System and method for identifying clinically similar clusters of daily continuous glucose monitoring (CGM) curves

Abstract

Embodiments relate to a system for processing glucose data through efficient glucose database management. A system includes a physical data storage region containing glucose measurement data and a representation of at least one cluster of the glucose measurement data, where the representation is approximated as an array of glucose curve vectors of clusters of a plurality of glucose curves segmented by a plurality of time ranges. The system includes a processor and computer memory configured with instructions stored thereon that, when executed, will cause the processor to: 1) receive a glucose measurement; 2) converting a glucose measurement value into a vector form; 3) using similarity metrics to compare newly received glucose measurements with the centroid of the cluster to search for a physical data store; 3) based on the comparison, classifying newly received glucose measurements with clusters with matched similarity metrics; and 4) giving a treatment regimen according to the newly accepted glucose measurements.

Description

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS

[0002] This patent application is related to and claims the benefit of priority of U.S. Provisional Application No. 63 / 443,918 filed on February 7, 2023 and U.S. Provisional Application No. 63 / 335,361 filed on April 27, 2022, the entire contents of each of which are incorporated by reference. Technical Field

[0003] Embodiments relate to systems that process glucose data through efficient glucose database management and use the categorized glucose data to monitor, analyze, influence, etc., the concentration of glucose levels in a fluid. Background Art

[0004] Glucose variability (GV) in diabetes reflects the underlying biobehavioral process of blood glucose (BG) fluctuations, which has two main dimensions: amplitude, which reflects the extent of BG excursions, and time, which reflects the frequency of BG changes and the rate of event progression. Over the past 20 years, the ability to observe this process has evolved from intermittent self-monitoring (e.g., BG determinations several times per day) to contemporary continuous glucose monitoring (CGM), which captures dense datasets of BG readings at equal time intervals (e.g., every 5 minutes). These datasets, known as time series, have opened up new possibilities for the analysis and optimal control of human metabolic systems in diabetes, including assessment of system dynamics, prediction of glycemic trends and events (e.g., impending hypoglycemia or hyperglycemia), and automated closed-loop control, often referred to as an “artificial pancreas.”

[0005] The widespread adoption of CGM technology inevitably generates large amounts of data; for example, our latest report on the real-life use of artificial pancreas systems was based on more than 1 billion data points. The diabetes data ecosystem plays an increasingly important role in supporting data sharing, virtual clinics, and remote access. Cloud databases accumulate this data and require the use of data science tools such as pattern recognition, neural networks, deep learning, and artificial intelligence, all of which can help improve treatment and create fully automated systems. The most promising application of cloud databases and data science tools is the use of adaptive technologies that can "learn" and personalize treatment for each individual. To do this, it is necessary to create an appropriate structure in the CGM data space that well represents the clinical significance of the CGM data curves while being simple, finite, and fixed so that the structure does not need to change with each new data set.

[0006] Many indices of glycemic control exist, and we discussed them in detail in a 2017 paper published in Nature Reviews Endocrinology. CGM-based indices should generally include some notion of the timing of CGM readings, not just their magnitude. Several existing measures, such as MAGE (mean magnitude of glucose excursions) and LBGI / HBGI (low and high BG indexes), have also been adapted for use with CGM: the adaptation of MAGE to CGM data follows the classical time-independent structure of the measure, so that in this case CGM is used only as a source of magnitude assessment; the adaptation of LBGI and HBGI account for differences between SMBG and CGM data. The mean of daily differences (MODD) was introduced as a measure of intraday variability, and the continuous overlapping net glycemic action (CONGA) was presented as a composite measure of the magnitude and timing of glycemic excursions captured over different time periods. The standard deviation of the rate of BG change was used as a marker of the stability of the metabolic system over time, based on the premise that more erratic BG changes are a sign of systemic instability. A range of standard deviations was introduced to reflect the GV contained within different clinically relevant periods of CGM data, and the clinical interpretation of various CGM-based indices of glucose variability was discussed. A review of statistical methods that can be used to analyze CGM data includes several plots, such as the Poincaré plot for system stability and variability grid analysis (VGA) for visualizing glucose fluctuations captured by CGM.

[12] VGA has also been used to characterize the effectiveness of closed-loop control algorithms. [4]

[13] A perspective published in Diabetes Care re-evaluated several methods for GV calculation and visualization in the context of the relationship between GV and hypoglycemia risk, and we refer readers to this article for a more detailed discussion of the interpretation of VGA and Poincaré plots for CGM data.

[0007] As the CGM field is overloaded not only with a large number of complex data sets but also with a large number of metrics used to assess various aspects of the CGM profile, the 2019 international consensus on time in range (TIR, typically 70-180 md / dL), in which we participated, proposed TIR as the primary metric for CGM-based glycemic control and set clinical goals for its use. Over the past 3 years, the “TIR metric system” has been widely adopted. The TIR system is based on the active glucose profile (AGP) and was introduced as a template for data presentation and visualization. The standardized CGM report, originally developed by Mazze et al., introduced core CGM metrics and goals and the 14-day integrated glucose profile as an integral component of clinical decision making. This recommendation was endorsed by the international consensus and is also cited in the American Diabetes Association’s 2019 Standards of Care and the AACE consensus on the use of CGM. The AGP report has now been adopted by most CGM device manufacturers in their CGM companion software. Examples of AGP reports and the TIR metric system are shown in Figure 2. Figure 2 shown.

[0008] The TIR metric system defines 5 times in range for blood glucose values. In addition to the AGP, these times in range are used to provide a numerical interpretation of the AGP graph. In one embodiment, for example, the ranges within these times in range are: Grade 2 Hypoglycemia - less than 54 mg / dL, Grade 1 Hypoglycemia - 54 to 69 mg / dL, In Target Range (TIR) ​​- 70 to 180 mg / dL, Grade 1 Hyperglycemia - 180 to 250 mg / dL, Grade 2 Hyperglycemia - greater than 250 mg / dL. Other embodiments of TIR are shown in accordance with consensus recommendations for different types of diabetes. Figure 3 middle.

[0009] like Figure 2 and Figure 3 As shown, neither the AGP nor the TIR metric systems represent the day-to-day variability of the CGM traces and do not provide a fixed, finite structure for a large number of daily CGM curves.An aspect of embodiments of the systems, methods, and computer readable media of the present invention takes the next step. Summary of the invention

[0010] Embodiments may relate to a system for processing glucose data through efficient glucose database management. The system may include a physical data storage area, which contains glucose measurement data and a representation of at least one cluster of glucose measurement data. The representation may be approximated as a blood glucose curve vector array of clusters of multiple glucose curves segmented by multiple time ranges. The system may include a processor and a computer memory, the processor and computer memory are configured with instructions stored thereon, and the instructions, when executed, will cause the processor to perform any method steps disclosed herein. The instructions may cause the processor to receive glucose measurements. The instructions may cause the processor to convert the glucose measurements into vector form. The instructions may cause the processor to use a similarity metric to compare the newly received glucose measurements with the centroid of the cluster to search the physical data storage area. The instructions may cause the processor to classify the newly received glucose measurements based on the comparison with the clusters with matching similarity metrics. The instructions may cause the processor to give a treatment plan based on the newly received glucose measurements.

[0011] Embodiments may relate to a method for processing glucose data for efficient glucose database management. The method may include receiving a glucose measurement. The method may include converting the glucose measurement into a vector form. The method may involve searching a physical data store by comparing a newly received glucose measurement with a centroid of a cluster using a similarity metric. The physical data store may contain glucose measurement data and a representation of at least one cluster of glucose measurement data. The representation may be approximated as a blood glucose curve vector of a cluster of multiple glucose curves segmented by multiple time frames. The method may involve classifying newly received glucose measurements based on comparing clusters with matching similarity metrics. The method may involve providing a treatment regimen based on the newly received glucose measurement. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Other features and advantages of the present disclosure will become more apparent when the following detailed description is read in conjunction with the accompanying drawings, in which like elements are represented by like reference numerals, and in which:

[0013] Figure 1A is an exemplary system that can be used to process glucose data through efficient glucose database management;

[0014] Figure 1B is an exemplary system that can be used to develop a glucose database for clustered data sets;

[0015] Figure 2 is an exemplary dynamic glucose profile with recommended time in range;

[0016] Figure 3It is a TIR international consensus recommendation, which is displayed as a CGM-based target visualization;

[0017] Figure 4 An exemplary single iteration of a process that may be used to identify and evaluate a candidate set of CSCs is shown;

[0018] Figure 5 An exemplary CGM-based target visualization for each of the 35 CSCs sorted by TIR is shown:

[0019] Figure 6 The points (dp) for k∈{T54, T70, TIR, T180, T250} are shown when using the daily CGM curves from the test dataset (but excluding the 1,169 daily CGM curves from healthy individuals). ik , CSC k (dp i ))'s exemplary scatter plot;

[0020] Figure 7 An exemplary individual (f i ), mean value (f G ), and fitted traces stratified by health status and treatment modality;

[0021] Figure 8 An exemplary frequency and cumulative frequency distribution of daily CGM curves in a test dataset of 35 CSCs, stratified by health status and T1D treatment modality, are shown;

[0022] Fig. 9 CSC index box plots showing exemplary daily curves for T1D-MDI, T1D-PMP, T1D-CLC, T2D-MDI, and healthy subgroups; pairwise comparisons between T1D-MDI, T1D-PMP, T1D-CLC, T2D-MDI, and healthy subgroups (Bonferroni corrected);

[0023] Fig.10 An exemplary two steps in an iterative process for determining the "best" set of CSCs are shown, where each step uses a different dataset (a training dataset for identifying a candidate set of CSCs and a validation dataset for evaluating the candidate set of CSCs);

[0024] Fig.11 An exemplary visualization of all 35 CSC centroids sorted by TIR is shown, with the centroids with the highest TIR on the left and the centroids with the lowest TIR on the right;

[0025] Fig.12shows an exemplary scatter plot of points obtained by classifying 141,867 daily CGM curves of a test data set;

[0026] Fig.13 Shows the point pair An exemplary Hexbin plot of wherein the plots for “all individuals”, “healthy individuals”, and “T1D-CSII individuals” use a logarithmic scale as the color scale;

[0027] Fig.14 is an exemplary 3-panel graph showing the progression of three individuals with T1D over 14 days;

[0028] Fig.15 is an exemplary 4-panel graph illustrating the ability of CSC sets to distinguish between health states and treatment modalities;

[0029] Fig.16A , Fig. 16B , Fig. 16C , Fig.16D , Fig.16E , Fig.16E , Fig.16F , Figure 16G , Fig.16H , Fig.16I and Fig.16J is an exemplary illustration of the relationship between CSC and AGP;

[0030] Fig.17 An exemplary high-level functional block diagram of an embodiment of the system is shown;

[0031] Fig.18 An exemplary network system is shown in which embodiments of the systems and methods may be implemented;

[0032] Fig.19 An exemplary block diagram is shown that illustrates a system including a computer system and associated Internet connection upon which embodiments may be implemented;

[0033] Fig. 20 illustrates an exemplary system in which one or more embodiments of the systems and methods may be implemented using a network or portions of a network or a computer; and

[0034] Fig.21 An exemplary block diagram is shown that illustrates one example of a machine upon which one or more aspects of embodiments of the systems and methods may be implemented. DETAILED DESCRIPTION

[0035] Embodiments may relate to a system 100 for processing glucose data through efficient glucose database management. The system 100 may include a physical data storage area 102, which contains glucose measurement data and a representation of at least one cluster of glucose measurement data. The representation may be approximated as a blood glucose curve vector array of clusters of multiple glucose curves segmented by multiple time ranges. The system 100 may include a processor 104 and a computer memory 106, the computer memory 106 being configured with instructions 108 stored thereon, which when executed will cause the processor 104 to implement any method steps disclosed herein. The instructions may cause the processor 104 to receive glucose measurements. The instructions may cause the processor 104 to convert the glucose measurements into vector form. The instructions may cause the processor 104 to search the physical data storage area 102 by comparing the newly received glucose measurements with the centroid of the cluster using a similarity metric. The instructions may cause the processor 104 to classify the newly received glucose measurements based on the comparison with the clusters having a matching similarity metric. The instructions may cause the processor 104 to give a treatment plan based on the newly received glucose measurements. The treatment plan may be a command signal, a modification signal, a suggestion, etc. for an insulin dose, a bolus dose, an exercise plan, a meal consumption plan, a medication plan, etc.

[0036] The instructions may cause the processor 104 to store the classification result of the newly received glucose measurement value in a data storage area 102, which communicates with other devices 110 (e.g., one or more of a predictive modeling system, a decision support system, an insulin delivery system, an insulin monitoring system, an automatic control system, etc.) configured to use the classification result as input. In addition or alternatively, the instructions may cause the processor 104 to send the classification result of the newly received glucose measurement value to other devices 110 (e.g., one or more of a predictive modeling system, a decision support system, an insulin delivery system, an insulin monitoring system, an automatic control system, etc.) configured to use the classification result as input. In addition or alternatively, the instructions may cause the processor 104 to use the classification result of the newly received glucose measurement value to monitor, analyze, or affect the concentration of the glucose level in the fluid.

[0037] In some implementations, the instructions may cause the processor 104 to receive glucose measurements from a glucose measurement device or data source 112 (eg, a glucose monitor / sensor, a continuous glucose monitor / sensor, an assay device, etc.).

[0038] In some embodiments, system 100 may include a glucose measurement device or data source 112 .

[0039] In some embodiments, the system 100 may include a data storage area 102 that communicates with other devices 110 (e.g., one or more of a predictive modeling system, a decision support system, an insulin delivery system, an insulin monitoring system, an automated control system, etc.). In some embodiments, the system 100 may include other devices 110 (e.g., one or more of a predictive modeling system, a decision support system, an insulin delivery system, an insulin monitoring system, an automated control system, etc.).

[0040] In some implementations, the instructions may cause the processor 104 to calculate a Euclidean distance between one or more newly received glucose measurements and one or more centroids as a similarity measure.

[0041] In some embodiments, the physical data storage area 102 may include multiple clusters. Clusters may be generated by generating an array of glucose measurements for each time frame. Multiple arrays may form blood glucose curve vectors. Arrays may be assigned weights. An iterative hierarchical clustering technique may be applied until one or more clusters that are approximately one or more blood glucose curve vectors are generated. Any one or more clusters in the multiple clusters may be defined by the centroid of the cluster.

[0042] In some embodiments, iterative hierarchical clustering techniques can calculate R by linear regression of the array. 2 value, and change the weights to maximize R 2 value.

[0043] In some embodiments, the plurality of time ranges may include five time ranges. For example, the plurality of time ranges may be 1) a level 2 hypoglycemia below glucose measurement value -1; 2) a level 1 hypoglycemia within the range of glucose measurement value -2 and glucose measurement value -3; 3) a target range (TIR) ​​within the range of glucose measurement value -4 and glucose measurement value -5; 4) a level 1 hyperglycemia within the range of glucose measurement value -6 and glucose measurement value -7; and 5) a level 2 hyperglycemia above glucose measurement value -8. In a non-limiting example, glucose measurement value -1 may be 54 mg / dl; glucose measurement value -2 may be 54 mg / dl; glucose measurement value -3 may be 70 mg / dL; glucose measurement value -4 may be 70 mg / dL; glucose measurement value -5 may be 180 mg / dL; glucose measurement value -6 may be 180 mg / dL; glucose measurement value -7 may be 250 mg / dL; and glucose measurement value -8 may be 250 mg / dL.

[0044] In some embodiments, the glucose measurement value may include multiple glucose curves for an individual. Each glucose curve may include multiple glucose measurement values ​​obtained within a predetermined time period. The instructions may cause the processor 104 to compile multiple glucose curves into a single glucose measurement time series for an individual. The instructions may cause the processor 104 to classify one or more glucose curves using one or more clusters to generate an index sequence representing the classification results of one or more glucose curves in a single glucose measurement time series.

[0045] In some embodiments, the instructions cause the processor 104 to generate a trace representing glucose variability of the individual using the index sequence.

[0046] In some implementations, the instructions may cause the processor 104 to generate an approximate dynamic glucose report (AGP) using the index sequence.

[0047] In some embodiments, one or more of the plurality of glucose curves may be a continuous glucose monitoring (CGM) curve comprising glucose measurements obtained over a 24-hour period. Additionally, one or more of the plurality of glucose curves of an individual may be a continuous glucose monitoring (CGM) curve comprising glucose measurements obtained over a 24-hour period.

[0048] Embodiments may relate to a method for processing glucose data for efficient glucose database management. The method may include receiving glucose measurements. The method may include converting the glucose measurements into vector form. The method may involve searching a physical data storage area 102 by comparing a newly received glucose measurement with a centroid of a cluster using a similarity metric. The physical data storage area 102 may contain glucose measurement data and a representation of at least one cluster of glucose measurement data. The representation may be approximated as a blood glucose curve vector of a cluster of multiple glucose curves segmented by multiple time ranges. The method may involve classifying newly received glucose measurements based on comparisons with clusters having matching similarity metrics. The method may involve providing a treatment regimen based on the newly received glucose measurements. The treatment regimen may be a command signal, a modification signal, a suggestion, etc. for an insulin dose, a one-time dose, an exercise plan, a meal consumption plan, a medication plan, etc.

[0049] In some embodiments, the method may include calculating a Euclidean distance between one or more newly received glucose measurements and one or more centroids as a similarity measure.

[0050] In some embodiments, the physical data storage area 102 may include multiple clusters generated by: 1) generating an array of glucose measurement values ​​for each time range, the multiple arrays forming a blood glucose curve vector; 2) assigning weights to the arrays; and 3) applying an iterative hierarchical clustering technique that changes the weights until one or more clusters that approximate one or more blood glucose curve vectors are generated; 4) defining clusters in the set of clusters by the centroids of the clusters.

[0051] The method may involve calculating R by linear regression of the array via an iterative hierarchical clustering technique. 2 value and change the weights to maximize R 2 value.

[0052] As can be appreciated from this disclosure, embodiments relate to systems 100 and methods for processing glucose data through efficient database management. This allows glucose data to be classified and the classified glucose data to be used to monitor, analyze, influence, etc. the concentration of glucose levels in a fluid. Some embodiments may relate to methods and systems for developing a database for classification, and some embodiments may relate to methods and systems for implementing data processing using a database.

[0053] An embodiment of the system 100 includes a processor 104 configured to construct a database of cluster data for glucose measurement classification and / or implement data processing for classifying glucose measurement values. The processor 104 can be any processor 104 disclosed herein. The processor 104 can be part of the machine 2000 (logic, one or more components, circuits (e.g., modules) or components) or communicate with the machine 2000. The processor 104 can be hardware (e.g., processor, integrated circuit, central processing unit, microprocessor, core processor, computer device, etc.), firmware, software, etc., which is configured to operate by executing instructions embodied in algorithms, data processing program logic, artificial intelligence programming, automatic reasoning programming, etc. It should be noted that the use of the processor 104 herein includes any one or combination of a graphics processing unit (GPU), a field programmable gate array (FPGA), a central processing unit (CPU), etc. The processor 104 may include one or more processing modules. The processing module may be a software or firmware operating module configured to implement any method steps disclosed herein. The processing module may be embodied as software and stored in a memory, which is operably associated with the processor 104. The processing module may be embodied as a web application, a desktop application, a console application, etc. Exemplary implementations of the processor 104 and the machine 2000 are discussed below.

[0054] The processor 104 may include or be connected to a computer or machine readable medium 2002. As discussed in more detail below, the computer or machine readable medium 2002 may include a memory 106. Any memory 106 discussed herein may be a computer readable memory configured to store data. The memory 106 may include volatile or non-volatile, temporary or non-temporary memory, and may be embodied as in-memory, active memory, cloud memory, etc. Implementations of the memory 106 may include processor modules and other circuits to allow data to be transmitted to and from the memory 106, which may include transmitting data to and from other components of the communication system. The transmission may be via hardwire or wireless transmission. The communication system may include a transceiver, which may be used in combination with a switch, a receiver, a transmitter, a router, a gateway, a waveguide, etc., to facilitate the transmission of controlled and coordinated signals and processing communications to any other component or combination of components of the communication system via a communication method or protocol. The transmission may be performed via a communication link. The communication link may be electronic based, optical based, optoelectronic based, quantum based, etc.

[0055] The computer or machine readable medium 2002 may be configured to store thereon one or more instructions 108. The instructions 108 may be in the form of an algorithm, program logic, etc. that causes the processor 104 to construct and / or implement a classification model.

[0056] Processor 104 can communicate with other processors of other devices 110 (for example, prediction modeling system, decision support system, insulin delivery system, insulin recommendation system, blood sugar state or insulin monitoring system, glucose or insulin management system, automatic control system, etc.) configured to use classification results as input. Any one of those other devices 110 can include any one of the exemplary processors disclosed herein. Any processor can have a transceiver or other communication device / circuit to facilitate the transmission and reception of wireless signals. Any one of the processors can include an application programming interface (API) as a software intermediary, which allows two applications to talk to each other. The use of API can allow the software of the processor 104 of system 100 to communicate with the software of the processor of other devices 110.

[0057] Any transmission between processors / devices / systems / modules can be a push operation, a pull operation, or a combination of the two. Any transmission can be a direct transmission between two components or a transmission via an intermediary. For example, the intermediary can be a memory, a database, a data storage area, etc. For example, data from one processor can be transferred to a database for storage before being transferred to another processor. As another example, the data can be transferred to an intermediate processor or processing module to process the data, format the data, encode the data, etc. before being transferred to another processor. Data transmission between components can be performed continuously, periodically, on some other predetermined schedule, as required by a control signal, based on the conditions satisfied by each algorithm function, etc.

[0058] Exemplary systems and methods for developing a database of clustered data sets

[0059] Embodiments may relate to a system 100 for developing a database to classify glucose data. System 100 may include a processor 104. System 100 may include a computer memory 106 having instructions 108 stored thereon, which when executed will cause processor 104 to implement any method steps disclosed herein. Instructions 108 may cause processor 104 to receive glucose curve data. Glucose curve data may include one or more glucose measurements. Glucose measurements may be measurements representing a time series of glucose level curves (e.g., patterns, behaviors, trends, etc.). Glucose curve data may be historical data, current data, and / or real-time data. Glucose curve data is received by processor 104. This may be performed continuously, periodically, or with some other predetermined schedules. Glucose curve data may be pulled by processor 104 from data source 112 and / or pushed to processor 104 from data source 112. The data source 112 can be a device (e.g., a glucose monitor / sensor, a continuous glucose monitor / sensor, an assay device, etc.) that generates glucose measurements or a data storage area 102 (e.g., a database) that stores glucose curve data. The glucose measurements can be glucose measurements of a fluid, such as an interstitial fluid, etc. The processor 104 can store the glucose curve data in a transient or persistent memory for later processing, or process the glucose curve data when the glucose curve data is received. For example, the processor 104 can receive the glucose curve data and aggregate the glucose curve data in storage. Aggregation can be based on the type of data, what the data represents, the time when the data is received, the time when the data is generated, etc., which can be embodied in metadata, for example.

[0060] Instruction 108 can make processor 104 generate a group of clusters from glucose curve data. As a non-limiting example, instruction can make processor 104 perform machine learning data mining technology, and this technology divides multiple groups of objects in glucose curve data into multiple classes of similar objects. Cluster can be configured to approximate multiple times in range of glucose curve data. For example, clustering can be performed so that one or more in the cluster are approximate to one or more times in range in glucose curve data. Time in range can be the duration of the glucose measurement value of glucose curve data having the value in the glucose measurement value range. For example, there can be a duration, during which glucose curve data has glucose measurement value G1 and G1 falls in glucose measurement value range xy. There can be multiple times in the range of setting. In other words, it is beneficial to know when and how many glucose curve data have the glucose measurement value that falls into the predetermined time in range. This information can be used to generate a cluster representing this information.

[0061] Instruction 108 can make processor 104 generate one or more cluster sets. Any set in the cluster set can be generated using hierarchical clustering technology. For example, for each time in range, a set of clusters can be generated by generating an array of glucose measurement values. One or more arrays can form a vector. For example, it is expected that there are five times in range, which will generate five arrays. More or less time in range (and arrays) can be used. One or more arrays (for example, all five arrays) can be used to generate vectors. Because the array includes glucose measurement values, the vector can be a blood glucose curve vector (or one or more blood glucose curve vectors). Weights can be assigned to the array, which can include assigning weights to one or more arrays. Weights can be values ​​from 0 to 1, such as weight functions or any other mathematical operators that give the array desired influence or effect. Any one or combination of weights can be determined by optimization functions, objective functions, cost functions, etc. Any one or combination of weights can be fixed or variable. Weights can be variable and are randomly set to arbitrary values ​​for the first or initial iteration. Weights can change at each iteration until optimal. For example, the instructions 108 may cause the processor 104 to apply an iterative hierarchical clustering technique that changes the weights until a set of clusters that approximates the blood glucose curve vector is generated. The approximation may be a best approximation, a desired approximation, an optimal approximation, etc. For example, the optimal approximation may be an approximation defined by an optimization function, an objective function, a cost function, etc.

[0062] Instruction 108 can make processor 104 define cluster (it can include any number of clusters) that define cluster set by the centroid of cluster.For example, a single cluster in cluster set can be defined by the single centroid of this single cluster.This can be done for one or more clusters.The centroid can be a statistical (weighted or unweighted) median, mean value, mode, etc.Therefore, each cluster can be defined by the value or variable representing the centroid of the cluster.Cluster set can be a set of values ​​or variables representing a cluster.As described herein, each value or variable represents the time in range of glucose curve data, or an approximation of the time in range.It is noteworthy that cluster data is an accurate representation (or representation) of the time in range of glucose data curve, but with a significantly reduced data set.

[0063] As can be appreciated, the system 100 can reduce the data required for glucose analysis, reduce the computing resources required for the system to process these data, etc. For example, a single number can be sent / processed instead of sending / processing a daily continuous glucose monitoring (CGM) curve (which is typically 288 data points). The system 100 can generate cluster data from any type of glucose measurement system (e.g., data from any type of measurement system, data from different glucose measurement systems, unnormalized data, etc.) and data related to one or more of type 1 diabetes, type 2 diabetes, etc. In other words, the system 100 may not know the type of data, the mode of measurement, etc. The system 100 improves robustness and accuracy because glucose curve data that is considered insufficient due to missing data, data from different data sources, or unnormalized data, etc., can be used to generate clusters.

[0064] Instruction 108 can make processor 104 store cluster set in data storage area 102.This can be physical data storage area 102.Data storage area 102 can communicate with other devices 110 (for example, one or more in prediction modeling system, decision support system, insulin delivery system, insulin monitoring system, automatic control system, etc.), and other devices 110 are configured to use one or more clusters in cluster set as input.For example, as explained in more detail below, decision support system can compare new glucose curve data with model, to classify new glucose curve data as falling within one or more time in range, and do so by assigning one or more centroid values ​​to new glucose curve data--for example, new glucose curve data can match with one or more clusters and give centroid cluster value matched with it.Alternatively, processor 104 can perform this function and send value to decision support system.Then this value is used as the representative or substitute of the time in range of glucose measurement value of glucose curve.Data storage area 102 can be a part of system 100 or a part of another system. Additionally or alternatively, system 100 may be a predictive modeling system, a decision support system, an insulin delivery system, an insulin monitoring system, an automated control system, etc., and use the data directly (e.g., avoiding the use of data storage area 102). The cluster set may be stored or used as a classification database. The database may be modified, learned, etc. based on additional or updated data.

[0065] Glucose curve data can include multiple glucose curves. System 100 can classify one or more glucose curves in multiple glucose curves by one or more clusters in cluster set. One or more glucose curves can include multiple glucose measurements obtained in a predetermined time period. For example, one or more glucose curves can be continuous monitoring glucose (CGM) curves. One or more CGM curves can include glucose measurements obtained in a 24-hour time period, which can include glucose measurements obtained every 5 minutes in a 24-hour time period. As described herein, system 100 can operate in the case of missing data. Therefore, although it can be expected that each CGM curve data includes 288 glucose measurements (e.g., 288 data points), system 100 can generate useful clusters with fewer data points.

[0066] When implementing the iterative hierarchical clustering technique, instructions 108 may cause processor 104 to calculate R by linear regression of each array. 2 Instructions 108 may cause processor 104 to change one or more weights to maximize R 2 The weights may be changed via an iterative or recursive process, which may be controlled by an optimization function, an objective function, a cost function, or the like.

[0067] It is contemplated that a plurality of times within range includes five times within range, although other numbers of times within range and other ranges of times within range may be used. These may be:

[0068] Grade 2 hypoglycemia below the glucose measurement -1;

[0069] Grade 1 hypoglycemia within the range of glucose measurement -2 and glucose measurement -3;

[0070] a target range (TIR) ​​between a glucose measurement value of -4 and a glucose measurement value of -5;

[0071] Grade 1 hyperglycemia within the range of a glucose measurement of -6 and a glucose measurement of -7; and

[0072] Grade 2 hyperglycemia above a glucose measurement of -8;

[0073] The times in the range can be:

[0074] Glucose measurement -1 was 54 mg / dl;

[0075] Glucose measurement -2 was 54 mg / dl;

[0076] Glucose measurement -3 was 70 mg / dL;

[0077] Glucose measurement -4 was 70 mg / dL;

[0078] Glucose measurement -5 was 180 mg / dL;

[0079] Glucose measurement -6 was 180 mg / dL;

[0080] Glucose measurement -7 is 250 mg / dL; and

[0081] Glucose measured -8 was 250 mg / dL.

[0082] When developing a database, glucose curve data may include glucose measurements from one or more individuals. Each individual may have one or more glucose curves. This robust data set may allow a model to be used to determine blood sugar trends, predict blood sugar states, use multivariate analysis of conditions and factors (e.g., dietary behavior, exercise behavior, medical conditions, age, sex, race, heart rate, respiratory rate, oxygen saturation, etc.) that are related to or cause blood sugar states, etc. For example, multivariate modeling techniques may be used to determine which conditions or factors statistically contribute to changes in blood sugar states, changes in hypoglycemia or hyperglycemia risk, etc., which may also be used to estimate the probability of changes in blood sugar states, changes in hypoglycemia or hyperglycemia risk, etc. Multivariate modeling techniques may include one or more of logistic regression with or without cubic splines, random forests, xgboost, support vector machines, nearest neighbors, artificial neural networks and / or long short-term memory (LSTM), multivariate analysis of variance (MANOVA), multivariate analysis of covariance (MANCOVA), principal component analysis (PCA), canonical correlation analysis, redundancy analysis (RDA), correspondence analysis (CA), canonical correspondence analysis (CCA), multidimensional scaling, discriminant analysis, linear discriminant analysis (LDA), clustering, recursive adaptive partitioning, vector autoregression, principal response curve analysis (PRC), etc. As a non-limiting example, the means, standard deviations, and / or cross-correlations of one or more of the conditions or factors and the cluster centroids may be fit with a logistic ridge regression model using cubic splines, for example, to generate an output that is an estimate or probability that a glycemic state will occur.

[0083] As mentioned above, each individual can have one or more glucose curves.If there are multiple glucose curves of an individual, instruction 108 can make processor 104 compile multiple glucose curves into a single glucose measurement time series of an individual---for example, a single glucose measurement time series of the entire glucose curve set of the individual.Instruction 108 can make processor 104 classify each glucose curve by one or more clusters in a cluster set to generate an index sequence representing the classification results of each glucose curve in a single glucose measurement time series.Instruction 108 can make processor 104 store the index sequence in data storage area 102 to become a part of a database.This can be completed for one or more individuals.Therefore, data storage area 102 can have an index sequence for each individual, and each sequence is an approximate value of the time within the range of the glucose measurement value in its corresponding time series.It should be noted that an individual can have the data of one or more time series.In addition, for the data of any single time series, one or more index sequences can exist.

[0084] The index sequence can allow the database to be used to determine blood glucose trends, predict blood glucose status, use multivariate analysis of conditions and factors (e.g., dietary behavior, exercise behavior, medical conditions, age, gender, race, heart rate, respiratory rate, blood oxygen saturation, etc.) that cause or involve individual blood glucose status, etc. In addition or alternatively, instructions 108 can cause processor 104 to generate a trace representing individual glucose variability using the index sequence. This can be done for one or more individuals. In addition, an individual can have one or more traces. Instructions 108 can cause processor 104 to store the trace in data storage area 102 to become part of the database.

[0085] Exemplary systems and methods for classifying glucose data

[0086] Embodiments may relate to a system 100 for classifying glucose data. The system 100 may be configured to classify glucose data using a database of cluster data to implement embodiments of the methods disclosed herein. The system 100 may include a processor 104. The system 100 may include a computer memory 106 having instructions 108 stored thereon, which when executed will cause the processor 104 to implement or apply embodiments of the methods disclosed herein. The instructions 108 may cause the processor 104 to receive glucose curve data including multiple glucose measurements. Optionally, the glucose data is of a single individual, so as to evaluate or assess the glycemic state of the individual by comparing the glucose curve data of the individual with the clusters in the database; however, the glucose curve data may be of one or more individuals. Optionally, the glucose curve data is recent data (e.g., data collected in real time or within the past 24 hours), but the glucose curve data may be historical, current and / or real-time data.

[0087] Instruction 108 can make processor 104 classify glucose curve data or a part thereof by comparing glucose curve data with the implementation of database. Database can include a group of clusters, which are configured to approximate one or more blood glucose curve vectors of individual and / or individual group to which individual belongs (for example, individual can be grouped by age, sex, race, medical condition, etc.). Blood glucose curve vector is an array of previously processed glucose curve data divided by multiple time in range. Previously processed glucose curve data include historical glucose data, but can also include current or real-time glucose data. Previously processed glucose data can be individual data, data of each individual in individual group (which can include or not include the data of the individual), data of each individual that can or cannot be in individual group, etc. One or more time in range can be a duration, during which the glucose measurement value of previously processed glucose curve data has a value within a certain glucose measurement value range.

[0088] The instructions 108 may cause the processor 104 to store the classification results of the glucose curve data in the data storage area 102, which communicates with other devices 110 configured to use the classification results as input (e.g., one or more of a predictive modeling system, a decision support system, an insulin delivery system, an insulin monitoring system, an automatic control system, etc.). Additionally or alternatively, the instructions 108 may cause the processor 104 to transmit the classification results of the glucose curve data to other devices 110 configured to use the classification results as input (e.g., one or more of a decision support system, an insulin delivery system, an insulin monitoring system, etc.). Additionally or alternatively, the instructions 108 may cause the processor 104 to use the classification results to monitor, analyze, and / or affect the concentration of the glucose level in the fluid.

[0089] Instruction 108 can make processor 104 classify glucose curve data by comparing glucose curve data with the centroid of the cluster in cluster set.For example, glucose curve data that matches accurately, approximately or similarly with the centroid can be classified as having a time pattern in range that is accurate, approximately or similar to the cluster to which the centroid belongs.Any individual's glucose curve data can have one or more classification results.Any individual's glucose curve data can include one or more glucose curves of the individual.One or more classification results can be arranged for any glucose curve.It is expected that the classified glucose curve has only one classification result (that is, it best matches with the centroid of a cluster).In the unlikely case that the similarity score is equal, the first match can be selected.

[0090] The model can have a set of clusters. For example, the number of clusters can be 35. Each set of clusters can use more or fewer clusters. There can be one or more sets of clusters. The number of clusters in one set of clusters can be the same or different from the number of clusters in another set of clusters. The number of clusters, the number of groups, etc. can be set by desired design criteria (e.g., optimization, computing resources, processing speed, accuracy, robustness, etc.). The comparison can be to compare the glucose curve data with one or more clusters (or centroids) within the same group, within different groups, within a single group of clusters, multiple groups of clusters, etc.

[0091] Instruction 108 can make processor 104 compare glucose curve data with one or more centroids using similarity measure.The cluster with the best similarity measure can be used to classify glucose curve data.Similarity measure can be a numerical value falling within a value range (e.g., from 0 to 1).Similarity measure 0 can indicate a match, and similarity measure 1 can indicate a mismatch between a degree of match grade between 0 and 1. Alternatively, similarity measure 1 can indicate a match, and similarity measure 0 can indicate a mismatch between a degree of match grade between 1 and 0. Other similarity measure schemes can be used.In an exemplary embodiment, instruction 108 can make processor 104 calculate the Euclidean distance between one or more glucose curve data points and one or more centroids as a similarity measure.For example, the distance can be normalized to fit within the range of 0 to 1.

[0092] Glucose curve data may include one or more glucose curves of an individual. Each glucose curve may include multiple glucose measurements obtained in a predetermined time period (e.g., within a 24-hour time period). Instructions 108 may enable processor 104 to compile multiple glucose curves into individual single glucose measurement time series. Instructions 108 may enable processor 104 to classify each glucose curve by one or more clusters in a cluster set to generate an index sequence representing the classification results of each glucose curve in a single glucose measurement time series. Instructions 108 may enable processor 104 to store the index sequence in a data storage area 102, which communicates with other devices 110 (e.g., one or more of a prediction modeling system, a decision support system, an insulin delivery system, an insulin monitoring system, an automatic control system, etc.) configured to use the classification results as input. Additionally or alternatively, instructions 108 may enable processor 104 to send the index sequence to other devices 110 (e.g., one or more of a prediction modeling system, a decision support system, an insulin delivery system, an insulin monitoring system, an automatic control system, etc.) configured to use the classification results as input. Additionally or alternatively, the instructions 108 may cause the processor 104 to use the classification results to monitor, analyze, and / or influence the concentration of glucose levels in the fluid.

[0093] Instructions 108 may cause processor 104 to generate an approximate dynamic glucose report (AGP) using the index sequence. Additionally or alternatively, other devices 110 (e.g., one or more of a predictive modeling system, a decision support system, an insulin delivery system, an insulin monitoring system, an automatic control system, etc.) may generate an approximate dynamic glucose report (AGP) using the index sequence.

[0094] As described herein, the glucose curve data may include a plurality of glucose curves, each glucose curve including a plurality of glucose measurements obtained within a predetermined time period. For example, each glucose curve may be a continuous glucose monitoring (CGM) curve including glucose measurements obtained within a 24-hour time period.

[0095] It is contemplated that a plurality of times within range includes five times within range, although other numbers of times within range and other ranges of times within range may be used. These may be:

[0096] Grade 2 hypoglycemia below -1 on the glucose measurement;

[0097] Grade 1 hypoglycemia within the range of glucose measurement -2 and glucose measurement -3;

[0098] a target range (TIR) ​​between a glucose measurement value of -4 and a glucose measurement value of -5;

[0099] Grade 1 hyperglycemia within the range of a glucose measurement of -6 and a glucose measurement of -7; and

[0100] Grade 2 hyperglycemia above a glucose measurement of -8;

[0101] The times in the range can be:

[0102] Glucose measurement -1 was 54 mg / dl;

[0103] Glucose measurement -2 was 54 mg / dl;

[0104] Glucose measurement -3 was 70 mg / dL;

[0105] Glucose measurement -4 was 70 mg / dL;

[0106] Glucose measurement -5 was 180 mg / dL;

[0107] Glucose measurement -6 was 180 mg / dL;

[0108] Glucose measurement -7 is 250 mg / dL; and

[0109] Glucose measured -8 was 250 mg / dL.

[0110] Any other device 110 may be configured to generate an output based on the classification result input. In some embodiments, the other device 110 may be part of the system 100, for example, the system 100 may include the other device 110. The classification result may be used by the processor 104 or the processor of the other device 110 to generate the following signals:

[0111] a. recommending or implementing a process to obtain additional data (e.g., generating a signal that additional patient data, insulin delivery data, metabolic data, etc. is needed);

[0112] b. Recommend or implement a process to initiate a signal for preventive or mitigating action (e.g., generate a signal to change insulin rate, change behavior, etc.);

[0113] c. Recommending or implementing a process to initiate a signal for enhanced monitoring (e.g., generating a signal to notify the user that the risk of hypoglycemia is increased and additional monitoring should be performed);

[0114] d. The signal is an alarm signal or a command signal sent to the insulin delivery device to modify the insulin rate or dosage, etc.

[0115] The system 100 or any other device 110 may include a display configured to generate a user interface. A user may control various aspects of the system 100 via the user interface. In addition, the user interface may display various aspects of the classification results and other outputs, generate graphical displays, audible, graphic or textual alerts, etc.

[0116] The system 100 may include a processor 104 combined with one or more data storage areas 102. The data storage area 102 may be configured to include multiple classification databases. For example, the system 100 may be configured to generate multiple classification databases. The processor 104 may be configured to use any one or combination of multiple classification databases. Each classification database may be generated based on available glucose data and other patient data, the expected availability of glucose data or other patient data, the quality of glucose data or other patient data (how reliable the data is), the frequency of glucose data or other patient data (how often it is generated or available), the dimension of glucose data or other patient data (how many attributes or variables the data has), etc. For example, a first classification database may be generated for a data set in which some types of data are scarce but other types of data are abundant, a second classification database may be generated for a data set in which some data have low reliability but other types of data have high reliability, etc. The type of patient data may include which data source 112 is received or attempted (or expected) to receive data, which attributes are included in the data, the number of attributes the data has, etc. A classification database may be generated for an expected data stream, thereby generating multiple classification databases. Multiple classification databases may be stored in one or more data storage areas 102. The processor 104 may communicate with the data store 102 to access any one or combination of a plurality of classification databases.

[0117] The processor 104 may be configured to switch from the first classification database to the second classification database for implementation based on at least one or more of the following: type of data, availability of data, reliability of data, etc. The processor 104 may detect the above changes (e.g., based on metadata) and switch the classification database.

[0118] The processor 104 can be configured to update the classification database based on the new data. As described above, the glucose curve data can be historical, current and / or real-time data, and can be received continuously, periodically, or on some other predetermined schedule, and can include information about glycemic episodes, treatments, etc. The system 100 can update any one or combination of the classification databases based on the updated data. The updated classification database can replace the classification database that already exists in the data storage area 102. Alternatively, if the updated classification database is sufficiently different or more suitable for the patient data scenario than any other existing classification database, the updated classification database can be added to the multiple classification databases.

[0119] It will be appreciated from this disclosure that one aspect of embodiments of the present invention provides, among other things, a system, method, and computer-readable medium for identifying clinically similar clusters in daily continuous glucose monitoring (CGM) curves.

[0120] One aspect of an embodiment of the present invention provides, inter alia, a system, method, and computer-readable medium for: a) constructing and then fixing a set of clinical similarity clusters (CSCs), wherein the set of clinical similarity clusters (CSCs) has the following properties: for any other daily continuous glucose monitoring (CGM) curve, there exists a clinical similarity cluster (CSC) that is approximately within the time range of the daily CGM curve; and b) determining an approximation of any daily CGM curve by the CSC.

[0121] One aspect of an embodiment of the system, method, and computer-readable medium of the present invention includes, for example, but not limited to, two steps. For example, the first step may include: constructing and then fixing a set of clinical similarity clusters (CSCs) having the following properties: for any other daily CGM curve, there is a clinical similarity cluster (CSC) that is approximately within the time range of the daily CGM curve, thereby retaining the key clinically relevant features of the daily CGM curve.

[0122] A set of clinically similar clusters can be defined using hierarchical clustering, where the weights of the input columns are varied until the set of CSCs has a desired performance when approximated to within-range times of a daily CGM curve. For example, a second step can include determining an approximation of any daily continuous glucose monitoring (CGM) curve by a CSC, which can involve calculating a similarity metric (e.g., Euclidean distance) between a candidate daily CGM curve and the centroid of each CSC, and selecting a single CSC with a minimum similarity metric value.

[0123] In one embodiment, when these steps are completed, any daily CGM curve can be mapped to a CSC, and then the individual's CSC sequence can be used as a surrogate for the individual's dynamic glucose profile (AGP) and the associated in-range time of the original daily CGM curve. In addition to the AGP and its related metrics, the sequence of CSCs also provides information about the timing and day-to-day variability of clinically relevant glycemic events for the patient. Potential applications of one aspect of embodiments of the systems, methods, and computer-readable media of the present invention include, but are not limited to, one or more of the following: (i) data structuring and dimensionality reduction; (ii) database indexing; (iii) compression / encryption of daily CGM curves; (iv) distinguishing between health conditions and treatment modalities; (v) CGM replacement for common clinical tests; (vi) CGM pattern recognition and prediction; or (vii) tracking disease progression.

[0124] One aspect of embodiments of the systems, methods, and computer-readable media of the present invention may be configured to, inter alia, work with daily CGM curves generated by sensors having different sampling resolutions, and with daily CGM curves having missing data up to a certain threshold. One of the significant advantages of one aspect of embodiments of the present invention is, but is not limited to, the ability to classify all CGM daily curves into a relatively small, finite, and fixed set of CSCs across patient groups and health states that describe the clinical state of those patients well. One aspect of embodiments of the present invention also adds a time-varying component to commonly accepted CGM data representations, such as, but not limited to, AGP and its associated time in range.

[0125] An aspect of one embodiment of the present invention provides a system, method, and computer-readable medium, inter alia, for identifying clinically similar clusters of daily CGM curves.

[0126] One aspect of an embodiment of the present invention provides a system, method and computer readable medium, particularly for providing classification results of daily CGM curves and clinical interpretation thereof.

[0127] One aspect of an embodiment of the present invention provides a system, method, and computer readable medium, inter alia, for defining a set of CSCs, wherein any daily CGM curve can be classified as one of the CSCs, and wherein the CSCs reliably approximate the clinical characteristics of the daily CGM curve.

[0128] One aspect of an embodiment of the present invention provides a system, method and computer readable medium, inter alia, for providing a CSC that clearly distinguishes between health states and treatment regimes and can be used as a representation of a person's blood glucose fluctuations over time.

[0129] As described herein, one aspect relates to a method for identifying clinically similar clusters of daily continuous glucose monitoring (CGM) curves. The method may include:

[0130] Get individual i, where each individual i generates a single CGM time series during the course of the study they participated in;

[0131] All daily CGM curves in a single time series for an individual are sorted, resulting in an indexed sequence s i (may be discontinuous) which shows the CSC that each daily CGM curve is classified as;

[0132] The sequence s i Each entry of corresponds to a single observation day, and the days are sorted by occurrence date, and where f i(t) is the number of unique CSCs visited by individual i after t days of observation;

[0133] where f i is the trace of the number of unique CSCs visited by individual i over time and provides a view of individual glycemic variability;

[0134] Provides a mean trace that shows the average behavior of individuals in a subgroup and can help highlight differences in behavior between different subgroups; where for a given subgroup G, the mean is

[0135]

[0136] Where |I| is the total number of individuals in the subgroup; not all individuals in the subgroup have their CSC index sequence s i have the same number of observation days; therefore, is defined only if there is a minimum number of sequences at time t; and wherein the minimum number of sequences is a function of the subgroup G; and

[0137] Health states and treatment modalities were distinguished as they tracked glycemic variability; wherein, on average, healthy individuals had the least glycemic variability, followed by individuals with T2D, and then individuals with T1D; and wherein, for individuals with T1D, individuals using MDI as a treatment modality had the greatest glycemic variability on average, followed by PMP individuals, and then CLC individuals.

[0138] The method may further include:

[0139] Data structuring and dimensionality reduction, wherein the large number of all possible daily CGM curves clinically represented by AGPs and their time in range are reduced to a finite and fixed set of CSCs;

[0140] Database indexing, where the database is indexed by a structure defined by the CSC, which will ensure fast and efficient searching for subgroups of similar daily CGM curves;

[0141] compressing and / or encrypting the daily CGM curve;

[0142] Differentiate between health states and treatment modalities; and

[0143] The ability of CSCs to differentiate health states with high fidelity serves as a surrogate for clinical testing by wearing a CGM for a designated number of days in a home setting, along with a predefined schedule of meals and physical activity, which will yield a diagnostic result.

[0144] One aspect relates to a method for a) constructing and then fixing a set of clinically similar clusters (CSCs) having the following properties: for any other daily continuous glucose monitoring (CGM) curve, there exists a clinically similar cluster (CSC) that approximates the time within range of the daily CGM curve, and b) determining an approximation of any daily CGM curve by the CSC, as described herein.

[0145] One aspect relates to a system for a) constructing and then fixing a set of clinically similar clusters (CSCs) having the following properties: for any other daily continuous glucose monitoring (CGM) curve, there exists a clinically similar cluster (CSC) that is approximately within the time range of the daily CGM curve, and b) determining an approximation of any daily CGM curve by the CSC, as described herein.

[0146] One aspect relates to a computer-readable storage medium having stored thereon computer-executable instructions which, when executed by one or more processors, cause one or more computers to perform functions for: a) constructing and then fixing a set of clinically similar clusters (CSCs) having the following properties: for any other daily continuous glucose monitoring (CGM) curve, there exists a clinically similar cluster (CSC) that is approximately within the time range of the daily CGM curve, and b) determining an approximation of any daily CGM curve by the CSC, as described herein.

[0147] One aspect relates to a method configured to present a two-step iterative process to identify a fixed set of clinically similar clusters (CSCs) of daily CGM curves. The two-step process uses hierarchical clustering on a training data set configured to identify a set of CSC candidates. The two-step process uses a validation data set configured to evaluate the performance of the set of CSC candidates. The CSCs are evaluated for their ability to accurately capture five different in-range times of the classified daily CGM curves. A fixed set of 35 CSCs, Ψ, is then used to classify daily CGM curves in a separate test data set, and wherein the results show that the set is robust and generalizes well.

[0148] In some embodiments, the distribution of daily CGM curves relative to different CSCs is displayed specific to health states and treatment modalities.

[0149] One aspect relates to a method for visualizing individual glycemic control. Clinically similar clusters (CSCs) can be used to visualize differences in glycemic control between individuals with the same health status and treatment modality to identify individuals who may need more personalized attention. In some embodiments,

[0150]

[0151] Among them, u i is the number of unique CSCs required to classify k daily CGM curves of individual i.

[0152] In some embodiments:

[0153] The total number of unique CSCs can be bounded (i.e., there are exactly 35 different CSCs) if k is large, then will tend to 0;

[0154] k is fixed to 28 daily curves (i.e. 4 weeks of data); and

[0155] Defined as The average value of The values ​​are calculated using a sliding window of k = 28 daily CGM curves for each The sliding window is 7 days ahead (1 week of data), and each sliding window must have at least 14 daily CGM curves (2 weeks of data) to calculate

[0156] In some embodiments, mi is the average CSC index of k daily CGM curves for individual i, and is m i The average value of the value, m i The values ​​are calculated using a sliding window of k = 28 daily CGM curves for each m i The generated sliding window is 7 days ahead and each sliding window must have at least 14 daily CGM curves to calculate

[0157] In some embodiments, any daily CGM curve can be approximated by one of 35 pre-fixed clinically similar clusters (or a specified number of pre-fixed clinically similar clusters). The approximation means that when a daily CGM curve is classified as a CSC, the CSC retains the information carried by the original daily CGM curve based on a time measurement system within the range. CSC expands, and to some extent completes, the interpretation of CGM data provided by the AGP / TIR system - where AGP / TIR is a static snapshot of 14 days (or a specified number of days) of data, and the CSC sequence derived from the same data tracks the progress of glycemic control over time. The time series of CSCs over 14 days (or a specified number of days) illustrates how stable or unstable a person's glycemic control is.

[0158] One aspect of embodiments of the systems, methods, and computer-readable media of the present invention generally relates to, but is not limited to, drugs and medical devices for insulin therapy of diabetes and other metabolic disorders, including, but not limited to, type 1 and type 2 diabetes, type 2 (T1D, T2D), latent autoimmune diabetes in adults (LADA), postprandial or reactive hyperglycemia, or insulin resistance. In an alternative embodiment, an aspect of an embodiment of the present invention defines and then fixes a set of clinically similar clusters (CSCs) having the following properties: for any other daily CGM curve, there is a clinically similar cluster (CSC) that approximates the time within the range of the daily CGM curve, thereby retaining the key clinically relevant features of the daily CGM curve. When the CSC is defined and fixed, any daily CGM curve can be mapped to the CSC, and then the individual's CSC sequence can be used as a substitute for the individual's dynamic glucose profile (AGP) and the associated time within the range of the original daily CGM curve. In addition to the AGP and its related metrics, the sequence of CSCs also provides information about the time and day-to-day variability of clinically relevant glycemic events of the patient. One of the significant advantages of one aspect of one embodiment of the present invention is, but is not limited to, the ability to classify all CGM daily curves into a relatively small, finite and fixed set of CSCs across patient groups and health states that well describes the clinical state of these patients.

[0159] Example

[0160] The following are examples of embodiments of developing, testing, and implementing the systems and methods disclosed herein. The following are intended to be illustrative only and not limiting.

[0161] Example 1

[0162] data

[0163] The data used in this work comes from:

[0164] The University of Virginia Diabetes Technology Center (i.e., DCLP1

[19] , DCLP3

[20] , DIAMOND1[2], DIAMOND2

[22] , DSS1

[23] , NIGHTLIGHT

[24] , and TRIALNET studies), and

[0165] Jebb Health Research Center sites (CITY

[25] , DCLP5

[26] , NDIAB

[27] , MDEX

[28] , REPLACE-BG

[29] , RT-CGM

[30] , SENCE

[31] , SEVHYPO

[32] , and WISDM

[33] studies).

[0166] Based on each of these studies, the procedure outlined in Section III.A of reference

[34] was used to process the CGM time series and define a daily CGM curve, where a daily CGM curve is a time series of 288 blood glucose data points collected every 5 minutes from midnight to midnight (24 hours). The 204,710 daily CGM curves from these 16 different studies were used to form three different datasets, each with a different purpose:

[0167] 1. Training data set: This data set is from D CLP 1. D CLP 3. D IAMOND 1. D IAMOND 2. D SS 1 and N IGHTLIGHT A total of 23,916 daily CGM curves were studied and used to define a candidate set of CSCs.

[0168] 2. Validation dataset: This dataset consists of 37,758 daily CGM curves, which are also from D CLP 1. D CLP 3. D IAMOND 1. D IAMOND 2. D SS 1 and N IGHTLIGHT The results are studied and used to a) evaluate the performance of each CSC candidate set and b) select the final and fixed CSC set.

[0169] 3. Test data set: This data set is from C ITY , D CLP 5. D IAMOND 2. N DIAB 、M DEX , R EPLACE-BG , R T-CGM , S ENCE , S EVHYPO 、T RIALNET and W ISDM A total of 143,036 CGM curves were studied and used to evaluate the robustness and generalizability of the final selected CSC set.

[0170] The studies represented healthy individuals, individuals with type 1 diabetes (T1D), and individuals with type 2 diabetes (T2D). The studies also represented a variety of treatment modalities, including multiple daily injections (MDI), insulin pumps (PMPs), and closed-loop control (CLC).

[0171] There were 2,462 subjects and a total of 204,710 daily CGM curves in the 16 datasets. The characteristics of the participants in each study are detailed in Table 1.

[0172] Table 1: Characteristics of the 16 datasets used in this work. Statistics are presented as mean (SD) unless otherwise stated. T1D denotes type 1 diabetes, T2D denotes type 2 diabetes, BMI denotes body mass index, CGM denotes continuous glucose monitoring, MDI denotes multiple daily injections, PMP denotes insulin pump, and CLC denotes closed-loop control. * indicates data not available at the subject level and therefore taken from the study protocol.

[0173]

[0174]

[0175] The majority (95.9%) of daily CGM curves were generated by T1D subjects treated with MDI, insulin pump (PMP), or closed-loop control (CLC). Two datasets focused on children with T1D (D CLP 5 and S ENCE ). IAMOND The 2 dataset contained data from participants with T2D who were receiving MDI therapy and accounted for 3.6% of the daily CGM curves generated. DIAB and T RIALNET The dataset contains data from a nondiabetic (healthy) population. Glycemic control, assessed by the mean HbAlc per participant at baseline, was 5.2% (N DIAB research) to 9.1% (C ITY Healthy populations generally have data for less than 7 days, whereas the vast majority of populations in other studies have data for an average of 5 weeks or more.

[0176] Identify clinically similar clusters of daily CGM curves

[0177] In one embodiment, for example, the 5 times within range are: within a 24 hour period, Grade 2 hypoglycemia [[T54]]: 54 mg / dL, Grade 1 hypoglycemia [[T70]]: 54 mg / dL to 69 mg / dL, within target range [[TIR]]: 70 mg / dL to 180 mg / dL, Grade 1 hyperglycemia [[T180]]: 180 mg / dL to 250 mg / dL, Grade 2 hyperglycemia [[T250]]: above 250 mg / dL. In other embodiments, other times within range may be used depending on specific clinical indicators for different populations (e.g., pregnant women with diabetes, where the recommended TIR is 63 mg / dl to 140 mg / d1). In one aspect of an embodiment of the invention, these times within range are used as input features for the proposed clustering algorithm.

[0178] When performing hierarchical clustering, the time in range of a single daily CGM curve is used as the input for the single daily CGM curve. The input is generated using all the daily CGM curves in the training dataset. The scipy.cluster.hierarchy Python module

[35] implements hierarchical clustering using a centroid algorithm for calculating the Euclidean distance between two rows of input. Because we want to ensure that the CSC faithfully captures the behavior below the time in range, we weight the T54 and T70 input columns greater than the TIR, T180, and T250 input columns. Each clinically similar cluster (CSC) will be a collection of daily CGM curves such that each daily CGM curve in the collection has substantially the same time in range.

[0179] Figure 4 An exemplary process for identifying and then evaluating a single set of CSC candidates is shown. For a given set of inputs (determined using the daily CGM curves in the training dataset and the weights selected for the T54 and T70 columns), the hierarchical clustering algorithm can produce a dendrogram that indicates the hierarchical relationships between the daily CGM curves in the training dataset. "Cutting" the dendrogram at a specific height can produce cluster groups with a specific number of clusters. The evaluation of each CSC candidate set can use the validation dataset. The centroid of the CSC is used to classify each daily CGM curve in the validation dataset. Using dp ijk is the in-range time k from the jth daily CGM curve of individual i, and CSC k (dp ij ) is the time k within the range of CSC for which the jth daily CGM curve of individual i is classified, where k∈{T54, T70, TIR, T180, T250}. In addition, let

[0180]

[0181] and

[0182]

[0183] Among them, J i is the set of all daily CGM curves for individual i. We want a single “optimal” set of CSCs where:

[0184] 1. Through each k point (dp ik , CSC k (dp i )) that maximizes the r-squared value of linear regression, and

[0185] 2. Ensure that the absolute value of the relative effect size of each k is less than 0.15, where the relative effect size is calculated by dp ijk and CSCk (dp ijk ) divided by the average of all dp ijk Calculate the standard deviation of .

[0186] We explored 9 different weights for the T54 and T70 input columns (i.e., 9 different sets of inputs to the hierarchical clustering algorithm), which resulted in a candidate set of 166 CSCs being evaluated. The final fixed set of CSCs had 35 clusters. Each CSC is defined by a centroid. The centroid of a given CSC is calculated using the daily CGM curves in the training dataset assigned to the CSC. The centroid of each CSC can be visualized as a CGM-based target.

[0187] Figure 5 A CGM-based target visualization associated with each of the 35 CSC centroids is shown. Figure 5 The CSC centroid visualization in is sorted by their TIR values ​​(from highest on the left to lowest on the right). Analysis of this figure shows that, as expected, no two CSC visualizations are identical. Figure 6 The points (dp ik , CSC k (dp i )) , Table 2 provides the relative effect size for each k∈{T54, T70, TIR, T180, T250} when the 35 CSCs are used for a set of 141,869 daily CGM curves from the test dataset, where the set does not include the 1,169 daily CGM curves belonging to healthy individuals in the test dataset. These results indicate that the final set of 35 CSCs faithfully represents the clinical characteristics of the daily CGM curves they approximate.

[0188] Table 2: Relative effect sizes and linear regression results for k∈{T54, T70, TIR, T180, T250} when using daily CGM curves from the test dataset (but not daily CGM curves from healthy individuals).

[0189]

[0190] Table 3 of the 2019 International Consensus on Time in Range

[14] defines guidelines for two diabetes groups:

[0191] 1. Adults with T1D or T2D, and

[0192] 2. Elderly / high-risk individuals with T1D or T2D.

[0193] Table 3: Fitted values ​​of parameters γ, λ and k in the modified Weibull equation for six different subgroups.

[0194]

[0195] CSC1 meets guidelines for adults with T1D or T2D, and CSC5 meets guidelines for older / high-risk individuals with T1D or T2D. Therefore, physicians have one target CSC for both conditions.

[0196] Tracking blood sugar fluctuations:

[0197] Unique CSCs accessed over time

[0198] Each individual i generates a single CGM time series during the study they participated in. Note that a CGM time series may have periods where CGM data is not collected (e.g., during a washout period of a study). Sorting all daily CGM curves in an individual's single time series generates an index sequence s i (may be discontinuous) which indicates the CSC that each daily CGM curve is classified as. i Each entry of corresponds to a day of observation, and the days are sorted by the date of occurrence. i (t) is the number of unique CSCs visited by individual i after t days of observation. i is a trace of the number of unique CSCs visited by individual i over time, and provides a view of the individual's glycemic variability. In general, a larger number of unique CSCs visited by an individual indicates increased glycemic variability (and therefore worse glycemic control). Note that in extreme cases, individuals with greater glycemic variability visit only a small number of CSCs, and those visited CSCs represent greater glycemic variability.

[0199] Average trace

[0200] The mean trace represents the average behavior of individuals in a subgroup and helps highlight differences in behavior between subgroups. For a given subgroup G, the mean is

[0201]

[0202] where |I| is the total number of individuals in the subgroup. Not all individuals in the subgroup have their CSC index sequence s i The number of days observed in t is the same. Therefore, it is defined only if there is a minimum number of sequences at time t. The minimum number of sequences is a function of the subgroup G.

[0203] In general, we want to estimate the long-term (more than 2-3 months) behavior of a subgroup, especially for those subgroups that lack long-term data. To achieve this, we fit a curve to the average trace of the subgroup. Figure 7 As shown in the first three plots of , the curve has the general shape of a cumulative density function, and we conjecture that modifying the Weibull cumulative distribution function (CDF)

[36] to take into account the non-unit upper bound will provide a good fit. The following modified Weibull CDF is used as input to the curve_fit function of the scipy.optimize SciPy module:

[0204]

[0205] Note that γ is capped so that it is no longer 1 in the Weibull CDF.

[0206] Figure 7 Individual, mean curves fitted by health status and treatment are shown. Figure 7 The gray dashed curves in the first three rows of the figure are the individual traces f i ,and Figure 7 The thick solid lines in the first three rows of the graph are the average traces Figure 7 The thick solid line in the bottom row is the average trace The bold dashed lines are modified Weibull curves fitted to each average trace (the parameters of these fitted curves can be found in Table 3). These curves again distinguish between health status and treatment mode because they track blood glucose fluctuations. As expected, on average, healthy individuals have the least blood glucose fluctuations, followed by individuals with T2D, and then individuals with T1D. For individuals with T1D, individuals using MDI as treatment mode have the greatest blood glucose fluctuations on average, followed by individuals taking PMP, and then individuals taking CLC.

[0207] The fitted modified Weibull curves reached thresholds after approximately 200 days of observation: the maximum number of unique CSCs accessed was higher in subjects with T1D than in subjects with T2D (13.3 CSCs vs 9.2 CSCs), and was higher in subjects treated with MDI than in subjects treated with PMP or CLC (14.4 CSCs vs 13.5 CSCs vs 8.3 CSCs, respectively).

[0208] application

[0209] Data structuring and dimensionality reduction

[0210] The large set of all possible daily CGM curves, as clinically represented by AGPs and their time in range, is reduced to a finite and fixed set of CSCs that can be used as input to decision support, clinical and automated treatment algorithms.

[0211] Database Index

[0212] A database indexed by a CSC-defined structure can ensure fast and efficient searching for subgroups of similar daily CGM curves. This can enable new capabilities in decision support or automated insulin delivery systems, such as algorithms that learn from a person's CGM patterns as well as from the patterns of other patients stored in a population database.

[0213] Compression / encryption of daily CGM curves

[0214] Instead of sending a daily CGM curve (typically 288 data points), a single number can be sent that identifies the CSC index of the original daily CGM curve. On the receiving end, a decoder equipped with this set of CSCs can reconstruct the AGP clinical features of the daily CGM curve, such as time in range and other clinical indicators, with the fidelity of preserving these indicators of the original daily CGM curve.

[0215] Distinguishing between health status and treatment

[0216] One application of CSCs is the ability to differentiate between health states and treatment modalities. Figure 8 The frequency distribution of 143,036 daily CGM curves in 35 different CSCs in the test data set is shown. The results in these figures are stratified by health status and treatment mode. It is clear that the frequency distribution is a function of the health status of the individual under consideration (i.e., healthy individuals or individuals with T1D or T2D). In addition, for individuals with T1D, the frequency distribution is a function of the treatment mode (MDI, PMP or CLC). As expected, the vast majority (94.6%) of daily CGM curves generated by healthy individuals are classified as CSC1, and more than 99% of daily CGM curves are classified as one of CSC1, CSC2 or CSC3. A comparison of health status is performed between T1D-MDI and T2D-MDI, so the comparison is fair-because the treatment mode is the same, and the difference in frequency distribution should be attributed to the difference in diabetes type (T1D vs. T2D).

[0217] To formally test the visual observation results, an independent sample Kruskall-Wallis test was performed between the frequency distributions of the T1D-MDI, T1D-PMP, T1D-CLC, T2D-MDI, and healthy subgroups, with a result of P < .001. Pairwise comparisons adjusted for Bonferroni correction for multiple testing showed adjusted significance of P < .05 for all pairwise comparisons except the comparison between T1D-PMP and T2D-MDI (see Fig. 9 ), which suggests that pump therapy brings the clinical outcomes of T1D-MDI patients close to those of T2D, while the clinical outcomes of T1D-CLC are better than those of T2D.

[0218] CGM Alternatives for Common Clinical Trials

[0219] Measuring fasting glucose levels, homeostasis (HOMA) assessments of insulin sensitivity and beta-cell function or the oral glucose tolerance test (OGTT) are common clinical methods used to assess a person’s glycemic health status. These and other common tests of glycemic function typically require a physician visit, blood draw, and laboratory analysis. In the case of the OGTT, several hours of testing are required in a clinical setting. While cumbersome, these tests are routinely needed in many situations, such as the frequent OGTTs in gestational diabetes. CSC’s ability to distinguish between health states with high fidelity could serve as an alternative to these clinical tests – a CGM worn for 10 days in a home setting, accompanied by a predefined schedule of meals and physical activity, would yield diagnostic results similar to those accepted in clinical practice, while greatly simplifying data collection.

[0220] CGM pattern recognition and prediction

[0221] The transition probability matrix describing the evolution of a patient on a predefined CSC is a natural tool to observe the progression of a disease or treatment. Pattern recognition or cyclic behavior is reflected by patterns or cycles detected in the transition probabilities from one state to the next. Short-term or long-term predictions of glycemic control are based on probabilistic patterns or repeated visits to a certain subspace of the Markov chain state space. The latter is the subject of semi-Markov chain theory, which is obtained by aggregating (lumping) the state space into related subsets, which is characterized by the fact that the duration spent in each subset is random.

[0222] Tracking disease progression over time

[0223] Disease progression is indicated by a transition to an undesirable state, whereas successful treatment optimization of drug titration is reflected by a transition to a clinically desirable state. In practical applications, the state space of the Markov chain is defined / aggregated to correspond to the CSC defined by one aspect of embodiments of the present systems, methods, and computer-readable media.

[0224] Example 2

[0225] The ability to track blood glucose has evolved from intermittent self-monitoring to contemporary continuous glucose monitoring (CGM). However, clinical decision making based on CGM is difficult because CGM produces large and complex data sets that require advanced analytics to provide insights. This work proposes a two-step iterative process that reduces the "clinical dimension" of the CGM data space by identifying a fixed set of clinically similar clusters (CSCs) such that daily CGM curves within each cluster convey similar clinical information. The two-step process uses hierarchical clustering of the training dataset to identify candidate sets of CSCs and uses linear regression and relative effect sizes to evaluate the ability of the candidate sets of CSCs to capture five different in-range times of daily CGM curves from the validation dataset. The best set of 35 CSCs identified using the validation dataset is then used to classify daily CGM curves in a separate test dataset. The results show that the set of CSCs is robust, generalizes well, but most importantly captures the clinical characteristics of daily CGM curves with high fidelity. This fixed set of CSCs enables tracking of an individual's daily glycemic control over time, facilitating the design of personalized treatments and potentially enabling automated treatment optimization by mapping optimal treatment responses to predefined rules for each CSC. The CSCs can also be used to visualize differences in glycemic control between individuals and differences between treatment modalities, thereby identifying individuals who may benefit from treatment adjustments.

[0226] Glucose variability (GV) in diabetes reflects the underlying biobehavioral process of blood glucose (BG) fluctuations, which has two main dimensions: amplitude, which reflects the extent of BG excursions, and time, which reflects the frequency of BG changes and the rate of event progression. Observation of this process has evolved from intermittent self-monitoring, which produces a few BG readings per day, to contemporary continuous glucose monitoring (CGM), which produces large data sets, time series of glucose readings at equal time intervals (e.g., every 5 minutes). The continuous development of CGM technology has inevitably generated a large amount of data. CGM time series data are used to gain insights that allow better treatment of diabetes, including risk stratification, prediction of events of interest (e.g., impending hypoglycemia or hyperglycemia), or automated closed-loop control, often referred to as an “artificial pancreas.” To improve the clinical utility of CGM data and simplify their interpretation, the 2019 International Consensus on Time in Range (TIR) ​​proposed TIR as the primary indicator of CGM-based glycemic control and set clinical goals for its use. This “TIR indicator system” has been widely adopted over the past 3 years. The TIR system is based on the dynamic glucose profile (AGP) and is introduced as a template for data presentation and visualization. Initially proposed by Mazze et al., for occasional self-test data, the standardized CGM report introduced core CGM indicators and targets and a 14-day composite glucose profile as an integral component of clinical decision making.

[0227] One area of ​​existing literature focuses on using CGM data to cluster subjects (e.g., cluster subjects into groups at higher risk for gestational diabetes) or to build classification models (e.g., classify subjects as healthy, prediabetic, or diabetic based on their CGM data). Acciaroli et al. used 25 CGM-based indices of glycemic variability as input to a 2-step binary logistic regression model. The model first classified subjects as healthy or unhealthy and then classified those subjects who were not classified as healthy in the first step as affected by impaired glucose tolerance (IGT) or type 2 diabetes (T2D). The model was able to distinguish between healthy subjects and those with IGT or T2D, and between those with both IGT and T2D. Bartolome et al. developed an algorithm called GlucoMine that is designed to discover individualized patterns in long-term CGM data (3–6 months of data) that are not evident in short-term data. Gecili et al. used CGM data to identify phenotypes of glycemic variability in type 1 diabetes (T1D) through functional data analysis. They concluded that these phenotypes can be used to optimize T1D management in subgroups of subjects at highest risk for adverse outcomes. Inayama et al. derived summary statistics from CGM data and then used hierarchical clustering of the summary statistics to classify 29 women into three groups (low glucose levels with less glucose variability, L; moderate glucose levels with moderate to high glucose variability, M; high glucose levels with high glucose variability, H). They found that women with gestational diabetes mellitus (GDM) tended to belong to the H group, so their clustering approach could help identify subgroups of women with GDM characteristics. Li et al. collected CGM time series data and decomposed them into trend, seasonal (daily), and random components. The trend component was then clustered using k-means clustering, which produced 5 clusters, 2 of which had an increasing trend, 2 with a decreasing trend, and 1 without a trend (unchanged). They then used these five clusters to classify the subjects into 3 groups: increasing, decreasing, and unchanged. After 6 months of glucose-lowering treatment, the fasting blood glucose values ​​of the subjects assigned to the increasing and decreasing groups increased and decreased, respectively. Tao et al. clustered the 24-hour CGM time series generated by T2D subjects with the goal of identifying subjects with different degrees of dysglycemia and clinical phenotypes. Mao et al. developed a pipeline for analyzing CGM data with the goal of identifying glycoforms: groups of subjects that differ in their degree of control, time spent in range, and the presence and timing of hyperglycemia and hypoglycemia. They say their approach “can be used to guide targeted interventions for patients with diabetes” in addition to other biometric data.

[0228] Other literature uses CGM data to build analytical frameworks that can then be used in a number of different applications (e.g., identifying different subtypes of patients). Hall et al. developed an "analytic framework that can group individuals based on specific patterns of glycemic response called 'glycotypes,' revealing heterogeneity or subphenotypes within traditional glucose regulation diagnostic categories." Matabeuna et al. used CGM data to derive glucose density, where glucose density is the distribution of glucose values ​​from CGM time series for individual subjects, claiming that glucose density is an extension of the time metric over range. They suggest that glucose density can be used in clinical practice to provide "a more accurate representation of an individual's glycemic profile," "identify different subtypes of patients based on their glycemic profile and other variables," and even "determine if there are statistically significant differences between patients receiving different interventions."

[0229] The ultimate goal of all the papers cited above is to classify individuals with diabetes into different subgroups, and all classification methods do not capture the daily variation of glycemic control within individuals. To more fully exploit the temporal structure of CGM data, we define daily CGM curves, i.e., 24-hour CGM time series, and use them to build a limited set of 483 representative daily curves or motifs, which they claim can be matched to almost any daily CGM curve. This set of motifs has been externally validated and can be considered fixed. The motifs reflect not only differences between individuals, but also differences in the daily variation of glycemic control of individuals.

[0230] The work presented in this paper establishes an analytical framework based on a “TIR metric system” that is able to classify the daily glycemic behavior of individuals. This classification of a day provides the basis for a large number of different analyses: it can be used to define a fixed number of groups if desired, but can also be used to follow individuals over time for modeling purposes, for clinical subgroup stratification and transitions from one subgroup to another, or for informing automated control strategies, among others. Furthermore, while the existing literature assumes that the subjects have a specific type of diabetes, the analytical framework presented in this paper is agnostic to the type of diabetes and is equally applicable to CGM data generated by healthy individuals - the observed patterns recognize the glycemic state of the individual without the need for a prior classification or diagnosis.

[0231] This paper outlines data and methods for identifying clinically similar clusters (CSCs), which include using training and validation datasets to obtain a fixed set of CSCs using the described method, and then evaluating the performance of the fixed set of clusters on a separate test set.

[0232] method

[0233] The method includes: training, validation, and testing datasets; input generation for a hierarchical clustering method that defines a candidate set for identifying clinically similar clusters (CSCs); and a two-step iterative process for identifying the "best" set of CSCs.

[0234] Dataset source

[0235] The data used in this work comes from:

[0236] 1. University of Virginia Diabetes Technology Center (D CLP 1[4], D CLP 3[5]、D IA 1

[21] , D IA 2『22]、D SS 1『23]、N TLT 『24] and T RLNT research), and

[0237] 2. Jaeb Health Research Center website 1 (i.e. C ITY

[25] D CLP 5[6]、N DIAB

[26] M. DEX

[27] R EPBG

[28] R TCGM [29,30], S ENCE

[31] S. EVHYPO

[32] and W ISDM [33, 34] studies). For these studies, the analysis, content, and conclusions presented in this work are solely the responsibility of the authors and have not been reviewed by any research group.

[0238] In each of these studies, the procedure outlined in our previous paper

[19] was used to process CGM time series and define daily CGM curves, where the daily CGM curve is a time series of 288 blood glucose data points collected every 5 minutes from midnight to midnight (24 hours). There were 2,462 subjects and a total of 204,710 daily CGM curves in the 16 datasets. These studies represented healthy individuals, individuals with T1D, and individuals with T2D. The studies also represented a variety of treatment modalities, including multiple daily injections (MDI), insulin pumps (CSII), and closed-loop control (CLC). The characteristics of the participants in each study are detailed in Table 4. The majority (95.9%) of daily CGM curves were generated by subjects with T1D who were treated with MDI, CSII, or CLC. Two datasets (D CLP 5 and S ENCE ) focuses on children with T1D. IAThe 2 dataset contained data from participants with T2D who were receiving MDI therapy and accounted for 3.6% of the daily CGM curves generated. DIAB and T RLNT The dataset contains data from a nondiabetic (healthy) population. Glycemic control, as assessed by the mean HbA1c for each participant at baseline, ranged from 5.2% (N DIAB Healthy populations generally have data for less than 7 days, while the vast majority of other studies have data for an average of 5 weeks or more.

[0239] The 204,710 daily CGM curves were used to form three different datasets, each with a different purpose: Training dataset: This dataset was collected from D CLP 1. D CLP 3. D IA 1. D IA 2. D SS 1 and N TLT A total of 23,916 daily CGM curves were studied and used to define the candidate set of CSCs.

[0240] Table 4: Characteristics of the 16 datasets used in this work. Statistics are presented as mean (SD) unless otherwise stated. T1D denotes type 1 diabetes, T2D denotes type 2 diabetes, BMI denotes body mass index, CGM denotes continuous glucose monitoring, MDI denotes multiple daily injections, CSII denotes insulin pump, and CLC denotes closed-loop control. * indicates data not available at the subject level and therefore taken from the study protocol.

[0241]

[0242] 1. Validation dataset: This dataset consists of 37,758 daily CGM curves, which are also from D CLP 1. D CLP 3. D IA 1. D SS 1 and N TLT The results are studied and used to a) evaluate the performance of each CSC candidate set and b) select the final and fixed CSC set.

[0243] 2. Test data set: This data set is from City, D CLP 5. D IA 2. M DEX 、N DIAB , R EPBG , R TCGM , S ENCE , S EVHYPO, T RLNT and W ISDM A total of 143,036 daily CGM curves were studied and used to evaluate the robustness and generalizability of the final selected CSC set.

[0244] Hierarchical clustering was performed to identify clinically similar clusters of daily CGM curves. That is, the TIR metric system defined time in range within 5 blood glucose value ranges, which are:

[0245] 1. Grade 2 hypoglycemia (T54): blood sugar is strictly lower than 54 mg / dl.

[0246] 2. Grade 1 hypoglycemia (T70): blood glucose greater than or equal to 54 mg / dL and strictly less than 70 mg / dL,

[0247] 3. Target range (TIR): blood glucose greater than or equal to 70 mg / dL and less than or equal to 180 mg / dL,

[0248] 4. Grade 1 hyperglycemia (T180): blood sugar strictly greater than 180 mg / dL and less than or equal to 250 mg / dL, and

[0249] 5. Grade 2 hyperglycemia (T250): blood sugar is strictly greater than 250 mg / dL.

[0250] The five in-range times for each daily CGM curve were used as input for hierarchical clustering, which was calculated using the scipy.cluster.hierarchy Python module

[35] , where the centroid method was used to calculate the Euclidean distance between two rows of input. The T54 and T70 input columns were multiplied by the weight ω. T54 ≥1 and ω T70 ≥1 to emphasize the importance of these two input columns in the clustering process and ensure that the time behavior below the range (i.e., T70 and T54) is captured with very high fidelity. Each clinically similar cluster is a collection of daily CGM curves such that each daily CGM curve in the collection has substantially the same time within the range. The centroid of each cluster of daily CGM curves identified by hierarchical clustering can define the CSC. In contrast to our previously published work on daily CGM curves (see '19, 20]), CSC ignores the intra-day time of blood glucose changes. Therefore, the daily CGM curve consisting of 288 data points is reduced to a single CSC centroid consisting of only 5 data points, one data point for each of the 5 range times, involving extracting the time information contained in the daily CGM curve.

[0251] Iterative process to identify the “best” set of CSCs

[0252] Fig.10An exemplary two-step iterative process for identifying the "best" set of CSCs is shown. Inputs for defining a candidate set of CSCs are generated using the 23,916 daily CGM curves in the training data set. For a given set of inputs (determined using the daily CGM curves in the training data set and the weights selected for the T70 and T54 columns), the hierarchical clustering algorithm can produce a dendrogram that indicates the hierarchical relationship between the times within the range of the daily CGM curves in the training data set. "Cutting" the dendrogram at a specific height can define a specific set of N clusters, and the centroid of each cluster is calculated using the daily CGM curves assigned to that cluster. This set of N clusters is the candidate set of CSCs that must be evaluated subsequently.

[0253] Evaluation of each CSC candidate set began by classifying the 37,758 daily CGM curves in the validation dataset using the CSC centroid. k (dp ij ) is the time k that the jth daily CGM curve of individual i is within the range, and c k (dp ij ) is the time k at which the CSC centroid of the jth daily CGM curve of individual i is classified as within the range, where k∈{T54, T70, TIR, T180, T250}. In addition, let

[0254]

[0255] And, make

[0256]

[0257] Among them, J i is the set of all daily CGM curves for individual i. Considering the above definition, the “optimal” set of CSCs will be:

[0258] 1. For each k∈{T54,T70,TIR,T180,T250} has a relative effect size less than or equal to δ, and

[0259] 2. For each k∈{T54, T70, TIR, T180, T250}, through the point set Maximize the linear regression value r with intercept 2 .

[0260] The relative effect size was calculated as follows:

[0261]

[0262] The denominator is all 37,758 a k (dp ij) values. This part of the process uses linear regression with an intercept (specifically the scipy.stats.linregress function from the SciPy

[35] Python package) so that r 2 Values ​​can be compared.

[0263] Table 5: Relative effect sizes and linear regression results for five different in-range times {T54, T70, TIR, T180, T250} when the 35 CSCs in Ψ are used to classify 37,758 daily CGM curves in the validation dataset (top) and 143,036 daily CGM curves in the test dataset (bottom). Note that the linear regression results presented are for a linear regression with a fixed intercept of 0.

[0264]

[0265] In this application, the relative effect size δ = 0.15 was an attempt to ensure that a) relatively infrequent hypoglycemic events were captured with reasonably high fidelity, i.e., exchanging daily CGM curves with CSCs did not result in more than 15% bias, and b) the collection of CSCs captured glucose dynamics without being influenced by the individual generating the daily CGM curves.

[0266] result

[0267] An “optimal” set of clinically similar clusters was determined, and the performance of this set for a test dataset of daily CGM curves is also presented.

[0268] Identification of collections of clinically similar clusters

[0269] Nine different weight combinations (ω T54 ,ω T70). The upper limit of 60 and the lower limit of 15 restricted the number of candidate sets of CSCs considered: no “cuts” of the hierarchical cluster dendrograms that resulted in CSC candidate sets with more than 60 or less than 15 CSCs were evaluated using the second step of the iterative process. 166 CSC candidate clusters were evaluated, and the final selected (fixed) set of SCSs Ψ had 35 clusters. The centroid of a given CSC was calculated using the daily CGM curves in the training dataset assigned to that CSC. The relative effect sizes and linear regression results obtained by evaluating Ψ using the 37,758 daily CGM curves of the validation dataset are presented in the upper half of Table 5. Note that the linear regression results are linear regressions with a fixed intercept of 0, performed using the sklearn.linear model.LinearRegression function of the scikit-learn Python package. When Ψ was used to classify the 37,758 daily CGM curves in the validation dataset, the point set for each k∈{T54, T70, TIR, T180, T250} These results indicate that the final set of 35 CSCs selected faithfully represents the clinical characteristics of the approximated daily CGM curves. In particular, the CSCs capture the relatively uncommon hypoglycemic component of the daily CGM curves with high fidelity.

[0270] The centroid of the CSC can be visualized as a CGM-based target and in Fig.11 The height of each color in the visualization corresponds to the percentage of time the centroid spends in each range. Fig.11 A CGM-based target visualization associated with each of the 35 CSC centroids in Ψ is shown, sorted by their TIR values ​​(highest on the left to lowest on the right). Inspection of this figure shows that, as expected, no two CSCs are the same, since no two CSC centroid visualizations are the same. Furthermore, further inspection shows that most centroids have a significant hypoglycemic (T54 and T70) component. This is a result of weighting the T54 and T70 input columns when generating the CSC candidate set in the first step of the iterative process.

[0271] The centroids of the 35 CSCs in Ψ were then used to classify the 143,036 daily CGM curves in the test dataset. Fig.12 The point set for each k∈{T54,T70,TIR,T180,T250} is plotted The associated linear regression results and relative effect sizes are provided in the lower half of Table 5. These results suggest that Ψ is robust and generalizes to data it has not seen before and in the context of healthy individuals for which it was not trained.

[0272] Visualize individual blood sugar control

[0273] CSC can be used to visualize differences in glycemic control between individuals with the same health status and treatment modalities, and thereby identify individuals who may need more personalized attention.

[0274]

[0275] where u i is the number of unique CSCs required to classify the k daily CGM curves of individual i. Because the total number of unique CSCs is bounded (i.e., there are only 35 different CSCs), if k is large, then will tend to 0. To overcome this problem, k is fixed to 28 daily curves (i.e. 4 weeks of data). Then Defined as using a sliding window of k = 28 daily CGM curves to calculate each Generated The average of the values, where the sliding window is 7 days ahead (1 week of data), and each sliding window must have at least 14 daily CGM curves (2 weeks of data) to calculate For example, if individual i has 40 daily CGM curves, will be and The average value of Use daily CGM curves 1 to 28 to calculate, Use daily CGM curves 8 to 35 to calculate, Calculated using daily CGM curves 15 to 40, and Calculated using daily CGM curves 22 to 40. Note that if the number of daily CGM curves in the sliding window is less than k (as in (which has only 18 daily CGM curves in the sliding window of ), the denominator in equation (4) is set to the number of daily CGM curves in the sliding window. Because different CSCs are different (i.e., by construction, no two CSC centroids have the same time in range), It is a measure of the degree of fluctuation in an individual's day-to-day blood sugar control. The value of is close to 1 / k, indicating that the individual does not visit a large number of different CSCs and therefore does not have large daily fluctuations in blood glucose control. A value of much greater than 1 / k indicates that the individual's daily blood sugar control fluctuates more.

[0276] Make m i is the average CSC index of k daily CGM curves of individual i, and The sliding window of k = 28 daily CGM curves is used to calculate each m iThe generated m i The average of the values, where the sliding window is 7 days ahead, and where each sliding window must have at least 14 daily CGM curves to calculate Because the CSCs are indexed so that the centroid of CSC1 has the highest TIR value and the centroid of CSC35 has the lowest TIR value, is a measure of daily glycemic control for individual i. Values ​​of closer to 1 indicate that the individual spends more time within the target range, while values ​​closer to 35 indicate that the individual spends less time within the target range.

[0277] Fig.12 The points k∈{T54, T70, TIR, T180, T250} are shown An exemplary scatter plot of , which was obtained by classifying 141,867 daily CGM curves of the test dataset using Ψ.

[0278] Fig.13 Draw a point Hexagonal box plots, 2D histograms, "where the bins are hexagonal and the color indicates the number of data points within each bin." For individuals with T1D or T2D, the points Only in |J i | ≥ 14, i.e., when individual i has at least 14 daily CGM curves. For healthy individuals, this constraint is relaxed to |J i |≥4, because typically these individuals have no more than 7 daily CGM curves (see NDIAB and TRLNT rows)—in this case, and Only a single sliding window is used for computation.

[0279] Fig.13 The top left figure in the figure plots the points for all individuals The minimum value again reflects the wide range of glycemic control exhibited by individuals. The other figures of the paper further illustrate this point: even when considering single health states and treatment combinations, there is still a wide range of glycemic control. As expected, the points for healthy individuals Located in the lower left corner of the figure. The numerical values ​​are due to the small number of days of CGM data available for healthy individuals; in general, we expect healthy individuals to have The differences in the figure show that, as expected, individuals with T1D who were treated with CLC had better glycemic control in general than those with T1D who were treated with CSII, who in turn had better glycemic control than those with T1D who were treated with MDI.

[0280] Note that if Smaller but If the index is larger, the individual does not visit a large number of different CSCs and does not spend a large amount of time in the target range. This is the worst case scenario, which can be recognized by the individual in the upper left corner of the "T1D One Individual" figure. Further analysis shows that this individual spent 89% of their 74 daily CGM curves in CSCs with an index greater than or equal to 30, and 53% of the 74 daily CGM curves were classified as CSC 35. This individual is a typical example of someone who may benefit from more personalized clinical care.

[0281] In this work, we propose a two-step iterative process to identify a fixed set of clinically similar clusters (CSCs) of daily CGM curves. The two-step process uses hierarchical clustering on a training dataset to identify a candidate set of CSCs, and uses a validation dataset to evaluate the performance of the candidate set of CSCs. In particular, the ability of CSCs to accurately capture five different ranges of time of the classified daily CGM curves is evaluated. A fixed set of 35 CSCs Ψ is then used to classify daily CGM curves in a separate test dataset, and the results show that the set is robust and generalizes well. In addition, the distribution of daily CGM curves with respect to different CSCs is shown to be specific to health states and treatment modalities.

[0282] There are multiple possible applications of CSCs. A large number of all daily CGM curves, as clinically represented by their range time, are reduced to a finite and fixed set of CSCs, which can be used as input to decision support, clinical and automated treatment algorithms. In addition, a database indexed by the structure defined by CSCs can ensure fast and efficient searches for subgroups of clinically similar daily CGM curves. CSCs also allow tracking of changes in an individual's daily glycemic control over time, which has the potential to enable new features in decision support, or inform automated insulin delivery systems, such as algorithms that learn from a person's CGM patterns and patterns of other patients stored in a population database. Finally, abstracting the typical 288 data points of a daily CGM curve into a single CSC index is both a compression and encryption of the data. These applications are the subject of further research.

[0283] The following 8 datasets were downloaded from https: / / public.jaeb.org / datasets / diabetes: ITY , DCLP 5. M DEX , R EPBG , R TCGM , S ENCE , S EVHYPO , W ISDM The analysis, content, and conclusions presented in this work are solely the responsibility of the authors and have not been reviewed by any research group (C ITY : CGM Intervention Study in Adolescents and Young Adults with T1D; D CLP 5: iDCL experimental research group; M DEX : T1D exchange low-dose glucagon exercise study group; R EPBG :REPLACEBG research group; R TCGM :JDRF CGM Research Group; S ENCE : Study Group on Strategies to Enhance New CGM Use in Early Childhood; S EVHYPO T1D Exchange Severe Hypoglycemia Study Group in Elderly People with Type 1 Diabetes; W ISDM : WISDM Study Group).

[0284] Table 6: T54 and T70 weight combination - weight ω T54 and ω T70 There are nine different combinations of which are used to emphasize the T54 and T70 input columns when performing hierarchical clustering.

[0285]

[0286] Example 3

[0287] Background: The use of CGM generates a large amount of data, but their interpretation remains more art than exact science. The international consensus on time in range (TIR) ​​proposed a widely accepted TIR metric system, which we now introduce by introducing a finite and fixed set of clinically similar clusters (CSCs) such that within a cluster, the TIR metric of daily CGM curves is homogeneous.

[0288] Methods: CSCs were defined and validated using 204,710 daily CGM curves from healthy, differently treated individuals with type 1 and type 2 diabetes (T1D, T2D). CSCs were defined using 23,916 daily CGM curves (training data) and a further 37,758 curves (validation data) to obtain the final fixed set of CSCs. Test data (143,036 curves) were used to establish the robustness and generalizability of CSCs.

[0289] Results: The final set of CSCs contained 35 clusters. Any daily CGM curve could be classified into a single CSC that faithfully approximated the common glucose metric of the daily CGM curve, as demonstrated by regression analysis with an intercept of 0 (R squared > 0.81, correlations > 0.9 for all TIRs and most other metrics). CSCs differentiated CGM curves of healthy, differently treated T2D, and T1D, and allowed tracking of daily changes in individual glycemic control over time.

[0290] Conclusions: Any daily CGM curve can be classified into one of [only] 35 pre-fixed CSCs, which enables many applications, such as algorithmic approaches to tabular data interpretation and treatment, CGM replacement for clinical testing, database indexing, pattern recognition, and tracking disease progression.

[0291] introduction

[0292] The widespread adoption of continuous glucose monitoring (CGM) technology inevitably generates a large amount of data; for example, two recent reports on the real-life use of artificial pancreas systems are based on more than 1.5 billion data points. Over the years, many glycemic control metrics have been introduced, with the overall goal of aggregating CGM data to convey meaningful clinical information. Some existing measures based on self-monitoring data, such as MAGE (mean amplitude of glucose excursions) and LBGI / HBGI (low and high BG index), have also been adapted for use with CGM: the adaptation of MAGE to CGM data follows the classical time-independent structure of the measure, so that in this case CGM is used only as a source of amplitude assessment; the adaptation of LBGI and HBGI accounts for differences between SMBG and CGM data. The mean of daily differences (MODD) was introduced as a measure of intraday variability, and the continuous overlapping net glycemic action (CONGA) was expressed as a composite indicator of the amplitude and timing of blood glucose (BG) fluctuations captured over different time periods. The standard deviation of the rate of BG change was used as a marker of the stability of the metabolic system over time, based on the premise that more erratic BG changes are a sign of system instability. Standard deviation arrays were introduced to reflect the glucose variability contained in different clinically relevant periods of CGM data, and the clinical interpretation of various CGM-based glucose variability metrics was discussed8. Early reviews of statistical methods that can be used to analyze CGM data included several plots, such as Poincaré plots for system stability and variability grid analysis (VGA) for visualizing glucose fluctuations captured by CGM and the efficacy of automated insulin delivery (AID). Reviews published in Biomedical Engineeringl1 and Diabetes Care12 evaluated methods for calculating and visualizing the relationship between glucose variability and hypoglycemia risk. The glucose management index (GMI) was introduced as a CGM-based approximation of the HbA1c test. The newest member of the family of glycemic control metrics is the glycemic risk index (GRI), which is based on the collective opinion of many physicians and attempts to balance the risks of hypoglycemia and hyperglycemia in a single metric.

[0293] As a result, the CGM field is overloaded not only with a large number of datasets, but also with a large number of metrics used to assess various aspects of glycemic control. For reference, a 2017 paper published in Nature Reviews Endocrinology discusses many (but not all) existing metrics in detail. The overall argument is that CGM-based metrics should generally include some notion of the timing of CGM readings, rather than just their amplitude. This is because CGM data represent a time series of equally spaced glucose observations—a property that enables analysis far beyond traditional MAGE, LBGI / HBGI, MODD, CONGA, GRI, or any other amplitude-based metric that has been employed over the years. For example, contemporary algorithms that enable AID are made possible solely because of the temporal information carried by the CGM data stream.

[0294] Recently, the confusion of multiple glycemic markers has been sorted out, showing that almost all glycemic control measures introduced over the years are described by [only] two “basic measures”—hyperglycemia exposure and hypoglycemia risk, which means that the quantitative representation of glycemic control is a rather simple two-dimensional construct. In 2019, the international consensus on time in range (TIR, usually 70-180 mg / dL) proposed TIR as the primary CGM-based measure of glycemic control and set clinical goals for its use. Given that time below range (TBR), TIR, and time above range (TAR) always add to 100%, the TIR metric system well represents the two-dimensional structure of glycemic control—TBR as an indicator of hypoglycemia risk and TAR (or equivalently, TIR) measuring hyperglycemia exposure. The system is based on the dynamic glucose profile (AGP), which was introduced as a data presentation and visualization template originally developed by Mazze et al. for self-monitoring data. Based on the AGP, standardized CGM reports now incorporate core CGM measures and targets as well as the 14-day composite glucose profile as an integral component of clinical decision making. This model has been recognized by the international consensus on TIR17 and is also cited by the American Diabetes Association's 2019 Standards of Care20 and the AACE consensus on CGM use21. AGP reports have now been adopted by many CGM device manufacturers in their CGM companion software and are proposed by international consensus as a standardized output for AID technology evaluation and presentation of clinical trial results.

[0295] The widely adopted TIR metric system defines 5 in-range times for CGM glucose values. These in-range times are used to provide a numerical interpretation of the AGP: Grade 2 hypoglycemia - less than 54 mg / dL, Grade 1 hypoglycemia - 54 to 69 mg / dL, Within target range (TIR) ​​- 70 to 180 mg / dL, Grade 1 hyperglycemia - 180 to 250 mg / dL, and above Grade 2 hyperglycemia - 250 mg / dL.

[0296] These boundaries may vary according to consensus recommendations for different types of diabetes, but the concept remains the same. The AGP / TIR representation thus establishes well a static visual and quantitative description – a snapshot of CGM data (typically 14 days). However, the TIR metric system does not reflect the day-to-day variability of CGM data (except for the AGP cloud) or the progression (improvement / worsening) of glycemic control over time. Given that the main advantage of CGM is to measure a time series of glucose values ​​and capture the evolution of glycemic control, it becomes critical to equip the TIR system with a time component. Therefore, this manuscript takes the next step in advancing the AGP / TIR concept by establishing a fixed and finite set of clinically similar clusters (CSCs) that accurately represent the population of all daily CGM curves with a relatively small number (N=35) of fixed CSCs and allow tracking of daily glucose changes over time in a look-up table format.

[0297] Materials and methods

[0298] data

[0299] Sixteen de-identified archival datasets were used in this work, as shown in Table 7 , which included demographic information and summary statistics of study participants, diabetes type (T1D, T2D) or health status, and treatment modality, such as multiple daily insulin injections (MDI), continuous subcutaneous insulin delivery (CSII) via insulin pump, or automated insulin delivery (AID).

[0300] Table 7: Characteristics of participants in the 16 studies used in this work.

[0301]

[0302]

[0303] Statistics are presented as mean (SD) unless otherwise stated. T1D denotes type 1 diabetes, T2D denotes type 2 diabetes, BMI denotes body mass index, CGM denotes continuous glucose monitoring, MDI denotes multiple daily injections, CSII denotes continuous subcutaneous insulin delivery via insulin pump, and AID denotes automated insulin delivery.

[0304] These data were collected during clinical trials conducted at the University of Virginia Diabetes Technology Center or were obtained from public databases at the Jaeb Center for Health Research in Tampa, Florida. The references to these studies are as follows: CITY,24DCLP1,25DCLP3,26DCLP5,27DIAMOND1,28DIAMOND2,29DSS1,30MDEX,31NDIAB,32NIGHTLIGHT,33REPLA-BG,34RTCGM,35SENCE,36SEVHYPO,37UVA-TRIALNET,38andWISDM39. In total, these datasets contain CGM traces from 2,462 individuals (52.6% female) and 204,710 daily CGM curves, or approximately 560 years of health, T1D, and T2D data. The majority (95.9%) of daily CGM curves were generated by people with T1D who were treated with MDI, CSII, or AID. Two studies focused on children with T1D (DCLP5 and SENCE). The DIAMOND2 study had participants with T2D who were treated with MDIs, accounting for 3.6% of daily CGM curves. The NDIAB and UVA-TRIALNET studies included data from non-diabetic patients. Glycemic control, assessed by mean HbA1c at baseline, ranged from 5.2% (NDIAB study) to 9.1% (CITY study) in non-diabetic patients. In healthy populations, there was generally less than 7 days of data, while the vast majority of participants in the diabetic studies had an average of 5 weeks or more of data.

[0305] Data is preprocessed and separated into training, validation and testing data:

[0306] A previously published protocol was used to process the CGM time series and define daily CGM curves, where a daily CGM curve is a time series of 288 blood glucose data points collected every 5 minutes from midnight to midnight (24 hours) - see Section III, Lobo et al. The daily CGM curves from these 16 different studies formed three different datasets, each with a different purpose:

[0307] 1. The training data consisted of 23,916 daily CGM curves sampled from the DCLP1, DCLP3, DIAMOND1, DIAMOND2, DSS1, and NIGHTLIGHT studies and was used to define a candidate set of CSCs.

[0308] 2. The validation data, consisting of 37,758 daily CGM curves sampled from the same 6 studies as the training data, was used to evaluate the performance of the CSC candidate set before selecting and fixing the final set of CSCs.

[0309] 3. The test data consists of 143,036 daily CGM curves sampled from CITY, DCLP5, DIAMOND2, NDIAB, MDEX, REPLACE-BG, RT-CGM, SENCE, SEVHYPO, UVA-TRIALNET, and WISDM studies to evaluate the robustness and generalizability of the final set of selected CSCs.

[0310] These datasets were constructed so that there was no overlap between the training, validation, and test data. In addition, the methods and CGM techniques used in the studies from which the test data were derived were different from those used for the training and validation data.

[0311] Analysis: Building the final set of CSCs and daily curve classifications

[0312] Step 1: Define and then fix a CSC set with the following properties: For any daily CGM curve, there are CSCs that approximate 5 standard in-range times for that daily CGM curve, abbreviated here as follows: T54 (percent CGM time below 54 mg / dl); T70 (percent CGM time below 70 mg / dl), TIR (percent CGM time within 70-180 mg / dl), T180 (percent CGM time above 180 mg / dl), and T250 (percent CGM time above 250 mg / dl). This approximation ensures that key clinically relevant features of the daily CGM curve are preserved. Hierarchical clustering is used to define the CSC set, where the weights of the inputs are varied until the CSC set has the desired performance approximated as a vector {T54, T70, TIR, T180, T250}.

[0313] Step 2: A process was developed for mapping daily CGM curves to their closest CSCs, which involves calculating a similarity metric (e.g., Euclidean distance) between the candidate daily CGM curves and the centroid of each CSC, and then selecting the CSC with the best similarity metric value. The similarity metric is calculated in a "space" defined by all possible vectors {T54, T70, TIR, T180, T250}. When these two steps are completed, any daily CGM curve can be mapped to a CSC at five in-range times that approximate the original daily CGM curve, and an individual's CSC sequence can be used as a surrogate for that individual's glycemic control progress.

[0314] Keeping in mind that the glycemic control space is essentially two-dimensional16, approximating the daily CGM curve with CSCs, i.e., minimizing the distance between the two according to {T54, T70, TIR, T180, T250}, ensures that any other glycemic control metric derived from the daily CGM curve will also be approximated by CSCs. In addition to metric approximation, the sequence of CSCs also provides information about the temporal and day-to-day variability of clinically relevant glycemic events for the patient. An extended mathematical description of the process described in this section is provided in the Supplementary Materials.

[0315] result

[0316] The final set of CSCs: a review and interpretation

[0317] The process described in the previous section produced a final set of 35 clinically similar clusters. The results in this section use test data as an external validation of the CSC method and as an illustration of its potential for clinical application. The CSC index indicates the degree of glycemic control, as represented by TIR alone. TIR is highest in CSCs with lower indices (e.g., 1, 2, 3) and lowest in CSCs with the highest indices (e.g., 33, 34, 35). For example, CSC 1 has a TIR = 85.4%, while CSC 35 has a TIR = 2.4%. However, CSCs with adjacent indices, although having similar TIRs, can be very different in terms of hypoglycemia or hyperglycemia. For example, in CSC 12, TIR = 46.4%, and the rest are distributed between grade 1 hypoglycemia and grade 2 hypoglycemia, with readings significantly below 54 mg / dl (T54 = 27.3%). In contrast, for CSC 13, TIR = 44.9%, but the remainder was due to hyperglycemia (T180 = 30.4% and T250 = 21.9%).

[0318] Table 8 lists all CSCs and their respective values ​​of {T54, T70, TIR, T180, T250}, i.e., the centroid defining each CSC, and the number / percentage of daily CGM curves associated with each CSC in the test data.

[0319] Table 8: List of all CSCs with their respective values ​​of {T54, T70, TIR, T180, T250}.

[0320]

[0321] Build the final set of CSCs and daily curve classification

[0322] When performing hierarchical clustering, the time {T54, T70, TIR, T180, T250} in the range is used as the input of a single daily CGM curve. The input is generated using all daily CGM curves in the training data. The implementation of hierarchical clustering using the centroid algorithm 17 of the scipy.cluster.hierarchy Python module is used to calculate the Euclidean distance between two rows of inputs. Because we want to ensure that CSC faithfully captures the behavior of the time below the range, we give the T54 and T70 input columns a weight greater than the TIR, T180 and T250 input columns. Each CSC is a collection of daily CGM curves so that each daily CGM curve in the collection has substantially the same time in the range. For a given set of inputs (determined using the daily CGM curves in the training data set and the weights selected for the very low columns and the lower columns), the hierarchical clustering algorithm produces a dendrogram that shows the hierarchical relationship between the daily CGM curves in the training data set.

[0323] "Cutting" the dendrogram at a specific height produces clusters with a specific number of clusters. Each set of CSC candidates is evaluated using the validation data. The centroid of the CSC is used to classify each daily CGM curve in the validation data. ijk is the in-range time k from the jth daily CGM curve of individual i, and CSC k (dp ij ) is the time k within the range of CSC that the jth daily CGM curve from individual i is classified as, where k∈(T54, T70, TIR, T180, T250}.

[0324] In addition,

[0325]

[0326] And make

[0327]

[0328] where j is the set of all daily CGM curves for individual i. A single “best” set of CSCs:

[0329] Through each k point set (dp ik , CSC k (dp i )) Maximize the r-squared value of the linear regression and ensure that the absolute value of the relative effect size for each k is less than 0.15, where the relative effect size is calculated by dp ijk and CSC k (dp ijk ) divided by the average of all dp ijk Calculate the standard deviation.

[0330] We explored 9 different weights for the very low and low input columns (i.e., 9 different input sets to the hierarchical clustering algorithm), which resulted in evaluating a candidate set of 166 CSCs. The final fixed set of CSCs had 35 clusters. Each CSC is defined by a centroid. The centroid of a given CSC is calculated using the daily CGM curves in the training data assigned to the CSC.

[0331] Clearly, “extreme hypoglycemia” CSCs such as #10 and #22 were rare (54 and 46 occurrences, respectively, out of 143,036 daily CGM curves), whereas “moderate hypoglycemia” clusters such as #6 and #11 (6,770 and 1,294 occurrences, respectively), or clusters primarily associated with hyperglycemia such as #34 and #35 (4,226 and 1,146 occurrences, respectively) were more common, which was expected as the majority of daily CGM curves in this analysis were from individuals with T1D receiving various treatments.

[0332] It is quite clear that the most common CSCs are #1 and #5 (Table 8), which is a function of including healthy people and those receiving advanced treatments, such as those with T1D receiving AIDs or those with T2D receiving CGM.

[0333] Track progress in blood sugar control over time

[0334] Over time, the sequence of daily CGM curves produced by each person can be presented as a sequence of CSCs that represent the progression of glycemic control for that individual over time. For example, a person with good glycemic control will visit fewer unique CSCs, typically with lower indices (indicating more time spent in range), while a person with erratic glucose changes will visit more unique CSCs, with those that are visited typically having higher indices (indicating less time spent in target range and more time spent above or below range). Fig.14 is a 3-panel graph showing the progression of three individuals with T1D over 14 days; Panel A (top row) shows data from a 6-year-old boy with a baseline HbA1c of 7.8% treated with MDI from the SENCE study, Panel B (middle row) shows data from a 7-year-old boy with a baseline HbA1c of 7.9% treated with CGM+CSII from the DCLP5 study 27, and Panel C (bottom row) shows data from a 9-year-old boy with a baseline HbA1c of 7.8% treated with AID from the DCLP5 study to represent the three treatment modalities. These individuals had the same sex, similar age, and essentially the same baseline HbA1c, but their trajectories diverged thereafter, with significantly different daily transitions between CSCs and the number / index of CSCs visited.

[0335] also, Fig.14 Includes respective AGP presented as 14-day CSC traces. Obviously, the sequence of CSCs faithfully represents the information carried by the AGPs, and adds information about daily changes in individual glycemic control, their worst or best days, or any treatment progress trends that may occur during the 14-day observation period. In addition, it is obvious that even for people treated with AIDs (Panel C) with very stable glycemic control of TIR=85.4% during the observation period, there is one day of significant hyperglycemia (Day 8, classified in CSC#5), one day of hypoglycemia (Day 13, classified in CSC#2), and two days of incomplete data (Days 2 and 12), when the AID system is not worn or malfunctions, these information are not conveyed by the AGP or TIR measurement system.

[0336] CSC visits as an indication of health status and treatment efficacy

[0337] Fig.15 It is a 4-panel diagram, which shows the ability of CSC collection to distinguish health status and treatment mode. One panel shows the average number (solid line) of unique CSCs visited by people with T1D, T2D and healthy people. Obviously, in the T1D population, the number of unique CSCs visited over time (e.g., 6 months) is the largest, while in the healthy population, only 3 unique CSCs are visited on average. Another panel presents the same trajectory as the first panel, but in this case distinguishes the treatment of individuals with T1D, i.e., MDI, CSII and AID. The dotted lines in these panels are Weibull distribution functions that have been fitted to the data, and these fitting curves are very close to the true trajectory. Weibull fitting has a certain probabilistic meaning, which is beyond the scope of this article. Another panel shows a box plot of the CSC index of the health state of T1D divided by treatment mode. The figure confirms that, on average, people with T1D treated with MDI reach the highest CSC index, in diabetic patients treated with AID, the CSC index is the lowest, and the healthy population only visits a few low-index CSCs (all below 10). Finally, another panel shows the statistical pentagons with Bonferroni correction for pairwise comparisons between all 5 considered conditions. As can be seen, all pairwise differences are statistically significant except for the difference between T1D and T2D treated with CSII.

[0338] The relationship between CSC, AGP and established blood sugar control indicators

[0339] Each CSC represents many daily CGM curves from different individuals. The relationship between the information carried by the CSC and the AGP is Fig.16A, which shows the two adjacent clusters #12 and #13 mentioned above. For each of these CSCs, we plotted a version of the AGP that, rather than clustering the individual's continuous daily CGM curves, clustered all daily CGM curves associated with each CSC. As expected, if you compare the AGP associated with CSC#12, the AGP associated with CSC#13 is shifted up, and the AGP "cloud" is visually similar. Although CSC#12 and #13 have similar TIRs (46.4% and 44.9%, respectively), they differ significantly in hypoglycemia and hyperglycemia, which is expected. Figures 16B-16J is similar to Fig.16A . The graph for all 35 CSCs reported herein is shown in . Comparison of the lowest and highest CSC indices (e.g. #1, #2 vs. #34, #35) clearly shows the effect of good glycemic control with features primarily associated with hyperglycemia. Particularly instructive are, for example, CSC #28 or #32, which show high volatility of glycemic control with significant hypoglycemia and hyperglycemia.

[0340] To confirm the view that CSCs faithfully represent the generally accepted index values ​​of glycemic control of their members' daily CGM curves, Table 9 presents univariate regression analyses with a zero intercept. In each regression, the dependent variable is the index calculated from the daily CGM curve, and the independent variable is the same index calculated from the CSC centroid associated with the daily CGM curve. As shown in Table 9, the slopes of all regressions are close to 1, indicating that the index values ​​calculated from CSCs and their associated daily CGM curves are located near the identity line. In addition, the R-squared values ​​are very high (usually above 0.81, which corresponds to correlations above 0.9), indicating that the indexes calculated from CSCs account for a large portion of the variance carried by the original daily CGM curves (in addition to the coefficient of variation, the reasons are mentioned in the discussion).

[0341] Table 9: Glycemic control indices calculated from daily CGM curves and their respective CSC.

[0342]

[0343] discuss

[0344] After years of focusing on single indicators of glycemic control (e.g., HbA1c), the medical community is increasingly recognizing that contemporary technologies (e.g., CGM) provide a means of providing in-depth insight into the state of the person and the dynamics of glucose fluctuations. The TIR indicator system accepted by the international consensus on TIR17 and its extension to AIDs and clinical trial reporting is an excellent step in this direction, as long as TIR does not become the next single marker of treatment efficacy alone and attention is turned to other parameters, such as the risk of hypoglycemia. Further simplifying the large number of glycemic indicators, it was shown that two basic dimensions are sufficient to capture the information carried by almost all indicators introduced to date: hyperglycemia exposure and hypoglycemia risk.

[0345] One exception is the coefficient of variation, which has been shown to be incompatible with the TIR metric system and is often controversial because it is the ratio of two quantities (SD and mean), both of which typically decrease with successful treatment. The TIR metric system covers both dimensions well, with some mathematical redundancy, as the TIR components add up to 100%, which is clinically OK but may challenge some statistical methods. In this regard, we can assume that there is now an established method for reviewing CGM data, including the AGP and its adjacent TIR metric, and that it is reliable enough to allow a framework to be built on top of it to classify and track daily CGM curves over time.

[0346] In this manuscript, we introduce such a framework—a set of 35 clinically similar clusters—such that any daily CGM curve can be assigned to a single CSC that will approximate the clinical impression conveyed by the raw CGM data. Two properties of the CSC set are important: (1) it is finite and not too large—only 35 CSCs provide a reasonable classification of the seemingly infinite number of daily CGM curves, at least with respect to the TIR metric, and (2) it is fixed and does not need to be recomputed with new data. These properties, finite and fixed, allow the set of CSCs to be the basis for many clinical, computational, and algorithmic applications, essentially allowing lookup tables to solve apparently complex treatment optimization problems. A non-exhaustive list of potential applications includes:

[0347] 1. Data structuring, dimensionality reduction, and database indexing: The continuum of all possible daily CGM curves clinically represented by AGP and TIR indicators is reduced to a finite and fixed set of CSCs, which can be used as input to decision support, clinical, and automated treatment algorithms. A database indexed by the structure defined by CSCs will ensure fast and efficient searches for subgroups of similar daily CGM curves. This can facilitate features in decision support or AID systems, such as algorithms that learn from an individual's CGM patterns as well as from the patterns of others stored in the database.

[0348] 2. Differentiating health states and treatment modalities: In another application of CSC, we can envision that based on observed CSC patterns, wearing a CGM for 10 to 14 days in a home environment, which may be accompanied by a predefined meal and physical activity schedule, can obtain diagnostic results similar to those accepted in clinical practice. This approach will greatly simplify data collection, replace some common laboratory tests, and enable telemedicine methods.

[0349] 3. CGM pattern recognition and prediction: The transition probability matrix describing the patient's evolution through a predefined set of CSCs is a natural tool for observing disease or treatment progression. Pattern recognition or cyclic behavior is reflected by patterns or cycles detected in the transition probabilities from one CSC to the next. Short-term or long-term predictions of glycemic control are based on probabilistic patterns or repeated visits to a subset of CSCs. The latter is the subject of semi-Markov chain theory, which is obtained by aggregating (lumping) the state space into related subsets, characterized by the random duration of the time spent in each subset.

[0350] 4. Tracking disease progression over time: Deterioration in glycemic control is indicated by a shift to an undesirable CSC, whereas successful treatment optimization or drug titration is reflected by a shift to a clinically desired CSC. In practical applications, the CSC set will be labeled or graded by clinical expectations (e.g., by TIR in Figure 1), and then the label / grade of each CSC is fixed and used to guide treatment away from risk and toward optimal control. Automation of the treatment process can be facilitated by tabulating thousands of daily CGM curves into a few CSCs.

[0351] The major finding of this manuscript is that any daily CGM profile can be approximated by one of 35 pre-fixed clinically similar clusters. Approximation means that when a daily CGM profile is classified as a CSC, the CSC systematically retains the information carried by the original daily CGM profile based on the time metrics within the range. Thus, the CSC extends and to some extent completes the interpretation of CGM data provided by the AGP / TIR system—while the AGP / TIR is a static snapshot of 14 days of data, the CSC series derived from the same data tracks the progression of glycemic control over time.

[0352] A time series of CSCs over 14 days illustrates how stable or unstable a person's glycemic control is. In particular, this visualization allows us to see individual days when glycemic control is radically different from what a person normally experiences. Because the collection of CSCs condenses the clinical impression conveyed by all possible daily CGM curves into a limited set of actionable tables, a multitude of clinical applications are enabled, including: database indexing, pattern recognition and tracking of treatment progress across a limited set of possibilities, lookup table data interpretation for decision support and AID algorithms, or CGM substitution for common clinical tests.

[0353] Fig.17 is an exemplary high-level functional block diagram of an embodiment of the present invention or an aspect of an embodiment of the present invention. Fig.17 As shown, processor 104 or controller communicates with glucose monitor or data source 112 and optionally with insulin delivery device (for example, other device 110). Glucose monitor or device communicates with experimenter 1600 to monitor the glucose level of experimenter 1600. Processor 104 or controller is configured to perform required calculation. Alternatively, insulin delivery device communicates with experimenter 1600 to deliver insulin to experimenter 1600. Processor 104 or controller is configured to perform required calculation. Glucose monitor and insulin delivery device can be implemented as separated device or single device. Processor 104 can be implemented locally in glucose monitor, insulin delivery device or independent device (or in any combination of two or more in glucose monitor, insulin device or independent device). A part of processor 104 or system can be remotely located so that the device is operated as a telemedicine device.

[0354] refer to Fig.18 In its most basic configuration, computing device 1700 typically includes at least one processor 104 and memory 106. Depending on the exact configuration and type of computing device, memory 106 may be volatile (such as RAM), non-volatile (such as ROM, flash memory, etc.), or some combination of the two.

[0355] In addition, computing device 1700 may also have other features and / or functions. For example, computing device 1700 may also include additional removable and / or non-removable storage devices, including but not limited to disks or optical disks or tapes and writable electrical storage media. Such additional storage is represented by removable storage 1702 and non-removable storage 1704. Computer storage media include volatile and non-volatile, removable and non-removable media, which are implemented in any method or technology for storing information such as computer-readable instructions, data structures, program modules or other data. Memory, removable storage devices and non-removable storage devices are all examples of computer storage media. Computer storage media include but are not limited to RAM, ROM, EEPROM, flash memory or other memory technology CDROM, digital versatile disk (DVD) or other optical storage, cassettes, tapes, disk storage or other magnetic storage devices, or any other medium that can be used to store desired information and can be accessed by the device. Any such computer storage medium can be part of the device or used in conjunction with the device.

[0356] The computing device 1700 may also include one or more communication connections 1708 that allow the device to communicate with other devices (e.g., other computing devices). The communication connection carries information in a communication medium. Communication media typically embody computer-readable instructions, data structures, program modules, or other data in a modulated data signal such as a carrier wave or other transmission member, and include any information transfer medium. The term "modulated data signal" refers to a signal that sets or changes one or more of its characteristics in a manner that encodes, executes, or processes information in a signal. By way of example and not limitation, communication media include wired media such as a wired network or a direct wired connection, and wireless media such as radio, RF, infrared, and other wireless media. As described above, the term computer-readable media as used herein includes both storage media and communication media.

[0357] In addition to stand-alone computing machines, embodiments of the present invention may also be implemented on a network system that includes multiple computing devices communicating with a networked device, such as a network with an infrastructure or an ad hoc network. The network connection may be a wired connection or a wireless connection. As an example, Fig.18A network system in which an embodiment of the present invention can be implemented is shown. In this example, the network system includes a computer 1706 (e.g., a network server), a network connection device 1708 (e.g., a wired and / or wireless connection), a computer terminal 1710, and a PDA (e.g., a smart phone) 1720 (or other handheld or portable devices, such as cellular phones, laptop computers, tablet computers, GPS receivers, mp3 players, handheld video players, pocket projectors, etc., or handheld devices (or non-portable devices) with a combination of these features). In one embodiment, it should be understood that module 1706 can be a glucose monitor device. In one embodiment, it should be understood that the modules listed as 1706 can be glucose monitoring devices, artificial pancreas and / or insulin devices (or other interventional or diagnostic devices). Any component can be multiple in number. Embodiments of the present invention can be implemented in any device of the system. For example, the execution of instructions or other desired processing can be implemented on the same computing device 1700. Alternatively, embodiments of the present invention can be executed on different computing devices of the network system. For example, certain desired or required processing or execution may be implemented on one of the computing devices (e.g., server 1706 and / or glucose monitoring device) of the network, while other processing and execution of instructions may be implemented on another computing device (e.g., terminal 1710) of the network system, and vice versa. In fact, certain processing or execution may be implemented at one computing device (e.g., server 1706 and / or insulin device, artificial pancreas or glucose monitor device (or other interventional or diagnostic device)); while other processing or execution of instructions may be performed at different computing devices that may or may not be networked. For example, certain processing may be performed at terminal 1706, while other processing or instructions may be passed to computing device 1700 where instructions are executed. This scenario may be particularly valuable, especially when a PDA device, such as through a computer terminal 1710 (or an access point in an ad hoc network), accesses the network. For another example, one or more embodiments of the present invention may be used to execute, encode or process software to be protected. The processed, encoded or executed software may then be distributed to customers. The distribution may be in the form of a storage medium (e.g., disk) or an electronic copy.

[0358] Fig.19 1800 and the associated Internet 1802, with which embodiments can be implemented. This configuration is typically used for computers (hosts) connected to the Internet 1802 and executing server or client (or combination) software. For example, a source program computer such as a laptop, a final destination computer, and a relay server, as well as any computer or processor described herein, can use Fig.19, and an Internet connection. System 1800 may be used as a portable electronic device such as a notebook / laptop computer, a media player (e.g., an MP3-based media player or video player), a cellular phone, a personal digital assistant (PDA), a glucose monitor device, an artificial pancreas, an insulin delivery device (or other interventional or diagnostic device), an image processing device (e.g., a digital camera or video recorder), and / or any other handheld computing device, or any combination of these devices. Note that although Fig.19 Various components of the computer system are shown, but they are not intended to represent any particular architecture or manner of interconnecting the components; as these details are not germane to the present invention. It should also be understood that network computers, handheld computers, cellular phones, and other data processing systems with fewer or perhaps more components may also be used. Fig.19 The computer system 100 may be, for example, an Apple Macintosh computer or Power Book, or an IBM compatible PC. The computer system 100 includes a bus 1804, an interconnect or other communication means for transmitting information, and a processor 104, typically in the form of an integrated circuit, coupled to the bus 1804 for processing information and for executing computer executable instructions. The computer system 100 also includes a main memory 106, such as a random access memory (RAM) or other dynamic storage device, coupled to the bus 1804 for storing information and instructions to be executed by the processor 104.

[0359] The main memory 106 may also be used to store temporary variables or other intermediate information during the execution of instructions to be executed by the processor 104. The computer system 100 also includes a read-only memory (ROM) 136 (or other non-volatile memory) or other static storage device coupled to the bus 1804 for storing static information and instructions for the processor 104. Storage devices 1808 such as magnetic disks or optical disks, hard disk drives for reading from and writing to hard disks, magnetic disk drives for reading from and writing to magnetic disks, and / or optical disk drives for reading from and writing to removable optical disks (such as DVDs) are coupled to the bus 1804 to store information and instructions. The hard disk drive, magnetic disk drive, and optical disk drive may be connected to the system bus via a hard disk drive interface, a magnetic disk drive interface, and an optical disk drive interface, respectively. The drives and their associated computer-readable media provide non-volatile storage of computer-readable instructions, data structures, program modules, and other data for general computing devices. Typically, the computer system 100 includes an operating system (OS) stored in non-volatile memory for managing computer resources and providing access to computer resources and interfaces to applications and programs. An operating system typically processes system data and user input and responds by allocating and managing tasks and internal system resources, such as controlling and allocating memory, determining the priority of system requests, controlling input and output devices, facilitating networking, and managing files. Non-limiting examples of operating systems are Microsoft Windows, Mac OS X, and Linux.

[0360] The term "processor" is intended to include any integrated circuit or other electronic device (or collection of devices) capable of performing operations on at least one instruction, including but not limited to reduced instruction set core (RISC) processors, CISC microprocessors, microcontroller units (MCUs), CISC-based central processing units (CPUs), and digital signal processors (DSPs). The hardware of such devices may be integrated onto a single substrate (e.g., a silicon "chip"), or distributed between two or more substrates. In addition, various functional aspects of the processor may be implemented separately as software or firmware associated with the processor.

[0361] The computer system 100 may be coupled to a display 1810, such as a cathode ray tube (CRT), a liquid crystal display (LCD), a flat screen monitor, a touch screen monitor or the like, via the bus 1804 for displaying text and graphical data to a user. The display may be connected via a video adapter for supporting the display. The display allows a user to view, input and / or edit information related to the operation of the system. An input device 1812 including alphanumeric keys and other keys is coupled to the bus 1804 for transmitting information and command selections to the processor 104. Another type of user input device is a cursor control 1814, such as a mouse, a trackball or cursor direction keys, for transmitting direction information and command selections to the processor 104 and for controlling cursor movement on the display 1810. The input device typically has two degrees of freedom in two axes, a first axis (e.g., x) and a second axis (e.g., y), which allows the device to specify a position in a plane.

[0362] Computer system 1800 can be used to implement the methods and techniques described herein. According to one embodiment, these methods and techniques are performed by computer system 1800 in response to processor 104 executing one or more sequences of one or more instructions contained in main memory 1816. Such instructions can be read into main memory 106 from another computer-readable medium, such as storage device 1808. Execution of the sequence of instructions contained in main memory 106 causes processor 104 to perform the processing steps described herein. In alternative embodiments, hard-wired circuits can be used instead of software instructions or in combination with software instructions to implement the arrangement. Therefore, embodiments of the present invention are not limited to any specific combination of hardware circuitry and software.

[0363] The term "computer-readable medium" (or "machine-readable medium") as used herein is an extensible term that refers to any medium or any memory that participates in providing instructions to a processor (such as processor 104) for execution, or any means for storing or transmitting information in a form readable by a machine (e.g., a computer). Such media can store computer-executable instructions to be executed by processing elements and / or control logic, as well as data manipulated by processing elements and / or control logic, and can take many forms, including but not limited to non-volatile media, volatile media, and transmission media. Transmission media include coaxial cables, copper wire, and optical fiber, including the wires that make up bus 1804. Transmission media can also take the form of acoustic or light waves, such as those generated during radio wave and infrared data communications, or other forms of propagated signals (e.g., carrier waves, infrared signals, digital signals, etc.). Common forms of computer readable media include, for example, a floppy disk, a foldable disk, a hard disk, magnetic tape or any other magnetic medium, a CD-ROM, any other optical medium, punch cards, paper tape, any other physical medium with a pattern of holes, a RAM, a PROM and EPROM, a Flash-EPROM, any other memory chip or cassette, a carrier wave as described below, or any other medium from which a computer can read.

[0364] Various forms of computer readable media may be involved in carrying one or more sequences of one or more instructions to the processor 104 for execution. For example, the instructions may initially be carried on a disk of a remote computer. The remote computer can load the instructions into its dynamic memory and send the instructions over a telephone line using a modem. A modem local to the computer system 100 can receive the data on the telephone line and use an infrared transmitter to convert the data into an infrared signal. An infrared detector can receive the data carried in the infrared signal, and appropriate circuitry can place the data on the bus 1804. The bus 1804 transfers the data to the main memory 1816, from which the processor 104 retrieves and executes the instructions. The instructions received by the main memory 1816 may optionally be stored on the storage device 1808 before or after execution by the processor 104.

[0365] The computer system 100 also includes a communication interface 1818 coupled to the bus 1804. The communication interface 1818 provides a two-way data communication coupling to a network link 1822, which is connected to a local network 1820. For example, the communication interface 1818 can be an integrated services digital network (ISDN) card or a modem to provide a data communication connection to a corresponding type of telephone line. As another non-limiting example, the communication interface 1818 can be a local area network (LAN) card to provide a data communication connection to a compatible LAN. For example, an Ethernet-based connection based on the IEEE 802.3 standard may be used, such as 10 / 100BaseT, 1000BaseT (Gigabit Ethernet), 10 Gigabit Ethernet (10GE or 10GbE or 10GigE, as standardized by IEEE standard 802.3ae-2002), 40 Gigabit Ethernet (40GbE), or 100 Gigabit Ethernet (100GbE, in accordance with Ethernet standard IEEE P802.3ba), as described in Cisco Systems, Inc. Publication number 1-587005-001-3 (6 / 99), "Internetworking Technologies Handbook", Chapter 7: "Ethernet Technologies", pages 7-1 to 7-38, which is incorporated in its entirety for all purposes as if fully set forth herein. In this case, the communication interface 1818 typically includes a LAN transceiver or modem, such as the Standard Microsystems Corporation (SMSC) LAN91C111 10 / 100 Ethernet transceiver described in the Standard Microsystems Corporation (SMSC) data-sheet “LAN91C111 10 / 100 Non-PCI Ethernet Single Chip MAC+PHY” Data-Sheet, Rev. 15 (02-20-04), which is incorporated in its entirety for all purposes as if fully set forth herein.

[0366] Wireless links may also be implemented.In any such implementation, communication interface 1818 sends and receives electrical, electromagnetic or optical signals that carry digital data streams representing various types of information.

[0367] The network link 1822 generally provides data communication to other data devices through one or more networks. For example, the network link 1822 can provide a connection to a host computer or to data equipment operated by an Internet Service Provider (ISP) 1824 through a local network 1820. The ISP 1824 in turn provides data communication services through the global packet data communication network Internet 1802. Both the local network 1820 and the Internet 1802 use electrical, electromagnetic or optical signals that carry digital data streams. The signals through the various networks and the signals on the network link 1822 and through the communication interface 1818 transmit the digital data to and from the computer system 100, which is an exemplary form of a carrier wave that transmits information.

[0368] The received code may be executed by processor 104 as it is received, and / or stored in storage device 1808 or other non-volatile storage for later execution. In this manner, computer system 100 may obtain application code in the form of a carrier wave.

[0369] The concept of identifying clinically similar clusters of daily continuous glucose monitoring (CGM) curves was developed by the inventors. The concept of performing the following operations was developed by the inventors: a) constructing and then fixing a set of clinically similar clusters (CSCs) having the following properties: for any other daily continuous glucose monitoring (CGM) curve, there is a clinically similar cluster (CSC) that is approximately within the time range of the daily CGM curve, and b) determining the approximation of any daily CGM curve by the CSC.

[0370] As can be seen from the algorithm and method requirements discussed herein, the process is easily applicable to identifying clinically similar clusters of daily continuous glucose monitoring (CGM) curves, and can be implemented and utilized with the relevant processors, networks, computer systems, the Internet, and components and functions according to the scheme disclosed herein. As can be seen from the algorithm and method requirements discussed herein, the process is easily applicable to devices that perform the following operations: a) construct and then fix a set of clinically similar clusters (CSCs), the set of clinically similar clusters (CSCs) having the following characteristics: for any other daily continuous glucose monitoring (CGM) curve, there is a clinically similar cluster (CSC) that is approximately within the time range of the daily CGM curve, and b) determine the approximate value of any daily CGM curve through the CSC, and can be implemented and utilized with the relevant processors, networks, computer systems, the Internet, and components and functions according to the scheme disclosed herein.

[0371] Fig. 20A system is shown in which a network or portion of a network or computer may be used to implement one or more embodiments of the invention. Although the glucose monitor, artificial pancreas or insulin device (or other interventional or diagnostic device) of the invention may be implemented without a network. Fig. 20 An exemplary system in which an embodiment of the present invention may be implemented is schematically illustrated. In one embodiment, a glucose monitor, artificial pancreas, or insulin device (or other interventional or diagnostic device) may be implemented locally by a subject (or patient) at home or other desired location. However, in alternative embodiments, it may be implemented in a clinic setting or an auxiliary setting. For example, a clinic setting 1900 provides a place for a physician (e.g., 1902) or a clinician / assistant to diagnose a patient (e.g., 1904) with a disease associated with glucose and related diseases and conditions. A glucose monitoring device 1906 may be used as a stand-alone device to monitor and / or test a patient's glucose level. It should be understood that although only the glucose monitoring device 1906 is shown in Fig. 20 The system of the present invention and any of its components can be Fig. 20 The system or components can be fixed to the patient or communicate with the patient according to expectations or needs. For example, a combination of a system or its components, which includes a glucose monitor device 1906 (or other related equipment or systems, such as a controller and / or an artificial pancreas, an insulin pump (or other interventional or diagnostic devices), or any other desired or required device or component), can be contacted, communicated or attached with the patient by tape or a pipeline (or other medical equipment or components), or can be communicated by a wired or wireless connection. This monitoring and / or testing can be short-term (such as a clinical visit) or long-term (such as a clinical hospitalization or family). A doctor (clinician or assistant) can use the glucose monitoring device output to perform appropriate actions, such as injecting insulin or feeding food for the patient, or other appropriate actions or modeling. Alternatively, the glucose monitoring device output can be transmitted to a computer terminal 1908 for immediate or future analysis. Transmission can be by cable or wireless or any other suitable medium. The glucose monitoring device output from the patient can also be transmitted to a portable device, such as a PDA 1910. The glucose monitoring device output with improved accuracy can be transmitted to the glucose monitoring center 1912 for processing and / or analysis. Such transmission can be achieved in a variety of ways, such as a network connection 1914, which can be wired or wireless.

[0372] In addition to the glucose monitoring device output, the error, parameters for accuracy improvement, and any accuracy-related information can be transmitted to, for example, a computer and / or glucose monitoring center 1912 for performing error analysis. Due to the importance of glucose sensors (or other interventional or diagnostic sensors or devices), this can provide centralized accuracy monitoring, modeling, and / or accuracy enhancement for a glucose center (or other interventional or diagnostic center).

[0373] Embodiments of the invention may also be implemented in a stand-alone computing device associated with a target glucose monitoring device, artificial pancreas, and / or insulin device (or other interventional or diagnostic device).

[0374] Fig.21 is a block diagram illustrating an example of a machine on which one or more aspects of embodiments of the present invention may be implemented. Fig.21 , one aspect of embodiments of the present invention includes, but is not limited to, systems, methods, and computer-readable media that provide the following: identifying clinically similar clusters of daily continuous glucose monitoring (CGM) curves, which shows a block diagram of an example machine 2000 on which one or more embodiments (e.g., the methods discussed) can be implemented (e.g., run).

[0375] refer to Fig.21 , one aspect of embodiments of the present invention includes, but is not limited to, systems, methods, and computer-readable media that provide the following: constructing and then fixing a set of clinical similarity clusters (CSCs) having the following characteristics: b) for any other daily continuous glucose monitoring (CGM) curve, there exists a clinical similarity cluster (CSC) that is approximately within the time range of the daily CGM curve; and b) determining an approximation of any daily CGM curve by the CSC, which shows a block diagram of an example machine 2000 on which one or more embodiments (e.g., the method discussed) can be implemented (e.g., run).

[0376] Examples of machine 2000 may include logic, one or more components, circuits (e.g., modules), or components. A circuit is a tangible entity configured to perform certain operations. For example, a circuit may be arranged in a specified manner (e.g., internally or relative to an external entity such as other circuits). In one example, one or more computer systems (e.g., stand-alone, client, or server computer systems) or one or more hardware processors (processors) may be configured by software (e.g., instructions, application portions, or applications) to operate as a circuit to perform certain operations as described herein. In one example, the software may (1) reside on a non-transitory machine-readable medium or (2) reside in a transmission signal. In one example, the software, when executed by the circuit's underlying hardware, causes the circuit to perform certain operations.

[0377] In one example, the circuit may be implemented mechanically or electronically. For example, the circuit may include dedicated circuits or logic that are specifically configured to perform one or more techniques such as those discussed above, such as including a dedicated processor, a field programmable gate array (FPGA), or an application-specific integrated circuit (ASIC). In one example, the circuit may include programmable logic (e.g., such as circuits contained within a general-purpose processor or other programmable processor) that can be temporarily configured (e.g., by software) to perform certain operations. It should be understood that the decision to implement the circuit mechanically (e.g., in a dedicated and permanently configured circuit) or in a temporarily configured circuit (e.g., configured by software) may be driven by cost and time considerations.

[0378] Thus, the term "circuit" is understood to include a tangible entity that is physically constructed, permanently configured (e.g., hardwired) or temporarily (e.g., temporarily) configured (e.g., programmed) to operate in a specified manner or perform a specified operation. In one example, given a plurality of temporarily configured circuits, it is not necessary to configure or instantiate each circuit at any one point in time. For example, where the circuit includes a general-purpose processor configured via software, the general-purpose processor can be configured as corresponding different circuits at different times. The software can configure the processor accordingly, for example, to construct a particular circuit at one point in time and a different circuit at a different point in time.

[0379] In one example, a circuit can provide information to other circuits and receive information from other circuits. In this example, a circuit can be considered to be communicatively coupled to one or more other circuits. In the case of multiple such circuits being present at the same time, communication can be achieved by signal transmission (e.g., by appropriate circuits and buses) connecting the circuits. In embodiments where multiple circuits are configured or instantiated at different times, communication between such circuits can be achieved, for example, by storing and retrieving information in a memory structure accessible to multiple circuits. For example, a circuit can perform an operation and store the output of the operation in a memory device communicatively coupled thereto. Then, another circuit can access the memory device at a later time to retrieve and process the stored output. In one example, a circuit can be configured to initiate or receive communication with an input or output device, and can operate on a resource (e.g., a collection of information).

[0380] The various operations of the method embodiments described herein may be performed at least in part by one or more processors that are temporarily or permanently configured to perform the relevant operations (e.g., by software). Such processors, whether temporarily or permanently configured, may construct processor-implemented circuits that operate to perform one or more operations or functions. In one example, the circuits mentioned herein may include processor-implemented circuits.

[0381] Similarly, the methods described herein may be implemented at least in part by a processor. For example, at least some of the operations of the method may be performed by one or more processors or circuits implemented by the processor. The execution of certain operations may be distributed between one or more processors, not only residing in a single machine, but also deployed on multiple machines. For example, one or more processors may be located in a single location (e.g., in a home environment, an office environment, or as a server cluster), while in other embodiments, the processors may be distributed in multiple locations.

[0382] The one or more processors may also operate to support execution of the related operations in a "cloud computing" environment or as "software as a service" (SaaS). For example, at least some operations may be performed by a group of computers (as an example of a machine including a processor), where the operations are accessible via a network (e.g., the Internet) and via one or more appropriate interfaces (e.g., an application program interface (API)).

[0383] Example embodiments (e.g., apparatus, systems, or methods) may be implemented in digital electronic circuitry, computer hardware, firmware, software, or any combination thereof. Example embodiments may be implemented using a computer program product (e.g., a computer program tangibly embodied in an information carrier or machine-readable medium for execution by, or to control the operation of, a data processing apparatus such as a programmable processor, a computer, or multiple computers).

[0384] A computer program may be written in any form of programming language, including compiled or interpreted languages, and it may be deployed in any form, including as a stand-alone program or as a software module, subroutine or other unit suitable for use in a computing environment. A computer program may be deployed to be executed on one computer or on multiple computers at one site or distributed across multiple sites and interconnected by a communication network.

[0385] In one example, the operations may be performed by one or more programmable processors executing a computer program to perform functions by operating on input data and generating output. Embodiments of the method operations may also be performed by, and example apparatus may be implemented as, special purpose logic circuitry (e.g., a field programmable gate array (FPGA) or an application specific integrated circuit (ASIC)).

[0386] The computing system may include a client and a server. The client and the server are usually far away from each other and usually interact through a communication network. The relationship between the client and the server is generated by means of a computer program that runs on a corresponding computer and has a client-server relationship with each other. In the implementation of the deployment of the programmable computing system, it should be understood that both hardware and software architectures need to be considered. Specifically, it should be understood that whether to implement certain functions in a permanently configured hardware (e.g., ASIC), in a temporarily configured hardware (e.g., a combination of software and a programmable processor), or in a combination of permanently and temporarily configured hardware can be a design choice. The hardware (e.g., machine 2000) and software architecture that can be deployed in an example implementation are described below.

[0387] In some examples, the machine 2000 may operate as a stand-alone device. In some examples, the machine 2000 may be connected (e.g., using a network) to other machines.

[0388] In a networked deployment, the machine 2000 can operate in the role of a server or a client machine in a server-client network environment. In one example, the machine 2000 can act as a peer machine in a peer-to-peer (or other distributed) network environment. The machine 2000 can be a personal computer (PC), a tablet PC, a set-top box (STB), a personal digital assistant (PDA), a mobile phone, a network appliance, a network router, a switch or a bridge, or any machine capable of executing (sequentially or otherwise) instructions specifying actions to be taken (e.g., performed) by the machine 2000. In addition, although only a single machine 2000 is shown, the term "machine" shall also be construed to include any collection of machines that individually or jointly execute a set (or multiple sets) of instructions to perform any one or more of the methodologies discussed herein.

[0389] The example machine (e.g., computer system) 2000 may include a processor 104 (e.g., a central processing unit (CPU), a graphics processing unit (GPU), or both), a main memory 106, and a static memory 106, some or all of which may communicate with each other via a bus 2020. The machine 2000 may also include a display unit 2002, an alphanumeric input device 2004 (e.g., a keyboard), and a user interface (UI) navigation device 2006 (e.g., a mouse). In one example, the display unit 2002, the input device 2004, and the UI navigation device 2006 may be a touch screen display. The machine 2000 may additionally include a storage device (e.g., a drive unit) 2008, a signal generating device 2010 (e.g., a speaker), a network interface device 2012, and one or more sensors 2014, such as a global positioning system (GPS) sensor, a compass, an accelerometer, or other sensors.

[0390] The storage device 2008 may include a machine-readable medium 2016 having stored thereon one or more sets of data structures or instructions 108 (e.g., software) embodying or utilized by any one or more of the methodologies or functionality described herein. The instructions 108 may also reside, completely or at least partially, within the main memory 106, within the static storage 106, or within the processor 104 during execution by the machine 2000. In one example, one or any combination of the processor 104, the main memory 106, the static storage 106, or the storage device 2008 may constitute a machine-readable medium.

[0391] Although the machine-readable medium 2016 is shown as a single medium, the term "machine-readable medium" may include a single medium or multiple media (e.g., a centralized or distributed database, and / or associated caches and servers) configured to store one or more instructions 108. The term "machine-readable medium" may also be considered to include any tangible medium that is capable of storing, encoding, or carrying instructions for execution by a machine and causing the machine to perform any one or more of the methods of the present disclosure, or capable of storing, encoding, or carrying data structures used by or associated with such instructions. Therefore, the term "machine-readable medium" may be considered to include, but is not limited to, solid-state memory and optical and magnetic media. Specific embodiments of machine-readable media may include non-volatile memory, for example, including semiconductor memory devices (e.g., electrically programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM)) and flash memory devices; magnetic disks, such as internal hard disks and removable disks; magneto-optical disks; and CD-ROM and DVD-ROM disks.

[0392] The instructions 108 may also be sent or received over the communication network 2018 using a transmission medium via a network interface device that utilizes any of a number of transmission protocols (e.g., Frame Relay, IP, TCP, UDP, HTTP, etc.). Example communication networks may include a local area network (LAN), a wide area network (WAN), a packet data network (e.g., the Internet), a mobile telephone network (e.g., a cellular network), a plain old telephone (POTS) network, and a wireless data network (e.g., a wireless network known as a cellular network). The IEEE 802.11 family of standards, known as The term "transmission medium" shall be deemed to include any intangible medium that is capable of storing, encoding, or carrying instructions for execution by the machine, and includes digital or analog communication signals or other intangible media to facilitate communication of such software.

[0393] Although example embodiments of the present disclosure are explained in detail in some cases herein, it should be understood that other embodiments are also contemplated. Therefore, the scope of the present disclosure is not intended to be limited to the details of the construction and arrangement of the components set forth in the following description or shown in the accompanying drawings. The present disclosure is capable of other embodiments and can be implemented or executed in various ways.

[0394] It should be understood that any element, component, section, subsection, or assembly described with reference to any specific embodiment above can be combined with any other embodiment described herein, integrated into any other embodiment described herein, or otherwise suitable for use with any other embodiment described herein, unless otherwise specifically stated or if it invalidates the device embodiment. Similarly, any step described with reference to a specific method or process can be integrated, incorporated, or otherwise combined with other methods or processes described herein, unless otherwise specifically stated or if it invalidates the method embodiment. In addition, multiple device embodiments or method embodiments can be combined with each other, incorporated, or otherwise integrated with each other to construct or develop further embodiments of the present invention described herein.

[0395] Should be understood that any assembly or module mentioned about any embodiment of the present invention discussed herein can be formed integrally or separately with each other.In addition, redundant functions or structures of assembly or module can be implemented.In addition, various assemblies can communicate locally and / or remotely with any user / clinician / patient or machine / system / computer / processor.In addition, various assemblies can communicate via wireless and / or hard wiring or other expectations and available communication devices, system and hardware.In addition, various assemblies and modules can be replaced with other modules or assemblies providing similar functions.

[0396] It should be understood that the devices and related components discussed herein can take on all shapes along the entire continuous geometric operating spectrum of the x, y and z planes to provide and meet anatomical, environmental and structural needs and operational requirements. In addition, the position and alignment of the various components can be varied as desired or required.

[0397] It should be understood that various sizes, dimensions, contours, stiffnesses, shapes, flexibility, and materials of parts of any components or parts in the various embodiments discussed throughout can be varied and utilized as desired or necessary.

[0398] It should be understood that although certain dimensions are provided on the aforementioned figures, the device can be constructed of various sizes, dimensions, contours, rigidities, shapes, flexibility and materials associated with components or portions of components of the device and can therefore be varied and utilized as desired or needed.

[0399] It must also be noted that, as used in this specification and the appended claims, the singular forms "a," "an," and "the" include plural referents unless the context clearly dictates otherwise. Ranges may be expressed herein as from "about" or "approximately" one particular value and / or to "about" or "approximately" another particular value. When such a range is expressed, other exemplary embodiments include from the one particular value and / or to the other particular value.

[0400] “Comprising” or “containing” or “including” means that at least the stated compound, element, particle or method step is present in the composition or article or method, but does not exclude the presence of other compounds, materials, particles or method steps, even if such other compounds, materials, particles or method steps have the same function as the stated compound, material, particle or method step.

[0401] When describing example embodiments, for the sake of clarity, terminology will be used. Each term is intended to encompass its broadest meaning as understood by those skilled in the art, and includes all technical equivalents that operate in a similar manner to achieve similar purposes. It should also be understood that reference to one or more steps of a method does not exclude the presence of additional method steps or intermediate method steps between those steps that are clearly identified. Without departing from the scope of the present disclosure, the steps of the method may be performed in an order different from the order described herein. Similarly, it should also be understood that reference to one or more components in a device or system does not exclude the presence of additional components or intermediate components between those components that are clearly identified.

[0402] Some references are cited in the reference list and discussed in the disclosure provided herein, which may include various patents, patent applications, and publications. The citation and / or discussion of such references is provided only to illustrate the description of the present disclosure, and it is not admitted that any such reference is "prior art" for any aspect of the present disclosure described herein. In terms of notation, "[n]" corresponds to the nth reference in the list. All references cited and discussed in this specification are incorporated herein by reference in their entirety to the same extent as if each reference was individually incorporated by reference.

[0403] It should be understood that, as discussed herein, a subject can be a human or any animal. It should be understood that an animal can be any suitable type, including but not limited to mammals, veterinary animals, livestock animals, or pet-type animals, etc. For example, an animal can be a laboratory animal (e.g., rats, dogs, pigs, monkeys), etc. that is specifically selected to have certain characteristics similar to humans. It should be understood that, for example, a subject can be any suitable human patient.

[0404] As used herein, the term "about" means approximately, in the region, roughly or in the vicinity. When the term "about" is used in conjunction with a numerical range, it modifies the range by extending above and below the boundary of the numerical value. Typically, the term "about" is used in this article to modify the numerical value above and below the value with a variance of 10%. In one aspect, the term "about" means the plus or minus 10% of the numerical value of the number used therewith. Therefore, about 50% means in the range of 45% to 55%. The numerical range described herein by endpoints includes all numerals and fractions (e.g., 1 to 5 includes 1, 1.5, 2, 2.75, 3, 3.90, 4, 4.24 and 5) contained in the range. Similarly, numerical ranges recited herein by endpoints include subranges contained within that range (e.g., 1 to 5 includes 1-1.5, 1.5-2, 2-2.75, 2.75-3, 3-3.90, 3.90-4, 4-4.24, 4.24-5, 2-5, 3-5, 1-4, and 2-4). It is also to be understood that all numbers and fractions thereof are contemplated to be modified by the term "about."

[0405] Additional description of various aspects of the present disclosure will now be provided with reference to the accompanying drawings, which form a part hereof and show by way of illustration specific implementations or examples.

[0406] The following references are incorporated herein by reference in their entirety.

[0407] 1.MDBreton and BPKovatchev, "One year real-world use of the control-IQ advanced hybrid closed-loop technology," Diabetes technology&therapeutics, vol.23, p.601-608, 2021.

[0408] 2.BPKovatchev, "Metrics for glycaemic control—from HbA 1c to continuous glucose monitoring," Nature Reviews Endocrinology, vol.13, p.425-436, 2017.

[0409] 3.B.Kovatchev and C.Cobelli,“Glucose variability:timing,riskanalysis,and relationship to hypoglycemia in diabetes,”Diabetes Care,vol.39,p.502-510,2016.

[0410] 4.C.Cobelli,C.Dalla Man,G.Sparacino,L.Magni,G.De Nicolao andB.P.Kovatchev,“Diabetes:models,signals,and control,”IEEEreviews in biomedicalengineering,vol.2,p.54-96,2009.

[0411] 5.B.P.Kovatchev,W.L.Clarke,M.Breton,K.Brayman and A.McCall,“Quantifying temporal glucose variability in diabetes via continuous glucosemonitoring:mathematical methods and clinical application,”Diabetestechnology&therapeutics,vol.7,p.849-862,2005.

[0412] 6.C.M.McDonnell,S.M.Donath,S.I.Vidmar,G.A.Werther and F.J.Cameron,“Anovel approach to continuous glucose analysis utilizing glycemic variation,”Diabetes technology&therapeutics,vo1.7,p.253-263,2005.

[0413] 7.P.A.Baghurst,“Calculating the mean amplitude of glycemic excursionfrom continuous glucose monitoring data:an automated algorithm,”Diabetestechnology&therapeutics,vol.13,p.296-302,2011.

[0414] 8.C.Fabris,S.D.Patek and M.D.Breton,“Are risk indices derived fromCGM interchangeable with SMBG-based indices7,”Journal of diabetes science andtechnologV,vol.10,p.50-59,2016.

[0415] 9.A.L.McCall,D.J.Cox,J.Crean,M.Gloster and B.P.Kovatchev,“A novelanalytical method for assessing glucose variability:using CGMS in type 1diabetes mellitus,”Diabetes technology&therapeutics,vol.8,p.644-653,2006.

[0416] 10.D.Rodbard,“New and improved methods to characterize glycemicvariability using continuous glucose monitoring,”Diabetes technology &therapeutics,vol.11,p.551-565,2009.

[0417] 11.D.Rodbard,“Interpretation of continuous glucose monitoring data:glycemic variability and quality of glycemic control,”Diabetes technology&therapeutics,vol.11,p.S-55,2009.

[0418] 12.W.Clarke and B.Kovatchev,“Statistical tools to analyze continuousglucose monitor data,”Diabetes technology&therapeutics,vol.11,p.S-45,2009.

[0419] 13.L.Magni,D.M.Raimondo,C.D.Man,M.Breton,S.Patek,G.De Nicolao,C.Cobelli and B.P.Kovatchev,“Evaluating the efficacy of closed-loop glucoseregulation via control-variability grid analysis,”Journal of diabetes scienceand technology,vol.2,p.630-635,2008.

[0420] 14.T.Battelino,T.Danne,R.M.Bergenstal,S.A.Amiel,R.Beck,T.Biester,E.Bosi,B.A.Buckingham,W.T.Cefalu,K.L.Close and others,“Clinical targets forcontinuous glucose monitoring data interpretation:recommendations from theinternational consensus on time in range,”Diabetes care,vol.42,p.1593-1603,2019.

[0421] 15.R.S.Mazze,D.Lucido,O.Langer,K.Hartmann and D.Rodbard,“Ambulatoryglucose profile:representation of verified self-monitored blood glucosedata,”Diabetes Care,vol.10,p.111-117,1987.

[0422] 16.R.M.Bergenstal,A.J.Ahmann,T.Bailey,R.W.Beck,J.Bissen,B.Buckingham,L.Deeb,R.H.Dolin,S.K.Garg,R.Goland and others,Recommendations forstandardizing glucose reporting and analysis to optimize clinical decisionmaking in diabetes:the Ambulatory Glucose Profile(AGP),Mary Ann Liebert,Inc.140 Huguenot Street,3rd Floor New Rochelle,NY 10801 USA,2013.

[0423] 17.A.D.Association,“7.Diabetes Technology:Standards of Medical Carein Diabetes—2019,”Diabetes Care,vol.42,pp.S71-S80,December2018.

[0424] 18.V.Fonseca and G.Grunberger,“Letter to the Editor,”EndocrinePractice,vol.23,pp.629-632,2017.

[0425] 19.B.Kovatchev,S.Anderson,D.Raghinaru,Y.Kudva,L.a.L.C.Laffel,J.Pinsker and others,“Randomized Controlled Trial of Mobile Closed-LoopControl,”Diabetes Care,vol.43,no.3,pp.607-615,2020.

[0426] 20.S.Brown,B.Kovatchev,D.Raghinaru,J.Lum,B.Buckingham,Y.Kudva,L.Laffel and others,“Six-Month Randomized,Multicenter Trial of Closed-LoopControl in Type 1 Diabetes,”New England Journal of Medicine,vol.381,no.18,pp.1707-1717,2019.

[0427] 21.R.Beck,T.Riddlesworth,K.Ruedy,A.Ahmann,R.Bergenstal,S.Haller,C.Kollman and others,“Effect of Continuous Glucose Monitoring on GlycemicControl in Adults With Type 1 Diabetes Using Insulin Injections:The DIAMONDRandomized Clinical Trial,”Journal of the American Medical Association,vol.317,no.4,pp.371-378,2017.

[0428] 22.R.Beck,T.Riddlesworth,K.Ruedy,A.Ahmann,S.Haller,D.Kruger,J.McGilland others,“Continuous Glucose Monitoring Versus Usual Care in Patients WithType 2 Diabetes Receiving Multiple Daily Insulin Injections,”Annals ofInternal Medicine,vol.167,no.6,pp.365-374,2017.

[0429] 23.A.Bisio,S.Anderson,L.Norlander,G.O'Malley,J.Robic,S.Ogyaadu,L.Hsuand others,“Impact of a Novel Diabetes Support System on a Cohort ofIndividuals With Type 1 Diabetes Treated With Multiple Daily Injections:AMulticenter Randomized Study,”DIabetes Care,vol.45,no.1,pp.186-193,2022.

[0430] 24.B.Kovatchev,L.Kollar,S.Anderson,C.Barnett,M.Breton,K.Carr,R.Gildersleeve and others,“Evening and overnight closed-loop control versus24 / 7 continuous closed-loop control for type 1 diabetes:a randomisedcrossover trial,”The Lancet Digital Health,vol.2,no.2,pp.e64-e73,2020.

[0431] 25.L.Laffel,L.Kanapka,R.Beck,K.Bergamo,M.Clements,A.Criego,D.DeSalvoand others,“Effect of Continuous Glucose Monitoring on Glycemic Control inAdolescents and Young Adults With Type 1Diabetes:A Randomized ClinicalTrial,”Journal of the American Medical Association,vol.323,no.23,pp.2388-2396,2020.

[0432] 26.M.Breton,L.Kanapka,R.Beck,L.Ekhlaspour,G.Forlenza,E.Cengiz,M.Schoelwer and others,“A randomized trial of closed-loop control in childrenwith type 1 diabetes,”New England Journal of Medicine,vol.383,no.9,pp.836-845,2020.

[0433] 27.V.Shah,S.DuBose,Z.Li,R.Beck,A.Peters,R.Weinstock,D.Kruger andothers,“Continuous Glucose Monitoring Profiles in Healthy NondiabeticParticipants:A Multicenter Prospective Study,”The Journal of ClinicalEndocrinology&Metabolism,vol.104,no.10,pp.4356-4364,2019.

[0434] 28.M.Rickels,S.DuBose,E.Toschi,R.Beck,A.Verdejo,H.Wolpert,M.Cumminsand others,“Mini-Dose Glucagon as a Novel Approach to Prevent Exercise-Induced Hypoglycemia in Type 1 Diabetes,”Diabetes Care,vol.41,no.9,pp.1909-1916,2018.

[0435] 29.G.Aleppo,K.Ruedy,T.Riddlesworth,D.Kruger,A.Peters,I.Hirsch,R.Bergenstal and others,“REPLACE-BG:A Randomized Trial Comparing ContinuousGlucose Monitoring With and Without Routine Blood Glucose Monitoring inAdults With Well-Controlled Type 1Diabetes,”Diabetes Care,vol.40,no.4,pp.538-545,2017.

[0436] 30.JDRF CGM Study Group,“Continuous glucose monitoring and intensivetreatment of type 1 diabetes,”New England Journal of Medicine,vol.359,no.14,pp.1464-1476,2008.

[0437] 31.L.Laffel,K.Harrington,A.Hanono,N.Naik,L.Ambler-Osborn,A.Schultz,L.DiMeglio and others,“A Randomized Clinical Trial Assessing ContinuousGlucose Monitoring(CGM)Use With Standardized Education With or Without aFamily Behavioral Intervention Compared With Fingerstick Blood GlucoseMonitoring in Very Young Children With Type 1 Diabetes,”Diabetes Care,vol.44,no.2,pp.464-472,2021.

[0438] 32.R.Weinstock,S.DuBose,R.Bergenstal,N.Chaytor,C.Peterson,B.Olson,M.Munshi and others,“Risk Factors Associated With Severe Hypoglycemia inOlder Adults With Type 1 Diabetes,”Diabetes Care,vol.39,no.4,pp.603-610,2016.

[0439] 33.A.Carlson,L.Kanapka,K.Miller,A.Ahmann,N.Chaytor,S.Fox,L.Kiblingerand others,“Hypoglycemia and Glycemic Control in Older Adults With Type 1Diabetes:Baseline Results From the WISDM Study,”Journal of Diabetes Scienceand Technology,vol.15,no.3,pp.582-592,2021.

[0440] 34.B.Lobo,L.Farhy,M.Shafiei and B.Kovatchev,“A data-driven approachto classifying daily continuous glucose monitoring(CGM)time series,”IEEETransactions on Biomedical Engineering,vol.62,no.2,pp.654-665,2022.

[0441] 35.P.Virtanen,R.Gommers,T.Oliphant,M.Haberland,T.Reddy,D.Cournapeauand others,“SciPy 1.0:Fundamental Algorithms for Scientific Computing inPython,”Nature Methods,vol.17,pp.261-272,2020.

[0442] 36.National Institute of Standards and Technology,“WeibullDistribution,”[Online].Available:https: / / www.itl.nist.gov / div898 / handbook / eda / section3 / eda3668.htm.[Ac cessed 13 12 2021].

[0443] 37.International Patent Application Serial No.PCT / US2022 / 032121,entitled“METHOD FOR STRUCTURING AND CLASSIFICATION OF CONTINUOUS GLUCOSEMONITORING(CGM)PROFILES”,filed June 03,2022.

[0444] 38.U.S.Utility Patent Application Serial No.17 / 829,754,entitled“METHOD FOR STRUCTURING AND CLASSIFICATION OF CONTINUOUS GLUCOSE MONITORING(CGM)PROFILES”,filed June 01,2022.

[0445] 39.U.S.Utility Patent Application Serial No.17 / 776,717,entitled“METHOD AND SYSTEM FOR MODEL-BASED TRACKING OF HEMOGLOBIN A1c FROM DAILYCONTINUOUS GLUCOSE MONITORING PROFILES”,filed May 13,2022.

[0446] 40.International Patent Application Serial No.PCT / US2020 / 060658,entitled“METHOD AND SYSTEM FOR MODEL-BASED TRACKING OF HEMOGLOBIN A1c FROMDAILY CONTINUOUS GLUCOSE MONITORING PROFILES”,filed November 16,2020;Publication No.WO 2021 / 097396,May 20,2021.

[0447] 41.U.S.Utility Patent Application Serial No.17 / 776,056,entitled“SYSTEM,METHOD AND COMPUTER READABLE MEDIUM FOR COMPRESSING CONTINUOUSGLUCOSE MONITOR DATA”,filed May 11,2022.

[0448] 42.International Patent Application Serial No.PCT / US2020 / 060366,entitled“SYSTEM,METHOD AND COMPUTER READABLE MEDIUM FOR COMPRESSINGCONTINUOUS GLUCOSE MONITOR DATA”,filed November 13,2020;Publication No.WO2021 / 097178,May 20,2021.

[0449] 43.U.S.Utility Patent Application Serial No.17 / 694,870,entitled“INSULIN MONITORING AND DELIVERY SYSTEM AND METHOD FOR CGM BASED FAULTDETECTION AND MITIGATION VIA METABOLIC STATE TRACKING”,filed March15,2022;Publication No.US 2022-0203020 A1,June 30,2022.

[0450] 44.U.S.Utility Patent Application Serial No.15 / 580,935,entitled“INSULIN MONITORING AND DELIVERY SYSTEM AND METHOD FOR CGM BASED FAULTDETECTION AND MITIGATION VIA METABOLIC STATE TRACKING”,filed December 08,2017;U.S.Patent No.11,311,665,issued April 26,2022.

[0451] 45.International Patent Application Serial No.PCT / US2016 / 036729,entitled“INSULIN MONITORING AND DELIVERY SYSTEM AND METHOD FOR CGM BASEDFAULT DETECTION AND MITIGATION VIA METABOLIC STATE TRACKING”,filed June 09,2016;Publication No.WO 2016 / 201120,December 15,2016.

[0452] 46.U.S.Utility Patent Application Serial No.17 / 683,676,entitled“System,Method and Computer Readable Medium for Dynamical Tracking of theRisk for Hypoglycemia in Type 1 and Type 2 Diabetes”,filed March 01,2022;Publication No.US 2022-0262519 A1,August 18,2022.

[0453] 47.U.S.Utility Patent Application Serial No.15 / 958,257,entitled“System,Method and Computer Readable Medium for Dynamical Tracking of theRisk for Hypoglycemia in Type 1 and Type 2 Diabetes”,filed April 20,2018;U.S.Patent No.11,289,201,issued March 29,2022.

[0454] 48.International Patent Application Serial No.PCT / US2016 / 058234,entitled“System,Method and Computer Readable Medium for Dynamical Tracking ofthe Risk for Hypoglycemia in Type 1 and Type 2Diabetes”,filed October 21,2016;Publication No.WO 2017 / 070553,April 27,2017.

[0455] 49.International Patent Application Serial No.PCT / US2022 / 017489,entitled“METHOD AND SYSTEM FOR QUANTITATIVE PHYSIOLOGICAL ASSESSMENT ANDPREDICTION OF CLINICAL SUBTYPES OF GLUCOSE METABOLISM DISORDERS”,filedFebruary 23,2022;Publication No.WO 2022 / 182736,September 01,2022.

[0456] 50.International Patent Application Serial No.PCT / US2022 / 017449,entitled“METHOD AND SYSTEM FOR MAPPING INDIVIDUALIZED METABOLIC PHENOTYPE TOA DATABASE IMAGE FOR OPTIMIZING CONTROL OF CHRONIC METABOLIC CONDITIONS”,filed February 23,2022;Publication No.WO2022182709,September 01,2022.

[0457] 51.U.S.Utility Patent Application Serial No.17 / 590,659,entitled“System,Method,and Computer Simulation Environment for In Silico Trials inPre-Diabetes and Type 2 Diabetes”,filed February 01,2022;Publication No.US2022-0230762 A1,July 21,2022.

[0458] 52.U.S.Utility Patent Application Serial No.13 / 380,839,entitled“System,Method,and Computer Simulation Environment for In Silico Trials inPre-Diabetes and Type 2 Diabetes”,filed December 25,2011;U.S.Patent No.11,238,990,issued February 01,2022.

[0459] 53.International Patent Application Serial No.PCT / US2010 / 040097,entitled“System,Method,and Computer Simulation Environment for In SilicoTrials in Prediabetes and Type 2 Diabetes”,filed June 25,2010;PublicationNo.WO 2010 / 151834,December 29,2010.

[0460] 54.International Patent Application Serial No.PCT / US2021 / 045936,entitled“METHOD AND SYSTEM FOR GENERATING A USER TUNABLE REPRESENTATION OFGLUCOSE HOMEOSTASIS IN TYPE 1 DIABETES BASED ON AUTOMATED RECEIPT OF THERAPYPROFILE DATA”,filed August 13,2021;Publication No.WO 2022 / 036214,February 17,2022.

[0461] 55.U.S.Utility Patent Application Serial No.17 / 339,153,entitled“Method and System for the Safety,Analysis,and Supervision of Insulin PumpAction and Other Modes of Insulin Delivery in Diabetes”,filed June 04,2021;Publication No.US 2021-0313036 A1,October 07,2021.

[0462] 56.U.S.Utility Patent Application Serial No.13 / 634,040,entitled“Method and System for the Safety,Analysis,and Supervision of Insulin PumpAction and Other Modes of Insulin Delivery in Diabetes”,filed September 11,2012;U.S.Patent No.11,069,434,issued July 20,2021.

[0463] 57.International Patent Application Serial No.PCT / US2011 / 028163,entitled“Method and System for the Safety,Analysis,and Supervision of InsulinPump Action and Other Modes of Insulin Delivery in Diabetes”,filed March 11,2011;Publication No.WO 2011 / 112974,September 15,2011.

[0464] 58.U.S.Utility Patent Application Serial No.17 / 333,161,entitled“METHOD,SYSTEM AND COMPUTER PROGRAM PRODUCT FOR CGM-BASED PREVENTION OFHYPOGLYCEMIA VIA HYPOGLYCEMIA RISK ASSESSMENT AND SMOOTH REDUCTION INSULINDELIVERY”,filed May 28,2021;Publication No.US 2021-0282677 A1,September 16,2021.

[0465] 59.U.S.Utility Patent Application Serial No.17 / 070,245,entitled“METHOD,SYSTEM AND COMPUTER PROGRAM PRODUCT FOR CGM-BASED PREVENTION OFHYPOGLYCEMIA VIA HYPOGLYCEMIA RISK ASSESSMENT AND SMOOTH REDUCTION INSULINDELIVERY”,filed October 14,2020;Publication No.US 2021 / 0038132 A1,February11,2021.

[0466] 60.U.S.Utility Patent Application Serial No.15 / 669,111,entitled“METHOD,SYSTEM AND COMPUTER PROGRAM PRODUCT FOR CGM-BASED PREVENTION OFHYPOGLYCEMIA VIA HYPOGLYCEMIA RISK ASSESSMENT AND SMOOTH REDUCTION INSULINDELIVERY”,filed August 04,2017;U.S.Patent No.10,842,419,issued November 24,2020.

[0467] 61.U.S.Utility Patent Application Serial No.14 / 015,831,entitled“CGM-Based Prevention of Hypoglycemia Via Hypoglycemia Risk Assessment and SmoothReduction of Insulin Delivery”,filed August 30,2013;U.S.Patent No.9,750,438,issued September 05,2017.

[0468] 62.U.S.Utility Patent Application Serial No.13 / 203,469,entitled“CGM-Based Prevention of Hypoglycemia via Hypoglycemia Risk Assessment and SmoothReduction Insulin Delivery”,filed August 25,2011;U.S.Patent No.8,562,587,issued October 22,2013.

[0469] 63.International Patent Application Serial No.PCT / US2010 / 025405,entitled“CGM-BASED PREVENTION OF HYPOGLYCEMIA VIA HYPOGLYCEMIA RISK ASSESMENTAND SMOOTH REDUCTION INSULIN DELIVERY”,filed February 25,2010;PublicationNo.WO 2010 / 099313 A1,September 02,2010.

[0470] 64.U.S.Utility Patent Application Serial No.17 / 181,888,entitled“SYSTEMS OF CENTRALIZED DATA EXCHANGE FOR MONITORING AND CONTROL OF BLOOD

[0471] 65.GLUCOSE”,filed February 22,2021;Publication No.US 2021-0169409 A1,June 10,2021.

[0472] 66.U.S.Utility Patent Application Serial No.15 / 109,682,entitled“SYSTEMS OF CENTRALIZED DATA EXCHANGE FOR MONITORING AND CONTROL OF BLOODGLUCOSE”,filed July05,2016;U.S.Patent No.10,925,536,issued February 23,2021.

[0473] 67.International Patent Application Serial No.PCT / US2015 / 010167,entitled“Central Data Exchange Node For System Monitoring and Control ofBlood Glucose Levels in Diabetic Patients”,filed January 05,2015;PublicationNo.WO2015103543,July 09,2015.

[0474] 68.U.S.Utility Patent Application Serial No.17 / 180,301,entitled“INFLUENCING END-STAGE RENAL DISEASE OUTCOMES THROUGH PREDICTINGPHYSIOLOGICAL PARAMETERS AND DETERMINING DOSING RECOMMENDATIONS”,filedFebruary 19,2021;Publication No.US 2021-0257096 A1,August 19,2021.

[0475] 69.U.S.Utility Patent Application Serial No.17 / 156,169,entitled“Method,System and Computer Program Product for Real-Time Detection ofSensitivity Decline in Analyte Sensors”,filed January 22,2021;PublicationNo.US 2021-0218481 A1,July 15,2021.

[0476] 70.U.S.Utility Patent Application Serial No.15 / 866,384,entitled“Method,System and Computer Program Product for Real-Time Detection ofSensitivity Decline in Analyte Sensors”,filed January 09,2018;U.S.PatentNo.10,903,914,issued January 26,2021.

[0477] 71.U.S.Utility Patent Application Serial No.14 / 266,612,entitled“Method,System and Computer Program Product for Real-Time Detection ofSensitivity Decline in Analyte Sensors”,filed April 30,2014;U.S.Patent No.9,882,660,issued January 30,2018.

[0478] 72.U.S.Utility Patent Application Serial No.13 / 418,305,entitled“Method,System and Computer Program Product for Real-Time Detection ofSensitivity Decline in Analyte Sensors”,filed March 12,2012;U.S.Patent No.8,718,958,issued May 06,2014.

[0479] 73.International Patent Application Serial No.PCT / US2007 / 082744,entitled“Method,System and Computer Program Product for Real-Time Detectionof Sensitivity Decline in Analyte Sensors”,filed October 26,2007;PublicationNo.WO / 2008 / 052199,May 02,2008.

[0480] 74.U.S.Utility Patent Application Serial No.11 / 925,689,entitled“Method,System and Computer Program Product for Real-Time Detection ofSensitivity Decline in Analyte Sensors”,filed October 26,2007;U.S.PatentNo.8,135,548,issued March 13,2012.

[0481] 75.U.S.Utility Patent Application Serial No.17 / 132,604,entitled“ACCURACY CONTINUOUS GLUCOSE MONITORING METHOD,SYSTEM,AND DEVICE”,filedDecember 23,2020;Publication No.US 2021-0113122 A1,April 22,2021.

[0482] 76.U.S.Utility Patent Application Serial No.15 / 510,878,entitled“ACCURACY CONTINUOUS GLUCOSE MONITORING METHOD,SYSTEM,AND DEVICE”,filed March13,2017;U.S.Patent No.10,881,334,issued January 05,2021.

[0483] 77.International Patent Application Serial No.PCT / US2015 / 045340,entitled“IMPROVED ACCURACY CONTINUOUS GLUCOSE MONITORING METHOD,SYSTEM,ANDDEVICE”,filed August 14,2015;Publication No.WO2016025874,February 18,2016.

[0484] 78.U.S.Utility Patent Application Serial No.16 / 789,901,entitled“Unified Platform For Monitoring and Control of Blood Glucose Levels inDiabetic Patients”,filed February 13,2020;Publication No.US-2020-0214629-A1,July 09,2020.

[0485] 79.U.S.Utility Patent Application Serial No.14 / 128,922,entitled“Unified Platform For Monitoring and Control of Blood Glucose Levels inDiabetic Patients”,filed December 23,2013;U.S.Patent No.10,610,154,issuedApril 07,2020.

[0486] 80.International Patent Application Serial No.PCT / US2012 / 043910,entitled“Unified Platform For Monitoring and Control of Blood Glucose Levelsin Diabetic Patients”,filed June 23,2012;Publication No.WO 2012 / 178134,December 27,2012.

[0487] 81.U.S.Utility Patent Application Serial No.16 / 588,881,entitled“Tracking the Probability for Imminent Hypoglycemia in Diabetes from Self-Monitoring Blood Glucose(SMBG)Data”,filed September 30,2019;PublicationNo.US-2020-0066410-A1,February 27,2020.

[0488] 82.U.S.Utility Patent Application Serial No.13 / 394,091,entitled“Tracking the Probability for Imminent Hypoglycemia in Diabetes from Self-Monitoring Blood Glucose(SMBG)Data”,filed March 02,2012;U.S.Patent No.10,431,342,issued October 01,2019.

[0489] 83.International Patent Application Serial No.PCT / US2010 / 047711,entitled“Tracking the Probability for Imminent Hypoglycemia in Diabetes fromSelf-Monitoring Blood Glucose(SMBG)Data”,filed September 02,2010;PublicationNo.WO 2011 / 028925,March 10,2011.

[0490] 84.U.S.Utility Patent Application Serial No.16 / 546,335,entitled“System Coordinator and Modular Architecture for Open-Loop and Closed-LoopControl of Diabetes”,filed August 21,2019;Publication No.US-2019-0374137-A1,December 12,2019.

[0491] 85.U.S.Utility Patent Application Serial No.13 / 322,943,entitled“System Coordinator and Modular Architecture for Open-Loop and Closed-LoopControl of Diabetes”,filed November 29,2011;U.S.Patent No.10,420,489,issuedSeptember 24,2019.

[0492] 86.International Patent Application Serial No.PCT / US2010 / 036629,entitled“System Coordinator and Modular Architecture for Open-Loop andClosed-Loop Control of Diabetes”,filed May 28,2010;Publication No.WO 2010 / 138848,December 02,2010.

[0493] 87.U.S.Utility Patent Application Serial No.16 / 451,766,entitled“TRACKING CHANGES IN AVERAGE GLYCEMIA IN DIABETICS”,filed June 25,2019;Publication No.US-2019-0318801-A1,October 17,2019.

[0494] 88.U.S.Utility Patent Application Serial No.14 / 769,638,entitled“METHOD AND SYSTEM FOR MODEL-BASED TRACKING OF CHANGES IN AVERAGE GLYCEMIA INDIABETES”,filed August21,2015;U.S.Patent No.10,332,615,issued June 25,2019.

[0495] 89.International Patent Application Serial No.PCT / US2014 / 017754,entitled“METHOD AND SYSTEM FOR MODEL-BASED TRACKING OF CHANGES IN AVERAGEGLYCEMIA IN DIABETES”,filed February 21,2014;Publication No.WO 2014 / 130841,August 28,2014.

[0496] 90.U.S.Utility Patent Application Serial No.17 / 180,301,entitled“INFLUENCING END-STAGE RENAL DISEASE OUTCOMES THROUGH PREDICTINGPHYSIOLOGICAL PARAMETERS AND DETERMINING DOSING RECOMMENDATIONS”,filedFebruary 19,2021;Publication No.US 2021-0257096 A1,August 19,2021.

[0497] 91.U.S.Utility Patent Application Serial No.16 / 126,879,entitled“Method,System and Computer Program Product for Evaluation of InsulinSensitivity,Insulin / Carbohydrate Ratio,and Insulin Correction Factors inDiabetes from Self-Monitoring Data”,filed September 10,2018;PublicationNo.US-2019-0019571-A1,January 17,2019.

[0498] 92.U.S.Utility Patent Application Serial No.12 / 665,149,entitled“Method,System and Computer Program Product for Evaluation of InsulinSensitivity,Insulin / Carbohydrate Ratio,and Insulin Correction Factors inDiabetes from Self-Monitoring Data”,filed December 17,2009;PublicationNo.2010 / 0198520,August 05,2010.

[0499] 93.International Patent Application Serial No.PCT / US2008 / 069416,entitled“Method,System and Computer Program Product for Evaluation of InsulinSensitivity,Insulin / Carbohydrate Ratio,and Insulin Correction Factors inDiabetes from Self-Monitoring Data”,filed July 08,2008;Publication No.WO2009 / 009528,January 15,2009.

[0500] 94.U.S.Utility Patent Application Serial No.15 / 580,915,entitled“System and Method for Tracking Changes in Average Glycemia in Diabetics”,filed December 08,2017;Publication No.US-2018-0313815-A1,November 01,2018.

[0501] 95.International Patent Application Serial No.PCT / US2016 / 036481,entitled“System and Method for Tracking Changes in Average Glycemia inDiabetics”,filed June 08,2016;Publication No.WO2016200970,December 15,2016.

[0502] 96.U.S.Utility Patent Application Serial No.14 / 902,731,entitled“SIMULATION OF ENDOGENOUS AND EXOGENOUS GLUCOSE / INSULIN / GLUCAGON INTERPLAY INTYPE 1DIABETIC PATIENTS”,filed January 04,2016;U.S.Patent No.10,169,544,issued January 01,2019.

[0503] 97.International Patent Application Serial No.PCT / US2014 / 045393,entitled“SIMULATION OF ENDOGENOUS AND EXOGENOUS GLUCOSE / INSULIN / GLUCAGONINTERPLAY IN TYPE 1DIABETIC PATIENTS”,filed July 03,2014;PublicationNo.WO2015003124,January 08,2015.

[0504] 98.U.S.Utility Patent Application Serial No.14 / 799,329,entitled“ACCURACY OF CONTINUOUS GLUCOSE SENSORS”,filed July14,2015;U.S.Patent No.10,194,850,issued February 05,2019.

[0505] 99.U.S.Utility Patent Application Serial No.12 / 065,257,entitled“Accuracy of Continuous Glucose Sensors”,filed February 28,2008;PublicationNo.2008 / 0314395,December 25,2008.

[0506] 100.International Patent Application Serial No.PCT / US2006 / 033724,entitled“Method for Improvising Accuracy of Continuous Glucose Sensors and aContinuous Glucose Sensor Using the Same”,filed August 29,2006;PublicationNo.WO07027691,March 08,2007.

[0507] 101.U.S.Utility Patent Application Serial No.14 / 419,375,entitled“COMPUTER SIMULATION FOR TESTING AND MONITORING OF TREATMENT STRATEGIES FORSTRESS HYPERGLYCEMIA”,filed February 03,2015;U.S.Patent No.10,438,700,issuedOctober 08,2019.

[0508] 102.International Patent Application Serial No.PCT / US2013 / 053664,entitled“COMPUTER SIMULATION FOR TESTING AND MONITORING OF TREATMENTSTRATEGIES FOR STRESS HYPERGLYCEMIA”,filed August 05,2013;Publication No.WO2014 / 022864,February 06,2014.

[0509] 103.U.S.Utility Patent Application Serial No.14 / 241,383,entitled“Method,System and Computer Readable Medium for Adaptive Advisory Control ofDiabetes”,filed February 26,2014;U.S.Patent No.11,024,429,issued June 01,2021.

[0510] 104.International Patent Application Serial No.PCT / US2012 / 052422,entitled“Method,System and Computer Readable Medium for Adaptive AdvisoryControl of Diabetes”,filed August 26,2012;Publication No.WO 2013 / 032965,March07,2013.

[0511] 105.U.S.Utility Patent Application Serial No.14 / 128,811,entitled“Methods and Apparatus for Modular Power Management and Protection ofCritical Services in Ambulatory Medical Devices”,filed December 23,2013;U.S.Patent No.9,430,022,issued August 30,2016.

[0512] 106.International Patent Application Serial No.PCT / US2012 / 043883,entitled“Methods and Apparatus for Modular Power Management and Protection ofCritical Services in Ambulatory Medical Devices”,filed June 22,2012;Publication No.WO 2012 / 178113,December 27,2012.

[0513] 107.U.S.Utility Patent Application Serial No.29 / 467,039,entitled“Alarm Clock Display of Personal Blood Glucose Level”,filed September 13,2013.

[0514] 108.International Patent Application Serial No.PCT / US2013 / 042745,entitled“INSULIN-PRAMLINTIDE COMPOSITIONS AND METHODS FOR MAKING AND USINGTHEM”,filed May 24,2013;Publication No.WO 2013 / 177565,November 28,2013.

[0515] 109.U.S.Utility Patent Application Serial No.13 / 637,359,entitled“METHOD,SYSTEM,AND COMPUTER PROGRAM PRODUCT FOR IMPROVING THE ACCURACY OFGLUCOSE SENSORS USING INSULIN DELIVERY OBSERVATION IN DIABETES”,filedSeptember 25,2012;U.S.Patent No.9,398,869,issued July 26,2016.

[0516] 110.International Patent Application Serial No.PCT / US2011 / 029793,entitled“METHOD,SYSTEM,AND COMPUTER PROGRAM PRODUCT FOR IMPROVING THEACCURACY OF GLUCOSE SENSORS USING INSULIN DELIVERY OBSERVATION IN DIABETES”,filed March 24,2011;Publication No.WO 2011 / 119832,September 29,2011.

[0517] 111.U.S.Utility Patent Application Serial No.13 / 393,647,entitled“System,Method and Computer Program Product for Adjustment of InsulinDelivery(AID)in Diabetes Using Nominal Open-Loop Profiles”,filed March 01,2012;Publication No.2012 / 0245556,September 27,2012.

[0518] 112.International Patent Application Serial No.PCT / US2010 / 047386,entitled“System,Method and Computer Program Product for Adjustment of InsulinDelivery(AID)in Diabetes Using Nominal Open-Loop Profiles”,filed August 31,2010;Publication No.WO 2011 / 028731,March 10,2011.

[0519] 113.U.S.Utility Patent Application Serial No.13 / 131,467,entitled“Method,System,and Computer Program Product for Tracking of Blood GlucoseVariability in Diabetes”,filed May 26,2011;U.S.Patent No.9,317,657,issuedApril 19,2016.

[0520] 114.International Patent Application Serial No.PCT / US2009 / 065725,entitled“Method,System,and Computer Program Product for Tracking of BloodGlucose Variability in Diabetes”,filed November 24,2009;Publication No.WO2010 / 062898,June 03,2010.

[0521] 115.U.S.Utility Patent Application Serial No.12 / 975,580,entitled“Method,System,and Computer Program Product for the Evaluation of GlycemicControl in Diabetes from Self-Monitoring Data”,filed December 22,2010;Publication No.2012 / 0004512,January 05,2012.

[0522] 116.U.S.Utility Patent Application Serial No.11 / 305,946,entitled“Method,System,and Computer Program Product for the Evaluation of GlycemicControl in Diabetes from Self-Monitoring Data”,filed December 19,2005;U.S.Patent No.7,874,985,issued January 25,2011.

[0523] 117.U.S.Utility Patent Application Serial No.10 / 240,228,entitled“Method,System,and Computer Program Product for the Evaluation of GlycemicControl in Diabetes from Self-Monitoring Data”,filed September 26,2002;U.S.Patent No.7,025,425,issued April 11,2006.

[0524] 118.International Patent Application Serial No.PCT / US2001 / 009884,entitled“Method,System,and Computer Program Product for the Evaluation ofGlycemic Control in Diabetes”,filed March 29,2001;Publication No.WO 01 / 72208,October 04,2001.

[0525] 119.U.S.Utility Patent Application Serial No.12 / 664,444,entitled“Method,System and Computer Simulation Environment for Testing of Monitoringand Control Strategies in Diabetes”,filed December 14,2009;U.S.Patent No.10,546,659,issued January 28,2020.

[0526] 120.International Patent Application Serial No.PCT / US2008 / 067725,entitled“Method,System and Computer Simulation Environment for Testing ofMonitoring and Control Strategies in Diabetes”,filed June 20,2008;PublicationNo.WO 2008 / 157781,December 24,2008.

[0527] 121.U.S.Utility Patent Application Serial No.12 / 516,044,entitled“Method,System,and Computer Program Product for the Detection of PhysicalActivity by Changes in Heart Rate,Assessment of Fast Changing MetabolicStates,and Applications of Closed and Open Control Loop in Diabetes”,filedMay 22,2009;U.S.Patent No.8,585,593,issued November 19,2013.

[0528] 122.International Patent Application Serial No.PCT / US2007 / 085588,entitled“Method,System,and Computer Program Product for the Detection ofPhysical Activity by Changes in Heart Rate,Assessment of Fast ChangingMetabolic States,and Applications of Closed and Open Control Loop inDiabetes”,filed November 27,2007;Publication No.WO2008 / 067284,June 05,2008.

[0529] 123.U.S.Utility Patent Application Serial No.12 / 159,891,entitled“Method,System and Computer Program Product for Evaluation of Blood GlucoseVariability in Diabetes from Self-Monitoring Data”,filed July 02,2008;U.S.Patent No.11,355,238,issued June 07,2022.

[0530] 124.International Patent Application Serial No.PCT / US2007 / 000370,entitled“Method,System and Computer Program Product for Evaluation of BloodGlucose Variability in Diabetes from Self-Monitoring Data”,filed January 05,2007;Publication No.WO07081853,July 19,2007.

[0531] 125.U.S.Utility Patent Application Serial No.11 / 943,226,entitled“Systems,Methods and Computer Program Codes for Recognition of Patterns ofHyperglycemia and Hypoglycemia,Increased Glucose Variability,and IneffectiveSelf-Monitoring in Diabetes”,filed November 20,2007;Publication No.2008 / 0154513,June 26,2008.

[0532] 126.U.S.Utility Patent Application Serial No.11 / 578,831,entitled“Method,System and Computer Program Product for Evaluating the Accuracy ofBlood Glucose Monitoring Sensors / Devices”,filed October 18,2006;U.S.PatentNo.7,815,569,issued October 19,2010.

[0533] 127.International Patent Application Serial No.US2005 / 013792,entitled“Method,System and Computer Program Product for Evaluating the Accuracy ofBlood Glucose Monitoring Sensors / Devices”,filed April 21,2005;PublicationNo.WO05106017,November 10,2005.

[0534] 128.U.S.Utility Patent Application Serial No.10 / 524,094,entitled“Method,System,And Computer Program Product For The Processing Of Self-Monitoring Blood Glucose(SMBG)Data To Enhance Diabetic Self-Management”,filedFebruary 09,2005;U.S.Patent No.8,538,703,issued September 17,2013.

[0535] 129.International Patent Application Serial No.PCT / US2003 / 025053,entitled“Managing and Processing Self-Monitoring Blood Glucose”,filed August08,2003;Publication No.WO 2004 / 015539,February 19,2004.

[0536] 130.International Patent Application Serial No.PCT / US2002 / 005676,entitled“METHOD AND APPARATUS FOR THE EARLY DIAGNOSIS OF SUBACUTE,POTENTIALLYCATASTROPHIC ILLNESS”,filed February 27,2002;Publication No.WO02 / 67776,September 06,2002.

[0537] 131.U.S.Utility Patent Application Serial No.09 / 793,653,entitled“METHOD AND APPARATUS FOR THE EARLY DIAGNOSIS OF SUBACUTE,POTENTIALLYCATASTROPHIC ILLNESS”,filed February 27,2001;U.S.Patent No.6,804,551,issuedOctober 12,2004.

[0538] 132.U.S.Utility Patent Application Serial No.10 / 069,674,entitled“Method and Apparatus for Predicting the Risk of Hypoglycemia”,filedFebruary22,2002;U.S.Patent No.6,923,763,issued August 02,2005.

[0539] 133.International Patent Application Serial No.US00 / 22886,entitled“METHOD AND APPARATUS FOR PREDICTING THE RISK OF HYPOGLYCEMIA”,filed August21,2000;Publication No.WO01 / 13786,March 01,2001.

Claims

1. A system for processing glucose data through efficient glucose database management, the system include: a physical data storage area containing glucose measurement data and a representation of at least one cluster of the glucose measurement data, wherein the representation is an array of blood glucose curve vectors approximating a cluster of a plurality of glucose curves segmented by a plurality of time ranges; and A processor and computer memory configured with instructions stored thereon which, when executed, will cause the processor to: receiving a glucose measurement; converting the glucose measurement value into a vector form; searching the physical data store by comparing a newly received glucose measurement to a centroid of a cluster using a similarity metric; Based on the comparison, classifying the newly received glucose measurement using a cluster having a matching similarity metric; and A treatment regimen is prescribed based on the newly received glucose measurement.

2. The system according to claim 1, in, The instructions cause the processor to perform one or more of the following: storing the classification result of the newly received glucose measurement in a data store that is in communication with one or more of a predictive modeling system, a decision support system, an insulin delivery system, an insulin monitoring system, or an automated control system configured to use the classification result as input; sending the classification result of the newly received glucose measurement to one or more of a predictive modeling system, a decision support system, an insulin delivery system, an insulin monitoring system, or an automated control system configured to use the classification result as input; or The classification result of the newly received glucose measurement is used to monitor, analyze or influence the concentration of the glucose level in the fluid.

3. The system according to claim 1, in, Instructions cause the processor to receive the glucose measurement value from a glucose measurement device.

4. The system according to claim 3, include: Glucose measuring device.

5. The system according to claim 2, include: a data store in communication with one or more of the predictive modeling system, the decision support system, the insulin delivery system, the insulin monitoring system, or the automated control system; or One or more of the predictive modeling system, the decision support system, the insulin delivery system, the insulin monitoring system, or the automated control system.

6. The system according to claim 1, in, The instructions cause the processor to: A Euclidean distance between one or more newly received glucose measurements and one or more centroids is calculated as the similarity measure.

7. The system of claim 1, comprising a plurality of clusters, wherein the plurality of clusters are generated by: generating an array of glucose measurement values ​​for each time range, wherein the plurality of arrays form a blood glucose curve vector; Assigning weights to arrays ； as well as applying an iterative hierarchical clustering technique that varies weights until one or more clusters that approximate one or more blood glucose curve vectors are generated; as well as A cluster in the plurality of clusters is defined by a centroid of the cluster.

8. The system according to claim 7, in, The iterative hierarchical clustering technique is calculated by linear regression of array R 2 value, and change the weights to maximize the R 2 value.

9. The system according to claim 1, in, The plurality of time ranges include five time ranges.

10. The system according to claim 1, in, The multiple time ranges include: Grade 2 hypoglycemia below the glucose measurement -1; Grade 1 hypoglycemia within the range of glucose measurement -2 and glucose measurement -3; a target range (TIR) ​​between a glucose measurement value of -4 and a glucose measurement value of -5; Grade 1 hyperglycemia within the range of a glucose measurement of -6 and a glucose measurement of -7; and Grade 2 hyperglycemia above a glucose measurement of -8.

11. The system according to claim 10, in, Glucose measurement -1 was 54 mg / dL; Glucose measurement -2 was 54 mg / dL; Glucose measurement -3 was 70 mg / dL; Glucose measurement -4 was 70 mg / dL; Glucose measurement -5 was 180 mg / dL; Glucose measurement -6 was 180 mg / dL; Glucose measurement -7 is 250 mg / dL; and Glucose measured -8 was 250 mg / dL.

12. The system according to claim 1, in, The glucose measurements include a plurality of glucose curves for the individual, each glucose curve including a plurality of glucose measurements obtained over a predetermined time period, wherein the instructions cause the processor to: compiling the plurality of glucose curves into a single glucose measurement time series for the individual; and One or more glucose curves are classified using the one or more clusters to generate an index sequence representing classification results of the one or more glucose curves in the single glucose measurement value time series.

13. The system according to claim 12, in, The instructions cause the processor to: A trace representing glucose variability for the individual is generated using the index sequence.

14. The system according to claim 12, in, The instructions cause the processor to: An approximate glucose report (AGP) is generated using the index sequence.

15. The system according to claim 1, in, One or more of the plurality of glucose profiles is a continuous glucose monitoring (CGM) profile that includes glucose measurements obtained over a 24 hour period.

16. The system according to claim 12, in, One or more of the plurality of glucose profiles of the individual is a continuous glucose monitoring (CGM) profile that includes glucose measurements obtained over a 24 hour period.

17. A method for processing glucose data for efficient glucose database management, the method include: receiving a glucose measurement; converting the glucose measurement value into a vector form; The physical data store is searched by comparing the newly received glucose measurement to the centroid of the cluster using a similarity metric, where The physical data storage area includes glucose measurement data and a representation of at least one cluster of the glucose measurement data, wherein the representation is a blood glucose curve vector approximating a cluster of a plurality of glucose curves segmented by a plurality of time ranges; Based on the comparison, classifying the newly received glucose measurement using clusters having matching similarity metrics; and A treatment regimen is prescribed based on the newly received glucose measurement.

18. The method according to claim 17, include: A Euclidean distance between one or more newly received glucose measurements and one or more centroids is calculated as the similarity measure.

19. The method according to claim 16, in, The physical data storage area includes a plurality of clusters, and the plurality of clusters are generated by: generating an array of glucose measurement values ​​for each time range, wherein the plurality of arrays form a blood glucose curve vector; assigning weights to the array; applying an iterative hierarchical clustering technique that varies weights until one or more clusters that approximate one or more blood glucose curve vectors are generated; as well as A cluster in the set of clusters is defined by the centroids of the clusters.

20. The method according to claim 19, include: The R was calculated by linear regression of the array via the iterative hierarchical clustering technique. 2 value, and change the weights to maximize the R 2 value.

Images