[0067] G i =α-βI i +e i ,i=1,2…,k (1)

[0068] Among them G i represents the fasting blood sugar on day 1, and i It is the insulin injection given on the corresponding day.

[0069] The methods used for model identification are described below. These methods were used to investigate whether the expected dose response could be detected from the data and whether outliers in the data had a significant impact on model identification. Subsequently, the parameters of the three model structures were determined and it was determined which model gave the best fit to the individual dose responses.

[0070] To estimate α and β, the LSQ method can be used. For all n points of an individual, it can be written in matrix form,

[0071] y=Xθ+e (2)

[0072] where y=[G1,G2,…,G n ] T , X=[1,x], where x=[I1,I2,…,I n ] T ,e=[e1,e2,…,e n ] T And θ=[α,β] T The ordinary LSQ estimation assumes that the residual

[0073]

[0074] where r = [r1, r2, …, r n ], normally distributed around the mean response around. are the estimated parameters and are obtained by minimizing the residual sum of squares,

[0075]

[0076] The solution is a vector, It is an estimate of the unknown parameter θ. By substituting expression (3) into the 2-norm, we get

[0077]

[0078] Minimizing this expression with respect to θ and therefore taking the derivative with respect to this parameter gives the normal equation

[0079]

[0080] And Solve to get the estimated value of the unknown parameter,

[0081]

[0082] The distribution of this parameter estimate is

[0083]

[0084] where σ 2 is the estimated noise covariance,

[0085]

[0086] And n and n θ are the number of data points and parameters, respectively.

[0087] The estimated noise covariance is used to calculate confidence intervals, e.g.,

[0088]

[0089] in is the predicted glucose value, z represents the confidence interval (eg, 90% vs. 95%), and n is the number of data points.

[0090] In the illustrated embodiment, all data points have the same weight in the normal LSQ method, however, the points can have different weights, for example, through the robust LSQ method:

[0091] In the ordinary LSQ method, it is assumed that all data points are of equal quality. However, given the level of variation in SMBG data, this may not be the case. In order to minimize the sensitivity to outliers and errors in the fit, a weighted LSW can be used, where instead of minimizing the term in (4), we minimize

[0092]

[0093] where w i is the residual r i The weights can be chosen in different ways using knowledge of the data.

[0094] The normal equation becomes

[0095]

[0096] Where W = diag(w1,w2,…,w m ), and Solve to get the estimated value of the unknown parameter,

[0097]

[0098] The distribution of the parameter estimate is then

[0099]

[0100] Robust LSQ is performed here using bisquare weighting to minimize the effect of outliers on the fit. The method is iterative and gives full weight to small residuals and zero weight to residuals larger than expected by random chance. The weights are calculated iteratively as follows

[0101]

[0102] where u i is the adjusted and normalized residual r of the weighted LSQ i .

[0103]

[0104] Here, h i is the residual r i The leverage ratio, i.e., the degree to which the ith residual affects the fit, K is the fine-tuning constant, and s is the robust variance,

[0105]

[0106] The advantage of bisquare weighting is that it fits the data in a similar way to the ordinary LSQ method while eliminating the impact of outliers.

[0107] Alternatively, an asymmetric weighting function that takes into account time or user error can be used, such as Figure 5 As shown, the left figure shows the asymmetric weighting function of the residuals in (13). The right figure shows the forgetting weights in (14) for an effective memory period of 10 days. It should be noted that only three of the seven points are shown. This is due to the structure of the SMBG data, where only the last three values ​​before dose adjustment are available.

[0108] If an individual mistakenly measures SMBG at another time point, the glucose concentration is expected to be equal to or higher than the actual pre-breakfast SMBG. This will generally result in a positive error in the SMBG measurement. It can therefore be expected that outliers caused by user error will tend to elevate the measured glucose concentration.

[0109] Taking into account user error and the severity of hypoglycemia, the weighting function is designed so that low SMBG values ​​are weighted more than high SMBG values. However, it should be kept in mind that insulin does affect glucose levels, so eliminating information about the dose response is undesirable. Therefore, the SMBG reading is weighted compared to other SMBG readings when the same insulin dose is given. Therefore, the weighting function is

[0110]

[0111] Among them G i is the ith SMBG measurement, I i is the corresponding insulin injection, and G(I i ) is the measurement G of all SMBGs for j = 1, 2, ..., n j , where I j =I i When there is only one SMBG measurement for the corresponding insulin dose, dim(G(I i ))=1 and w i = 1. The weighting function is Figure 5 Shown on the left side of .

[0112] As described above, the determination of compliance or non-compliance can be used to provide insulin dosage recommendations for the subject in an effective and safe manner. More specifically, in the system described above, the method performed may include the following additional steps: receiving a dosage guidance request (DGR), and determining whether the subject has complied with the regimen in terms of insulin injections within a predetermined amount of time before receiving the DGR. If the subject complies, the system will provide an updated dosage recommendation (e.g., 48 units as described above) based on the received FBGH and IH, or if the subject does not comply, the system will maintain the current dosage recommendation (e.g., 44 units as described above).

[0113] In the above description of the exemplary embodiments, different structures and means for providing the described functions for the different components have been described to the extent that the concept of the invention will be readily understood by a skilled reader. The detailed construction and specifications of the different components are considered to be the object of a normal design procedure performed by a skilled person along the lines set forth in this specification.

Claims

1. A diabetes management system adapted to determine a subject's compliance with treatment according to a basal insulin regimen, the system adapted to receive: - regimen data, which lists the current dose size and prescribed injection frequency, - Multiple fasting blood glucose (FBG) measurements taken over the course of time for the subject to establish a FBG history (FBGH), wherein each respective glucose measurement in the multiple glucose measurements comprises: (i) FBG value, and (ii) the corresponding FBG timestamp, - a plurality of insulin injection data sets during the whole or part of said time course, thereby establishing an insulin injection history (IH), each injection data set comprising: (i) injection volume, and (ii) an injection timestamp indicating when the injection occurred during the stated time course, The system comprises one or more processors and a memory, the memory comprising instructions, which when executed by the one or more processors, perform a method comprising the following steps: - determining if one or more insulin injection data sets have not been received in accordance with the prescribed injection regimen over a period of time and are therefore missing, and if the insulin injection data sets are missing, then: - Calculate the expected dose-response FBG value for each missed injection based on the received FBGH and IH and using the dose-response algorithm, - determining, for a given confidence interval, whether the received FBG value corresponding to the missing insulin injection corresponds to the calculated dose-response FBG value, and - if the received FBG value corresponds to the calculated dose-response FBG value, then the subject is determined to be compliant with the basal insulin regimen, or - If the received FBG value does not correspond to the calculated dose-response FBG value, it is determined that the subject is not complying with the basal insulin regimen.

2. The diabetes management system of claim 1, wherein the predicted dose-response FBG value is based on received FBGH and IH and regimen dose data.

3. The diabetes management system according to claim 1 or 2, further adapted to provide insulin dosage recommendations to the subject, the method comprising the following additional steps: - Receive Dose Guidance Request (DGR), - determining whether the subject has been compliant with the regimen with respect to insulin injections for a predetermined amount of time prior to receiving the DGR, - If the subject is compliant, provide updated dosing recommendations based on the FBGH and IH received, or -If the subject is not compliant, maintain current dosing recommendations.

4. The diabetes management system of any one of claims 1-3, wherein the predetermined amount of time before receiving a DGR is the time since the last time a DGR was raised.

5. The diabetes management system according to any one of claims 1-4, wherein the FBG value is derived from received CGM data.

6. A method for determining a subject's compliance with treatment according to a basal insulin regimen, comprising the steps of: - obtain regimen data which lists the current dose size and prescribed injection frequency, - obtaining multiple fasting blood glucose (FBG) measurements of the subject over the course of time to establish a FBG history (FBGH), wherein each respective glucose measurement in the multiple glucose measurements comprises: (i) FBG value, and (ii) the corresponding FBG timestamp, - acquiring a plurality of insulin injection data sets during the whole or part of the time course, thereby establishing an insulin injection history (IH), each injection data set comprising: (i) injection volume, and (ii) an injection timestamp indicating when the injection occurred during the time period, - determining if one or more insulin injection data sets have not been received in accordance with the prescribed injection regimen over a period of time and are therefore missing, and if the insulin injection data sets are missing, then: - Calculate the expected dose-response FBG value for each missed injection based on the received FBGH and IH and using the dose-response algorithm, - determining, for a given confidence interval, whether the received FBG value corresponding to the missing insulin injection corresponds to the calculated dose-response FBG value, and - if the received FBG value corresponds to the calculated dose-response FBG value, then the subject is determined to be compliant with the basal insulin regimen, or - If the received FBG value does not correspond to the calculated dose-response FBG value, it is determined that the subject is not complying with the basal insulin regimen.

7. The method of claim 6, wherein the predicted dose-response FBG value is based on received FBGH and IH and protocol dose data.

8. The method according to claim 6 or 7, further adapted to provide insulin dosage recommendations to the subject, the method comprising the following additional steps: - Receive Dose Guidance Request (DGR), - determining whether the subject has been compliant with the regimen with respect to insulin injections for a predetermined amount of time prior to receiving the DGR, - If the subject is compliant, provide updated dosing recommendations based on the FBGH and IH received, or -If the subject is not compliant, maintain current dosing recommendations.

9. A method according to any one of claims 6-8, wherein the predetermined amount of time before receiving a DGR is the time since the last time a DGR was raised.

10. A method according to any one of claims 6-9, wherein the FBG value is derived from received CGM data.

Images