Systems, devices, and methods related to medication dosage guidance

Abstract

Systems, devices, and methods for determining a medication dose for a patient or user are provided. The dose determination can take into account recent and / or historical analyte levels of the patient or user. The dose determination can also take into account other information about the patient or user, such as physiological information, dietary information, activity, and / or behavior. A variety of different dose determination implementations are set forth, involving a number of different aspects of the system or environment in which the implementation can be implemented.

Description

[0001] Cross-reference to related applications

[0002] This application claims priority and benefit to U.S. Provisional Application No. 62 / 882,249, filed August 2, 2019; U.S. Provisional Application No. 62 / 979,578, filed February 21, 2020; U.S. Provisional Application No. 62 / 979,594, filed February 21, 2020; U.S. Provisional Application No. 62 / 979,618, filed February 21, 2020; and U.S. Provisional Application Serial No. 63 / 058,799, filed July 30, 2020, all of which are hereby expressly incorporated herein by reference in their entirety. Technical Field

[0003] The topics described herein generally relate to systems, apparatus, and methods related to drug dosage guidance, such as determining insulin dosages for treating elevated glucose levels caused by diabetes. Background Technology

[0004] For individuals with diabetes, detecting and / or monitoring analyte levels, such as glucose, ketones, lactate, oxygen, and hemoglobin AIC, can be extremely important for their health. Patients with diabetes may experience a variety of complications, including loss of consciousness, cardiovascular disease, retinopathy, neuropathy, and nephropathy. Monitoring glucose levels in diabetic patients is often necessary to ensure they are maintained within a clinically safe range, and this information can also be used to determine whether and / or when insulin is needed to lower their glucose levels, or when supplemental glucose administration is required to raise them.

[0005] Growing clinical data demonstrate a strong correlation between the frequency of glucose monitoring and glycemic control. Despite this correlation, many individuals diagnosed with diabetes do not monitor their glucose levels as frequently as they should due to a combination of factors, including inconvenience, the randomness of testing, pain associated with glucose testing, and costs.

[0006] For patients who rely on medications (such as insulin) to treat or control diabetes, there is a desire for systems, devices, or methods capable of automatically providing medication dosage guidance as needed in an easily accessible manner, utilizing glucose information collected by an analyte monitoring system. It is also desirable for such systems, devices, or methods to take into account the physiology, diet, activity, and / or behavior of the user or patient being treated when providing such medication dosage guidance, thereby improving accuracy and reliability. Furthermore, in some cases, it is also desirable for such systems, devices, or methods to automatically deliver the selected medication dose.

[0007] For these and other reasons, there is a need to improve the systems, methods, and devices related to drug dosage guidance. Summary of the Invention

[0008] This document provides exemplary embodiments of systems, apparatus, and methods relating to providing drug dosage guidance and, in some embodiments, drug delivery. According to one aspect, the various embodiments described herein include a dosage guidance system comprising a display device, a sensor control device, and a drug delivery device. The dosage guidance system may include a dosage guidance application (e.g., software) that determines dosage guidance and outputs it to the patient (e.g., recommendations regarding dosage, correction, and titration). Furthermore, according to some embodiments, the dosage guidance system can learn the patient's medication dispensing strategy during a learning period in which key dosage parameters can be estimated. According to some embodiments, once the system is configured with the patient's current medication dispensing strategy, the dosage guidance system can also provide guidance for titration and correction. The dosage guidance system can also provide guidance for different dietary dosage regimens. For example, in some embodiments, the dosage guidance system can provide dosage guidance at the start of a meal, before a meal, or after a meal. The dosage guidance system can also provide guidance for complex meals (e.g., desserts) or for "touch-up" doses to address the problem of high postprandial glucose levels. Exemplary systems and safety features of the dosage guidance system are also described.

[0009] The various implementations provided herein include improved software features or graphical user interfaces for use with analyte monitoring systems that are highly intuitive, user-friendly, and provide rapid access to the user's physiological information. More specifically, these implementations allow users (or HCPs) to quickly determine appropriate drug therapy based on information related to the user's physiological condition, historical dosing patterns, and other factors, eliminating the burden of sifting through large amounts of analyte data. Furthermore, some GUIs and GUI features allow users (and their caregivers) to better understand and improve the user's dosing patterns and subsequent hypoglycemic and hyperglycemic episodes. Similarly, many other implementations provided herein include improved software features for dosing guidance systems that improve upon these features by providing dosing guidance to the user through safe titration strategies that minimize hypoglycemic episodes, by varying dosing guidance based on when the dose is taken relative to the start of a meal (e.g., before, at, or after the start of a meal), by considering realities affecting the dosing strategy, by providing postprandial alerts based on predicted probabilities rather than thresholds, and so on. Other improvements and advantages are also provided. Various structures of these devices are described in detail through the various implementations, which are merely examples.

[0010] Other systems, apparatuses, methods, features, and advantages of the subject matter described herein will become apparent to those skilled in the art upon examination of the following drawings and detailed description. All such additional systems, apparatuses, methods, features, and advantages are to be included in this specification, within the scope of the subject matter described herein, and protected by the appended claims. Features of exemplary embodiments not expressly enumerated in the claims should not be construed as limiting the appended claims. Attached Figure Description

[0011] By studying the accompanying drawings, the details of the subject matter described herein, its structure, and its operation become readily apparent, wherein the same reference numerals denote the same parts. The parts in the drawings are not necessarily drawn to scale, but rather to emphasize the principles of the invention. Furthermore, all descriptions are intended to convey concepts, wherein relative dimensions, shapes, and other detailed properties may be shown schematically rather than formally or precisely.

[0012] Figure 1A and Figure 1B This is a block diagram of an exemplary implementation of a dose guidance system.

[0013] Figure 2A This is a schematic diagram depicting an exemplary embodiment of a sensor control device.

[0014] Figure 2B This is a block diagram depicting an exemplary embodiment of a sensor control device.

[0015] Figure 3A This is a schematic diagram depicting an exemplary embodiment of a drug delivery device.

[0016] Figure 3B This is a block diagram depicting an exemplary embodiment of a drug delivery device.

[0017] Figure 4A This is a schematic diagram depicting an exemplary embodiment of a display device.

[0018] Figure 4B This is a block diagram depicting an exemplary embodiment of a display device.

[0019] Figure 5 This is a block diagram depicting an exemplary embodiment of a user interface device.

[0020] Figure 6 This is an example glucose pattern report.

[0021] Figure 7 This is a flowchart depicting an exemplary implementation of a processing flow involving a part of a learning-based dosage guidance application for estimating patient insulin dosage practices.

[0022] Figure 8A This is a flowchart illustrating an exemplary implementation of a processing flow for operation via a dose guidance application used to evaluate dietary bolus titrations for multi-day injection (MDI) dose therapy.

[0023] Figure 8B This is a flowchart depicting an exemplary implementation of a processing flow operated by a dose-guided application for glucose pattern analysis (GPA).

[0024] Figure 8C This is an exemplary implementation of a graph depicting information used to determine the risk of hypoglycemia and other measures for GPA.

[0025] Figures 8D to 8H This is a flowchart illustrating various exemplary implementations of an algorithm for evaluating dietary bolus titrations used in MDI insulin dose therapy.

[0026] Figures 9A to 9C This is a flowchart depicting an exemplary implementation of a method for determining dosage guidance using physiological dosing algorithms.

[0027] Figure 10A This is a flowchart illustrating an exemplary implementation of a processing flow operated via a dose-guided application for correction factor titration.

[0028] Figures 10B to 10C It is a flowchart that shows the aspect of responding to the output data of the analyte data control user interface device for adjusting the correction factor.

[0029] Figure 11 This is a flowchart illustrating an exemplary implementation of a method for determining dosage guidelines for administration at the start of a meal or before the start of a meal.

[0030] Figure 12A This is a flowchart illustrating an exemplary implementation of a method for determining dosage guidance for administration after the start of a meal.

[0031] Figure 12B This is a flowchart illustrating an exemplary implementation of an alternative method for determining dosage guidance for administration after the start of a meal.

[0032] Figure 12C This is a flowchart illustrating an exemplary implementation of an alternative method for determining dosage guidance for administration after the start of a meal.

[0033] Figure 13 This is a flowchart illustrating an exemplary implementation of a method for determining dosage guidance for compound dietary administration.

[0034] Figure 14 This is a flowchart illustrating an exemplary implementation of a method for determining dosage guidance for corrective dosing.

[0035] Figure 15 This is a flowchart depicting an exemplary implementation of a method for warning subjects.

[0036] Figure 16A This is a flowchart depicting an exemplary implementation of a method for covering dose guidance settings.

[0037] Figure 16B This is a flowchart depicting an exemplary implementation of a method for sensor fault detection.

[0038] Figure 16C This is a flowchart depicting an exemplary implementation of a method for detecting changes in a dispensing strategy.

[0039] Figure 16D This is a flowchart depicting an exemplary implementation of a method for detecting differences in medication dispensing strategies.

[0040] Figure 16E This is a flowchart depicting an exemplary implementation of a method for managing multiple doses.

[0041] Figure 16F This is a flowchart depicting an exemplary implementation of a method for adjusting insulin dosage guidance.

[0042] Figure 16G This is a flowchart depicting an exemplary implementation of a method for managing medication dispensing strategies.

[0043] Figure 17A This is a flowchart depicting an exemplary implementation of a method for recommending insulin dose titration.

[0044] Figure 17B This is a flowchart depicting another exemplary implementation of a method for recommending insulin dose titration.

[0045] Figure 17C This is a flowchart depicting an exemplary implementation of a method for recommending insulin dose titrations associated with overnight periods.

[0046] Figure 17D This is a flowchart depicting an exemplary implementation of a method for recommending insulin dose titrations associated with dinner time.

[0047] Figure 17E This is a flowchart depicting an exemplary implementation of a method for recommending insulin dosage.

[0048] Figure 17FThis is a flowchart depicting an exemplary implementation of a method for determining whether insulin delivery is abnormal.

[0049] Figure 17G This is an exemplary implementation of a graph of an exemplary tracking pair.

[0050] Figure 18A This is a block diagram depicting an exemplary implementation of a site rotation system.

[0051] Figure 18B This is a flowchart of an exemplary embodiment of a method for monitoring drug injection sites. Detailed Implementation

[0052] Before describing the subject matter of this invention in detail, it should be understood that the invention is not limited to the specific embodiments described herein, and therefore variations are possible. It should also be understood that the terminology used herein is for describing particular embodiments only and not for limitation, as the scope of the invention is defined only by the appended claims.

[0053] Typically, embodiments of the present invention include systems, apparatus, and methods related to drug dosage guidance. Dosage guidance can be based on a broad array of information and user-specific categories of information, such as the user's current and previous analyte levels, the user's current and previous diet, the user's current and previous physical activity, the user's current and previous medication history, and other physiological information about the user. According to one aspect of the embodiments, dosage guidance provided by the systems, apparatus, and methods of the present invention can be based not only on the various categories of information but also on the predictive impact of such information on the user's future analyte levels.

[0054] The dosage guidance function can be implemented as a dosage guidance application (DGA), which includes software and / or firmware instructions stored in the memory of a computing device for execution by at least one processor or its processing circuitry. The computing device may be owned by a user or a healthcare professional (HCP), and the user or HCP may interact with the computing device through a user interface. According to some embodiments, the computing device may be a server or trusted computer system accessible via a network, and the dosage guidance software may be presented to the user as an interactive webpage via a browser running on a local display device (with a user interface) communicating with the server or trusted computer system via a network. In this and other embodiments, the dosage guidance software may execute across multiple devices, or may execute partly on the processing circuitry of the local display device and partly on the processing circuitry of the server or trusted computer system. Those skilled in the art will understand that when a DGA is described as performing an action, this action is performed according to instructions stored in computer memory (including instructions hard-coded in read-only memory), which, when executed by at least one processor of at least one computing device, cause the DGA to perform the action. In all cases, instead of being executed via instructions stored in memory, the action can be performed by hardware (e.g., dedicated circuitry) that is hardwired to implement the action.

[0055] Furthermore, as used herein, a system implementing DGA can be referred to as a dose guidance system. A dose guidance system can be configured to provide dose guidance only, or it can be a multifunctional system where dose guidance is only one aspect. For example, in some embodiments, the dose guidance system can also monitor the user's analyte levels. In some embodiments, the dose guidance system can also deliver the drug to the user, such as using an injection or infusion device. In some embodiments, the dose guidance system can monitor both the analyte and the drug.

[0056] The embodiments and other implementations described herein represent improvements in the field of computer-based dosing determination, analyte monitoring, and drug delivery systems. Specific features and potential advantages of the disclosed embodiments are further discussed below.

[0057] Before describing the dosage guidance implementation method in detail, it is necessary to first describe an example of a dosage guidance system on which a dosage guidance application can be implemented or through the dosage guidance system.

[0058] Exemplary implementation of a dose guidance system

[0059] Figure 1AThis is a block diagram illustrating an exemplary embodiment of a dose guidance system 100. In this embodiment, the dose guidance system 100 is capable of providing dose guidance, monitoring one or more analytes, and delivering one or more drugs. This multifunctional example serves to illustrate the high degree of interconnectivity and performance available through system 100. However, in the embodiments described herein, analyte monitoring components, drug delivery components, or both may be omitted if desired.

[0060] Here, system 100 includes: a sensor control device (SCD) 102 configured to collect analyte level information from a user; a drug delivery device (MDD) 152 configured to deliver drug to a user; and a display device 120 configured to present information to a user and receive input or information from a user. The structure and function of each device will be described in detail herein.

[0061] System 100 is configured for highly interconnected and highly flexible communication between devices. Each of the three devices 102, 120 and 152 can communicate directly with each other (without intermediate electronics) or indirectly with each other (e.g., via cloud network 190, or via another device and then via network 190). Figure 1A The bidirectional communication capabilities between devices and between a device and network 190 are illustrated by double-sided arrows. However, those skilled in the art will understand that any one of the one or more devices (e.g., SCD) is capable of unidirectional communication, such as broadcast, multicast, or advertising communication. In each case, whether bidirectional or unidirectional, communication can be wired or wireless. The protocols governing communication on each path can be the same or different, and can be proprietary or standardized. For example, wireless communication between devices 102, 120, and 152 can be performed according to Bluetooth (including Bluetooth Low Energy) standards, Near Field Communication (NFC) standards, Wi-Fi (802.1lx) standards, mobile phone standards, or other standards. All communication on various paths can be encrypted, and Figure 1A Each device can be configured to encrypt and decrypt those communications being sent and received. In each case, Figure 1A The communication path can be direct (e.g., Bluetooth or NFC) or indirect (e.g., Wi-Fi, mobile phone, or other Internet protocols). Implementations of system 100 do not require cross-platform communication. Figure 1A The ability to communicate along all the paths shown.

[0062] In addition, although Figure 1AA single display device 120, a single SCD 102, and a single MDD 152 are shown; however, those skilled in the art will understand that system 100 may include any of the aforementioned devices. By way of example only, system 100 may include a single SCD 102 that communicates with multiple (e.g., two, three, four, etc.) display devices 120 and / or multiple MDDs 152. Alternatively, system 100 may include multiple SCDs 102 that communicate with a single display device 120 and / or a single MDD 152. Furthermore, each of the multiple devices may be of the same or different device types. For example, system 100 may include multiple display devices 120, including smartphones, handheld receivers, and / or smartwatches, each of which may communicate with SCD 102 and / or MDD 152, as well as with each other.

[0063] Analytical data can be transmitted between each device within system 100 in an autonomous manner (e.g., automatically sent according to a schedule) or in response to a request for analyte data (e.g., a request for analyte data is sent from the first device to the second device, and then the analyte data is sent from the second device to the first device). Other technologies for transmitting data can also be used to accommodate more complex systems, such as cloud network 190.

[0064] Figure 1B This is a block diagram depicting another exemplary embodiment of the dosage guidance system 100. Here, system 100 includes an SCD 102, an MDD 152, a first display device 120-1, a second display device 120-2, a local computer system 170, and a trusted computer system 180 accessible by a cloud network 190. The SCD 102 and MDD 152 are capable of communicating with each other and with the display device 120-1, which can act as a communication hub to collect information from the SCD 102 and MDD 152, process and display that information as needed, and transmit some or all of the information to the cloud network 190 and / or the computer system 170. Conversely, the display device 120-1 can receive information from the cloud network 190 and / or the computer system 170 and transmit some or all of the received information to the SCD 102, MDD 152, or both. The computer system 170 can be a personal computer, server terminal, laptop computer, tablet, or other suitable data processing device. Computer system 170 may include or present software for data management and analysis, and for communicating with components in system 100. Computer system 170 can be used by users or medical professionals to display and / or analyze analyte data measured by SCD 102. Furthermore, although... Figure 1BA single SCD 102, a single MDD 152, and two display devices 120-1 and 120-2 are shown. Those skilled in the art will understand that system 100 may include any one of the plurality of devices described above, wherein each plurality of devices may include devices of the same or different types.

[0065] Still refer to Figure 1B According to some implementations, the trusted computer system 180 can be physically or virtually connected via a secure link within the scope of the manufacturer or distributor of the components of system 100, and can be used to perform authentication of devices of system 100 (e.g., devices 102, 120-n, 152), securely store user data, and / or act as a server serving data analysis programs (e.g., accessible via a web browser) for analyzing users' measurement analyte data and drug history. The trusted computer system 180 can also act as a data hub for routing and exchanging data among all devices communicating with system 180 via cloud network 190. In other words, all devices of system 100 capable of communicating with cloud network 190 (e.g., directly or indirectly via another device using an Internet connection) can also communicate with all other devices of system 100 capable of communicating directly or indirectly with cloud network 190.

[0066] Display device 120-2 is depicted as communicating with cloud network 190. In this example, device 120-2 may be owned by another user who is authorized to access the analyte and medication data of the person wearing SCD 102. For example, as an example, the person owning display device 120-2 could be the parent of a child wearing SCD 102, or, as another example, the caregiver of an elderly patient wearing SCD 102. System 100 may be configured to transmit analyte and medication data about the wearer to another user authorized to access that data via cloud network 190 (e.g., via trusted computer system 180).

[0067] Exemplary Implementation of Analyte Monitoring Device

[0068] The analyte monitoring function of the dose guidance system 100 can be implemented by including one or more devices capable of collecting, processing, and displaying user analyte data. Exemplary embodiments of such devices and methods of using them are described in International Publication No. WO2018 / 152241 and U.S. Patent Publication No. 2011 / 0213225, both of which are incorporated herein by reference in their entirety for all purposes.

[0069] Analyte monitoring can be performed in a variety of different ways. A “continuous analyte monitoring” device (e.g., a “continuous glucose monitoring” device) can transmit data from a sensor control device to a display device automatically, continuously, or repeatedly, with or without prompting, such as according to a schedule. Another example is a “flash analyte monitoring” device (e.g., a “flash glucose monitoring” device, or simply a “flash device”), which can transmit data from a sensor control device in response to a user’s request for data via a display device (e.g., a scan), such as using near field communication (NFC) or radio frequency identification (RFID) protocols.

[0070] An analyte monitoring device that utilizes a sensor configured to be partially or wholly placed within a user's body is called an in vivo analyte monitoring device. For example, an in vivo sensor can be placed within a user's body such that at least a portion of the sensor is in contact with a bodily fluid (e.g., interstitial fluid (ISF), such as dermal fluid in the dermis or subcutaneous fluid beneath the dermis, blood, or others), and the concentration of the analyte in that fluid can be measured. In vivo sensors can utilize various types of sensing technologies (e.g., chemical, electrochemical, or optical). Some systems utilizing in vivo analyte sensors can also operate without the need for manual finger-prick calibration.

[0071] "In vitro" devices are those that bring a sensor into contact with a biological sample outside the body (or more precisely, "ex vivo"). These devices typically include a port for receiving an analyte test strip carrying the user's bodily fluids, which can be analyzed to determine the user's blood glucose levels. Other ex vivo devices have been proposed that attempt to non-invasively measure the user's internal analyte levels, such as using optical techniques capable of measuring internal bodily analyte levels without mechanically penetrating the user's body or skin. In vivo and ex vivo devices typically include in vitro capabilities (e.g., an in vivo display device that also includes the test strip port).

[0072] While the detection and measurement of concentrations of other analytes are within the scope of this invention, the subject matter of the invention will be described with reference to sensors capable of measuring glucose concentrations. These other analytes may include, for example, ketones, lactate, oxygen, hemoglobin AIC, acetylcholine, amylase, bilirubin, cholesterol, human chorionic gonadotropin (hCG), creatine kinase (e.g., CK-MB), creatine, DNA, fructosamine, glutamine, growth hormone, hormones, peroxides, prostate-specific antigen, prothrombin, RNA, thyroid-stimulating hormone, troponin, etc. The concentrations of drugs may also be monitored, such as antibiotics (e.g., gentamicin, vancomycin, etc.), digoxin, drugs of abuse, theophylline, and warfarin. The sensor may be configured to measure two or more different analytes at the same or different times. In some embodiments, a sensor control device may be coupled to two or more sensors, one of which is configured to measure a first analyte (e.g., glucose), while another or more sensors are configured to measure one or more different analytes (e.g., any analyte described herein). In other embodiments, a user may wear two or more sensor controllers, each capable of measuring a different analyte.

[0073] The embodiments described herein can be used with all types of in vivo, in vitro, and ex vivo devices capable of monitoring the aforementioned analytes and other analytes.

[0074] In various embodiments, sensor operation can be controlled by SCD 102. The sensor can be mechanically and communicatively coupled to SCD 102, or it can be communicatively coupled to SCD 102 using only wireless communication technology. SCD 102 may include electronics and a power supply that implements and controls the analyte sensing performed by the sensor. In some embodiments, the sensor or SCD 102 may be self-powered, thus eliminating the need for a battery. SCD 102 may also include communication circuitry for communicating with another device (e.g., a display device) local to or non-local to the user's body. SCD 102 may be located on the user's body (e.g., attached to or otherwise placed on the user's skin, or carried in the user's clothing, etc.). SCD 102 may also be implanted in the user's body along with the sensor. The function of SCD 102 can be divided into a first component implanted in the body (e.g., a component controlling the sensor) and a second component located on or outside the body (e.g., a relay component that communicates with the first component and external devices such as a computer or smartphone). In other embodiments, SCD 102 can be external to the body and configured to non-invasively measure the user's analyte levels. Depending on the actual implementation or method, the sensor control device may also be referred to as a "sensor control unit," "on-body electronics" device or unit, "on-body" device or unit, "in-body electronics" device or unit, "in-body" device or unit, or "sensor data communication" device or unit.

[0075] In some embodiments, the SCD 102 may include a user interface (e.g., a touchscreen) and be capable of processing analyte data and displaying the calculated analyte level to the user. In such cases, the dose guidance implementation described herein may be implemented entirely or partially directly by the SCD 102. In various embodiments, the physical morphology of the SCD 102 is minimized (e.g., to minimize its appearance on the user's body), or the sensor control may be inaccessible to the user (e.g., if fully implanted), or other factors may lead the user to desire a display device that can be used to read analyte levels and interface with the sensor control.

[0076] Figure 2A This is a side view of an exemplary embodiment of SCD 102. SCD 102 may include sensor electronics ( Figure 2BThe housing or mounting 103 may be electrically coupled to the analyte sensor 101, which is configured herein as an electrochemical sensor. According to some embodiments, the sensor 101 may be configured to be partially located within a user's body (e.g., through the outermost surface of the skin), within which the sensor 101 may have fluid contact with the user's bodily fluids and be used, together with sensor electronics, to measure the user's analyte-related data. An attachment structure 105, such as an adhesive patch, may be used to secure the housing 103 to the user's skin. The sensor 101 may extend through the attachment structure 105 and protrude from the housing 103. Those skilled in the art will understand that other forms of attachment to the body and / or housing 103, besides adhesives or alternatives to adhesives, may be used and are fully within the scope of this invention.

[0077] The SCD 102 can be applied to the body in any desired manner. For example, an insertion device (not shown), sometimes referred to as an applicator, can be used to position all or part of the analyte sensor 101 across the outer surface of the user's skin and into contact with the user's bodily fluids. In doing so, the insertion device can also position the SCD 102 on the skin. In other embodiments, the insertion device may first position the sensor 101, and then accompanying electronics (e.g., wireless transmission circuitry and / or data processing circuitry, etc.) may be manually or by means of mechanical means subsequently coupled to the sensor 101 (e.g., inserted into an mounting). Examples of insertion devices are described in U.S. Patent Publications 2008 / 0009692, 2011 / 0319729, 2015 / 0018639, 2015 / 0025345, 2015 / 0173661, and 2018 / 0235520, all of which are incorporated herein by reference in their entirety for all purposes.

[0078] Figure 2B This is a block diagram depicting an exemplary embodiment of an SCD 102 having an analyte sensor 101 and sensor electronics 104. The sensor electronics 104 can be implemented in one or more semiconductor chips, such as application-specific integrated circuits (ASICs), processors or controllers, memories, programmable gate arrays, etc. Figure 1BIn this embodiment, the sensor electronics 104 includes advanced functional units including: an analog front-end (AFE) 110 configured to interface with the sensor 101 in an analog manner and convert analog signals to digital form and / or from digital form to analog signals (e.g., using an A / D converter); a power supply 111 configured to power components of the SCD 102; processing circuitry 112; a memory 114; timing circuitry 115 (e.g., an oscillator and a phase-locked loop for providing a clock or other timing to components of the SCD 102); and communication circuitry 116 configured to communicate with one or more devices external to the SCD 102 (e.g., display device 120 and / or MDD 152) in a wired and / or wireless manner.

[0079] SCD 102 can be implemented in a highly interconnected manner, wherein power supply 111 and Figure 2B Each component shown is coupled to the others, and those components that communicate or receive data, information or commands (e.g., AFE 110, processing circuitry 112, memory 114, timing circuitry 115, and communication circuitry 116) can be communicatively coupled to each of the other such components via, for example, one or more communication connections or buses 118.

[0080] Processing circuitry 112 may include one or more processors, microprocessors, controllers, and / or microcontrollers, each of which may be discrete chips or distributed among multiple different chips (and portions thereof). Processing circuitry 112 may include on-board memory. Processing circuitry 112 may interface with communication circuitry 116 and perform analog-to-digital conversion, encoding and decoding, digital signal processing, and other functions that help convert data signals into formats suitable for wireless or wired transmission (e.g., in-phase and quadrature). Processing circuitry 112 may also interface with communication circuitry 116 to perform the reverse functions required to receive wireless transmissions and convert them into digital data or information.

[0081] Processing circuit 112 can execute instructions stored in memory 114. These instructions enable processing circuit 112 to process raw analyte data (or preprocessed analyte data) and achieve a final calculated analyte level. In some embodiments, when executing instructions stored in memory 114, these instructions can cause processing circuit 112 to process raw analyte data to determine one or more of the following: the calculated analyte level, the average calculated analyte level over a predetermined time window, the rate of change of the calculated analyte level over a predetermined time window, and / or whether the calculated analyte metric exceeds a predetermined threshold condition. These instructions can also cause processing circuit 112 to read and act on received transmissions to adjust the timing of timing circuit 115, to process data or information received from other devices (e.g., receiving correction information, encryption or authentication information from display device 120, etc.), to perform tasks such as establishing and maintaining communication with display device 120, interpreting voice commands from the user, and enabling communication circuit 116 to transmit. In an implementation of SCD 102 that includes a user interface, instructions can cause the processing circuitry 112 to control the user interface, read user input from the user interface, display information on the user interface, format data for display, etc. The functions encoded in the instructions described herein can be implemented by SCD 102 using hardware or firmware design that does not depend on executing the stored software instructions to perform these functions.

[0082] Memory 114 may be shared by one or more different functional units present within SCD 102, or may be distributed among two or more of them (e.g., as independent memories present in different chips). Memory 114 may also be its own separate chip. Memory 114 is non-transitory and may be volatile (e.g., RAM) and / or non-volatile memory (e.g., ROM, flash memory, F-RAM, etc.).

[0083] The communication circuit 116 may be implemented as one or more components (e.g., transmitter, receiver, transceiver, passive circuit, encoder, decoder, and / or other communication circuitry) that perform functions for communication on a corresponding communication path or link. The communication circuit 116 may include or be coupled to one or more antennas for wireless communication.

[0084] The power supply 111 may include one or more batteries, which may be rechargeable or disposable. It may also include power management circuitry to regulate battery charging and monitor the use of the power supply 111, increase power, perform DC-DC conversion, etc.

[0085] In addition, temperature readings or measurements on the skin or from sensors can be collected by an optional temperature sensor (not shown). These readings or measurements can be transmitted from the SCD 102 (individually or as aggregated measurements over time) to another device (e.g., display device 120). However, temperature readings or measurements can be used in conjunction with software routines executed by the SCD 102 or the display to correct or compensate for analyte measurements output to the user, instead of, additionally, actually outputting temperature measurements to the user.

[0086] Exemplary Implementation of Drug Delivery Device

[0087] The drug delivery function of the dose guidance system 100 can be implemented by including one or more drug delivery devices (MDDs) 152. The MDD 152 can be any device configured to deliver a specific dose of drug. The MDD 152 may also include a device for transmitting data about the dose to the DGA, such as a pen cap, even if the device itself may not deliver the drug. The MDD 152 can be configured as a portable injection device (PID) that can deliver a single dose per injection, such as a bolus. The PID can be a basic, manually operated syringe where the drug is either pre-loaded into the syringe or must be drawn from a container into the syringe before injection. However, in most embodiments, the PID includes electronics for interfacing with the user and performing drug delivery. PIDs are often referred to as drug pens, although a pen-like appearance is not required. PIDs with user interface electronics are often referred to as smart pens. PIDs can be used to deliver a single dose and then discard, or they can be durable and reusable for delivering multiple doses over a day, week, or month. Users practicing multiple daily injection (MDI) treatment regimens often rely on PIDs.

[0088] MDD may also include a pump and an infusion set. The infusion set includes a tubular cannula that is at least partially located within the recipient's body. The tubular cannula is in fluid communication with the pump, which delivers the medication into the recipient's body in small, repetitive increments over time through the cannula. The infusion set can be applied to the recipient's body using an infusion set applicator, and the infusion set typically remains implanted for 2–3 days or longer. The pump unit includes a user interface and electronics for controlling the slow infusion of the medication. Both the PID (Infusion Device) and the pump can store the medication in a drug reservoir.

[0089] The MDD 152 can be part of a closed-loop system (e.g., an artificial pancreas system that does not require user intervention), a semi-closed-loop system (e.g., an insulin loop system that requires minimal user intervention, such as for confirming dose changes), or an open-loop system. For example, the SCD 102 can repeatedly and automatically monitor analyte levels in diabetic patients, and the dosage guidance implementation described herein can use this information to automatically calculate or otherwise determine an appropriate drug dose for controlling analyte levels in diabetic patients, and subsequently deliver that dose to the diabetic patient's body. This calculation can be performed in the MDD 152 or any other device in system 100, and the resulting determined dose can then be transmitted to the MDD 152.

[0090] In various embodiments, the dosage guidance provided herein will be for the type of insulin (e.g., rapid-acting (RA), short-acting insulin, intermediate-acting insulin (e.g., NPH insulin), long-acting (LA), ultra-long-acting insulin, and mixed insulin), and will be the same drug delivered by MDD 152. Types of insulin include human insulin and synthetic insulin analogs. Insulin may also include premixed formulations. However, the dosage guidance embodiments presented herein and the drug delivery capabilities of MDD 152 can be applied to other non-insulin drugs. Such drugs may include, but are not limited to, exenatide, exenatide extended-release, liraglutide, lixinatide, sumaglutide, pramlintide, metformin, SLGT1-i inhibitors, SLGT2-i inhibitors, and DPP4 inhibitors. Dosage guidance embodiments may also include combination therapy. Combination therapy may include, but is not limited to, insulin and glucagon-like peptide-1 receptor agonists (GLP-1RAs), insulin and pramlintide.

[0091] For ease of description of the dosage guidance implementation described herein, the MDD 152 is generally described as a PID, specifically in the form of a smart pen. However, those skilled in the art will readily understand that the MDD 152 can also be configured as a pen cap, a pump, or any other type of drug delivery device.

[0092] Figure 3A This is a schematic diagram depicting an exemplary embodiment of an MDD 152 configured as a PID, specifically a smart pen. The MDD 152 may include a housing 154 for electronics, an injection motor, and a drug reservoir (see [link to documentation]). Figure 3BThe drug can be delivered from the drug reservoir through the needle 156. The housing 154 may include a removable or detachable cap or cover 157, which, when attached, can cover 156 when not in use and then be removed for injection. The MDD 152 may also include a user interface 158, which may be implemented as a single component (e.g., a touchscreen for outputting information to and receiving input from the user) or as multiple components (e.g., a touchscreen or display combined with one or more buttons, switches, etc.). The MDD 152 may also include an actuator 159, which can be moved, pressed, touched, or otherwise triggered to initiate the delivery of the drug from the internal reservoir through the needle 156 into the recipient's body. According to some embodiments, the cap 157 and actuator 159 may also include one or more safety mechanisms to prevent removal and / or actuation, thereby mitigating the risk of harmful drug injection. Details of these and other security agencies are described in U.S. Patent Publication No. 2019 / 0343385, which is incorporated herein by reference in its entirety for all purposes.

[0093] Figure 3B This is a block diagram depicting an exemplary embodiment of an MDD 152 having electronics 160 connected to a power source 161 and an electric injection motor 162, which in turn is connected to the power source 161 and a drug reservoir 163. A needle 156 is shown in fluid communication with the reservoir 163, and a valve (not shown) may be present between the reservoir 163 and the needle 156. The reservoir 163 may be permanent or removable and may be replaced with another reservoir containing the same or different drugs. The electronics 160 may be implemented on one or more semiconductor chips (e.g., application-specific integrated circuits (ASICs), processors or controllers, memories, programmable gate arrays, etc.). Figure 3B In some implementations, electronic device 160 may include advanced functional units including processing circuitry 164, memory 165, communication circuitry 166 configured to communicate with one or more devices (such as display device 120) outside MDD 152 via wired and / or wireless means, and user interface electronics 168.

[0094] The MDD 152 can be implemented in a highly interconnected manner, where power supply 161 and Figure 3B Each component shown is coupled to the others, and those components that communicate or receive data, information or commands (e.g., processing circuitry 164, memory 165, and communication circuitry 166) can be communicatively coupled to each of the other such components via, for example, one or more communication connections or buses 169.

[0095] Processing circuitry 164 may include one or more processors, microprocessors, controllers, and / or microcontrollers, each processor being a discrete chip or distributed among multiple different chips (and portions thereof). Processing circuitry 164 may include on-board memory. Processing circuitry 164 may interface with communication circuitry 166 and perform analog-to-digital conversion, encoding and decoding, digital signal processing, and other functions that help convert data signals into formats suitable for wireless or wired transmission (e.g., in-phase and quadrature). Processing circuitry 164 may also interface with communication circuitry 166 to perform the reverse functions required to receive wireless transmissions and convert them into digital data or information.

[0096] Processing circuitry 164 can execute software instructions stored in memory 165. These instructions can cause processing circuitry 164 to receive selections or provision of a specified dose from a user (e.g., input via user interface 158 or received from another device), process commands to deliver a specified dose (such as signals from actuator 159), and control motor 162 to induce the delivery of the specified dose. These instructions can also cause processing circuitry 164 to read and act on received transmissions to process data or information received from other devices (e.g., calibration information, encryption, or authentication information received from display device 120, etc.) to perform tasks such as establishing and maintaining communication with display device 120, interpreting voice commands from the user, and enabling communication circuitry 166 to transmit data. In embodiments where MDD 152 includes user interface 158, the instructions can then cause processing circuitry 164 to control the user interface, read user input from the user interface (e.g., input of a drug dose for administration or confirmation of a recommended drug dose), display information on the user interface, format data for display, etc. The functions coded in the instructions described herein can be implemented by MDD 152 using hardware or firmware design that does not depend on executing the stored software instructions to accomplish these functions.

[0097] Memory 165 may be shared by one or more different functional units present within MDD 152, or may be distributed among two or more of them (e.g., as separate memories present within different chips). Memory 165 may also be its own separate chip. Memory 165 is non-transitory and may be volatile (e.g., RAM) and / or non-volatile memory (e.g., ROM, flash memory, F-RAM, etc.).

[0098] Communication circuitry 166 may be implemented as one or more components (e.g., transmitter, receiver, transceiver, passive circuitry, encoder, decoder, and / or other communication circuitry) that perform functions for communication on a corresponding communication path or link. Communication circuitry 166 may include or be coupled to one or more antennas for wireless communication. Details of exemplary antennas can be found in '385 Publication, which is incorporated herein by reference in its entirety for all purposes.

[0099] Power source 161 may include one or more batteries, which may be rechargeable or disposable. It may also include power management circuitry to regulate battery charging and monitor the use of power source 161, increase power, perform DC-DC conversion, etc.

[0100] The MDD 152 may also include an integrated or connectable in vitro glucose meter, including an in vitro test strip port (not shown) for receiving in vitro glucose test strips for in vitro blood glucose measurement.

[0101] Exemplary embodiments of display devices

[0102] Display device 120 can be configured to display information related to system 100 to a user and to accept or receive input from a user also related to system 100. Display device 120 can display recently measured analyte levels to the user in any number of formats. The display device can display the user's historical analyte levels as well as other metrics describing the user's analyte information (e.g., time within range, dynamic glucose profile (AGP), hypoglycemia risk level, etc.). Display device 120 can display medication delivery information, such as historical dose information and dosing time and date. Display device 120 can display alarms, warnings, or other notifications related to analyte levels and / or medication delivery.

[0103] Display device 120 may be dedicated to system 100 (e.g., an electronic device designed and manufactured primarily for interfacing with analyte sensors and / or drug delivery devices), or it may be a device of a multi-functional general-purpose computing device, such as a handheld or portable mobile communication device (e.g., a smartphone or tablet), or a laptop computer, personal computer, or other computing device. Display device 120 may be configured as a mobile smart wearable electronic component, such as smart glasses, or a smartwatch or wristband. For example, display devices and variations thereof may be referred to as “reader devices,” “readers,” “handheld electronic devices” (or handheld devices), “portable data processing” devices or units, “information receivers,” “receiver” devices or units (or simply receivers), “relay” devices or units, or “remote” devices or units.

[0104] Figure 4AThis is a schematic diagram depicting an exemplary embodiment of the display device 120. Here, the display device 120 includes a user interface 121 and therein holding display device electronics 130 (… Figure 4B The user interface 121 can be implemented as a single component (e.g., a touchscreen capable of input and output) or multiple components (e.g., a display and one or more devices configured to receive user input). In this embodiment, the user interface 121 includes a touchscreen display 122 (configured to display information and graphics and accept user input via touch) and input buttons 123, both of which are connected to the housing 124.

[0105] Display device 120 may have software stored thereon (e.g., provided by the manufacturer or downloaded by the user, in the form of one or more "applications" or other software packages), which interfaces with SCD 102, MDD 152, and / or the user interface. Additionally, or alternatively, web pages displayed on a browser or other internet interface software that can execute on display device 120 may affect the user interface.

[0106] Figure 4B This is a block diagram of an exemplary embodiment of a display device 120 having display device electronics 130. Here, the display device 120 includes a user interface 121, which includes a display 122 and input components 123 (e.g., buttons, actuators, touch switches, capacitive switches, pressure-sensitive switches, micro-wheels, microphones, speakers, etc.), processing circuitry 131, memory 125, communication circuitry 126 (configured for communication with one or more other devices external to / from the display device 120), power supply 127, and timing circuitry 128 (e.g., an oscillator and phase-locked loop for providing a clock or other timing to components of the SCD 102). Each of the aforementioned components can be implemented as one or more different devices, or can be combined into a multifunctional device (e.g., processing circuitry 131, memory 125, and communication circuitry 126 integrated on a single semiconductor chip). The display device 120 can be implemented in a highly interconnected manner, wherein power supply 127 and... Figure 4B Each component shown is coupled, and those components that communicate or receive data, information, or commands (e.g., user interface 121, processing circuitry 131, memory 125, communication circuitry 126, and timing circuitry 128) can be communicatively coupled to each of the other such components via, for example, one or more communication connections or buses 129. Figure 4B It is a simplified representation of typical hardware and functions located within a display device, and those skilled in the art will readily recognize that it may also include other hardware and functions (e.g., codecs, drivers, glue logic).

[0107] Processing circuitry 131 may include one or more processors, microprocessors, controllers, and / or microcontrollers, each processor being a discrete chip or distributed among multiple different chips (and portions thereof). Processing circuitry 131 may include on-board memory. Processing circuitry 131 may interface with communication circuitry 126 and perform analog-to-digital conversion, encoding and decoding, digital signal processing, and other functions that help convert data signals into formats suitable for wireless or wired transmission (e.g., in-phase and quadrature). Processing circuitry 131 may also interface with communication circuitry 126 to perform the reverse functions required to receive wireless transmissions and convert them into digital data or information.

[0108] Processing circuitry 131 can execute software instructions stored in memory 125. These instructions enable processing circuitry 131 to process raw analyte data (or pre-processed analyte data) to achieve appropriate analyte levels suitable for display to the user. These instructions can cause processing circuitry 131 to read, process, and / or store dosage instructions from the user, and because the dosage instructions are transmitted to MDD 152. These instructions can cause processing circuitry 131 to execute user interface software suitable for presenting a graphical user interface screen to the user, in order to configure system parameters (e.g., alarm thresholds, notification settings, display preferences, etc.), present current and historical analyte level information to the user, present current and historical drug delivery information to the user, collect other non-analyte information from the user (e.g., information about ingested meals, performed activities, taken medications, etc.), and present notifications and alarms to the user. These instructions can also cause the processing circuit 131 to transmit to the communication circuit 126, enabling the processing circuit 131 to read and act on the received transmissions, read input from the user interface 121 (e.g., inputting the drug dosage to be administered or confirming the recommended drug dosage), display data or information on the user interface 121, adjust the timing of the timing circuit 128, process data or information received from other devices (e.g., analyte data, calibration information, encryption or authentication information received from the SCD 102, etc.), perform tasks to establish and maintain communication with the SCD 102, interpret voice commands from the user, etc. The functions coded in the instructions described herein can be implemented by the display device 120 using hardware or firmware design that does not rely on executing the stored software instructions to perform these functions.

[0109] The memory 125 may be shared by one or more different functional units present within the display device 120, or may be distributed among two or more of them (e.g., as independent memories present in different chips). The memory 125 may also be its own separate chip. The memory 125 is non-transitory and may be volatile (e.g., RAM) and / or non-volatile memory (e.g., ROM, flash memory, F-RAM, etc.).

[0110] The communication circuit 126 may be implemented as one or more components (e.g., transmitter, receiver, transceiver, passive circuitry, encoder, decoder, and / or other communication circuitry) that perform functions for communication on a corresponding communication path or link. The communication circuit 126 may include or be coupled to one or more antennas for wireless communication.

[0111] The power supply 127 may include one or more batteries, which may be rechargeable or disposable. It may also include power management circuitry to regulate battery charging and monitor the use of the power supply 127, increase power, perform DC conversion, etc.

[0112] The display device 120 may also include one or more data communication ports (not shown) for wired data communication with external devices such as computer system 170, SCD 102, or MDD 152. The display device 120 may also include an integrated or attachable in vitro glucose meter, which includes an in vitro test strip port (not shown) for receiving in vitro glucose test strips for performing in vitro blood glucose measurements.

[0113] Display device 120 can display measured analyte data received from SCD 102 and can also be configured to output alarms, warnings, glucose levels, etc., which can be visual, audible, tactile, or any combination thereof. In some embodiments, SCD 102 and / or MDD 152 can also be configured to output alarms or warnings in a visible, audible, tactile, or combined form. Further details and other display embodiments can be found, for example, in U.S. Patent Publication No. 2011 / 0193704, the entire contents of which are incorporated herein by reference for all purposes.

[0114] Exemplary implementations related to dosage guidance

[0115] The following exemplary embodiments relate to dose guidance functionality provided by dose guidance system 100. In various embodiments, the dose guidance functionality will be implemented as a set of software instructions stored and / or executed on one or more electronic devices. This dose guidance functionality will be referred to herein as a dose guidance application (DGA). In some embodiments, the DGA is stored, executed, and presented to the user on the same single electronic device. In other embodiments, the DGA may be stored and executed on one device, but presented to the user on different electronic devices. For example, the DGA may be stored and executed on trusted computer system 180, but presented to the user via a webpage displayed by an internet browser running on display device 120.

[0116] Therefore, there are many different implementations, which are related to the number and type of electronic devices used to store, execute, and present the DGA to the user. Regarding presentation to the user, the device configured to perform this capability will be referred to herein as User Interface Device (UID) 200. Figure 5 This is a block diagram depicting an exemplary embodiment of UID 200. In this embodiment, UID 200 includes a housing 201 coupled to a user interface 202. The user interface 202 is capable of outputting information to and receiving input or information from a user. In some embodiments, the user interface 202 is a touchscreen. As shown here, the user interface 202 includes a display 204, which may be a touchscreen, and input components 206 (e.g., buttons, actuators, touch-sensitive switches, capacitive switches, pressure-sensitive switches, microwheels, microphones, touchpads, soft keys, keyboards, etc.).

[0117] Many of the devices described herein can be implemented as UID 200. For example, in various embodiments, display device 120 will be used as UID 200. In some embodiments, MDD 152 can be implemented as UID 200. In embodiments where SCD 102 includes a user interface, SCD 102 can be implemented as UID 200. Computer system 170 can also be implemented as UID 200.

[0118] MDI detection dose Strategy

[0119] Turning now to the aspect of DGA, more specifically, DGA can utilize knowledge of a patient's medication strategy and analyte levels to provide precise dosing guidance. This document describes exemplary implementations for automating the detection of patient medication strategies, which can simplify and accelerate DGA setup. Medication strategies can be detected based on many characteristics of the monitored drug (e.g., insulin) dose. For example, the implementation can identify a dose as a basal dose or bolus based on an MDD 152 used for administration. Some patients may have more than one MDD 152. For example, a patient may have one MDD for administering long-acting insulin (e.g., a basal dose) and another for administering rapid-acting insulin (e.g., a dietary dose). The count (e.g., number of doses) and timing of each basal dose administration can also be used to categorize basal strategies as "single" or "split" basal medication strategies. For example, in a "split" basal medication strategy, a daily 20U basal dose can be split into two 10U doses, one administered at bedtime and the other upon waking.

[0120] When administered in close, continuous bolus administration, the system can attempt to distinguish between the original meal dose, an increase in the original meal dose, or a corrective dose for inter-meal hyperglycemia. When the DGA rapidly detects a small dose followed by a larger dose, both occurring immediately after the start of a meal, the DGA can combine these doses as a single meal dose, even if the first dose may have been the initiating dose not yet injected into the patient. Subsequently, if a dose occurs long after a known meal and / or is labeled as a meal dose (group), the DGA can label the later dose as a corrective dose for post-meal hyperglycemia, or as an increase in the previous meal dose to address additional food consumption. When a meal event is identified, based on the meal detector algorithm or user-input meal event, the DGA can use the magnitude of the previous dose event and its timing relative to the currently detected meal to help characterize whether the previous dose was the first of several meal doses, rather than a correction for inter-meal hyperglycemia. It is assumed that the corrective dose is less than the meal dose. Furthermore, if the time elapsed between the previous dose and the current dietary event is sufficiently long, it is reasonable to assume that the two events are unrelated to the same glucose shift event, eliminating the possibility that the previous dose was the first of several doses for a given diet. Therefore, if the earlier dose is sufficiently smaller than dietary doses recorded within that window in previous days and sufficiently distant from the current diet, the earlier dose can be classified as a corrective dose event.

[0121] DGA can be configured to use real-time dietary detection algorithms and dose timing to identify additional doses in addition to the basal dose as breakfast, lunch, and / or dinner bolus doses, and / or correction doses. DGA can also be configured to use the number of daily bolus doses to identify dosing strategies as basal only, basal plus 1, basal plus 2, etc.

[0122] These different cases and aspects of DGA are discussed in more detail elsewhere in the manual.

[0123] On-Boarding

[0124] To enhance the safety profile of DGA, HCP can approve learned insulin dosage parameters and subsequent titrations calculated by DGA. DGA implementations include various methods of interaction between HCP and DGA, providing HCP with relevant evidence for approval of recommended dose learning and titration in a concise and informative manner, thus improving workflow.

[0125] For diabetic patients already on an insulin dosing regimen, the HCP can utilize existing reports to understand the patient's glucose patterns and identify users who can benefit from dosing guidance. The implementation of DGA provides a learning period (e.g., when using DGS 100) that allows for the categorization of a patient's dosing strategies and trends. If combined insulin and glucose data further confirm the user as a good candidate for DGA—for example, a candidate for DGA capable of learning their specific dosing strategy—the insulin dosing parameters learned during the learning period can be used as initial conditions for dosing guidance, which can be titrated as needed by the DGA. HCP notification methods for initializing and titrating dosing parameters for DGA can also be provided. This approach helps both the HCP and the user by simplifying DGA activation and titration, while also helping to ensure that the DGA is used only by those it has indicated. When the DGA fails to learn a patient's dosing parameters, the DGA can indicate a dosing inconsistency, which the HCP can use to resolve the dosing inconsistency.

[0126] The first step in identifying potential DGA users can involve an introductory analysis of their glycemic control through the patient's glucose concentration profile. To facilitate access to as many DGA users as possible, this process does not consider the user's current glucose monitoring methods.

[0127] For diabetic patients currently using SCD 102, glucose pattern reports may be available, including key metrics, glucose concentration curves (e.g., dynamic glucose profile (AGP)), patterns identified for different times of day, and titration and lifestyle recommendations to improve situations where glucose levels are consistently outside the target range. Pattern identification can be achieved using the GPA algorithm, as described in more detail elsewhere. Figure 6As shown, an exemplary glucose pattern report 250 is presented. Those skilled in the art will understand that the glucose pattern report 250 may be a graphical user interface output to a display of a computing device. The glucose pattern report 250 may include a time-in-range (TIR) ​​display 252 that shows the percentage of time a patient's glucose level is below a target range (e.g., below 70 mg / dL), within the target range (e.g., 70–180 mg / dL), and above the target range (e.g., above 180 mg / dL). The TIR display 252 may also report the amount of time a patient's glucose level is below a low threshold (e.g., below 54 mg / dL) (which is below the low boundary of the target range) or above a high threshold (e.g., above 250 mg / dL) (which is above the high boundary of the target range). The TIR display 252 may include a histogram in which different ranges are displayed in different colors. For example, times below the target range may be displayed in red, times within the target range may be displayed in green, and times above the target range may be displayed in yellow or orange. The glucose pattern report 250 may also display the average glucose level 254 over the reported time period 264 (e.g., approximately 14 days). The glucose mode report 250 can also display a glucose concentration curve 256, such as an AGP (Active Glucose Profile). The glucose concentration curve 256 is a graph of glucose data over a reporting period, where various data points can be color-coded to correspond to whether the glucose analyte level is below, within, or above the target range. This color coding can correspond to the color coding of the TIR display 252. Boxes 258 around different portions of the glucose concentration curve 256 highlight the mode (e.g., high, low, medium or medium-high, medium or medium-low, and combinations thereof), which are detected according to the GPA algorithm and discussed elsewhere in the specification.

[0128] If the patient's current treatment is known (e.g., basal plus RA insulin, basal only, basal plus SU, etc.), medication considerations 260 can also be provided in the glucose pattern report 250. Medication guidance can be provided in the form of text recommendations. General recommendations regarding insulin dose titration can be provided based on identified high and low glucose patterns, highlighted in boxes 258 in the glucose concentration curve 256. However, this general recommendation can be determined without needing data on the actual insulin dose administered. Recommendations can generally follow the rule of reducing any low patterns before reducing the high patterns. If the glucose pattern report contains recommendations related to insulin dose titration, the glucose pattern report 250 can also include recommendations that the patient is a good candidate for DGS 100, which facilitates communication between the HCP and the patient before transitioning to the learning phase.

[0129] When the GPA algorithm has identified a pattern with high variability, self-care considerations 262 can be displayed in the glucose pattern report 250. Alternatively, the glucose concentration distribution 256 may be so highly variable that the logic following the report cannot make specific recommendations, but instead defaults to the user negotiating lifestyle or treatment changes with their HCP.

[0130] For individuals not currently using a device or system (e.g., SCD 102) associated with the application capable of generating glucose pattern reports 250 as described above, HCP can suggest monitoring the patient through different devices or systems, thereby enabling the generation of report 250 or similar reports. For example, a patient may wear an SCD 102 to collect glucose data over multiple days or weeks, wherein the SCD 102 is configured in a masked or blinded mode, where the user cannot access the measured glucose levels and therefore cannot change his or her behavior during this period. Based on this data, a glucose pattern report can be generated. If recommended insulin titrations are included in the glucose pattern report, then glucose pattern report 250 may also include a recommendation that the patient is a good candidate for DGS 100 and may suggest a learning period for medication dispensing strategies.

[0131] During the learning phase, the MDD 152 can be combined with a glucose sensing system used for initial screening to provide a more complete picture of insulin-intensive diabetes management. The learning phase can utilize algorithms, as described elsewhere in this document, to detect the user's insulin dispensing strategy. During the learning phase, the DGA can be configured to determine how the user determines mealtime doses. For example, the DGA can determine whether the user is determining mealtime doses based on carbohydrate counting techniques, empirical techniques, such as a technique where the user learns an appropriate dose based on past experience with diets or similar diets, whether the user is dispensing a fixed amount of insulin for dietary use, whether the user is changing their insulin dietary dose based on pre-meal glucose levels (determined by fixed dispensing, carbohydrate counting, or empirical dosing), whether the user is taking into account residual insulin from previous injections (IOB) when determining doses (determined by fixed dispensing, carbohydrate counting, or empirical dosing), calculating carbohydrate counting or empirical dosing based on pre-meal glucose levels, or other techniques. The DGA can also determine whether the user's mealtime doses are fixed based on diet type (e.g., breakfast, lunch, and dinner) or whether the mealtime doses are changing. Determination that mealtime doses are changing can be an indication of the user's mealtime doses based on carbohydrate counting techniques. DGA can also determine if a user is adjusting their mealtime dose to address high pre-meal glucose levels. In some implementations, DGA can also determine target glucose levels, indicating that the user is adjusting or correcting their mealtime dose when their levels are above or predicted to be above the target glucose level. DGA can also determine which meals are associated with insulin doses. DGA can also identify patterns of missed meal doses. For example, DGA can detect whether a user has missed a meal- or time-related insulin dose at least twice or at least three times within a period of time (e.g., one or two weeks).

[0132] The learning period can last for any duration sufficient to obtain the required information. In various implementations, this period is at least two days, more preferably a week or longer (e.g., 14 days), and can vary depending on the extent to which the DGA is able to learn trends. The results can be compiled into summary reports for both the user and the physician.

[0133] Learning methods

[0134] Manual configuration of the DGS 100 may require HCP time, and HCP may not have sufficient time available. Furthermore, even if HCP time is available, configuration can be complex and error-prone. To mitigate these issues, a Patient Parameter Initialization (PI) module can be included in the DGA where no setup is required or only minimal setup is needed. The PI module learns the patient's medication dispensing strategy, which may include, for example, basal only, basal plus 1, basal plus 2, etc., and parameterizes the patient's medication dispensing practice to configure dose guidance settings via the DGA.

[0135] According to one aspect of the implementation, the learning process of the PI module may include automatically configuring patient dosage guidance settings based on observed data. Once these settings have been successfully learned, the DGS 100 can enter a guidance mode, where the patient can request dosage guidance and receive notifications regarding medication dispensing. During the learning process prior to the guidance mode, the DGA can process glucose and insulin data collected by the patient's SCD 102, UID 202, and / or other devices, and determine medication dispensing information based on the processed data.

[0136] Medication information may include, for example, dosage duration, dietary dosage type, dosage parameters, and dosage range. Dosage duration may include, for example, basal dose plus BF, basal dose plus LU, basal dose plus DI, basal dose plus BF / LU, basal dose plus BF / DI, basal dose plus LU / DI, and basal dose plus 3, where BF represents "breakfast," LU represents "lunch," and DI represents "dinner." Other regimens may also be included, such as afternoon snack dosage. Dietary dosage type may include, for example, fixed dietary dosage or variable dietary dosage. Dosage parameters may include, for example, the nominal fixed dose or carbohydrate ratio per meal, pre-meal correction factor (CF), and post-meal CF. Dosage range may include an estimate of the minimum dietary dose.

[0137] For each of the above-described medication information types, the DGA can determine whether the accumulated data is sufficient to determine the medication information. In some implementations, the patient's SCD 102 can be configured to operate for a predefined time period, such as 14 days. In these implementations, the DGA can determine, after the predetermined time period (or earlier if the sensor stops operating before the end of that time period), whether the available analyte and medication data are sufficient to determine each of the above-described medication information. If so, the DGA can perform parameterization method 300 and allow the start of a dose-guided mode. In another implementation, during the learning period (e.g., once daily), the DGA can periodically determine whether the data is sufficient to determine each of the above-described medication information. In either case, when sufficient data has been collected, the DGA can end the learning period, perform parameterization, and begin a guided period. If not, the DGA can continue the learning process.

[0138] refer to Figure 7The DGA can be configured to execute method 300 on a suitable computing device, such as UID 200, SCD 102, MDD 152, individually or in any combination. Program instructions for executing method 300 can be grouped into a PI module or any other suitable code configuration. In summary, method 300 may include: in step 302, classifying each of the drug doses received by the patient during the analysis period based on data characterizing the analyte and the drug doses received by the patient during the analysis period by the DGA. Method 300 may also include: in step 304, grouping each dose in one of a set of mealtime groups. The method may also include: in step 306, generating patient dose parameters, at least in part, by applying data from each mealtime group to a model. The method may include: in step 308, storing the dose parameters in computer memory to configure dose guidance settings. In the embodiments described herein, the analyte may be glucose, or may include an indicator of the patient's glucose level, and the drug may be insulin, or may include insulin. Dosage guidance settings can be used by the DGA to develop dosage guidance or provided to interface devices, such as UID 200 or a healthcare physician terminal, for output. More detailed aspects of each operation in method 300 are described below. As used herein, "PI module" refers to one or more parts of the DGA that perform the operations of method 300 and any auxiliary operations. The PI module is not limited to a specific configuration and can include various configurations of computer code.

[0139] On one hand, classification operation 302 may include classifying each dose of a drug (e.g., insulin) as one of a dietary dose, a corrected dose, and / or a vague dose. If the DGA cannot classify a drug dose as a dietary dose or a corrected dose to a defined confidence level, the DGA may classify the dose as vague, and the use of said dose in generating dose parameters for dose guidance may be omitted.

[0140] DGA can perform drug dosage classification through a series of two operations, which we refer to as feature extraction and classification. This is combined with... Figure 7In conjunction with this, classification operation 302 may include generating a feature matrix that associates a set of classification features with each dose. In some implementations, the DGA may configure a vector of insulin injection timestamps, a data file including analyte measurements from the patient's SCD 102, and results from a dietary monitoring algorithm module discussed elsewhere herein as input to a function that outputs a feature matrix for insulin dose classification. The number of rows in the feature matrix may indicate the number of injections or equivalent drug dispensing events during the correlation analysis. Each row in the feature matrix may be, or may include, a feature vector for a single dispensing event. In implementations for classifying insulin injections, each vector may include elements described below, referred to herein as classification features. The DGA may determine each element of the feature vector based on a corresponding segment of glucose monitoring data over a time range relative to the insulin injection time, for example, between -2.5 and 1.5 hours.

[0141] In implementation, classification features may include the time of drug administration for each dose, such as insulin injection recorded by MDD152 or by the time of day during which the patient injects insulin using UID 200.

[0142] Classification features can also include temporally filtered analytical values, such as glucose values ​​filtered using Savitsky-Golay filters, low-pass filters, band-pass filters, non-parametric smoothing filters such as locally estimated scatter plot smoothing, or other filters. For example, the Savitsky-Golay filter can be second-order with a frame length of 7 at a 15-minute sampling interval.

[0143] Classification features may also include the rate of change of analyte values ​​closest to the time of drug administration, such as the rate of change of analyte (e.g., glucose) values ​​calculated by linear regression of five analyte data points centered on the data point closest to the time of drug administration (e.g., injection) (e.g., using a 15-minute sampling interval).

[0144] Categorical features may also include a left-hand area under the curve (AUC) measure, which indicates the integral difference between multiple analyte values ​​and the analyte value closest to the drug time at the time interval preceding the drug time. For example, to obtain the left-hand AUC measure, DGA can calculate the left-hand AUC measure by collecting all data points from filtered analyte data within a time window (e.g., 2.5 hours), counting from the injection time, then calculating the difference between the average analyte value of the collected data points and the data point closest to the injection time (i.e., the reference data point), and multiplying that difference by the duration of the time window to calculate the incremental AUC.

[0145] Categorical features may also include a right-hand AUC metric, which indicates the integral difference between multiple analyte values ​​and the analyte value closest to the drug time at a time interval following the drug time. For example, DGA can compute the right-hand AUC metric by collecting all data points from filtered analyte data within a time window (e.g., 1.5 hours), counting from the injection time, then calculating the difference between the average analyte value of the collected data points and the data point closest to the injection time (reference data point), and multiplying that difference by the duration of the time window to calculate the right-hand incremental AUC.

[0146] Categorical features can also include the elapsed time between medication times. For example, for each injection time, the DGA can calculate the elapsed time between the previous injection time and the current injection time by subtracting the previous injection time from the current injection time. For the first injection time in the insulin log, since no previous injection time is available, the DGA can calculate the elapsed time from the first SCG time data point to the current injection time. Furthermore, for example, the DGA can calculate the elapsed time between the current and next injection time by subtracting the current injection time from the next injection time. For the last injection time in the insulin log, since no next injection time is available, the DGA can calculate the elapsed time from the current injection time to the last SCG time data point. In both forward and reverse calculations, if the elapsed time is greater than a predetermined maximum value, such as 12 hours, the DGA can set the elapsed time value to be equal to the maximum time.

[0147] Categorical features may also include the probability of a meal initiation within a defined interval prior to the administration time, for example, the maximum probability of a meal initiation within a time window prior to injection (e.g., 1.5 hours). This probability can be calculated by the meal detection module, as described elsewhere in this document.

[0148] Categorical features may also include the most likely time interval elapsed since the most recent meal start, for example, the elapsed time from the point of maximum meal start probability (e.g., determined by the meal detection module) relative to the injection time.

[0149] Categorical features may also include the probability of starting a meal within a defined interval after the time of medication, for example, the maximum probability of starting a meal within 2 hours after injection (determined by the meal detection module).

[0150] Categorical features may also include the most likely time interval until the next meal, such as the predicted elapsed time from the injection time to the point of maximum meal start probability after meal injection (e.g., determined by the meal detection module).

[0151] As noted, calculating some categorical features includes estimating the time of each meal consumed by the patient during the analysis period, and the methods for estimating meal times are described in more detail below. In short, estimating the time of each meal may also include generating a feature matrix from the DGA based on time-related analyte data, where the feature matrix associates a set of analyte (e.g., glucose) data features with each distinct region classified as rising, falling-preceding, and falling. This set of analyte data features may be, or may include, the maximum rate of change of analyte, the maximum acceleration of analyte, the analyte value at the point of maximum acceleration, the duration of the region, the height of the region, the maximum deceleration, the average rate of change within the region, and the time of the maximum acceleration of analyte. The estimation may also include generating the estimated meal times based on the feature matrix using the algorithm described below.

[0152] The following paragraphs describe more detailed aspects of the traceable meal detection algorithm for Method 300 or other applications. Other aspects of Method 300 are then described. DGA can perform traceable meal detection based on time-related analyte data by executing one or more code modules, such as a feature extraction module and a meal detection module. When executed by DGA, the feature extraction module allows DGA to receive a glucose time series as input and output a feature matrix that will be passed to the traceable meal detection module to detect glucose shifts in response to meal events.

[0153] DGA can perform feature extraction using the following operations, which can be divided into a sequence of three sub-operations: smoothing, segmentation, and extraction.

[0154] In the smoothing suboperation, DGA can use a Savitzky-Golay filter (2nd order) to smooth the analyte (e.g., glucose) time series and calculate the rate of change and acceleration at each analyte data point. The frame length parameter of the filter can be the number of data points collected within a first time interval (e.g., 60 minutes); therefore, sampling is interval-dependent. DGA can calculate the rate of change by averaging the forward and backward differences of the smoothed analyte values ​​between the point of interest and points at a second interval before and after it (e.g., 15 minutes), where the second interval is smaller than the first interval, for example, equal to 1 / 4 of the first interval. Similarly, DGA can calculate the acceleration by averaging the forward and backward differences of the analyte rate of change between the point of interest and points at a second interval before and after it (e.g., 15 minutes).

[0155] In the segmentation operation, DGA can segment a smooth analyte trace into monotonically increasing (i.e., rising) and decreasing (i.e., falling) regions. Each rising region is considered a candidate region for glucose shift in response to a dietary event.

[0156] In the extraction sub-operation, DGA can extract features from the data, such as 16 features, which can be or may include features from each rising region (e.g., eight (8) features), the preceding falling region (e.g., four (4) features), and the subsequent falling region (e.g., four (4) features). Features that DGA can extract from rising regions include, for example: 1) the maximum rate of change of analyte, 2) the maximum acceleration of analyte, 3) the analyte value at the point of maximum acceleration of analyte (reference point), 4) the duration of the rising region (the elapsed time from the reference point to the last point of the region), 5) the height of the region (the difference in smoothed analyte values ​​between the last point and the reference point), 6) the maximum deceleration (negative acceleration with the largest absolute value), 7) the average rate of change within the region (height / duration), and 8) the time of the data at the reference point. Also, for example, the four (4) features extracted from the preceding and subsequent falling regions may include: 1) the height of the falling region, 2) the duration of the falling region, 3) the average rate of change within the region (height / duration), and 4) the maximum absolute value of the rate of change of glucose. The number of rows in the feature matrix output by the feature extraction module can be the same as the number of rising regions in a smooth glucose time series.

[0157] According to another aspect of the implementation, the traceable diet detection module can take a feature matrix as input and output a binary detection result for each rising region. Such output may include a binary classification result and a probability value that each rising region is an analyte (e.g., glucose) shift in response to a dietary event. The DGA can assign the probability value of each rising region to its reference point. For example, in some implementations, the pre-trained machine learning model for diet detection can be implemented by scikit-learn using RandomForestCrystalifier (https: / / scikit-learn.org / stable / modules / generated / sklearn.ensemble.RandomForestClassifier.html). The diet detection module can detect diet-induced postprandial glucose shifts based on multiple decision trees constructed and optimized during training. In alternative implementations, the DGA can build the pre-trained model based on alternative classification algorithms such as gradient boosting, Ada boosting, artificial neural networks, linear discriminant analysis, and additional trees.

[0158] Refer again Figure 7In method 300, classification operation 302 can take the patient's feature matrix as input and output a binary classification result for each relevant pharmacological event (e.g., for each insulin injection). For example, DGA can output binary data, where "1" represents a dietary dose and "0" represents a non-dietary dose. According to some implementations, classification operation 302 can use dietary detection results, in which case dietary detection can be performed before insulin dose classification. As noted for retrospective meal detection, classification operation 302 can include a pre-trained machine learning model, such as the model implemented by scikit-learn using RandomForestClassifier (cited above). The machine learning model implemented by DGA can classify based on tree-building rules and thresholds for various features in each decision tree optimized during training. Alternatively, the model can also be trained using other machine learning algorithms, including gradient boosting, Ada boosting, artificial neural networks, linear discriminant analysis, and additional trees. After DGA successfully classifies each dose, it can proceed to determine the dosage duration and dosage parameters.

[0159] In step 304, method 300 may include the DGA grouping each dose within one of a set of mealtime clusters or groups. For example, the DGA may determine a dispensing strategy by performing cluster analysis on the timing of dietary doses of medication (e.g., injections). The DGA may execute a clustering module that takes the injection time as input, implemented using the K-means algorithm in conjunction with the elbow method, and outputs the optimal number of clusters K (maximum 3) and cluster indices for each injection time. The optimal number of clusters K may be the number of dietary doses ingested by the patient each day. Using the clustering index for each injection, the DGA can group the dietary doses into K groups based on the clustering index.

[0160] The DGA can identify these groups as breakfast, lunch, or dinner (B, L, D) as follows: For each group, the DGA can determine the typical time of day (TOD) by calculating the median TOD of that group. Alternatively, the DGA can use some other centroid measure. If K = 3, the DGA can associate breakfast with the group after the longest time interval between the typical group TODs. Then the next group is lunch, and the last group is dinner. If K = 2, the DGA can use assumed rules about the time between each meal to estimate which group is associated with breakfast, lunch, or dinner. For example, if two groups are more than six (6) hours apart from each other, then the DGA can identify these groups as breakfast and dinner. Otherwise, if the first group occurs before 10 a.m., then the DGA can identify that group as breakfast and dinner; otherwise, it is identified as lunch and dinner. In an alternative implementation, after the DGA identifies the typical time for the meal event, the DGA can prompt the user to identify the meal associated with each typical time. As another example, in an alternative implementation, the DGA can combine the two methods described herein by estimating the meal association and then prompting the user for confirmation. Another alternative approach could include analyzing glucose data to identify diets and clustering them to detect typical meal times. This is useful for differentiating diets when K=2; that is, identifying diets for which no intake dose was taken.

[0161] Once the doses are grouped into mealtime clusters, in step 306, the DGA can perform dose parameter generation for the patient, at least in part, by applying data from each mealtime group to the model. For example, for each meal group (B, L, D), the DGA can pair each corresponding pre-meal glucose level group with the corresponding meal dose. The DGA can fit each group with an appropriate model, such as a linear function with zero slope, a linear function with non-zero slope, a piecewise linear function connected at a single point, or a nonlinear function that approximates a connected piecewise model but has smooth curvature around the connection points. Other models are also suitable.

[0162] DGA performs model fitting and parameter estimation by minimizing the sum of squared residuals (SSR) with respect to the model parameters. DGA can then use a search algorithm to find the optimal parameters that minimize the SSR. For linear models, DGA can use the Nelder-Mead simplex method for fitting. For nonlinear models, DGA can use the Levenberg-Marquardt algorithm. That is, DGA can use the Nelder-Mead simplex numerical optimization method for linear models and the Levenberg-Marquardt optimization method for nonlinear models. Alternative methods for fitting data to these models are also possible.

[0163] When the number of iterations during optimization exceeds the convergence criterion, the model cannot fit, and DGA can exclude unfit models as candidate models. Furthermore, DGA can apply certain rules to minimize uncertainty in parameter estimation, such as requiring at least three pre-meal glucose data points greater than the estimated glucose threshold to validate the estimated correction factor; requiring at least three pre-meal glucose data points less than the estimated glucose threshold to validate the estimated fixed dose; requiring a 95% confidence interval for the parameter intercept to exclude zeros; or requiring a 95% confidence interval for the model slope to exclude zeros.

[0164] Insufficient data may lead to model failure, thus excluding certain models as candidate models. DGA can use the Akaike Information Criterion (AIC) to evaluate each model and select the model with the lowest AIC value as the preferred model for each meal group.

[0165] Once a model for each meal timing cluster is selected, the DGA can determine dosage parameters based on the selected model for that cluster, including, for example, a fixed dose of insulin, a target glucose level, and a correction factor. The DGA can determine the target glucose level and correction factor as single values ​​for all groups, as described in more detail in the following paragraphs. In an alternative implementation, the DGA can determine the target glucose level and correction factor separately for each group and use these individually determined parameters for downstream dose guidance operations.

[0166] According to another aspect of the implementation, the DGA can form a combined set of data to obtain a more accurate correction factor for the patient. For example, after fitting dose data to various models for each meal group to select the best model and estimate the fixed dose of insulin, the DGA can subtract the fixed dose of insulin from the relevant meal dose for each meal group. The remaining non-zero values ​​correspond to doses with correction amounts. These non-zero values ​​can then be combined from all three meal groups (B, L, D) to form a combined set. If the fixed dose of insulin for a group cannot be determined, the DGA can exclude the data for that group from the combined set. The system can then repeat the operation used to find the best-fit model for the combined set, or the same model identified when analyzing the groups individually can be used. The use of this combined set approach assumes that the patient has the same (or constant) correction factor and target glucose in all meals, and the combined set provides a larger sample size for potentially more accurate fitting. After the DGA determines the target glucose level and correction factor based on the best-fit model, the DGA has completed the estimation of the dose parameters. Then, in step 308, the DGA can store the dose parameters in computer memory used to configure dose guidance settings.

[0167] On the other hand, DGA can determine whether a patient is potentially counting carbohydrates (e.g., changing their dietary dosage to address carbohydrate consumption) by comparing the AIC value of the preferred model to a threshold (such as 50, 75, or 100). If the AIC value is greater than the threshold, DGA can determine that the patient is counting carbohydrates and request confirmation from the patient via UID 200.

[0168] In alternative implementations, one or more of the above-described operations may be omitted and replaced by requesting the patient or HCP to provide information manually, or by retrieving information from another source such as EMR or another software program. However, method 300 should be applicable to a variety of applications without requiring more information than that available from SCD and MDD.

[0169] User feedback during the learning period

[0170] Exemplary implementations of methods for obtaining user feedback during or after the learning phase of the DGA will now be described. During the initial learning phase of using the DGA, the user may be prompted for feedback. User feedback can provide the user with an indication that the system is in progress. The DGA may prompt the user for feedback (e.g., input or confirmation) regarding any aspect of the dosage guidance, including missing information about: dosage, analyte history, patient behavior or activity, usual dispensing strategies, type of specific dosage, confirmation that the dosage type or strategy determined by the DGA (e.g., learned by the system) is correct, etc.

[0171] During (or after) the learning period, the DGA may output a prompt or other indication on the UID 200 requesting user feedback. This feedback may relate to a dispensing strategy, such as a strategy associated with the type of insulin action (e.g., long-acting and / or short-acting or rapid-acting). If the feedback (or other determination) indicates that a long-acting strategy is being used, the DGA may monitor the patient’s basal dispensing pattern for the first time period, such as the first three (3) days, to classify each dose or dispensing pattern as a single or separate dose type, and / or characterize its dose by time period (e.g., single dose in the morning, single dose in the evening, or divided doses (e.g., twice a day, morning and evening). The DGA may also determine trends regarding dose (e.g., median, average) and related dose changes. Based on this information, the DGA may generate the expected basal dose. After the first time period, if the actual dose administered (e.g., automatically registered by MDD152 or entered by the user) differs from the expected dose, the user may be prompted for feedback.

[0172] According to one aspect of the implementation, the user can be prompted in many different situations. For example, the DGA can be configured to detect missed doses, such as when the user does not take a dose or bolus dose within the time period for which a previous basal dose or bolus dose was administered. If a missed dose is detected, the DGA can be configured to request input from the user regarding whether a basal dose was administered within that time period. According to some implementations, the DGA can also be configured to detect differences in dosing timing. For example, the DGA can be configured to detect when the user administers a basal dose at a different time of day than when the previous basal dose was administered (e.g., a basal dose usually administered in the morning is administered in the evening). When such a difference in dosing timing is detected, the DGA can be configured to request input from the user regarding whether the basal dose was administered at a different time period. In another aspect of the implementation, the DGA can also be configured to detect when additional doses have been administered. For example, the DGA can be configured to detect changes in the number of basal doses administered throughout the day. In yet another aspect of the implementation, the DGA can be configured to detect whether the dosing strategy on the first day (e.g., administering one basal dose) differs from the dosing strategy on the second day (e.g., administering two basal doses). When a different dispensing strategy is detected, the DGA can be configured to request input from the user regarding whether the user has already adopted the dispensing strategy to be used the following day as input for the new dispensing strategy. In another aspect of the implementation, the DGA can also be configured to detect whether a different dose was administered. For example, the DGA can be configured to detect whether a first dose administered during a time period of the day is different (less than or greater than) a previous dose administered during a time period of the previous day. When a different dose is detected, the DGA can be configured to request input from the user regarding whether the user has changed the dosage.

[0173] The user's response to these prompts can allow the DGA to confirm that it has identified the correct pattern (e.g., the user confirms that they missed their morning basal dose, but they usually take it) or give the user an opportunity to correct the pattern (e.g., the user informs the DGA before taking the dose that they are adjusting their basal dose based on their glucose levels).

[0174] In addition to the above-mentioned tips, DGA may also include tips regarding dose classification for rapid-acting insulin preparation strategies. Dose classification may include, but is not limited to, bolus, correction, split dose, bolus + correction, and bolus + carbohydrate count + correction classification.

[0175] The DGA can prompt the user in many different situations regarding rapid-acting insulin dosing. The DGA can be configured to detect whether a dose unrelated to a meal has been administered. For example, the DGA can be configured to determine whether a dose was ingested during a period when no meal was identified or detected. If the DGA detects that a dose was ingested and no meal was detected during the dosing period (e.g., approximately 1 hour after dosing), the DGA can request input from the user regarding the reason for the dosing (e.g., because they ate, because they lowered their glucose, or because they completed an earlier meal dose). The DGA can also be configured to detect whether a meal dose does not match a previous meal dose associated with the same meal type. For example, the DGA can be configured to determine whether a bolus dose associated with a first meal type and administered within a day differs from a previous bolus dose associated with the first meal type and administered within a day the previous day. In cases where such a difference in bolus dose is detected, the DGA can be configured to determine the reason for the different dose. For example, the DGA can be configured to determine the difference between the pre-meal glucose values ​​associated with the bolus dose and the previous bolus dose to determine whether the detected difference is a correction. DGA can also request input from users about the reasons for differences in bolus doses (e.g., because they eat less / more food and / or they are correcting for hyperglycemia, and / or they are correcting for other factors).

[0176] In addition to enabling DGA to determine what type of rapid-acting dose is ingested throughout the day, this allows DGA to help determine when the expected dose is reached. After a learning period without prompting, DGA can provide these prompts to the user if the dose differs from the expected dose, thus improving the DGA model of the user's medication dispensing strategy.

[0177] For both long-acting and rapid-acting doses, DGA can be designed to reduce the number of prompts over time and based on user response. Prompts can be emphasized frequently in the initial phase and gradually reduced as a recurring pattern is observed.

[0178] Glucose pattern analysis and dietary bolus titration for MDI insulin dosing therapy

[0179] Exemplary implementations of methods for determining dietary bolus titration will now be described. Once the system learns (or is configured with) a patient's current medication dispensing strategy, it can provide titration guidance for multiple daily injection (MDI) medication dispensing therapy. For patients using fixed-meal dispensing, a fixed dose can be titrated (e.g., for breakfast, lunch, dinner, snacks, etc.). For patients undergoing carbohydrate counting, carbohydrate ratios can be titrated against these same meals or against different times of day. Patients using empirical dispensing can titrate their doses based on each meal. The titration guidance from the DGA can provide recommendations for changing the dose or carbohydrate ratio in a specific direction. The amount of change can be an appropriate percentage change, such as 5%, 10%, 15%, etc. Dosage guidance can also include the starting meal dose. For example, if a patient is on a basal + 1 (e.g., lunch dose) regimen and breakfast shows a high pattern, the DGA can provide recommendations for administering RA insulin at breakfast.

[0180] DGAs may require the definition of medication categories, such as time of day (TOD), meal type (e.g., breakfast), and meal composition (e.g., milk-containing cereal). For example, a medication category could be a time of day, defined by a time period associated with the dietary insulin dose for that time of day. As another example, the post-breakfast time period could be defined as beginning when, for example, a dietary insulin dose is taken within a defined time period of day (e.g., between 5 a.m. and 10 a.m.) and ending after a prescribed postprandial time (e.g., 6 hours later), or at the time of taking the next insulin dose, whichever is earlier. One or more measures may be needed to define whether the postprandial glycemic response is nominal or requires correction, or to rank postprandial glycemic patterns as more or less favorable than another. Low glucose (LLG) measures and median glucose likelihood can be used to quantify the degree of risk of hypoglycemia and hyperglycemia, respectively.

[0181] U.S. Patent Publication No. 2018 / 0188400 ('400 Publication) describes an implementation for deriving and determining a risk measure in a glucose pattern analysis (GPA) suitable for use in DGA implementations, which is incorporated herein by reference in its entirety for all purposes. This implementation utilizes, in particular, central trend (e.g., mean, median, etc.) and variability data from multiple time periods to determine a risk measure corresponding to the degree of hypoglycemia risk (“hypoglycemia risk”). This implementation is summarized herein, and a more detailed description of this implementation and its variations is available by reference to the '400 Publication.

[0182] Alternatives to the implementation described in '400 Publication are listed in U.S. Patent Publication No. 2014 / 0350369, which is incorporated herein by reference in its entirety for all purposes. For example, instead of using intermediate values ​​and variability, the method can use any two statistical measures that define the data distribution. As described in '369 Publication, the statistical measures can be based on a glucose target range (e.g., G... LOW =70mg / dL and G HIGH =140 mg / dL). Commonly used measurements related to the target range are time within the target range (TIR) ​​and time above the target range (t). AT ) and time below target (t) BT If glucose data is modeled as a predetermined threshold G... LOW and G HIGH If the distribution is such that t is γ, then t can be calculated. AT and t BT For the threshold, the algorithm can also define t. BT_HYPO If t BT If the threshold is exceeded, the patient can be identified as being at high risk for hyperglycemia. For example, high risk for hyperglycemia can be defined as whenever t... BT For G LOW =70 mg / dL is greater than 5%. Similarly, a measure t can be defined. AT_HYPER If t AT If the levels exceed a certain threshold, the patient can be identified as being at risk of hyperglycemia. This can be addressed by adjusting the glucose levels. LOW or t BT_HYPO 、 or G HIGH or t AT_HYPER This is used to adjust for the degree of risk of hypoglycemia and hyperglycemia. Three measurements are used: TIR, t... BT and t AT Any two of these can be used to define the control grid. These alternatives (and other options) can be used to determine risk metrics for the DGA implementation described herein.

[0183] The DGA implementation described herein operates based on a quantitative assessment of a user's analyte data during a TOD period. This quantitative assessment can be performed in various ways. For example, the implementation described herein can assess analyte data over a multi-day period to determine one or more measures describing the associated risk of that analyte data to the corresponding TOD. These measures can then be used to categorize the analyte data from the TOD period into one of several patterns. For example, these patterns can indicate common or prevalent glucose behavior or trends in that TOD. The DGA implementation may use any number of two or more patterns. For ease of reference herein, these patterns are referred to as glucose pattern types, and the implementation described herein will refer to the use of three glucose pattern types (e.g., low pattern, high / low pattern, high pattern), although other implementations may use only two or more types, and these types may differ from those described herein.

[0184] For example, using a fixed dietary dose, once the DGA has learned the dosing strategy and dosage or carbohydrate proportions, titration assessment can begin, which can be divided into four titration categories: overnight, post-breakfast, post-lunch, and post-dinner. For each of these categories, the DGA can map the two measures mentioned above (LLG and median glucose) to four logical “pattern” variables according to the GPA method described below. Figure 8A An exemplary method 400 performed by a DGA for evaluating dietary bolus titration for multiple daily injection (MDI) medication administration is illustrated. Method 400 may include: at 402, the DGA determining the analyte pattern type of at least one TOD by executing a glucose pattern analysis (GPA) algorithm, the GPA algorithm receiving time-related analyte data as input from a sensor control device worn by the patient during the analysis period. Method 400 may further include, at 404, the DGA performing a recommendation algorithm selecting an MDI medication recommendation based on the analyte pattern type and the patient's defined medication administration strategy during the analysis period. Method 400 may further include, at 406, the DGA storing an indicator of the recommended action in computer memory for output to at least one of a UID 200 or MDD 152 administering the medication to the patient. UID 200 may use the indicator of the recommended action to control a user interface, for example, by displaying a human-readable expression of the indicator on a display, or by generating an audio output expressing the indicator in human language. MDD 152 may use the indicator to adjust or maintain the next relevant dose administration. More details about method 400 will be described below.

[0185] Figure 8BThis is a flowchart describing an exemplary implementation of GPA method 410, which can be implemented as the GPA algorithm referenced in 402. Method 410 can be performed for specific TOD periods, which can be a full day (e.g., a 24-hour period) or a portion of a day described by time blocks (e.g., three 8-hour periods) or user activities (e.g., meals, exercise, sleep, etc.). In various implementations, multiple TOD periods can correspond to meals (e.g., after breakfast, after lunch, after dinner) and sleep (e.g., overnight). These TOD periods can correspond to fixed times of day when activities typically occur (e.g., 5:00 AM to 10:00 AM after breakfast), where such time blocks can be set by the user or can depend on meals or activities that have actually been performed, such as those determined by automatic detection of meals or activities, or by user instructions (e.g., with UID 200).

[0186] The DGA can independently perform method 410 for each TOD period to obtain a separate pattern assessment for that period. At 412, the DGA can determine the central tendency and variability values ​​from user analyte data for a specific TOD period. For example, user analyte data can be obtained from the user's own records or the records of the user's healthcare professional, or user analyte data can be collected by the DGS 100. The analyte data preferably spans multiple time periods (e.g., two days, two weeks, one month, etc.) to ensure sufficient data exists for reliable determination within the TOD period. In other embodiments, the method can be performed in real-time on limited data. The DGA can use any type of central tendency measure associated with the central tendency of the data, including but not limited to the median or mean. You can also use any desired variability measure, including but not limited to a variability range across the entire dataset (e.g., from minimum to maximum), a variability range across most of the data but smaller than the entire dataset to reduce the importance of outliers (e.g., from 90% to 10%, from 75% to 25%), or a variability range targeting a specific asymmetric range (e.g., low-range variability, which may span, for example, from the central trend value or a range close to the central trend value to lower values ​​of the data, such as 25%, 10%, or the minimum). The choice of measures representing central trend and variability can vary depending on the implementation.

[0187] In 414, DGA can assess hypoglycemic risk measures (“hypoglycemic risk”) based on central tendency values ​​and variability values. (See reference...) Figure 8C A method for determining the risk of hypoglycemia is described, and an exemplary implementation of a framework for determining the risk of hypoglycemia and other measures is shown. Although Figure 8CThe aim is to transmit the frame to the reader; however, the frame can be implemented electronically in a variety of different ways, such as with software algorithms (e.g., mathematical formulas, a set of if-else statements, etc.), lookup tables, firmware, combinations thereof, or other methods.

[0188] Figure 8C This is a graph of central trend versus variability (e.g., low-range variability), which can be used to assess or identify regions or areas that maintain or correspond to a specific TOD (Time of Day) with defined central trend and variability data pairs. Any number of two or more regions can be used. In this embodiment, the data pairs may correspond to target region 425 or one of three hypoglycemic risk regions: low-risk region 426, moderate-risk region 428, or high-risk region 430. A first hypoglycemic risk function (e.g., a curve or linear boundary), referred to as the moderate-risk function 422, distinguishes between low-risk region 426 and moderate-risk region 428. A second hypoglycemic risk function, referred to as the high-risk function 424, distinguishes between moderate-risk region 428 and high-risk region 430. The central trend and variability data pairs can be evaluated against or compared with regions to determine a measure of hypoglycemic risk for the corresponding TOD period.

[0189] Hypoglycemic risk functions 422 and 424 can be explicitly implemented in the DGA as mathematical functions (e.g., polynomials) or implicitly implemented, such as by defining each region by the pairs it contains, using query tables, a set of if-else statements, threshold comparisons, or others. Hypoglycemic risk functions 422 and 424 can be pre-loaded into the DGA, downloaded from a trusted computer system 480, or set by another party such as the HCP. Once implemented in the DGA, hypoglycemic risk functions 422 and 424 can be considered fixed or can be tuned by the user or the HCP. Exemplary methods for determining hypoglycemic risk functions are described in '400 Publication'.

[0190] In 416, DGA can assess a measure of hyperglycemic risk (“hyperglycemic risk”) based on a central trend value. In this implementation, hyperglycemic risk can be assessed by comparing the central trend value for a specific TOD period with a central trend target or threshold 432. The magnitude and / or sign of the difference between the central trend value and the target 432 can identify the amount of hyperglycemic risk. For example, if the central trend value is less than the target 432 (e.g., a negative value), a low hyperglycemic risk may exist. If the central trend value exceeds the target 432 by a amount less than the threshold (e.g., 5%, 10%, etc.) (e.g., a positive value), a moderate hyperglycemic risk may exist. If the central trend value exceeds the target 432 by a value greater than the threshold, a high hyperglycemic risk may exist. Using three discrete groups (e.g., low, moderate, high) for hyperglycemic risk is just one example; any number of two or more groups can be used.

[0191] In other embodiments, at 416, DGA can assess a hyperglycemic risk measure, which may be prior to assessing the hypoglycemic risk at 414. Alternatively, in another embodiment, the assessment of hypoglycemic risk at 414 and the assessment of hyperglycemic risk at 416 can be performed simultaneously.

[0192] Other metrics, such as variability risk, can also be assessed. For example, a variability value less than a first variability threshold 434 indicates low variability risk, a variability value greater than the first variability threshold 434 and less than a second variability threshold 436 indicates moderate variability risk, and a variability value greater than the second variability threshold 436 indicates high variability risk. Again, using three discrete groups for variability risk is just one example. DGA can use any number of two or more groups.

[0193] In step 418, the DGA can determine the pattern type for the TOD period based on one or more risk measures assessed. In one exemplary implementation, the pattern determination can be assessed using a hypoglycemia risk measure and a hyperglycemia risk measure. If the hypoglycemia risk measure is high, then the pattern can be set to a low pattern. Otherwise, if the hypoglycemia risk is moderate and the hyperglycemia risk is high or moderate, then the pattern can be set to a high / low (or medium) pattern. Otherwise, if the hyperglycemia risk is high or moderate and the hypoglycemia risk is low, then the pattern can be set to a high pattern. If both the hyperglycemia and hypoglycemia risks are low, then the identified pattern can be problem-free (e.g., displaying and outputting an "OK" message).

[0194] Therefore, method 410 is an example of how the DGA outputs one of multiple pattern types for each TOD period. The number of pattern types within the pattern type itself may differ from those described in this embodiment (e.g., low, high / low, high). Once the pattern type for the TOD period is determined, the DGA can store an indicator of the pattern type in a memory location for use in determining titration recommendations. See again... Figure 8A In 404, DGA can continue to determine titration recommendations after completing the GPA for each relevant TOD period.

[0195] Recommendation methods can branch based on pattern type (e.g., low, high / low, high) and other factors, including TOD period, dispensing strategy, adherence to the strategy (e.g., missed doses), and availability of sufficient data for evaluation. DGA makes titration recommendations only when sufficient data is available for the corresponding TOD period. For example, if the amount of available data is less than a threshold, such as the number of separate days less than a threshold number (e.g., 5 days), and greater than the minimum fraction of available data (e.g., 90%), DGA may omit evaluation and generate an error message.

[0196] Figures 8D to 8H An example branch of a recommendation algorithm or method for determining dose titration recommendations based on the above input information is shown. Other branches are also useful. Figure 8D Branch 440 of the TOD recommendation methodology is shown, which has sufficient available data and possible reasons for a low pattern type, which includes one or more of the following: above the optimal basal dose, dietary dose, pre-meal correction dose, or post-meal dose. At 442, the DGA assesses whether the overnight TOD period pattern type is low. If the pattern is low, then at 444, the DGA generates a recommendation to reduce all relevant doses by an equal amount (e.g., 10%), which includes at least the basal dose and optionally one or more of the dietary dose, pre-meal correction dose, or post-meal dose. Titration recommendations for low patterns may include, for overnight TOD periods, generating a recommendation at 444 to reduce the long-acting insulin dose or basal rate. At 446, if any other TOD period has a low pattern, then at 448, the DGA may generate a recommendation to reduce the fixed dietary dose used only for the relevant TOD period.

[0197] In this implementation 440, if at least one low pattern is present, no titration guidance is provided for any high-pattern TOD period. The idea here is to emphasize the prevention of hypoglycemia and to increase the dose only when the risk of hypoglycemia is low across all TOD periods. Furthermore, in some cases, when a TOD period has a high pattern, this may be caused by a previous TOD period with a low pattern, and the patient may be overeating to compensate—therefore resolving the low pattern itself can help resolve the subsequent high pattern. At 449, if the pattern is not high, process 440 waits or terminates without generating a recommendation or transfer to a high-pattern assessment 450.

[0198] Therefore, for high / low patterns, the DGA does not generate titration guidance. If no titration guidance can be provided for a TOD period where data is sufficient for all timeframes, the DGA may inform the patient that glucose variability needs to be addressed before further titration guidance can be given. Additionally, the DGA may provide reports to the patient's HCP to consider alternative medications or therapies that can address glucose variability.

[0199] Figure 8E The diagram illustrates the procedure for DGA to generate titration recommendations for high-pattern insulin when there is no low-pattern TOD period. At 452, if the overnight period has a high pattern and there are no other periods with moderate hypoglycemia risk, then at 454, DGA may increase the long-acting insulin dose or basal rate recommendation. At 456, if the overnight period has a high pattern and there is at least one other non-dinner period with moderate hypoglycemia risk, then at 458, DGA may reduce the dietary insulin dose associated with any period with moderate hypoglycemia risk. At 460, if the overnight TOD period has no moderate hypoglycemia risk and no high pattern, then at 462, DGA may generate a recommendation to increase the dietary insulin dose associated with the first TOD period having a high pattern. At 464, if the overnight period has a moderate hypoglycemia risk and the only postprandial period with a high pattern is dinner, then at 466, DGA may generate a recommendation to increase the long-acting insulin dose or basal rate. If there is a moderate risk of hypoglycemia during the overnight period, but not during the late dinner period, then at 462, DGA can generate a recommendation to increase the dietary insulin dose associated with the first TOD period with a high pattern.

[0200] In alternative implementations, pre-meal glucose may be higher or lower than the target glucose level (e.g., 120 mg / dL). The glucose data for each meal, which helps in calculating measures of hypoglycemia and hyperglycemia risk, can be modified to compensate for the effects of previous meals or conditions that do not affect glucose due to the current meal. DGA can modify these data by subtracting an offset, resulting in a starting glucose level that is the target level. Alternatively, DGA can modify these data using a “triangle” function, where, for the meal start time, the difference between the starting glucose and the target glucose is subtracted, but this modification decreases over time; or linearly continues for a specified time (e.g., three (3) hours), or another decay function.

[0201] Alternatively, the function itself can be a function of the dietary start glucose level or glucose trend, and / or a function of the previous dietary dose ingested.

[0202] According to another aspect of the implementation, the algorithm for generating dietary bolus titration recommendations becomes more complex when considering other factors, such as missed dietary bolus doses, missed basal doses, postprandial corrections, and preprandial corrections. If these factors exist, the algorithm for providing appropriate recommendations may need to exclude some data while still meeting the data sufficiency threshold for providing guidance.

[0203] For example, refer to Figure 8FIf a high pattern is detected at 461, and some of these days were missed meal doses, then the missed meal dose days are excluded at 463, and GPA analysis 410 is repeated. If a high pattern is subsequently detected at 465, then at 467 the dose can be increased based on patterns identified in other TODs, or at 469 further input or return can be awaited. Alternatively, the system can only assess high patterns using data where missed meal dose days have been excluded. Figure 8F Algorithm 470 with this branching mode is shown in the diagram. If the system detects a low mode at 473, it can execute the low mode algorithm 472 described in the following paragraph. If the system does not detect either a high or low mode, it can return to box 469 for further input or return.

[0204] In section 472, regarding missed meal doses, if the DGA detects a low pattern during the TOD period, and if the missed meal dose occurs for several days during that TOD period, then the DGA may generate a dose reduction recommendation. Recommendations may include, for example, reducing a fixed portion or a corrective dose portion.

[0205] Regarding missed basal doses, if the DGA detects a low pattern 473 in the overnight TOD, the missed basal dose should not affect the dose titration logic. Similarly, if a low pattern is detected in the TOD outside of the overnight period, the missed basal dose should not affect the dose titration logic.

[0206] If the DGA detects a high pattern in a TOD using data including at least one day (or TOD) with a missed basal dose 461, then at 463 data for any one or more days (or TODs) with a missed basal dose can be excluded, and pattern analysis 410 can be repeated. Subsequent actions may depend on the specific TOD in which the high pattern was detected. For example, if the DGA detects a high pattern in an overnight TOD in data including at least one day with a missed basal dose, then data for any one or more days with a missed basal dose can be excluded, and pattern analysis can be repeated. If a high pattern is detected in an overnight TOD when one or more days with a missed basal dose have been excluded, the basal dose can be increased because the overnight TOD results can be used as guidance for adjusting the basal dose. If a high pattern is detected in a TOD other than an overnight period, then days with a missed basal dose can be excluded, and pattern analysis can be repeated. If a high pattern is detected when one (or more) days with a missed dose have been excluded, then the dietary dose associated with the TOD having the high pattern can be analyzed for titration, as described herein. In either case, the logic flow 470 is as follows: Figure 8F As shown.

[0207] Figure 8GAn example of the recommended logical flow 474 for developing a post-meal correction is shown. If, after GPA 410, the DGA detects a low pattern 479 of TOD within a few days including post-meal correction, then the following analysis can be used to titrate the correction or post-meal dose. At 475, if the DGA first detects a low pattern 479, it can exclude data from days without post-meal correction, first testing at 487 whether sufficient data is available. If insufficient data is available, the DGA can execute the error recovery routine 489, for example, displaying an error message. If sufficient data is available, the DGA can repeat pattern analysis 410. Subsequently, if the DGA detects a low pattern, then at 476, the DGA can reduce the post-meal correction dose (i.e., increase the correction factor), depending on the pattern analysis results in other TODs. Subsequently, if the DGA does not detect a low pattern, then it can reduce the meal dose at 477.

[0208] For all embodiments described herein, a modification of the correction dose in one direction (e.g., titration) can be achieved by modifying the correction factor in the opposite direction. These two parameters are inversely related, such that a decrease in the correction factor can lead to an increase in the correction dose, and an increase in the correction factor can lead to a decrease in the correction dose. Therefore, in all embodiments described herein, DGA can be recommended or achieved by modifying the correction factor or by modifying the correction dose. Thus, the extent to which the correction factor is modified or titrated is described herein allows these embodiments to be configured to achieve the same effect by reversing the modification of the correction dose; conversely, the extent to which the correction dose is modified or titrated is described herein allows these embodiments to achieve the same effect by reversing the modification of the correction factor. Given this interchangeability, both options are available for each embodiment described herein, although these options are not described for every embodiment merely for ease of description.

[0209] Furthermore, or alternatively, starting with the original data set 491, at 478, the DGA can exclude days with missed meal corrections. After finding sufficient data at 487, if pattern analysis 410 of these data at 490, which excludes days with post-meal corrections, does not indicate a low pattern, then the DGA can recommend reducing the post-meal correction dose 476. Otherwise, the DGA can achieve... Figure 10B Logic 510, which can lead to a reduction in mealtime insulin or a dose-guided pre-meal correction section recommendation. If no low level of DGA is detected at 479 and no high glucose pattern is detected at 492, it can wait for further input or return at 469. If a high pattern is detected in DGA at 492, then it can proceed to box 471 ( Figure 8H It can process 480.

[0210] Reference Figure 8H If the DGA detects a high pattern 493 within a few days of TOD including postprandial correction, then it can perform the following procedure 480 to develop recommendations for titration correction and dietary dosage. At 481, the DGA can include data from days of missed doses and postprandial correction and repeat pattern analysis 410. If the DGA subsequently detects a high pattern at 494, it can at 482 increase the postprandial correction dose (i.e., decrease the correction factor), depending on the pattern analysis results in other TODs. If it does not detect a high pattern at 494, it can at 495 check for a low pattern; if a low pattern is detected, it returns to... Figure 8G 474, otherwise wait for further input or a return at 469. Although in Figure 8H As not shown in the diagram, after excluding any data from GPA 410 and before executing GPA, DGA can test the adequacy of the data and execute an error recovery routine if the available data is insufficient.

[0211] In an alternative, or further, starting with the raw data set at 493, if pattern analysis at 483 on the data excluding days with missed meal doses indicates a high pattern along either branch 2.1 or 2.2, then the DGA can proceed to step 480 as follows. On branch 2.1, if pattern analysis at 484 on the data excluding days with postprandial correction (i.e., data with only bolus doses) does not indicate a high pattern at 497, then the DGA can at 482 generate a recommendation to increase the postprandial correction dose, depending on pattern analysis of other TODs. Otherwise, the DGA can… Figure 10C The process of 550 leads to recommendations to increase prandial insulin or pre-prandial correction.

[0212] In branch 2.2, at 485 if pattern analysis including only postprandial-corrected days does not indicate a high pattern at 496, then DGA may be increased with either prandial insulin or preprandial-corrected insulin, according to... Figure 10C The process is 550. If no high mode is detected at 496, the DGA can return to box 484.

[0213] If titration recommendations for correction factors from different TODs are conflicting, and if the patient is currently using the same correction factor for all TODs, then DGA may increase the correction factor. If none of the three components—dietary dose, pre-meal correction, and post-meal correction—are optimal, then process 480 may first increase the dietary dose. Pre-meal correction can be up-titrated after dietary dose titration. Post-meal correction can be up-titrated after dietary dose and pre-meal correction have been applied.

[0214] If, during subsequent analysis, the DGA generates a "increase correction factor" recommendation for a TOD using the aforementioned method, it will now result in a recommendation of "do not change correction factor." Conversely, if a TOD generated using the aforementioned method with a "decrease correction factor" recommendation still receives a recommendation of "decrease correction factor," then different TODs may be optimized by using different correction factors.

[0215] Although Figures 8D to 8H Aspects of various recommendation algorithms 404 used in method 400 are shown, but it should be understood that these are merely examples. Various other algorithms may also be suitable.

[0216] Dietary bolus titration lag

[0217] An exemplary implementation of a method for mitigating oscillations around the optimal dietary bolus dose will now be described. When the DGA has reached or is close to reaching the optimal titration, situations may arise where titration is requested but is not actually necessary. That is, the measure used to determine whether titration is needed may have some error or variability, and this variability may cause the titration algorithm to oscillate around the optimal dose. A patient parameter convergence tracking (PPC) module can be included in the DGA to mitigate this oscillation problem.

[0218] In one implementation, the PPC module can collect past information from the user, including data sufficient to determine the impact of various events on the patient's glucose levels at different times. The PPC module can be configured to calculate and track how multiple outcome measures change over time. Outcome measures may include, but are not limited to, hypoglycemic risk, hyperglycemic risk, and time within range. The PPC module can also be configured to calculate and track changes in dose guidance parameter estimates and / or dose guidance recommendations. Dose guidance parameter estimates may include, but are not limited to, estimated insulin sensitivity factor / insulin correction factor, estimated insulin-to-carbohydrate ratio, and estimated average carbohydrates per meal. Dose guidance recommendations include, but are not limited to, recommended basal insulin dose, recommended dietary dose, recommended supplemental dose, and recommended correction dose.

[0219] In one implementation, the DGA's PPC module can be configured to create a correlation for each outcome metric. This correlation can be a multidimensional correlation. The PPC module can be configured to map changes in each dose guidance parameter and / or guidance recommendation over time to changes in the outcome metric over time. The PPC module can also be configured to track the gradient of a specific metric. For example, if changes in multiple dose guidance parameters and / or guidance recommendations result in a small predicted change in the outcome metric, the PPC module can be configured to determine that adjustment recommendations are delayed for a period of time to prevent unnecessary adjustment requests to the user.

[0220] In another implementation, the PPC module can be configured to use multiple gradients to compare each of the multiple gradients with a set of predetermined gradient thresholds. The PPC module can be configured to determine that the adjustment recommendation is delayed when the number of gradients exceeding its thresholds is not greater than a predetermined value. The predetermined value can be a general or user-specific value.

[0221] Physiological Dosage Guidance Algorithm

[0222] Exemplary implementations of methods for determining dose guidance based on physiologically relevant processes will now be described. Many model-based control systems, specifically those using model predictive control (MPC) algorithms, are based on black-box diagrams and lack a rigorous physiological basis. As a result, these models may fail to account for certain pharmacokinetic and pharmacodynamic differences among different insulin analogs used to enhance MDI therapy. Furthermore, for MDI therapy, a given insulin dose must manage the user's blood glucose for a period of time, while many currently used dose guidance algorithms refer to their use in conjunction with an insulin pump, which delivers only one insulin analog (typically, a fast-acting one) to the user in a continuous flow. When communicating with the CGM, the MPC algorithm receives glucose feedback frequently (e.g., every 5 minutes) regarding the current insulin delivery rate and can subsequently adjust the pump flow rate in real time to maintain normal blood glucose levels. Therefore, while current MPC algorithms can drive frequent changes in pump-driven insulin delivery based on glucose feedback, they are not suitable for MDI therapy, where insulin delivery is less frequent and therefore requires prediction over longer time periods. As these levels become longer (e.g., on the order of hours), drug pharmacokinetics and pharmacodynamics can play a greater role in establishing accurate predictions. This necessitates the design of a more physiology-based approach for insulin dosing guidance and glucose control.

[0223] This paper describes an insulin dose-guided algorithm that considers physiologically relevant processes such as insulin diffusion, subcutaneous pharmacokinetics (PK), and glucose-insulin kinetics. Parameters are determined by numerically solving a minimal model of modified insulin absorption and glucose control during a “learning phase” (see Bergman et al., 1978 and Dalla Man et al., 2007, which are incorporated herein by reference in their entirety). Once these parameters are determined, the user-specific minimal model can be solved for mealtime and corrected dose to determine the optimal insulin dose at that time.

[0224] As described similarly in other embodiments, the DGA can receive or otherwise access glucose values ​​and trends from the glucose monitoring system. The DGA can be configured to ask the user if they will soon have medication dispensed, and can also, or alternatively, be configured to ask the user which diet they will need dosage guidance for. The DGA can be configured to output dosage recommendations based on an algorithm (e.g., a physiological dosage guidance algorithm or another algorithm described herein). The DGA can then be configured to observe whether the user adheres to the dosage guidance and can also track the resulting glucose trace for consideration in future dosage guidance.

[0225] DGA can be configured to include a physiological dose guidance algorithm that determines the optimal dose to output in dose guidance. In one implementation, dose guidance can be provided to users with a multiple daily insulin DI (MDI) regimen consisting of a once-daily long-acting analog and a corrected 3x daily rapid-acting dietary analog. This method can be applied to other MDI strategies, such as basal MDI strategies with only a single rapid-acting injection or basal MDI strategies with only a single rapid-acting injection.

[0226] In an exemplary implementation, such as Figure 9A As described in the flowchart, starting from step 904, in exemplary method 900, DGA can automatically generate possible bolus insulin doses (u1, u2, ..., u3) based on past medication history. n In step 906, the physiological dose guidance algorithm can process possible insulin doses and generate and output the glucose time progression (g1, g2, ..., g) for each bolus insulin dose input. n (e.g., a data array). In step 908, the DGA can then calculate the cost function value (C1, C2, ..., C) for each glucose time process. n The cost function can be defined as the time that minimizes the time outside a range (e.g., outside the range of approximately 70 mg / dL to approximately 180 mg / dL) for each glucose time course. In step 910, the DGA can determine the optimal insulin bolus dose that minimizes the cost function value.

[0227] The optimal insulin bolus dose (u) delivered at time t can be determined from a planned glucose trace that follows simulated insulin dispensing and carbohydrate dietary input (if needed). Using a simple bolus calculation equation (below), this dose can be broken down into components such as a glucose correction component, a dietary coverage component, and the insulin active component.

[0228]

[0229] However, physiological dose guidance algorithms may not use equation (1) to determine the optimal dose. Instead, physiological dose guidance algorithms can use the separate components of the equation to help users and HCPs understand how much of the dose is being used relative to dietary coverage at the current glucose level. In the equation, u is the optimal insulin bolus dose. The first term in the equation is enclosed in parentheses. It is the correction part and refers to the part used to correct the user's current glucose (BG(t)) and target value (BG) at time t. target Any deviation between these values. The correction factor (CF) represents the user's specific sensitivity to insulin, i.e., how effective a single insulin unit is in lowering the user's blood sugar. The second term (CHO*IC) represents the portion of the insulin dose required to cover the diet, if needed. The CHO value represents the amount of carbohydrates in the upcoming meal, while the IC value represents the subject-specific insulin:carbohydrate ratio, i.e., how many grams of carbohydrates can be covered by a given insulin dose. Finally, the IOB value refers to the "active insulin," which can still provide therapeutic benefit to avoid any situation where insulin buildup occurs. This value can be based on insulin pharmacokinetic models as well as endogenous insulin production. target The values ​​of IC and total daily insulin dose (TDD) can be defined by the user's HCP. In one implementation, the initial conditions for CF can be defined according to the "1800 rule" defined as 1800 / TDD. Furthermore, in one implementation, CF can be refined based on data from the calibration dose during the learning phase.

[0230] Physiological dose guidance algorithms can use different methods to determine dose guidance for corrected dose, dietary dose, and basal dose. However, for all three types of dose guidance, DGA can assume that rapid-acting insulin analogs (such as insulin lispro, insulin aspart, or insulin glutalis) will be used, and that all three have similar subcutaneous PK distributions, making their respective durations of action equivalent.

[0231] Corrected dose

[0232] When requesting dosage guidance for the corrected dose from the DGA, CHO = 0 because the corrected dose is independent of diet, and no new carbohydrates are considered. Therefore, the second term in equation (1) equals zero. Thus, the components of the corrected dose are the glucose correction portion and IOB.

[0233]

[0234] In one exemplary implementation, such as Figure 9BAs described in the flowchart, in exemplary method 191, starting from step 912, once a corrected dose is indicated, the DGA can generate multiple insulin dose candidates. In one implementation, multiple insulin dose candidates can be generated based on the subject's past insulin dose history.

[0235] In step 914, the DGA can determine multiple glucose time courses corresponding to multiple insulin dose candidates. In one implementation, the DGA can calculate a subject-specific glucose time course using a physiological dose-guided algorithm that modifies a minimal model of the possible dose range. Each dose candidate can be considered as the sum of the injected insulin dose and the currently active long-acting insulin dose. The amount of IOB can be calculated by the pharmacokinetics specific to the subject's long-acting analogue and can be the same across all candidate input doses. Numerical differences between all candidate insulin doses can be attributed to differences in the rapid-acting component of the total dose.

[0236] In step 916, DGA can calculate multiple cost function values ​​corresponding to multiple glucose time periods. In one embodiment, for each glucose time period, different time measures within a range can be calculated and used to determine a cost function value describing the risk of exceeding the target range (e.g., from about 70 mg / dL to about 180 mg / dL).

[0237] An exemplary cost function is shown in Equation 5 below, but many other forms can also be used. The area under the curve (AUC) is used in these calculations to combine the magnitude and duration of hyperglycemia / hypoglycemia events.

[0238]

[0239]

[0240] C = w hypo (AUC hypo，calc -AUC hypo，threshold ) 2 +w hyper (AUC hyper，calc -AUC hyper，threshold ) 2 (5)

[0241] Threshold AUC hypo and AUC hyper It can be defined as the minimum acceptable time and duration spent in any treatment course. Each term in C can have an associated weighting factor w, and the total set of weighting factors w is {w1, w2, ..., w...}. nThe sum should be 1. This weighting allows for prioritizing protection against hypoglycemic events. The insulin dose associated with the minimum cost function value can be a dose-guided recommended dose from the DGA output.

[0242] In step 918, the DGA can determine an optimal insulin dose, wherein the optimal insulin dose has the lowest cost function value among a plurality of cost function values. In one embodiment, the DGA can be configured to determine a dose candidate that minimizes the associated cost function C for out-of-range times as the optimal insulin dose.

[0243] In step 920, the DGA can output dosage guidance including the determined optimal insulin dose.

[0244] In another implementation, instead of simulating a glucose trace from a prior set of possible insulin doses and selecting the optimal option, in an alternative implementation, a DGA can provide an initial dose prediction, from which glucose levels and cost function values ​​can then be derived. Constraint minimization of the cost function can then be performed using gradient-based or non-gradient (e.g., genetic algorithm) methods. Therefore, each insulin dose prediction can be selected after the first dose to minimize the cost function.

[0245] Once the optimal insulin dose is determined, it can be further subdivided to help users understand what aspects need to be covered by correction and how much insulin is currently active.

[0246] Dietary dosage

[0247] When requesting dosage guidance for a dietary dose from the DGA, similar to the calculation of the corrected dose described above, the DGA can use modifications to a minimal model of the possible dose range to calculate the user-specific glucose time course. Because the dosage guidance is diet-related, unlike the process used to determine the corrected dose, the DGA can also consider additional dietary carbohydrate information when determining the dosage guidance. Because the dosage guidance is diet-related, CHO is greater than zero, and the composition of the dietary dose can include additional dietary carbohydrate information (CHO*IC). Therefore, according to equation (1), the composition of the dietary dose includes a glucose corrected portion, a dietary coverage portion, and IOB.

[0248] In one exemplary implementation, such as Figure 9CAs described in the flowchart, in exemplary method 921, starting from step 922, in response to a user's inquiry regarding dietary dosage guidance, the DGA can determine the distribution of carbohydrate values ​​in the diet. The distribution of carbohydrate values ​​may include central trend carbohydrate values, low carbohydrate values ​​below the central trend carbohydrate values, and high carbohydrate values ​​above the central trend carbohydrate values. In one implementation, the diet can be represented by a diet-specific distribution of carbohydrate values ​​using known descriptive statistics to describe the central trend (e.g., mean or median) and variations (e.g., standard deviation, coefficient of variation, 25 / 75 quartiles). These data can be transformed to provide these summary statistics. Thus, each diet can have its own carbohydrate distribution.

[0249] In one implementation, the central trend carbohydrate value can be the mean or the median. In one implementation, a low carbohydrate value can be, for example, 25% of the glucose data, alternatively 30%, alternatively 35%, alternatively between 20% and 40%, or μ-σ if the data is normally distributed. In one implementation, a high carbohydrate value can be, for example, 75% of the glucose data, alternatively 80%, alternatively 65%, alternatively 60%, alternatively between 60% and 80%, or μ+σ if the data is normally distributed.

[0250] In step 924, the DGA can identify multiple insulin dose candidates for each of the central trend carbohydrate value, low carbohydrate value, and high carbohydrate value. In one implementation, multiple insulin dose candidates can be generated based on the subject's historical insulin dose history.

[0251] In step 926, the DGA can determine multiple glucose time courses corresponding to multiple insulin dose candidates for each of the central trend carbohydrate value, low carbohydrate value, and high carbohydrate value. In one implementation, the DGA can use a modification of a minimal model to calculate a subject-specific glucose time course, the minimal model being used for possible dose ranges using a physiological dose guidance algorithm, as described above regarding corrected doses. In one implementation, each dose candidate can be considered as the sum of the injected insulin dose and the currently active long-acting insulin dose. The amount of IOB can be calculated from the pharmacokinetics specific to the subject's long-acting analogue and can be the same across all candidate input doses. Numerical differences between all candidate insulin doses can be attributed to differences in the rapid-acting component of the total dose.

[0252] In step 928, the DGA can calculate multiple cost function values ​​corresponding to multiple glucose time processes for each of the central trend carbohydrate value, low carbohydrate value, and high carbohydrate value. In one embodiment, for each glucose time process, different time measures within a range can be calculated and used to determine a cost function value describing the risk of exceeding a target range (e.g., from about 70 mg / dL to about 180 mg / dL). While many other forms can be used, in one embodiment, the DGA can be configured to calculate multiple cost function values ​​using an exemplary AUC analysis described with respect to corrected dose calculations. In other embodiments, for each glucose time process, different glucose values ​​at predetermined percentages of glucose values ​​within a predetermined (rolling) time window can be determined. For example, as an alternative to the percentage of low blood glucose time or an AUC below a low glucose threshold, a 5% value (or another percentage less than 30%) can be calculated within a time window and compared to a low glucose threshold. A time window can be a fixed function of the time of day, such as 9 a.m. to 11 a.m., relative to the start and / or end of an event, such as approximately 30 minutes after a meal, or up to approximately 300 minutes after a meal, or other definitions. A different low percentage with its corresponding low glucose threshold can also be used. As an alternative to the percentage of time with high blood glucose or to an AUC above the high glucose threshold, a 90% value (or another percentage greater than 70%) can be calculated within a time window and compared to the high glucose threshold.

[0253] In step 930, the DGA can determine an optimal insulin dose for each of the central trend carbohydrate value, low carbohydrate value, and high carbohydrate value. The optimal insulin dose can be the dose with the lowest cost function value among multiple cost function values ​​for each of the central trend carbohydrate value, low carbohydrate value, and high carbohydrate value.

[0254] In step 932, the DGA may output multiple dose guidelines, including an optimal insulin dose determined for each of the central trend carbohydrate value, low carbohydrate value, and high carbohydrate value. The dose guideline associated with the central trend carbohydrate value may correspond to the median dietary dose guideline. The dose guideline associated with the low carbohydrate value may correspond to the "below normal" dietary dose guideline. The dose guideline associated with the high carbohydrate value may correspond to the "above normal" dietary dose guideline. In one implementation, the DGA may output a series of dietary dose guidelines to illustrate the current glucose value and the distribution of dietary carbohydrate values. Any extreme cases may be modified or deviated from as needed. For example, the optimal insulin dose associated with a "above normal" diet may need to be based on 60% rather than 75% to avoid postprandial hypoglycemia due to over-injection. Using equation (1), the optimal insulin dose may be further subdivided to show the amount used for dietary or pre-meal glucose levels to the user and HCP. In one implementation, each component of the dose guideline (e.g., glucose correction portion, dietary coverage portion, and IOB) may be output and displayed to the user and / or HCP.

[0255] Basic dose

[0256] The basal dose of long-acting insulin analogs is administered 1-2 times daily, depending on the drug chosen. For example, insulin glargine has a duration of action of 24 hours and is administered once daily, while insulin detemir has a duration of action of 12 hours and is administered twice daily. Because these timescales are much longer than the 4-5 hour duration of action associated with rapid-acting analogs, different subcutaneous insulin PK distributions can be included for long-acting analogs. The time course of action is the difference between the basal dose scenario and the other two scenarios. Similar to the corrected dose, the basal dose is independent of diet but is present in all meals due to its prolonged pharmacokinetic distribution.

[0257] In one exemplary embodiment, reference Figure 9B The flowchart, starting from step 912, shows that once the DGA is prompted with basal dose guidance, it can generate multiple basal insulin dose candidates. In one implementation, multiple basal insulin dose candidates can be generated based on the subject's historical basal dose.

[0258] In step 914, the DGA can determine multiple glucose time courses corresponding to multiple basal insulin dose candidates. In one implementation, the DGA can use a physiological dose-guided algorithm to modify a minimal model of the possible dose range to calculate a subject-specific glucose time course. The amount of IOB can be calculated from the pharmacokinetics of the subject-specific insulin analog dose and can be the same across all candidate input doses.

[0259] In one implementation, the DGA can be configured to extend multiple glucose time processes relative to time to reflect the extended duration of action (e.g., 12 or 24 hours) associated with long-acting analogs. Because basal insulin is typically a once-daily medication, long-acting basal guidance can be developed to simultaneously prompt the user daily. Similar to corrected dose and dietary dose determination, the DGA can be configured to generate blood glucose profiles for basal dose candidates. Each of the multiple glucose time processes may include three dietary events, each including an associated rapid-acting dose input to simulate a shift during the 24-hour window of basal action. Each dietary event may include a carbohydrate input and a rapid-acting insulin input. In one implementation, the value of the insulin input may be based on a subject-specific median rapid-acting simulation requirement from MDD 152 data. In one implementation, the dietary input may be represented by a central trend carbohydrate amount (e.g., median or average carbohydrate amount), as described in mealtime dose guidance determination. In one implementation, the dietary events may be identical for each basal simulation event within a single dose guidance. In one implementation, the basal dose is the only variable input value.

[0260] In step 916, the DGA can calculate multiple cost function values ​​corresponding to multiple glucose time processes. In one embodiment, for each glucose time process, different time measures within a range can be calculated and used to determine a cost function value describing the risk of exceeding the target range (e.g., from about 70 mg / dL to about 180 mg / dL). In one embodiment, the cost function analysis for the multiple glucose time processes generated for multiple baseline dose candidates can be the same as described with respect to the corrected dose analysis. In another embodiment, the cost function analysis for the multiple glucose time processes generated for multiple baseline dose candidates can be different from the cost function analysis used for the corrected dose analysis.

[0261] In step 918, the DGA can determine the optimal basal insulin dose, wherein the optimal basal insulin dose has the lowest cost function value among a plurality of cost function values. In one embodiment, the DGA can be configured to determine a basal dose candidate that minimizes the out-of-range time-related cost function C as the optimal insulin dose.

[0262] In step 920, the DGA can output dosage guidance including the determined optimal basal insulin dose.

[0263] Correction factor titration

[0264] An exemplary implementation of a method for determining correction factor titration will now be described. Although physician-initiated, insulin dispensing for diabetic patients has largely been part of patient management throughout the course of the disease. Typically, these management approaches are highly experienced and rely on patients learning from trial and error protocols to improve diabetes management. For even more quantitative assays, MDIs have largely relied on simple methods, such as insulin bolus calculators, which depend on frequent blood glucose measurements and values ​​such as the insulin:carbohydrate ratio or insulin correction factor, which are difficult for patients to determine and understand from their perspective. The algorithmic subroutine described herein aims to utilize dense glucose data from CGM devices and insulin dispensing information from Bluetooth-enabled insulin pens to titrate previously learned patient-specific dispensing parameters to optimal levels during passive observation, thereby providing personalized dosing guidance with minimal user-evolving input from the user.

[0265] To provide initial dosage guidance, as described elsewhere in the instructions, DGAs can undergo a “learning phase” in which user-specific dosing parameters are determined, such as a fixed postprandial dose, target glucose, and insulin correction factor. This article combines... Figure 7 A method for the learning phase is described. When using the system, these parameter values ​​can be used as initial estimates and can be further titrated to better patient-specific values. This approach of initial learning and subsequent continuous parameter titration allows the system to respond to disease progression and external changes in lifestyle and medications that the user may experience. Because the increasingly popular insulin-based diabetes treatments increase both endogenous insulin production and insulin sensitivity, the user's correction factor cannot remain static for the entire system lifetime. This paper describes a method for titrating the user's insulin correction factor (also known as insulin sensitivity).

[0266] As used in various implementations of DGA, the user's correction factor does not need to be a single value, but rather a unique set of distinct values ​​for a given meal type (e.g., breakfast, lunch, dinner). The unit of the correction factor can be mg / dL glucose per insulin unit. Because small doses can cause a significant drop in glucose, a high correction factor signal indicates that the user is highly sensitive to insulin. Conversely, a small correction factor suggests that the user is less sensitive to insulin. An effective mealtime insulin dose should restore postprandial blood glucose to a safe range within its efficacy window. Assuming the conventions of dosing calculators, such as those used in combination with... Figures 8A to 8HThe precise correction factor should facilitate insulin dosing that addresses both dietary carbohydrate intake and pre-meal (pre-meal) glucose elevation, thereby returning glucose levels to a safe and stable target value. Therefore, if the GPA is plotted against a specific metric quantifying glucose readings over a meal-related time period (e.g., a metric quantifying meal-related glucose data or offset, such as a set time after dosing (e.g., 4 hours), or a percentage of glucose levels during the meal (e.g., 5%, 10%)), then theoretically, the best fit for these data is a line with zero slope and a y-intercept equal to the user's target glucose level. In practice, all real-time insulin doses will bring the user's blood glucose levels back to the target value or range.

[0267] A zero-slope insulin-postprandial glucose relationship represents an ideal dosing efficacy profile. It can be assumed that the DGA will have prior knowledge of the user's fixed dose, for example, through user input or transmission via MDD 152. When the dose is greater than the fixed dose, the system can then observe the relationship between insulin dose and postprandial glucose from SCD 102 or other sources. The DGA can use one or more of several methods to assess dose efficacy, including but not limited to the following:

[0268] Determine the centroid (median / mean) of postprandial glucose levels. This method uses the difference between the centroid of postprandial glucose values ​​excluding any corrected insulin dose and the centroid of postprandial glucose values ​​including the corrected insulin dose. Ideally, the two centroids should be zero or show no significant difference. A positive difference exceeding a predetermined threshold indicates that the correction factor is less than the optimal value. Similarly, a negative difference less than the predetermined threshold indicates that the correction factor is higher than the optimal value.

[0269] Apply linear fitting to the data If the fit meets certain quality metrics, DGA can be statistically tested to determine if there is a statistically significant difference between the best-fit slope and the ideal situation, where the slope is zero. If the slope is less than zero, too much insulin is dispensed. This corresponds to a correction factor that is too low and needs to be increased. If the slope is greater than zero, postprandial glucose is elevated, indicating that the insulin dose can be increased. This corresponds to a correction factor that is too high and needs to be decreased. The amount of these values ​​that should be titrated can depend on the fit slope. Steep slopes may require more aggressive titration than those close to zero. This titration method is independent of fixed-dose titration and can be performed in parallel with fixed-dose titration.

[0270] Determine the area under the postprandial glucose / insulin curveAn ideal dosing regimen will have a constant area, representing a return to constant glucose levels after all doses. Therefore, DGA can assess any variation within this area as an indicator of an incorrect correction factor. Integral methods can also be used to determine whether mealtime dose components corresponding to a fixed dietary intake are effective or require further titration. An increased integral level relative to the ideal will indicate how much the fixed dose can be increased to bring the postprandial value closer to the target level.

[0271] During a prescribed learning phase, such as 14 days based on the characteristics of the SCD or other factors, the DGA can learn the user's medication strategy and other user-specific parameters, including but not limited to the amount of dose, correction factor, and insulin duration of action (IAT) / active insulin (IOB). Titration of the correction factor may be necessary for any one or more of the following reasons: (1) insufficient correction factor, (2) changes due to treatment intervention or environmental changes. For bolus doses that only include correction (without carbohydrate counting), the DGA can analyze postprandial glucose values ​​to titrate the correction factor. For example, if the postprandial glucose peak is greater than 180 mg / dL or the difference between the postprandial peak and the preprandial value is greater than a threshold, and there is no risk of postprandial hypoglycemia, the correction factor can be titrated in each measurement cycle. For example, the correction factor can be titrated at predetermined values, such as 1-unit increments or 2-unit increments. For separate corrected doses (e.g., without a diet), when glucose levels are high, medication is dispensed to treat hyperglycemia, and the difference between the value at the time of dispensing and the glucose value at a predetermined time (e.g., four (4) hours) after dispensing) can be used to titrate the correction factor.

[0272] A mealtime insulin dose can consist of two distinct amounts: a) a portion used to cover the portion of the meal to be consumed, and b) a portion used to address pre-meal glucose levels above the target range. The meal-related portion is typically fixed (e.g., such that the portion used for breakfast is always a specific amount that can be titrated over time) or variable to match the amount of carbohydrates the patient expects to consume. In the examples below, a fixed meal portion is assumed. However, the description below can be applied to a variable portion, where the actual portion is determined by a fixed carbohydrate / insulin ratio that is itself titrable. The portion used herein to address pre-meal glucose refers to the corrected dose or corrected portion determined by a pre-meal correction factor.

[0273] Pre-meal correction factors can be used to control MDD 152. As described herein, reducing or lowering the titration correction factor is comparable to increasing the correction dose. For example, Figure 10AA method 500 for providing a pre-meal correction factor in response to analyte data is illustrated for the control of MDD 152. At 502, the method 500 may include determining the analyte pattern type for at least one TOD period by executing a GPA algorithm, the GPA algorithm receiving time-related analyte data as input from a patient-worn sensor control device during the analysis period. At 504, the method 500 may include determining the pre-meal correction factor based on the analyte pattern type and the patient's defined dispensing strategy during the analysis period by at least one processor executing the algorithm. At 506, the method 500 may include storing an indicator of the pre-meal correction factor in computer memory by at least one processor for output to at least one user or drug dispensing device. In contrast to the linear fitting implementation described above, the implementation described herein is a unified method in which fixed-dose and correction factor titration can be performed simultaneously by a technique or software function (e.g., a single logic tree).

[0274] In related aspects, Figure 8B The above-described method 410 for GPA classifies a patient’s glucose pattern at different TOD stages as high, low, or high / low. Figure 8A The related description discloses how DGA can use the output from GPA410 to provide MDI dosage guidance recommendations. Figures 8D to 8H The related description provides examples of algorithms for DGA to provide specific MDI dosage guidance recommendations based on the glucose pattern type output by GPA 410 and the patient's defined dosing strategy for the analysis period. These recommendations can be provided for automated or semi-automated control of MDD 152, or for user interface operation to guide manual dosage control.

[0275] Figure 10B Recommended methods 510 for assessing titrated dietary doses and pre-meal corrections to increase high analyte levels are shown, for example, whether or not pre-meal corrections are performed if the GPA indicates a low pattern. In related aspects, Figure 8GAn example of a recommended algorithm 474 for developing postprandial correction is shown. When executed by a DGA, algorithm 474 requires considering whether preprandial correction should be implemented if the GPA indicates a low pattern, taking into account analyte data that does not include data from TOD periods with missed bolus doses. Other low glucose pattern conditions 512 for TOD periods can also be suitable to trigger the execution of method 510. If both the dietary dose and preprandial correction are above optimal values, method 510 can output a preprandial correction factor that reduces only the dietary dose. If a low pattern exists, it may be due to a high fixed dose, a high preprandial correction, or both. In one embodiment, the fixed dose and preprandial correction can be titrated sequentially, with the fixed dose titrated first as the basis for the dose. Once the fixed dose is titrated, if a low pattern is still observed, the correction dose can be titrated. In another embodiment, the fixed dose and preprandial correction can be titrated in parallel.

[0276] If a low pattern is detected during the TOD period, then at 514, the DGA can determine whether the dietary dosage of the defined dispensing strategy includes pre-meal correction. If so, then at 534, the DGA can exclude days with missed boluses from the original dataset or pre-meal correction, for example, by including only a portion of the analyte data on days with dietary dosages and no missed boluses. At 536, the DGA can test whether enough data remains after excluding data on days with missed boluses or pre-meal correction to reach the minimum confidence level. If enough data exists, then at 410, the DGA can repeat GPA 410 on the input analyte dataset, which excludes data on days with missed boluses or pre-meal correction.

[0277] At 538, the DGA can determine whether the subsequent glucose pattern remains low; if so, at 540, only the dietary portion of the insulin dose is reduced. Otherwise, at 542, if the pattern is not low, the DGA can reduce the pre-meal correction factor, which, when implemented by the user or MDD, can result in a reduction of the corresponding pre-meal correction dose.

[0278] If at 536, the DGA determines that there is insufficient data to determine the pattern, then at 516, the DGA can include data for days with dietary doses and pre-meal corrections. Then, at 518, the DGA can retest the sufficiency of the dataset. If the data is sufficient, then at 410, the DGA can repeat the GPA for the amplified dataset. At 540, if the obtained glucose pattern is not low, the DGA can reduce the recommended dietary portion for each dose of the relevant TOD period without lowering the pre-meal correction factor. If the DGA determines the glucose pattern is low at 520, then at 526, the DGA can include only the dietary dose data. Then, at 530, the DGA can retest the sufficiency of the dataset. If the dataset is insufficient, then at 528, the DGA can reduce both the dietary dose and the correction dose. If the dataset is sufficient, the DGA can run the GPA at 410 and determine if the glucose pattern is low at 532. If the glucose pattern is low, then at 540, the DGA can reduce the recommended dietary portion for each dose of the relevant TOD period without lowering the pre-meal correction factor. If the glucose pattern is not low, at 542, GPA can reduce the pre-meal correction factor, which, when implemented by the user or MDD, can result in a reduction of the corresponding pre-meal correction dose.

[0279] Figure 10C Recommended method 550 for assessing titrated dietary doses and pre-meal correction to reduce high analyte levels is shown, for example, if GPA 410 indicates a high pattern with or without pre-meal correction. In related aspects, Figure 8H An example of algorithm 480 for generating postprandial correction recommendations is shown. When executed by DGA, if GPA indicates a high pattern, algorithm 480 requires consideration of whether preprandial correction should be implemented. Other high glucose pattern conditions 552 for TOD periods can also be suitable to trigger the execution of method 550. If neither the dietary dose nor preprandial correction is optimal, then method 550 can first increase the dietary dose. High glucose patterns can be processed sequentially such that the correction dose can be titrated after the dietary dose is titrated (when the pattern is not high and the data is uncorrected) to avoid hypoglycemia arising from increasing the fixed dose and decreasing the correction factor, which is similar to increasing the correction dose at a given blood glucose level. High patterns can be addressed first by increasing the fixed dose. If the high pattern persists and occurs only when the correction dose is included, the correction factor can be titrated.

[0280] If a high pattern is detected in a specific TOD, then at 554, the DGA can determine whether the analyte data includes a dietary dose with pre-meal correction. If pre-meal correction is not included, then at 564, the DGA can then exclude data from the original dataset for days with pre-meal correction and days with missed bolus doses, and repeat the glucose pattern analysis at 410. At 566, the DGA can determine whether the subsequent glucose pattern remains high. If not, then at 572, the DGA can add a pre-meal correction factor, which, by outputting to the UID or MDD, can result in a reduction of the pre-meal correction dose. Otherwise, if the glucose pattern is high, then at 562, the DGA can add a dietary portion recommendation for each dose in the relevant TOD period. If pre-meal correction is determined to be included at 554, then at 556, the DGA can include data for days with pre-meal doses and pre-meal corrections, and exclude data from the original dataset for days with missed bolus doses, and repeat the glucose pattern analysis at 410. At 558, the DGA can determine whether the subsequent glucose pattern remains high. If the pattern is not high, then at 562, DGA can increase the recommended dietary portion for each dose during the relevant TOD period. If the pattern is not high at 558, DGA can exclude data from days with pre-meal correction at 564, and exclude data from days with missed bolus doses, and repeat the pattern analysis and subsequent steps above at 410.

[0281] Although Figures 10B to 10C Aspects of various algorithms for determining pre-meal correction factors are illustrated, but it should be understood that these are examples. Various other algorithms may also be suitable.

[0282] Insulin dosage depends on the timing of administration (at the start of a meal, before a meal, or after a meal). Dosage guidance

[0283] An exemplary implementation of a method for determining dosage guidance based on dosing timing will now be described. To maximize glycemic control, insulin preparation should be time-synchronized with meals so that the peak of circulating insulin coincides with the postprandial glucose rise. However, patients sometimes forget to take their mealtime insulin before meals. The mismatch between the subsequent meal-related glucose rise and insulin action can lead to uncertainty about the appropriate dose. Taking the meal dose after the start of a meal, rather than exactly before, can cause hypoglycemia because active insulin can still circulate once the postprandial glucose rise is exhausted. Furthermore, patients correcting for glucose trends may tend to increase their dose due to an upward glucose trend at the start of a meal. This effect can also cause hypoglycemia. To minimize timing-related hypoglycemic episodes, the DGA can be configured to determine and consider the preparation time delay when providing dosage guidance. A positive time delay indicates an insulin dose administered after the start of a meal. A zero time delay indicates that the insulin dose is administered approximately simultaneously with the start of a meal. A negative delay indicates that the insulin dose is administered before the start of a meal.

[0284] In various implementations, the DGA can be configured to include methods for real-time detection of missed meal doses to determine whether meal dose guidance calculations should be “on-time” (i.e., zero or negative delay) or “lagging dose” (i.e., positive delay). In some implementations, the DGA can also be configured to determine and output an estimate of the meal start time that can be used for late dose calculation. In some implementations, if no mealtime insulin dose is detected within +x or -y minutes from the estimated meal start time, the DGA can be configured to notify the patient that they may have forgotten to take the dose.

[0285] Algorithm for real-time detection of missed meal dosages

[0286] In various implementations, the DGA can be configured to use a real-time dietary detection algorithm to detect missed meal doses. This document describes systems and methods for real-time detection of missed meal doses and subsequent alerting of the patient. The process for detecting missed meal doses can be performed periodically (e.g., whenever new glucose data becomes available to the system). Alternatively, the process can be performed whenever an appropriate “missed dose” alert is issued to the patient, or whenever an alarm is enabled.

[0287] In one exemplary embodiment, real-time dietary monitoring can be performed using a feature extraction module and a dietary monitoring module. The feature extraction module can receive CGM data points one at a time as they become available. When the feature extraction module detects an increase in glucose levels, it can extract multiple features and transmit these features to the dietary monitoring module for further dietary monitoring.

[0288] In one implementation, the feature extraction module is configured to perform data smoothing whenever a new glucose data point is received, by fitting the data within a time window and using a quadratic function counting backwards from the current data point. The time window may be approximately 60 minutes. The feature extraction module may be configured to store the fitted value at the center of the time window as the current smoothed data. The feature extraction module may also be configured to store the coefficients of the linear and quadratic terms of the fitted value at the center of the time window as the most recent glucose rate of change and acceleration values, respectively. In addition to being configured to store the fitted value at the center point, the feature extraction module may also be configured to store the fitted value at the most recent point used for feature extraction. The feature extraction module may be configured to compare the current smoothed glucose value with a previous smoothed glucose value (e.g., the smoothed glucose value immediately preceding the current smoothed glucose value) to determine whether the smoothed glucose data is increasing or decreasing. The feature extraction module may be configured to extract multiple features and then pass these features to the dietary detection module after the feature extraction module determines that the comparison between the current and previous smoothed glucose values ​​indicates an increase.

[0289] The feature extraction module can be configured to extract multiple features from two segments in smoothed data. These two segments can be the current rising segment and the previous falling segment. The multiple features extracted from the current rising segment can include, but are not limited to: 1) maximum acceleration, 2) time of the point of maximum acceleration, 3) glucose value of the point of maximum acceleration, 4) height calculated from the difference in glucose values ​​between the current time point (fitted value) and the point of maximum acceleration (reference point), 5) duration of the current segment calculated from the elapsed time from the reference point to the current point, 6) average rate of ascent of the current segment calculated by dividing the height by the duration, 7) maximum increase in acceleration (the increase in acceleration at a given time point is obtained by subtracting the acceleration at the previous point from the acceleration at that point), and 8) area under the curve (subtracting the glucose value at the reference point from the average glucose value, and then multiplying the difference by the duration of the segment). Multiple features extracted from the previous descent segment may include, but are not limited to: 1) duration, 2) altitude, 3) average descent rate (altitude / duration), 4) maximum descent rate (maximum absolute value of the rate of change), and 5) maximum deceleration (maximum absolute value of the acceleration). The feature extraction module can be configured to pass multiple extracted features to the dietary dosage module.

[0290] The dietary detection module can be configured to receive feature vectors as input and output a binary detection result indicating whether the current rising segment is a dietary response glucose shift. The dietary detection module can also be configured to output a probability value with the binary detection result. In one implementation, the pre-trained machine learning model in the dietary detection module can be implemented by scikit-learn using RandomForestClassifier (https: / / scikit-learn.org / stable / modules / generated / sklearn.ensemble.RandomForestClassifier.html). The dietary detection module can be configured to detect dietary starts based on tree construction rules and feature thresholds for each feature in each tree, which can be optimized during training. In one implementation, the pre-trained model can also be built based on alternative classification algorithms, including gradient boosting, Ada boosting, artificial neural networks, linear discriminant analysis, and additional trees.

[0291] The dietary detection model can also be configured to estimate the start time of the meal if it is detected. In one implementation, the start time of the meal can be estimated as the point in time when the glucose level increases most rapidly, traced back from the detection point within a time window of approximately 1.25 hours. For example, if a missed meal is detected by the algorithm at 1:15 PM, the model can trace back to approximately 12:00 PM to determine the meal start time. This is achieved by using a quadratic function, i.e., y = ax 2 The formula +bx+c fits five data points centered on the data point of interest, allowing calculation of the glucose acceleration at each point. The fitting parameter "a" represents the acceleration at the data point of interest. The increase in glucose acceleration at time point k can be defined as a(k+l)-a(k).

[0292] The dietary monitoring model can also be configured to output a notification about a missed dose to the user on UID 200 if no mealtime insulin dose is detected within a time period near the estimated start of the meal. In one implementation, if the estimated meal start time is less than 2 hours prior, the notification can also indicate that the patient can still receive dosage guidance for the meal and dose slightly later.

[0293] Details of other types of dietary detection methods and algorithms are described in U.S. Patent Publication No. 2017 / 0185748 and PCT Application Serial No. PCT / US2020 / 12134, which are incorporated herein by reference in their entirety.

[0294] Dosage instructions for administration at the start of a meal.

[0295] When the dispensing time delay is negative or zero (i.e., the dosage guideline is for administration at or before the start of a meal), the DGA can be configured to take into account the length of time that an insulin bolus administered before the start of a meal can be effective, and thus consider other factors that can alter the risk of hypoglycemia and hyperglycemia. For example, the DGA can be configured to include risk factors associated with circadian rhythms.

[0296] In one exemplary implementation, such as Figure 11 As described in the flowchart, in exemplary method 600, the DGA can provide dosage guidance to the subject at the start of a meal, for example, when the time delay is negative or zero. Starting at step 602, the DGA can determine a first dosage guideline for the meal to be administered at a time on the day the meal begins. The first dosage guideline can be a fixed dietary dose (with or without correction) or can be based on the carbohydrate content of the meal.

[0297] In step 604, the DGA can then determine whether a risk of hypoglycemia exists, at least based on the first dose guidance and the time of day when the dose is to be administered. The DGA can be configured to determine hypoglycemia risk by referencing a risk map. In one embodiment, prior user data (e.g., glucose level and insulin dose data), population data, or a combination of prior user and population data can be used to develop a risk map relating to time of day, day of week, and / or other available patterns. In one embodiment, the DGA can be configured to use the risk map to identify the edges of the distribution closest to the highest risk, rather than identifying typical behavior.

[0298] In step 606, the DGA may output a second dose instruction different from the first dose instruction. The second dose instruction may be output on UID 200 and includes a lower dosage than the first dose instruction. The second dose instruction may also be associated with a lower risk of hypoglycemia compared to the first dose instruction. For example, if the first time of day has a higher risk of hypoglycemia in its recent timeframe than another (second) time, and the user requests dose instruction at the first time of day at or before the start of a planned meal, the DGA may be configured to output a dose instruction lower than the nominal recommended dietary dose instruction (e.g., a fixed dietary dose (with or without correction) or a carbohydrate-based dietary dose).

[0299] When the dosage is started before or after a meal (i.e., the preparation time delay is positive), the DGA can be configured to determine the preparation time delay and output dosage guidance that takes into account the preparation time delay and any associated risks.

[0300] In one exemplary implementation, such as Figure 12AAs described in the flowchart, in exemplary method 607, starting from step 608, the DGA can determine a first dose guideline for the diet in response to an inquiry from the subject, wherein the first dose guideline is determined for administration to the subject at the start of the diet. In one embodiment, the DGA can be configured to assume that an initial optimal mealtime dose is entered at the start of the diet. Thus, in one embodiment, when a user opens the DGA and requests dose guideline, the algorithm can first calculate the optimal insulin dose if administration is at the start of the diet. The optimal insulin dose for the diet can be determined in various ways, including but not limited to, a fixed dietary dose based on carbohydrate content (with or without correction) or a dietary dose.

[0301] In step 610, the DGA can determine whether a time delay exists between the start of a meal and the subject's inquiry (i.e., the start of dose guidance). In one embodiment, the DGA can be configured to identify the medication dispensing time delay using multiple methods. In one embodiment, the DGA can be configured to determine an estimated meal start. In one embodiment, the DGA can be configured to use the aforementioned meal detection algorithm to identify the medication dispensing time delay, the meal detection algorithm detecting the estimated meal start, and the DGA can be configured to record the time between the start and the user-initiated dose guidance. In another embodiment, the DGA can be configured to identify a medication dispensing time delay between the start of a meal and the start of the insulin bolus dose, which can be recorded by the user. In this embodiment, the DGA can be configured to prompt the user for input regarding the medication dispensing time delay. For example, the DGA can be configured to ask the user whether the requested dose guidance is for a meal or for a corrective dose to correct for high glucose. If the requested dose guidance is for a meal dose, the DGA can be configured to ask whether eating has started, and if so, how long ago. This input can then be used as a time delay between the meal and the dose for dose guidance.

[0302] In step 612, the DGA may, in response to the determination of the time delay, determine whether a risk of hypoglycemia exists, based at least on the first dose guidance and the time delay. In one implementation, the DGA may determine whether the first dose guidance and time delay place the patient in a high-risk hypoglycemia zone or in a hypoglycemia-risk zone relative to the rest of the user population.

[0303] In one implementation, the DGA can refer to a multidimensional surface map to determine hypoglycemia risk. In one implementation, the map can be population-based. For example, the map may include observations of all DGA users until sufficient data can be collected to personalize recommendations for a specific user based on requested dose guidance. In one implementation, the DGA can be configured to receive and store multiple types of data related to the DGA user's mealtime dose. These multiple types of data related to mealtime dose may include, but are not limited to, dose-time delays, suggested or recommended dose guidance, administered dose, and glucose time series over a predetermined period after the administered dose. Glucose time series may have fixed time intervals between glucose value samples (e.g., from CGM), time intervals of variation, or combinations thereof. Postprandial glucose data in the glucose time series can be used to calculate a hypoglycemia metric for a meal / dose onset. This hypoglycemia metric may be, but is not limited to, times below approximately 70 mg / dL, times below approximately 54 mg / dL, or calculated risk factors such as the hypoglycemic index (LBGI). The hypoglycemia metric may also be normalized to the glucose concentration value at the start. Dose guidance and administered dose can be treated as variables based on the difference between them, or they can be treated separately. Multidimensional surface maps can then be derived, where the resulting measure of hypoglycemia is a function of various variables. To account for different meal sizes and circadian rhythm effects, DGA can create multiple multidimensional surface maps, including those for each of breakfast, lunch, and dinner. Once the maps are established, DGA can also determine cutoff values ​​for acceptable hypoglycemia risk. An important variable used in this system can be the meal-to-dose time delay.

[0304] The DGA can be configured to output dose guidance on UID 200 based on whether a hypoglycemic risk is determined. In step 614, if the DGA determines that the first dose guidance and time delay do not place the subject at risk of hypoglycemia, then the DGA can output the first dose guidance. In step 616, the DGA can output a second dose guidance in response to the determination of a hypoglycemic risk, wherein the second dose guidance is associated with a lower hypoglycemic risk compared to the first dose guidance. The second dose guidance can be output on UID 200. In one embodiment, if the DGA determines that there is a hypoglycemic risk for the first dose guidance at the time delay, the DGA can search for changes in the dose along the Δdose / hypoglycemic risk equipotential line of the multidimensional surface plot that reduce the risk below a predetermined cutoff value. This dose change can then be applied to the first dose guidance to provide updated dose guidance, thereby minimizing hypoglycemia. The system can continue to collect information to refine and / or update the multidimensional surface plot as a function of multiple variables, such as the dose-time delay, the suggested or recommended dose guidance, the administered dose, and the glucose time series over a predetermined period after dose administration.

[0305] In another implementation, the DGA can be configured to determine delayed dose guidance that does not take into account endogenous insulin production. A problem with using conventional dietary dose calculations when determining dose guidance after a meal is that the user's glucose levels are typically elevated even when insulin has been dispensed. If the user ingests the dietary dose after the meal has started and uses their current glucose level to determine the dietary dose, the user may inject too much insulin, potentially leading to hypoglycemia. The DGA can be configured to determine a dose guidance used for dosing after the meal has started, which mitigates the problem of elevated glucose by considering the user's glucose level at the estimated time of meal start.

[0306] In one exemplary implementation, such as Figure 12B As shown in the flowchart, in exemplary method 617, starting from step 618, the DGA can receive an inquiry for dosage guidance for a meal with a start time. In step 620, the DGA can determine whether the inquiry for dosage guidance was received after the meal started. In one embodiment, in response to an inquiry for dosage guidance from the user, the DGA can use the late dose detection algorithm described above to determine whether the dose is late. In one embodiment, the late dose detection algorithm can determine and output an estimated meal start time, which the DGA can compare with the time from which the inquiry was received from the user.

[0307] In step 622, the DGA can determine the user's glucose level in relation to the meal start time. In one implementation, the DGA can determine the glucose value associated with the meal start time (e.g., the glucose value most recent in time to the meal start time).

[0308] In step 624, the DGA can output a late dose guideline, which includes a dietary dose guideline and a corrective dose guideline. In one embodiment, the corrective dose guideline can be based on the glucose level determined at the start of the meal. The corrective dose guideline can be determined using a push calculator for correcting hyperglycemia and can include a glucose correction portion. As previously stated, with respect to equation (1), the glucose correction portion of the corrective dose guideline can be determined based on the following formula, where (BG(t)) is the current glucose value, BG target It is the target glucose.

[0309]

[0310] In one implementation, to determine the glucose correction portion for late dosing guidance, the current glucose value is a glucose value associated with the estimated meal start time. In one implementation, the correction factor may be a user's insulin sensitivity factor. In one implementation, late dosing guidance is dietary dosing guidance plus correction dosing guidance. Dietary dosing guidance can be determined in several ways. In one implementation, dietary dosing guidance may be determined based on the estimated carbohydrate content of the diet. In another implementation, dietary dosing guidance may be based on a "fixed" or "fixed + correction" dosing strategy, where the user ingests a predetermined amount of insulin for the dietary dose, regardless of the carbohydrate content of the diet.

[0311] In another implementation, the DGA can be configured to determine delayed medication refills that result in the production of endogenous insulin. A problem with using conventional dietary dosing calculations when refilling medication after a meal is that patients with type 2 diabetes typically still produce their own (endogenous) insulin. Therefore, if a user takes a delayed ingestion after a meal, the user may have additional "active insulin" produced by the user's own pancreas, and not taking this into account in dosing calculations could lead to eventual hypoglycemia. The DGA can be configured to determine dosing guidance that takes into account the additional endogenous insulin production.

[0312] In one exemplary implementation, such as Figure 12C As described in the flowchart, in exemplary method 630, starting from step 632, the DGA may determine a first dose guideline for the diet in response to an inquiry from the subject. In one embodiment, the first dose guideline is determined to administer the medication to the subject at the start of the diet. In other embodiments, the first dose guideline may be a fixed dietary dose (with or without correction) or a dietary dose determined at least based on the carbohydrate content of the diet.

[0313] In step 634, the DGA can calculate the time delay between the start of the meal and the inquiry from the subject. In one embodiment, the delayed dose detection algorithm described above can be used to calculate the time delay. In one embodiment, the delayed dose detection algorithm can estimate the detected start time of the meal and can calculate the time delay by comparing the time when the DGA receives the dose guidance inquiry from the user with the estimated start time. The estimated start time can be determined by the DGA, as described with respect to other embodiments.

[0314] In step 636, the DGA can determine a factor corresponding to the estimated amount of endogenous insulin and the amount of time delay, in response to determining that the time delay is greater than zero (>0). In one embodiment, the factor can be a fraction. In one embodiment, the factor can be determined by the amount of time between the query and the estimated meal start time, for example, from a general lookup table having small values ​​that decrease as the time delay of meal start increases.

[0315] In step 638, the DGA can output a second dose guideline, which is related to the first dose guideline and a factor. The second dose guideline can be output on UID 200. The second dose guideline can be calculated by multiplying the first dose guideline by a factor determined in step 636, which can be a fraction. According to some implementations, for example, a unit value less than the fraction can indicate endogenous insulin production up to the time of injection. In one implementation, the fraction can be based on a simulated type 2 metabolic response to a diet. While this may be unknown to each subject, adding this value to the late dose fraction can provide a dose that takes into account endogenous production to provide a modified dose that keeps the user's postprandial blood glucose within a safe range.

[0316] Dosage guidelines for complex diets (e.g., desserts)

[0317] In some cases, patients will take prandial insulin in the amount that covers the planned meal, but later decide to eat additional food as “supplements” or desserts. The usual strategy is to re-dose in a new amount, regardless of glucose levels and trends, which is often done in practice. However, there may be situations where this is not the correct course of action. For example, the patient may not have initially eaten enough to cover the original dose, so the new dose may be too large, potentially leading to a hypoglycemic episode. Conversely, if the patient takes a conservative approach and does not administer any additional insulin to cover the additional food, a hyperglycemic episode may occur. In one implementation, the DGA can provide dosing guidance for compound meals by (1) confirming that the user is continuing to extend the original meal with more food, (2) informing the user of any risk of hypoglycemia before administering the additional dose, and (3) monitoring such risks postprandially.

[0318] DGA can be configured to output an initial query to the user on UID 200 to confirm that the diet is being extended with supplemental food. It is assumed that if the user wishes to receive dosage guidance, they will first open the dosage guidance application. In one exemplary implementation, such as Figure 13As described in the flowchart, in exemplary method 660, starting from step 662, the DGA can be configured to determine whether the user's inquiry for dosage guidance occurred within the time period of the first attack, wherein the first attack includes a meal with a start time. The start time of this time period can be the meal start time determined by a real-time meal detection algorithm, and the length of this time period can also be determined by the real-time meal detection algorithm. The start time of the meal can be determined in several ways, including but not limited to the last detected meal (e.g., by a real-time meal detection algorithm), or by combining meal detection to provide the last dosage guidance and / or insulin dose.

[0319] As shown in step 664, if a user requests dosage guidance from the DGA over a period of time, the DGA may request input from the user to confirm whether the diet has been extended with additional food intake. The DGA may output a prompt or other indication requesting user feedback on UID 200. In one embodiment, the DGA may also request input from the user to confirm whether the purpose of the dosage guidance is to correct hyperglycemia. In another embodiment, the DGA may prompt the user with an explanation of the recent dosage guidance request and may optionally provide selectable options as answers. These options may include, for example, an option to make an additional request in response to an extended diet, and an option to make an additional request to correct hyperglycemia not related to additional food intake.

[0320] As shown in step 666, the DGA can determine the risk of hypoglycemia from the start time of the meal. In one embodiment, to avoid hypoglycemic episodes due to insulin dose buildup, the DGA can be configured to determine the risk of hypoglycemia by identifying a point in the user's current blood glucose offset before the meal is extended by the additional food. The DGA can also be configured to create a forward projection of the glucose level from the identified point in the current offset to determine whether the user's glucose level is still rising or falling. As a safety measure to avoid insulin buildup, the DGA can be configured not to provide dosage guidance until the glucose level reaches a local postprandial maximum. For example, if dosage guidance for a compound meal is requested while the user's glucose is still elevated, the DGA can output a notification that guidance cannot be provided at this time for safety reasons. In one embodiment, if a risk of hypoglycemia is identified, in step 668, the DGA can be configured to notify the user not to administer any dose or to take extreme care when preparing the medication. In another embodiment, if no risk of hypoglycemia exists, in step 670, the DGA can be configured to notify the user to prepare the medication for the additional food according to the HCP recommendation. Furthermore, in one implementation, if the DGA detects that an additional dose has been delivered, the DGA can be configured to recommend that the user check their glucose levels at least two hours later to ensure there is no hypoglycemia.

[0321] Dosage guidance methods for correcting doses (intake doses)

[0322] In some cases, external factors can affect the efficacy of insulin dosing, resulting in a dose that has a greater hypoglycemic effect than expected. In these cases, users can conservatively estimate their insulin dose to avoid hypoglycemia. Individuals may also conservatively take their medication if they are unsure whether they have enough insulin for a particular type or size of diet. For example, a diet type that is higher in fat and / or protein than a user typically consumes may be. To allow for conservative dosing, the DGA can be configured to provide the user with postprandial dosing guidance and an initial dietary insulin dose following a meal. In one implementation, the DGA can be configured to perform at least four functions after the initial dietary dosing guidance and administration. The DGA can be configured to confirm that the user has administered an amount less than the dose indicated in the initial dietary dosing guidance. The DGA can then be configured to identify the risk of hypoglycemia before recommending an additional dose and to inform the user of this risk. The DGA can also be configured to provide additional dosing guidance and monitor the risk of subsequent hypoglycemia for a period of time after the administration of the additional dosing guidance.

[0323] In one exemplary implementation, the user will ask the DGA for a recommended dose at or near the start of a meal. Figure 14 In the exemplary method 700, starting from step 702, the DGA may output a first dose instruction in response to a first user query. The first dose instruction may be output on the UID 200. In one implementation, the first dose instruction may be calculated to administer the medication at the start of a meal. The first dose instruction may also be a fixed-dose dietary instruction (with or without correction). Alternatively, the first dose instruction may be determined based on the carbohydrate content of the meal.

[0324] However, when administering insulin, the user can choose whether to adhere to the recommended dosage guidelines. If the administered dose differs from the recommended dose, the system will record the difference. As in step 704, the DGA can determine whether the administered first dose differs from the first dosage guidelines. The administered first dose can be lower or higher than the first dosage guidelines. In one embodiment, the administered first dose is lower than the first dosage guidelines. The DGA can be configured to record when the administered dose is lower than the recommended dosage guidelines.

[0325] In step 708, the DGA can determine whether a second user inquiry regarding the second dose instruction was received within the time period of administering the first dose. This time period can be determined by a real-time dietary monitoring algorithm, or it can be a predetermined time starting from the administration of the first dose.

[0326] In step 710, if a second user inquiry is determined to have been received within that time period, the DGA may request input from the user to determine whether the second user inquiry is for adjusting high glucose levels after a meal. The DGA may output a prompt requesting user feedback or other notification on UID 200. For example, if the DGA is subsequently asked for further dose recommendations after a conservative dose (e.g., less than the recommended dose) within a certain time period from the first dietary dose, the DGA may use the inquiry time and a record of medication mismatch to inform the user of the reason for the second dose instruction. In one embodiment, the DGA may be configured to request input to determine whether the subsequently requested dose is to cover additional mealtime food or to address the problem of high postprandial glucose. If the user requests a second dose instruction to cover additional mealtime food, the DGA may follow the process described in the "Dose Instructions for Complex Meals" section elsewhere in the specification.

[0327] If a user requests second dose guidance to correct postprandial hyperglycemia, as shown in step 712, the DGA can determine the risk of hypoglycemia by at least determining whether the user's blood glucose is elevated. In one embodiment, to avoid hypoglycemic episodes due to insulin dose accumulation, the DGA can be configured to determine the risk of hypoglycemia by determining the point of the user's current blood glucose offset. The DGA can also be configured to create a forward projection of the glucose level from the determined point of the current offset to determine whether the user's glucose level is still rising or falling. In one embodiment, as a safety measure to avoid insulin buildup, the DGA can be configured not to provide dose guidance until the glucose level reaches a local postprandial maximum. If a corrective dose recommendation is requested while the user's glucose is still elevated, the DGA can be configured to provide notification that guidance cannot be provided at this time for safety reasons. In one embodiment, the DGA can be configured to provide additional reasons for not providing dose guidance. For example, the DGA can be configured to provide a recent estimate of active insulin, which, if available, can be determined from a combination of population-based parameters and user-specific parameters.

[0328] In one implementation, once the DGA determines that the user's blood glucose is decreasing, the DGA can be configured to calculate the risk of future hypoglycemia without correcting for a "touch-up" dose. In one implementation, the DGA can be configured to calculate the hypoglycemia risk by calculating a positive projection of the current glucose level to check for possible hypoglycemic episodes. In another implementation, the DGA can be configured to analyze historical events to observe the frequency with which a given dose induces postprandial hypoglycemia for a given pre-meal glucose range. In one implementation, at step 716, if the DGA determines that the current hypoglycemia risk is above a predefined threshold, the DGA can be configured to output a recommendation on UID 200 that insulin should no longer be ingested at this time.

[0329] In step 714, if the DGA determines that the user has no current risk of hypoglycemia, the DGA can be configured to calculate and output a dose guide as if the subsequent dose were post-meal correction. Therefore, in one implementation, since the meal has been completed, the subsequent dose can be determined without taking into account any new carbohydrate intake. As discussed with respect to equation (2), the DGA can be configured to calculate the second dose (corrected dose guide) as a function of the glucose correction portion, subtracting the remaining amount of active insulin (IOB) from the initial meal dose.

[0330]

[0331] in This is the glucose correction section.

[0332] In one implementation, the DGA can be configured to calculate the dose-guided glucose correction portion as the current glucose (BG(t)) and the target glucose (BG). targetThe difference between the values ​​is divided by a correction factor. The correction factor can be the user's insulin sensitivity factor (ISF), a measure used to gauge the extent to which a single insulin unit lowers fasting blood glucose. ISF can be personalized for each user during algorithm learning. In one implementation, population-based statistics of insulin pharmacokinetics (PK) can be used to estimate IOB for the initial estimate. Rapid-acting insulin analogs typically reach peak plasma concentrations within approximately 45 minutes, after which their curve decays exponentially, but this period may be shorter if the user is ingesting an ultra-rapid insulin analog. This time window can fall within a dormant period where DGA cannot provide guidance due to elevated glucose levels. IOB can be estimated from this curve by directly measuring the exponential decay or by estimating a linear decay from peak insulin concentration to pre-meal values. The IOB value reflects the current decline in glucose levels since the initial insulin dose and can be subtracted from the glucose correction amount to minimize insulin buildup. When an additional dose is delivered, the application user interface will recommend that the user scan their glucose two hours later to ensure there is no hypoglycemia.

[0333] Postprandial hypoglycemia and hyperglycemia alert methods for dietary insulin therapy

[0334] DGA can be configured to generate and / or output alarms, or otherwise notify users in advance of predicted or possible future episodes of hypoglycemia and / or hyperglycemia. These alarms will allow users to take action to keep their blood glucose levels within the normal range, a primary goal of diabetes management.

[0335] Compared to threshold-based alerting methods, DGA's predictive alerting method can predict the probability, timing, and severity of hypoglycemic / hyperglycemic episodes. Several advantages are associated with predictive alerting methods. Predictive alerting methods overcome the problem of setting an optimal threshold, one of the most common issues with threshold-based alerts. Setting the threshold too low leads to too many false alarms, while setting it too high results in failure to alert patients in a timely manner. Furthermore, predictive alerting methods can provide patients with more specific information about an impending glycemic episode, including its probability, timing, and severity. The additional specificity provided in the alert allows patients to take more appropriate actions. Additionally, predictive alerting methods can offer patients personalized choices of appropriate actions.

[0336] In one exemplary implementation, such as Figure 15As described in the flowchart, in exemplary method 720, starting from step 722, the DGA can be configured to receive multiple data sets including time-series data of an analyte (e.g., glucose) and event data. The DGA can be configured to learn specific patterns from past data of each individual patient. The multiple data sets received by the DGA may include glucose time-series data and other event markers and associated timestamps. In one implementation, the higher-order derivatives and integrals of the glucose time-series data are also relevant inputs to the alerting system. Furthermore, the multiple data sets may include, but are not limited to, patient location data, calendar date data, TOD data, and stress level data. Event data may include, but are not limited to, dietary data, snack data, exercise data, and medication dispensing data, as well as associated timestamps for each event. Medication dispensing data may include bolus insulin doses and amounts, and / or basal insulin doses and amounts.

[0337] In step 724, the DGA can be configured to process at least a portion of multiple data sets to determine the probability, timing, and severity of future hypoglycemic or hyperglycemic episodes. In one implementation, past records for each event type can be summarized to generate a prediction of the most likely occurrence in the near future. Future glycemic episodes can be predicted by referencing TOD (Time of Day), the time of week, or correlations with other events. For example, the DGA can be configured to predict associations between exercise events prior to or following a specific meal on a specific day of the week.

[0338] DGA can be configured to predict future hypoglycemic or hyperglycemic episodes using a glucose prediction algorithm. In one implementation, the algorithm can be implemented using a natural Bayesian classifier. In another implementation, the algorithm can use a long short-term memory (LSTM) architecture combining recurrent neural networks (RNNs), random forests, or a combination of various methods. In one implementation, the machine learning model is initially trained using glucose time-series data collected from clinical studies and real-world databases to develop a population-based model. In one implementation, the population-based model can be an initial model for each patient at the starting point, and this model can learn subject-specific patterns as the algorithm is continuously trained using data from patients. Therefore, the performance of DGA can be improved with each subject used.

[0339] In step 726, the DGA can be configured to warn the patient of a predicted future hypoglycemic or hyperglycemic episode. In one embodiment, the DGA can be configured to output an alarm on the UID 200 to notify the patient, allowing them to take action in response to the alarm, such as by consuming carbohydrates to treat a future hypoglycemic episode or by ingesting insulin to treat a future hyperglycemic episode. In another embodiment, the DGA can be configured to couple the alarm system with a dispensing algorithm to suggest an appropriate dosage. In one embodiment of predicting a future hyperglycemic episode, the DGA can be configured to output an alarm on the UID 200 that includes recommended dosage guidance. In another embodiment, the DGA can be configured to provide a display on the UID 200 that the user can access when they want to determine whether a postprandial insulin dose is recommended. For example, if a high probability of future hypoglycemia is calculated, the DGA can be configured to output dosage guidance indicating that additional insulin is not recommended. The DGA can also be configured to output a recommendation on the UID 200 independent of dosage guidance. For example, the DGA can be configured to output recommendations for carbohydrate consumption, recheck glucose levels after a short period (e.g., 15 minutes), set reminders to check glucose after a preset or user-configured time, and / or enable a hypoglycemia threshold alarm. In addition to outputting alarms, the DGA can also be configured to include a display detailing future hypoglycemia / hyperglycemia episodes.

[0340] In one implementation, a specific seizure prediction can be qualified by the probability of occurrence. Different levels of probability can drive different system outputs. For example, alarm activation may require a higher probability than the postprandial dosage guidance requested by the patient.

[0341] In another implementation, the DGA can be configured to have adjustable sensitivity or specificity for its prediction method. In one implementation, different levels of sensitivity and / or specificity allow the user or the DGA to select a level of sensitivity and / or specificity suitable for the user's current level of engagement. For example, when a user needs to focus on other aspects of their life, they can choose higher specificity, so that they are only alerted to very urgent situations. In another implementation, higher sensitivity can be selected. For example, when a user decides to allocate more time to improving glucose management, a prediction system with higher sensitivity can be proactively selected to prevent emergencies. In one implementation, as with the learning module itself, sensitivity and specificity settings can initially be based on population data. As more user choices are recorded, pattern recognition can attempt to factor in the number of days in a week (e.g., a discrete 7-day week, or weekdays versus weekends), the time of day, and the potential density of activities recorded in the user's calendar to assess the most likely sensitivity and specificity setting preferences at any given time.

[0342] System characteristics

[0343] DGS 100 can incorporate systematic considerations for common events occurring during insulin-intensive diabetes management. Current insulin bolus dosage calculators do not account for these real-world events, preventing users from adjusting recommended doses based on their best judgment. While individuals with diabetes should be able to modify ideal dose recommendations to accommodate real-world occurrences, such considerations can place a significant cognitive burden on them. To reduce this burden and create a more user-friendly dispensing system, DGS 100 utilizes insulin and glucose data to appropriately adjust doses to real-world situations. These system features are further enhanced by the ability for physicians to modify dispensing parameters.

[0344] HCP rewrite dose guidance settings

[0345] DGA can provide dosing guidance based on individualized parameters such as fixed dose, target glucose, correction factor, and duration of insulin action. Users can view these different parameters in various ways. For example, users can view the parameters on UID 200 through the settings tab within the DGA, or as a viewable detail of the dose recommendation. Although users can view these parameters as part of the recommended dose, these values ​​can be strictly informative and may not be editable by the user.

[0346] Conversely, the DGA can be configured to allow user HCPs to view these values ​​as part of, for example, a clinically-oriented system web application, and can also be configured to allow HCPs to edit one or more of these parameters. The web application can display patient performance metrics through a series of patient glucose values ​​(such as glucose concentration curves) and insulin dosage statistics. If the HCP wants to modify dosage recommendations to accommodate changes in other treatment regimens, the clinician-oriented web application can allow the HCP to edit user-specific dosage parameters, including but not limited to those mentioned above.

[0347] In one exemplary implementation, such as Figure 16AAs described in the flowchart, in exemplary method 800, in step 801, the HCP may adjust at least one parameter used to provide insulin dose guidance to a subject in a dose guidance application to create a new insulin dose guidance. For example, the HCP may adjust at least one of a fixed dose, target glucose, correction factor, and duration of insulin action. Before adjusting said at least one parameter to create a new insulin dose guidance, the HCP may review glucose concentration curves and insulin statistics to determine if the subject experiences any low or high glucose patterns at any time of day, so as to notify them, if necessary, which insulin dose guidance needs adjustment. High or low glucose patterns may be determined by GPA, as described elsewhere in the specification.

[0348] After the HCP changes at least one parameter, in step 802, the new insulin dosage guidance can be communicated to the subject. Furthermore, an explanation of the changed dosage guidance can be provided to the subject. In one implementation, one or both the user and the HCP can approve any changes to the dosage guidance before the changes take effect. Parameters and dosage guidance can remain fixed for a certain period (e.g., 14 days to align with the sensor's lifespan), although any unchanged parameters can still be allowed to change as part of the algorithm's continuous learning process.

[0349] In step 804, the DGA can determine whether the subject experienced any hypoglycemic episodes for a period of time, such as 14 days, after at least one parameter was adjusted. If any hypoglycemia is observed during this period, the HCP can be immediately notified to determine whether the parameters should be restored to their values ​​before rewriting or adjustment. At the end of this period, the HCP can be asked to review the subject's performance during that period, such as the subject's glucose concentration curve and insulin statistics, and confirm whether the adjustments to the medication parameters should be maintained. If the HCP determines that the medication parameter adjustments should be maintained, then these values ​​can be used as initial conditions for titrating future doses. The history of all previous data associated with those values ​​can be negated or very lightly weighted in any future dose recommendations.

[0350] If the HCP only verbally informs the user of a medication change without updating the DGA, for example via a web application, discrepancies may exist between the output dosage instructions and the administered dosage. If such discrepancies are consistently observed over an extended period (e.g., a three-day period), the DGA can notify both the user and the HCP to inquire whether any changes in the dosage have occurred. If both the user and the HCP confirm, the algorithm can employ the strategy described above. Furthermore, the system can prompt the user to input the reason for this ongoing change.

[0351] The DGA can also include a "10%" reduction button. If selected by the HCP in association with a specific dose recommendation, the dose recommendation will be automatically reduced by 10%, for example, 1 U. For example, the HCP might want to reduce a patient's insulin dose due to a change in the dosage of a non-insulin medication. This dose change is known to make the patient more sensitive to insulin. Therefore, the patient's current insulin regimen may be too effective in lowering glucose and causing hypoglycemia. As a precaution, the HCP can reduce the insulin dose. In this example, the DGA can be titrated based on a forward adjustment value.

[0352] Detection of active glucose sensor malfunctions using insulin and glucose data

[0353] Although the glucose sensor 101 may have built-in fault detection to notify the user to remove and replace the current sensor, the DGS 100 can also be configured to provide fault detection via the total daily insulin dose.

[0354] In one exemplary implementation, such as Figure 16B As described in the flowchart, in exemplary method 807, starting from step 808, insulin dose data of the subject can be received from MDD 152.

[0355] In step 810, the DGA can determine whether multiple recommended insulin doses differ from multiple previous insulin doses administered over a period of time. Each of the multiple insulin dose guidelines and each of the multiple previous insulin doses administered are associated with a TOD (Time of Day) period, and each of the multiple insulin dose guidelines can be compared with one of the multiple previous insulin doses administered and associated with the same TOD period to determine a difference. If the dose guideline differs from the corresponding previous dose administered within the same time period, a difference is detected. If sensor 102 reads a higher glucose level compared to past sensors, the recommended dose may be similarly high due to the increased high glucose correction. Conversely, if the sensor 102 reading is significantly lower than previous sensor readings, the recommended dose can be drastically reduced to prevent hypoglycemia.

[0356] In step 811, the DGA can determine whether a new sensor has been connected to SCD 102 within the time immediately preceding the start of the period in which the difference was detected. If it is determined that a new sensor was recently connected to the SCD, then in step 812, the DGA can recommend replacing the sensor. While there are many reasons for high or low sensor glucose readings, if these abnormal glucose levels and dosage guidance begin coinciding with the placement of a new sensor and continue for a period of time until sensor 102 wears out, then a faulty sensor may be the root cause of this deviation. In this case, the DGA can notify the user to replace glucose sensor 102.

[0357] In response to changes in adjuvant therapy or new insulin formulations, medication strategies may change.

[0358] For patients with type 2 diabetes, intensive therapeutic diabetes management is often highly complex. In addition to insulin, patients frequently take various medications that supplement or synergistically improve glucose homeostasis in cases of progressive insulin loss. Changes in adjuvant therapy, such as secretagogues or growth hormone-based therapies, can affect endogenous insulin production and tissue sensitivity to insulin. Therefore, any changes in adjuvant therapy can influence the efficacy of subsequent exogenous insulin and should be considered part of any insulin dosing guidelines. A similar situation arises if a user switches to a different insulin (e.g., from rapid-acting to ultra-rapid-acting or from once-daily basal dose to twice-daily basal dose).

[0359] When the HCP changes a patient's adjuvant therapy or insulin type, a change in medication dispensing strategy can occur. In this case, as mentioned above, the HCP can provide adjustments to insulin dispensing parameters to minimize any hypoglycemic episodes. However, in some situations, the DGA may not be informed of any changes in adjuvant therapy or medication dispensing recommendations. In such cases, the system can monitor trends in insulin dose-related differences over a period of time.

[0360] In one exemplary implementation, such as Figure 16C As described in the flowchart, in exemplary method 813, starting from step 814, insulin dose data of the subject can be received from MDD 152.

[0361] In step 816, the DGA may detect a trend of difference associated with multiple insulin doses administered within a first time period of the day. This trend may include, but is not limited to, differences between the recommended dose and the administered dose, and differences in efficacy of a given dose (in terms of the magnitude or lifespan of the response) relative to a previous administration at that dose value. If any trend of difference is observed over a period of time, then in step 818, the DGA may provide notification of the trend of difference to the user, HCP, or both. The time period may be approximately 2 days, alternatively approximately 3 days, or alternatively approximately 4 days. In one implementation, the DGA may also output a prompt on the UID 200 to the HCP and / or user to confirm treatment changes that explain the trend of difference.

[0362] When the trend of difference is that the administered dose consistently differs from the recommended dose from the DGA, the HCP can rewrite the user's medication parameters, as discussed in the "HCP Rewriting Dosage Guidance Settings" section above. When the trend of difference is that the recommended dose shows a significant and consistent difference in postprandial glucose control compared to previous dosing, insulin efficacy may have changed. This change in insulin efficacy can be in magnitude (indicating a change in adjuvant therapy) or in duration (indicating a change in insulin analogs).

[0363] In the event of a change in adjunctive medication, once the change is confirmed by the user and HCP, the DGA can be configured to enter a conservative mode, where the dosage recommendation can be a small fraction of the previous dosage guidance. The DGA can then titrate the dispensing parameters and amounts to optimize for these new conditions. In the event of a change in insulin type (e.g., from rapid-acting to ultra-rapid-acting), the magnitude of the response should not change, as the market difference between the two lies in the rapid onset / shift of the drug. Instead, the duration of the response can change. To offset this, population-based values ​​for the new insulin type can be used to estimate the duration of insulin action until the system can determine a new individualized value for it.

[0364] User intake differs from recommended dosage guidelines.

[0365] While the DGA can recommend doses to users, the DGS 100 cannot guarantee strict adherence to dosage guidelines. The DGA can record any discrepancies between the administered dose and the recommended dose and detect trends in these discrepancies. Observed persistent trends are discussed in the sections “HCP Rewriting Dosage Guidance Settings” and “Changes in Dosing Strategies in Response to Adjunctive Therapy or New Insulin Types” above.

[0366] In one exemplary implementation, such as Figure 16D As described in the flowchart, in exemplary method 820, starting from step 822, the DGA can detect differences related to the insulin dose administered to the subject during the first time period compared to the dose guidance provided for the first time period. The DGA can compare the dose guidance with insulin dose data received from MDD 152 to determine whether a dose different from that indicated in the dose guidance was administered to the user. The dose administered may be a different amount of insulin ingested or a different type of insulin ingested compared to the provided dose guidance.

[0367] A specific instance of ingesting a dose different from the dosage guidance occurs when a user ingests the correct amount of the wrong insulin (e.g., injecting a long-acting instead of a rapid-acting, or vice versa). For any medication administration scenario, the DGA can calculate the optimal dose for a given insulin analog type; that is, rapid-acting for mealtime and corrected doses, and long-acting for basal doses. To avoid any errors, the DGA can output the insulin type and amount during dosage guidance on the UID 200. The MDD 152 can record and transmit the amount and type of insulin administered. The DGA can collect information about which insulin type was used for injection and detect any discrepancies. When a discrepancy is detected, a notification can be made to the user on the UID 200. This notification can instruct the DGA to detect the use of a different insulin type and require user confirmation.

[0368] In the event that the user has been given the wrong type of insulin, an insulin mismatch notification can be sent to UID 200 to alert the user of the possibility of a severe hypoglycemic episode even before a hypoglycemic alarm is triggered, thus prompting the user to respond as soon as possible. For example, if a rapid-acting dose is substituted for a long-acting dose, severe hypoglycemia may be imminent. The reasoning is threefold: (1) as the name suggests, rapid-acting insulin has a more pronounced effect than long-acting insulin shortly after administration; (2) a once-daily long-acting dose can be much larger than a single postprandial rapid-acting dose; and (3) depending on the timing of the last pre-meal dose, an accidental rapid-acting bolus can lead to insulin buildup. If a long-acting dose is substituted for a rapid-acting mealtime dose, the outcome may be even more unpredictable. Assuming a slower pharmacokinetics and distribution of action compared to rapid-acting insulin, postprandial blood glucose levels may rise rapidly after injection following an accidental administration of long-acting insulin. Depending on the timing of the previous long-acting dose, there is a possibility of insulin buildup and subsequent hypoglycemia, specifically when the current dose reaches its maximum plasma concentration approximately 6 hours after injection. The user can still be informed of the possibility of severe hypoglycemia in a notification on UID 200. In either case, the DGA cannot make any recommendations regarding insulin dosage until the injected insulin activity value is close to zero (e.g., approximately 5-6 hours for rapid-acting insulin and approximately 12-24 hours for long-acting insulin). Glucose data during these periods can also be flagged by the system and not used for further parameter refinement and dose titration.

[0369] In step 824, the DGA can detect any hypoglycemic episodes associated with the insulin dose administered in the first time period. For example, a hypoglycemic episode can be identified when glucose levels drop below 70 mg / dL.

[0370] In step 826, the DGA may notify the subject and / or HCP of any hypoglycemic episodes associated with the insulin dose administered during the first time period. The notification may be output to UID 200. When the DGA observes that a user is experiencing hypoglycemia due to a sustained intake of insulin exceeding the recommended dose, the DGA may output a predictive hypoglycemic alert to UID 200 to notify the user a period of time before the actual hypoglycemic event itself, thereby mitigating the associated side effects at that moment. However, if hypoglycemia occurs due to a sustained (e.g., defined as three independent episodes, or defined as four independent episodes) dispensing exceeding the recommended value, a notification may be sent to the user via the system telephone application and to the HCP via a clinically-oriented web application to warn of both trends.

[0371] DGA can note any differences that do not appear to be part of a broader trend, and a given dose can be incorporated into the system to continuously learn an individual's dosing patterns. As with any dose administration, postprandial measurements can be noted. Whenever a different dose is associated with postprandial hypoglycemia (e.g., less than 70 mg / dL) or any other negative outcome, the dose can be flagged for both the user and the HCP to help them each develop improved dosing methods. This flag can take the form of a note in the HCP's glucose pattern report or a note in the user's DGA insulin log.

[0372] How the system is responsible for a series of doses

[0373] To ensure accurate dose recording, the DGA should be able to correctly interpret instances of multiple insulin injections occurring within a short period. These instances may include priming (potentially multiple times) before the actual dose, and multiple injections recommended for a given dose.

[0374] In one exemplary implementation, such as Figure 16E As described in the flowchart, in exemplary method 827, starting from step 828, insulin dose data of the subject can be received from MDD 152.

[0375] In step 830, the DGA can detect the administration of multiple insulin doses. The multiple insulin doses include at least a first dose and a last dose, wherein the last dose is administered within the time period of the first dose. If multiple doses are administered in a short period, at least one of the administered doses can be a trigger dose. An optimal method of operation in insulin delivery is to use a new needle for each injection and trigger each new needle with insulin before the dose is administered. This is achieved by dispensing a small volume (typically 2U) of insulin into the air until insulin is visible at the needle tip, indicating that the needle is full of injection fluid. To distinguish the trigger volume, the DGA can assume that: (1) the trigger dose is typically much smaller than the actual dose, (2) the trigger dose is the same amount each time, and (3) the time elapsed between the trigger volume and the dose is small. Alternatively, in the case of multiple doses administered in a short period, the multiple doses can actually be part of separate doses due to the large volume that must be injected. For example, a user's MDD 152 may have less insulin remaining than the requested dose, thus requiring a cartridge change and subsequent additional injections.

[0376] In step 832, if the first criterion is met, the DGA may record the last dose as the administered dose. The first criterion is met if the last dose is administered within approximately 1 minute of the first dose, or alternatively within approximately 2 minutes, or alternatively within approximately 3 minutes. In this case, the DGA may consider the first dose and any subsequent intermediate doses (except the last dose) as priming doses, and therefore not as the actual administered doses. Thus, the DGA can be configured to record only the last dose as the administered dose. Because priming doses typically involve small volumes of insulin, in another implementation, the first condition may be met if the first dose is significantly smaller than the last dose. For example, the first condition may be met if the first dose is approximately one-tenth, or alternatively approximately one-fifth, or alternatively approximately one-quarter, or alternatively approximately one-third of the last dose.

[0377] Alternatively, in step 833, if the second criterion is met, the DGA may record the amount of the first and last dose combination as the amount of the dose administered. The second criterion can be met when a single dose is delivered via multiple smaller injections over a period longer than the triggering period (e.g., 1–3 minutes, e.g., about 4–35 minutes, alternatively about 5–30 minutes). For example, a user's MDD 152 may have less insulin remaining than the requested dose, thus requiring cartridge replacement and subsequent additional injections. In this case, the DGA may not interpret these injections as separate events, but rather as two actions within the same dispensing event. For this purpose, the DGA may use a waiting time in the dispensing record, recording the insulin dose value only approximately 30 minutes after the first dose is administered. For example, if the DGA recommends a 10U mealtime dose, but only 4U remains in the insulin cartridge, there will be two separate injections: one for the initial 4U and another for the remaining 6U. If a new needle is used for the second injection, an intermediate triggering dose will also occur. When the initial 4U dose is administered, the DGS 100 can continue searching for the remaining 6U dose. If the 6U dose arrives within 30 minutes of the first dose, the DGA can treat it as a 10U dose. In cases where one or more trigger doses are detected between actual doses, the DGA can consider all doses between the first and last doses as trigger doses and therefore not included in the final dose. When communicating with the MDD 152, the DGA can be configured to input the remaining insulin cartridge volume along with insulin dose information. By knowing the amount of insulin remaining in the cartridge, the DGA can be configured to predict split doses due to cartridge switching and even notify the user of the remaining insulin volume.

[0378] In another implementation, the DGA can be configured to treat the priming dose and subsequent doses as separate doses and add the two doses together to obtain a single mealtime value that includes the priming dose. If the priming dose is much smaller than the mealtime dose, the effect of the priming dose on the dose-guided algorithm titration is negligible. This would resemble the logic for separate doses described below. While not precise, it is reasonable to assume that for patients with type 2 diabetes and increased insulin resistance, the priming volume is much smaller than the injection volume.

[0379] Titration with glucose data from missed doses

[0380] To provide optimal medication recommendations for the underlying disease, DGA can continuously improve its estimates of user-specific medication parameters. Therefore, it is necessary to identify appropriate data streams upon which the algorithm learns. To avoid confounding results, DGA can be configured to learn solely based on insulin and glucose data aligned with the user's own clinically recommended medication strategy. These strategies include, but are not limited to: basal dose only, basal dose plus one prandial rapid-acting insulin dose, basal dose plus two prandial rapid-acting insulin doses, and multiple intraday injection strategies consisting of basal dose plus three prandial rapid-acting insulin doses.

[0381] In one exemplary implementation, such as Figure 16F As described in the flowchart, in exemplary method 833, starting from step 834, insulin dose data of the subject can be received from MDD 152.

[0382] In step 836, the DGA can detect missed insulin doses, where the insulin dose has a time interval related to the duration of action. For example, the DGA can first identify the user's medication strategy during initial learning before providing dosage guidance. Using automated meal detection methods and data from a Bluetooth-connected MDD 152, the system can also identify meal events and their associated doses. Therefore, the DGA can determine whether a dose was missed for a given meal. Missed basal doses can be detected when there is no reported data from the user's long-acting MDD 152.

[0383] In step 838, the DGA may ignore glucose analyte data associated with the time period for determining adjustments to insulin dosage guidance. Compared to the past, missing a mealtime dose can result in elevated blood glucose and an insulin bolus. These altered mealtime doses and the glucose levels following the missed meal will deviate the current dose titration determined by the DGA for a given dosing strategy. Similarly, missing a basal dose can result in sustained glucose elevation during insulin action, typically assumed to be one day. Therefore, the DGA algorithm may include only glucose and insulin data from the meal accompanying the insulin dose. For rapid-acting insulin, the duration of action can be approximately 4 hours, alternatively approximately 5 hours, alternatively approximately 6 hours, or alternatively approximately 4 hours to approximately 6 hours. For example, if a breakfast dose is missed, glucose data obtained 4 hours after the start of breakfast may not be included in the dose titration. For long-acting insulin, the duration of action can be approximately 18 hours, alternatively approximately 20 hours, alternatively approximately 24 hours, or alternatively approximately 20 hours to approximately 24 hours. Because long-acting insulin helps maintain normal blood glucose levels between meals and prevents diabetic ketoacidosis, the system algorithm may not include any missed basal dose data during the duration of action. For example, glargine insulin has a reported 24-hour duration of insulin action. If a user taking glargine misses their daily basal dose, all data from the subsequent 24 hours will not be used by the system for dose titration.

[0384] Users are encouraged not to miss a dose.

[0385] Adherence to well-tied insulin dosing regimens can improve diabetes management by reducing hyperglycemia associated with missed doses and hypoglycemia caused by overcompensation correction. DGA can be configured to provide users with actionable, easily interpretable data highlighting the positive impact of dose adherence. One such approach is to provide periodic updates that compare glucose levels or another relevant statistic within a range of times with missed doses to those without.

[0386] In one exemplary implementation, such as Figure 16G As described in the flowchart, in exemplary method 840, starting from step 841, insulin dose data of the subject can be received from MDD 152.

[0387] In step 842, the DGA can detect missed insulin doses. In step 846, the DGA can determine the amount of time a subject's glucose level remains within a target range (TIR) ​​for a first time period and a second time period. The target range can be set by the DGA, the user, or the HCP. The target range can be between about 70 mg / dL and about 180 mg / dL, alternatively between about 70 mg / dL and about 190 mg / dL, or alternatively between about 70 mg / dL and about 200 mg / dL. The first and second time periods can be the same amount of time, wherein the first time period does not include missed insulin doses and the second time period does include missed insulin doses. Therefore, a comparison of TIRs including and excluding missed doses can be prepared.

[0388] In step 848, the DGA can inform the user of the TIR determined in the first and second time periods. If the difference between the two determined TIRs (i.e., the TIR of the first time period minus the TIR of the second time period) is greater than a threshold amount, a positive message can be displayed to the user on UID 200 to encourage good medication behavior and educate the user about the benefits of dose adherence. One example is informing the user that taking all doses for a day results in a 10% increase in time within the range compared to a day of missed meal doses.

[0389] Safety features

[0390] The exemplary implementation of the safety features of the DGS 100 described herein prioritizes user safety during guidance and titration. Current insulin bolus dose calculators utilize static reconstitution parameters to calculate dosage recommendations. Any updates to these values ​​by the user or HCP are an attempt, and in some cases, an erroneous process can induce hypoglycemia. As an automated process, the DGA can titrate these parameters using the user's insulin and glucose data, providing dosage guidance tailored to the user and adaptable to their needs. Safety measures can be incorporated to ensure that the recommended titration does not result in post-reconstitution hypoglycemia.

[0391] Safety Titration Method

[0392] The titration logic for DGA can be configured so that dose titration does not increase until the hypoglycemic period of the day is relieved. When HCP is titrated manually, the HCP may want to increase some doses and decrease others to complete the titration as quickly as possible, thus reducing the time taken. However, since the HCP time is not involved, automated systems may titrate less aggressively, i.e., take longer (and be more conservative and safer). Therefore, when titration begins, the hypoglycemic patient may initially experience a higher mean glucose, but once the hypoglycemia subsides, the insulin dose can be safely increased to achieve the glucose-lowering target.

[0393] One problem with detecting high glucose patterns after meals is that the previous meal may have postprandial glucose, causing the next meal to start with high glucose. If this happens, the high starting glucose may incorrectly indicate a high pattern for the next meal. To address this, DGA titration strategies, if necessary, can include titrating the overnight dose first before titrating the meal dose. Furthermore, meal doses can be titrated sequentially in the order of the earliest meal with a high glucose pattern before sequentially titrating meals later in the day with a high glucose pattern. For example, it is recommended to titrate the insulin dose associated with the overnight time first. Next, for any high glucose pattern detected at any postprandial time, a dose recommendation associated with lunch can be provided before providing a dose recommendation associated with dinner, and a dose recommendation associated with breakfast can be provided before that. Titrating the previous meal first ensures that preprandial high glucose for the next meal is minimized, reducing the chance that the recommended titration will affect or interfere with the titration of the next meal.

[0394] In the implementation described herein, the DGA can detect high and low modes, as described in relation to the GPA elsewhere in the specification.

[0395] For the embodiments described herein, the DGA can be configured to recommend changes (e.g., increasing or decreasing) in the insulin dose. The recommended amount of change can be any desired amount of insulin, such as a fraction of an insulin unit (0.1 unit, alternatively 0.5 units), alternatively a single unit (1.0 unit), alternatively two or more insulin units (2.0 or more), or any combination thereof. For ease of description, the embodiments described herein will refer to adjustments in one-unit intervals.

[0396] DGA can perform the various steps described in the safety titration implementation in a variety of different ways. For example, the steps can be performed before each meal, alternatively at the beginning of the day, alternatively at the end of the day, alternatively daily, alternatively every other day, alternatively every two days, alternatively whenever the user asks the DGA for a dose recommendation, or a combination thereof.

[0397] In one exemplary implementation, the DGA can access measured glucose data (e.g., from SCD 102). The DGA can determine whether a high glucose pattern exists during the overnight period. If a high pattern is detected during the overnight period, the DGA can modify the dosing guidance. For example, if doing so is safe and does not cause any low glucose pattern at any time of day, the DGA can increase the amount of drug in the dosing guidance of the basal dose. The DGA can then determine whether a high glucose pattern exists in at least one postprandial period of the day. If a high pattern is detected, the DGA can increase the amount of drug in the dosing guidance associated with the earliest of the at least one postprandial period of the day. Thereafter, the DGA can be configured such that the DGA can increase the amount of drug in a dosing guidance associated with the next earliest period of the at least one postprandial period of the day in which a high glucose pattern is determined.

[0398] In another exemplary implementation, such as Figure 17A As described in the flowchart, in exemplary method 850, in step 852, the DGA may output a first dose instruction in response to the detection of a low glucose pattern in the subject's analyte data during a first time period of at least one day, wherein the first dose instruction is less than the previous dose during said at least the first time period of the day. The first dose instruction may be output to UID 200. If a low glucose pattern is detected, the DGA may output a dose instruction associated with the time period in which the low glucose pattern was detected. The DGA may be configured such that it does not recommend reducing the insulin dose to address the high glucose pattern until the low glucose pattern is no longer detected.

[0399] In step 856, the DGA may output a second dose guideline in response to the detection of a hyperglycemic pattern in the subject's analyte data during the overnight period, wherein the second dose guideline is less than the previous dose during the overnight period. For example, the DGA may recommend a basal dose increase if the basal dose increase is safe, such as if such an increase would not lead to a hypoglycemic pattern at another time of day.

[0400] In step 860, the DGA may output a third dose instruction in response to the detection of a hyperglycemic pattern in at least one postprandial time period in the subject's analyte data, wherein the third dose instruction is less than the previous dose in at least one postprandial time period. In one embodiment, the DGA may be configured to detect all hyperglycemic patterns found at any time of day. In another embodiment, the DGA may be configured to detect whether a hyperglycemic pattern exists in the post-breakfast time period, then the DGA may be configured to detect whether a hyperglycemic pattern exists in the post-lunch time period, and then the DGA may be configured to detect whether a hyperglycemic pattern exists in the post-dinner time period.

[0401] If a hyperglycemic pattern is identified in more than one postprandial period, the third dose guidance from the DGA can be associated with the earliest postprandial period with the hyperglycemic pattern that occurred that day. For example, if a hyperglycemic pattern is detected after both breakfast and lunch, the DGA can increase the recommended insulin dose associated with breakfast before recommending an increase in the recommended insulin dose associated with lunch. Furthermore, the DGA can reassess whether the hyperglycemic pattern in the postprandial period has been alleviated before recommending an increase in the recommended insulin dose associated with lunch.

[0402] In another exemplary implementation, such as Figure 17B As described in the flowchart, in exemplary method 862, starting from step 864, the DGA can detect a low glucose pattern at any time of day. As described elsewhere, the low pattern can be detected based on GPA. If a low glucose pattern is detected, then in step 866, the DGA can output a first dose guideline for the time of day in which the low glucose pattern was detected. The first dose guideline may include a lower amount of medication compared to a previous dose administered during the time of day in which the low glucose pattern was detected. The DGA can be configured not to recommend any increase in insulin dose to resolve the high glucose pattern until the low glucose pattern is no longer detected at any time.

[0403] In step 868, the DGA can detect the presence of a high glucose pattern during the overnight period. If a high glucose pattern is detected, then in step 870, the DGA can output a second dose guideline for the basal dose. For example, if the DGA has determined that such a change is safe, e.g., that such an increase will not cause a low glucose pattern at another time of day, the second dose guideline may include a higher dose of medication than the existing basal dose. In one embodiment, the second dose guideline may include a higher dose of medication than the basal dose administered the previous day.

[0404] In step 872, the DGA can detect the presence of a high glucose pattern in the post-breakfast period. If a high glucose pattern is detected, then in step 874, the DGA can output a third dose guideline. The third dose guideline can increase the recommended insulin dose associated with breakfast; that is, the third dose guideline can include a higher amount of medication than the previous post-breakfast dose. In one embodiment, the third dose guideline can include a higher amount of medication than the post-breakfast dose administered the previous day.

[0405] In step 876, the DGA can detect the presence of a high glucose pattern during the post-lunch period. If a high glucose pattern is detected, then in step 878, the DGA can output a fourth dose guideline. The fourth dose guideline can increase the recommended insulin dose associated with lunch; that is, the fourth dose guideline can include a higher amount of medication than the previous post-lunch dose. In one embodiment, the fourth dose guideline can include a higher amount of medication than the post-lunch dose administered the previous day.

[0406] In step 880, the DGA can detect the presence of a high-glucose pattern in the post-dinner period. If a high-glucose pattern is detected, in step 882, the DGA can output a fifth dose guideline, which increases the recommended insulin dose associated with dinner; specifically, the fifth dose guideline may include a higher amount of medication than the previous post-dinner dose. In one embodiment, the fifth dose guideline may include a higher amount of medication than the post-dinner dose administered the previous day. In one embodiment, if doing so is safe, for example, if it can be increased without causing a low-glucose pattern in the overnight period, the DGA may only increase the recommended insulin dose associated with dinner.

[0407] In another exemplary implementation, such as Figure 17C As described in the flowchart, in exemplary method 883, in step 885, DGA can detect the presence of a high glucose pattern during the overnight period, as explained with reference to GPA.

[0408] If a high glucose pattern is detected during the overnight period, then in step 886, the DGA can determine whether an increase in the basal dose would induce a low glucose pattern at any time of day. If an increase in the basal dose would induce a low glucose pattern at any time of day, then in step 887, the DGA can output a first dose guideline for the time period of the day identified as having a low glucose pattern. The first dose guideline can reduce the recommended insulin dose associated with the time period identified as having a low glucose pattern; that is, the first dose guideline can include a lower amount of drug than the previous dose administered at the same time of day. In one embodiment, the first dose guideline can include a lower amount of drug than the dose administered the previous day at the same time of day.

[0409] If the increase in the basal dose is not determined to cause a hypoglycemic pattern at any time period, then in step 888, the DGA may output a second dose guidance. The second dose guidance may increase the recommended basal dose to address the hyperglycemic pattern during the overnight period; that is, the second dose guidance may include a higher amount of medication than the previous basal dose. In one embodiment, the second dose guidance may include a higher amount of medication than the basal dose administered the previous day. The DGA may then be configured such that, after the DGA outputs the first dose guidance, this reduces the recommended insulin dose associated with the time period of the day associated with the time period determined to have a hypoglycemic pattern in step 887, and then in step 888, the DGA may perform the next titration iteration, and then in step 885, determine whether a hyperglycemic pattern exists during the overnight period.

[0410] In another exemplary implementation, such as Figure 17D As described in the flowchart, in exemplary method 890, starting from step 891, the DGA can detect the presence of a hyperglycemic pattern during the post-dinner period. If a hyperglycemic pattern is detected during the post-dinner period, then in step 892, the DGA can determine whether the insulin dose associated with the dinner can be safely increased. For example, if the insulin dose associated with the dinner would cause a hypoglycemic pattern during the overnight period, then it is not safe to increase the dinner dose.

[0411] If it is safe to increase the insulin dose associated with dinner, in step 893, the DGA can output a first dose guideline. The first dose guideline can increase the recommended insulin dose associated with dinner to address the high glucose pattern in the post-dinner period; that is, the first dose guideline can include a higher amount of medication than the previous post-dinner dose. In one embodiment, the first dose guideline can include a higher amount of medication than the post-dinner dose administered the previous day.

[0412] If it is not safe to increase the dinner dose, in step 894a, the DGA may output a second dose guideline. The second dose guideline may reduce or lower the recommended basal insulin dose; that is, the second dose guideline may include a lower amount of medication than the previous basal dose. In one embodiment, the second dose guideline may include a lower amount of medication than the basal dose administered the previous day. The DGA may also be configured such that, after reducing the recommended basal dose in step 894a, the DGA may then perform the next titration iteration in step 894b, and then determine in step 891 whether a hyperglycemic pattern exists in the post-dinner period.

[0413] MDD connectivity issues

[0414] Accurate dosage guidance requires DGAs to have access to the most recent glucose analyte data and insulin dosage data. Missing data in either area can lead to inaccurate recommendations, potentially resulting in severe hypoglycemia.

[0415] In one exemplary implementation, such as Figure 17E As shown, in exemplary method 895, starting from step 896, the DGA may receive or otherwise access the subject's insulin data (e.g., from MDD 152). For example, the DGA may check recent insulin delivery information by requesting delivery information from various sources, including but not limited to MDD 152, MDD-related applications, or interfaces that store recent insulin delivery information (e.g., MDD application web servers), or by checking the memory of various applications to obtain recent insulin delivery information.

[0416] In step 897, the DGA can determine whether data associated with the last dose administered to the subject has been received. This step is particularly applicable to implementations where the device or software responsible for recording dose administration differs from the DGA or the device performing the DGA. In implementations where the DGA is automatically provided with dose administration data (e.g., the DGA is performed by the MDD 152), this step may not be applicable.

[0417] The DGA can determine whether it has recently available data based on various factors. In one implementation, this determination can be based on time intervals in the received insulin dose data. For example, if the user is taking a full multiple-injection regimen daily (basal + 3 mealtime boluses), the DGA can communicate with the MDD 152 or its associated application at least once every 6 hours. If no communication is received at that time, the DGA can be configured to determine that the MDD 152 needs to connect to the DGA before dosing guidance can be given. In one implementation, if the time interval since the last dose was received is longer than the assumed time between meals, the DGA can assume that no data associated with the last dose has been received. For example, the assumed time between meals could be approximately 5 hours, alternatively approximately 6 hours, alternatively approximately 6.5 hours, alternatively approximately 7 hours, alternatively approximately 7.5 hours, or alternatively approximately 8 hours. In another implementation, the DGA can detect whether Bluetooth communication is enabled between the user's display device 120 (e.g., a smartphone) and the MDD 152. For example, Bluetooth communication may not be enabled on device 120, the MDD 152, or both. In another implementation, the DGA can detect whether the power supply associated with the drug delivery device needs to be replaced.

[0418] If it is determined that no data associated with the last dose administered to the subject has been received, then in step 898, the DGA may notify the user that dosage guidance cannot be provided. In one embodiment, the UID 200 may display a message to the user indicating that dosage guidance cannot be provided until the DGA receives the most recent insulin delivery information. In another embodiment, the DGA may also generate, and the UID 200 may display, a prompt to notify the user to turn on Bluetooth on the display device 120, the MDD 152, or both. In yet another embodiment, the DGA may also indicate that the battery of the MDD 152 needs to be replaced. Furthermore, the DGA may determine the remaining battery life in the MDD 152 and, when the battery life falls below a certain threshold, such as less than 10%, output a warning that can be displayed on the UID 200 to the patient. In another implementation, the DGA may also output a notification displayed on the UID 200 to inform the user of the last recorded insulin dose and timestamp, and may also serve as a warning to the user that any dosage guidance will not be based on any dose that may follow the last dose and timestamp.

[0419] If data relating to the final dose administered to the subject is received, in step 899, the DGA may output dosing guidance to the UID200 based on the received glucose analyte data and insulin dose data.

[0420] Recommendations for additional testing if insulin delivery is abnormal.

[0421] DGA can be configured to run statistics corresponding to various measurements of insulin administration and the measured glucose levels. Correlation between various insulin and glucose measures can be used to identify anomalies in the DGA, which may include, but are not limited to, incorrect recording of insulin doses, potentially low or high glucose readings, and decreased or increased insulin resistance.

[0422] The insulin measure used can be a rolling insulin measure. In one embodiment, the rolling insulin measure can be the total insulin dose over a period of time. The period can be approximately 24, approximately 48, or approximately 72 hours. Furthermore, the total insulin dose can be the total dose of both long-acting and rapid-acting insulin over that period. In another embodiment, the rolling insulin measure can be the active insulin over the time elapsed since the start of the meal. Such a measure can have different predetermined insulin-related parameters for different types of mealtime insulin, such as the DIA or duration of insulin action.

[0423] The glucose metric used can be a rolling glucose metric. In one embodiment, the rolling glucose metric can be a rolling average glucose, a rolling median glucose, or a rolling pattern glucose. In another embodiment, the rolling glucose metric can be a dietary start-normalized glucose AUC or dietary glucose variation (dietary δ).

[0424] In another exemplary implementation, such as Figure 17F As described in the flowchart, in exemplary method 849, in step 851, the DGA can determine a first rolling insulin measure associated with a first time based on insulin dose data. In step 853, the DGA can determine a first rolling glucose measure associated with a first time based on glucose dose data. The first rolling insulin measure and the first rolling glucose measure can be correlated together to form a first complementary pair.

[0425] Many different complementary pairs can be formed for rolling insulin measures and rolling glucose measures. For example, in one embodiment, the rolling insulin measure of the total insulin dose over a rolling time period can be paired with one of the rolling mean glucose, rolling median glucose, or rolling pattern glucose over the same or similar time window. In another embodiment, complementary pairs (such as the glucose pattern for rolling the last 48 hours and the total insulin dose delivered during the rolling last 48 hours) can also follow the same general procedure described above. In another embodiment, the IOB (Intake of Bounds) of a specific elapsed time since the start of a meal can be paired with one of the glucose AUC (Average Usage Value) normalized at the start of the meal or the dietary δ (Dietary Scale Value), for example, the glucose variability caused by the meal. The time range for IOB and AUC pairings can include, but is not limited to, approximately 60 minutes postprandial, alternatively approximately 120 minutes, alternatively approximately 150 minutes, or alternatively approximately 300 minutes.

[0426] In step 855, the DGA can refer to a data space comprising a first region, a second region, and a third region to determine which of the three regions contains the first complementary pair. The data space can be defined by multiple complementary pairs, each including a rolling insulin measure and a rolling glucose measure associated with the same time. For multiple complementary pairs, the paired values ​​can be collected at regular time intervals, such as approximately every 2 hours, alternatively approximately every 6 hours, alternatively approximately every 12 hours, alternatively approximately every 24 hours, or alternatively other intervals balanced between appropriate data density and minimum data storage requirements. In some embodiments, the number of stored complementary pairs can be maintained in a first-in, first-out (FIFO) buffer implemented in software or hardware. Among the complementary pairs in the FIFO buffer, correlations can be made between complementary pairs, similar to fitting a curve to a scatter plot of insulin and glucose measure pairs collected over time. The correlation can be a curve with a predetermined structure (e.g., a third-order polynomial) having one or more parameters potentially determined based on paired data in the FIFO buffer, and one or more parameters predetermined based on prior population data. This curve can be described as the nominal expected relationship between observations. In addition to the nominal expected relationship, two safety boundary curves with predetermined structures can be constructed above and below the nominal expected relationship curve. Some parameters can be determined based on paired data in the FIFO buffer, while other parameters can be predetermined based on prior population data. To improve numerical stability, the nominal curve, upper curve, and lower curve can be correlated with parameter fits for different time ranges using prior rules. To cover the insulin-glucose balance check, any other cases can be interpolated from the existing time range. The nominal expected relationship curve and the two safety boundary curves can form three regions in the data space. The first region is defined as the area between the upper and lower safety boundary curves and includes the nominal expected relationship curve. The second region is defined as the area above the upper safety boundary curve, and the third region is defined as the area below the lower safety boundary curve.

[0427] In step 857, the DGA may, in response to determining that the second or third region contains a complementary pair including a first rolling insulin measure and a first rolling glucose measure, output a notification regarding the examination of at least one of SCD 102 or MDD 152.

[0428] Figure 17G An example of complementary pair tracking is shown, where the rolling insulin metric is the estimated IOB 90 minutes postprandial, and the rolling glucose metric is the meal start-normalized glucose AUC 90 minutes postprandial. 867 complementary pairs were analyzed (e.g., in...). Figure 17GThe most recent pair (represented by solid circles) is excluded from the fit to the curve of the nominal expected relationship 861 and represents events of interest, such as the user's glucose-insulin balance at 90 minutes postprandial. The most recent complementary pair 867 is analyzed to determine which region it corresponds to. Its position relative to the two safety boundaries 863, 865 can be examined to determine which region contains the complementary pair 867. For example, Figure 17G The solid circle 867 shown appears "below" the lower safety boundary in region 873 of the third region. Several possibilities exist for this situation: (1) the recorded insulin amount may be incorrectly higher than the actual amount of insulin delivered, (2) the glucose reading may be too low, or (3) other confounding factors may be present, such as decreased insulin resistance due to exercise, or a significantly different dietary composition. Depending on the actual combination of events, if the most recent pair is too far below the lower safety boundary, there may be a risk of a false alarm of impending hypoglycemia or a failure to detect postprandial hyperglycemia.

[0429] Conversely, when the complementary pair of interest is located above the upper safety boundary in the second region 871, one or more of the opposite scenarios may occur: (1) the recorded insulin amount may be incorrectly lower than the actual amount of insulin delivered, including missed dietary doses; (2) the glucose reading may be high; or (3) other confounding factors, such as increased insulin resistance due to disease, or the consumption of significantly different dietary compositions. Depending on the actual combination of events, if the most recent pair is too far above the upper safety boundary, there may be a risk of false alarms of impending hyperglycemia or undetected postprandial hypoglycemia.

[0430] If the complementary pair maps above the upper safety boundary in the second region or below the lower safety boundary in the third region, sufficient research data has been used to a priori establish that the DGA100 integrity check is likely reliable. Therefore, the DGA can instruct the user to perform self-monitoring of blood glucose (SMBG). If blood glucose measurement (BGM) readings can be connected to the DGS100 or otherwise input to the DGA, one or more threshold comparisons not covered in this discussion can be used to determine whether sensor 102 needs replacement. Otherwise, the DGA can instruct the user to check the MDD 152 or smart insulin pen cap for possible causes of error.

[0431] Incorporate trends into betting calculations

[0432] Traditional dosing calculators determine the dosage based on the difference between the user's current glucose level and the target glucose level. In these traditional calculators, the current glucose value is treated as a discrete snapshot of time, without considering trends in glucose levels at that point in time. However, during periods of high glucose variability, the recommended dosage can vary considerably because the current glucose level can increase or decrease sharply from one time point to the next.

[0433] In December 2018, the Endocrine Society published unified guidelines for the use of glucose trend arrows in diabetes management (see YCKudva et al., “Approach to Using Trend Arrows in the FreestyleLibre Flash Glucose Monitoring Systems in Adults,” Journal of the Endocrine Society, Vol. 2, pp. 1320–1337, 2018, which is incorporated herein by reference in its entirety). Glucose trend arrows can be categorized into five types based on the rate of glucose change: (1) rapid increase (>2 mg / dL / min), (2) increase (between 1 and 2 mg / dL / min), (3) slow change (not exceeding 1 mg / dL / min), (4) decrease (decline between 1 and 2 mg / dL / min), or (5) rapid decrease (decline >2 mg / dL / min).

[0434] To attempt to account for these rapid changes, a trial-and-error approach can be developed for each trend arrow category, which, in addition to the current glucose value, will incorporate the rate of glucose change to calculate the recommended dose. Based on the reported rate of change, the recommended dose, using the additional rate of change term, can be increased or decreased to accommodate dynamic glucose responses. It is envisioned that these trial-and-error approaches will be developed to minimize the incidence of post-dose hypoglycemia, thereby improving the duration of glucose within the range.

[0435] Insulin site rotation

[0436] The method described in this article facilitates proper site rotation for insulin injection by automatically detecting the injection sites of insulin pen needles and infusion sets.

[0437] The most common method of insulin delivery is administration of insulin to subcutaneous tissue via discontinuous injection or continuous infusion. Proper technique for insulin injection and infusion set placement requires “rotating” the injection site, i.e., circulating it through different areas of the body to avoid any type of local skin reaction caused by repeated and continuous needle insertion. Two common consequences associated with a lack of rotation at the injection site are scar tissue formation and lipomatosis. Scar tissue is fibrous tissue that forms due to persistent trauma or improper healing of an initial injury, characterized by the presence of collagen-dense, avascular tissue. Clinically, lipomatosis is defined as the localized accumulation of fat deposits at the insulin injection site. While scar tissue is a common problem in injectable therapy, lipomatosis is almost the only condition associated with subcutaneous insulin administration. Both exist as indurated nodules beneath the skin and are associated with largely avascular local areas, which negatively impacts systemic insulin uptake and action. Therefore, their effects on insulin absorption are often combined. These effects include decreased insulin absorption (up to 40%), increased total daily insulin dose (TDD), and decreased glycemic control. The prevalence is also very high, estimated to occur in approximately 50% of people on insulin therapy.

[0438] Currently, there is no technology-assisted solution to address the problem of poor injection site rotation. Site rotation depends on patient education by physicians and diabetes educators, as well as subsequent patient adherence.

[0439] As described herein, the DGS 100 may include a site rotation application that can be configured to detect a general injection area and, if the same location is detected repeatedly, provide the user with guidance to rotate to a new injection site. By presenting site rotation information to the user as part of the dosage guidance protocol, the DGS 100 can facilitate better injection practice, higher insulin potency, and longer duration of action.

[0440] In one implementation, such as Figure 18A and Figure 18B As shown, in exemplary method 275, starting from step 276, the position rotation application can be configured to determine a first distance 270 between an electronic device (such as display device 120 (e.g., a smartphone)) and SCD 102, and a second distance 272 between the electronic device and MDD 152. In one embodiment, the position rotation application can use Bluetooth LTE communication (BLE) between three independent devices: (1) display device 120 (e.g., a smartphone), (2) SCD 102, and (3) MDD 152 (e.g., a connected pen needle). Figure 18AAs shown, display device 120 can be configured to serve as a central point for a communication connection. When requesting dose guidance from the DGA, display device 120 can determine a first distance 270 between display device 120 and SCD 102, and a second distance 272 between display device 120 and MDD 152. In one embodiment, the first distance 270 can be determined by the strength of a first signal between display device 120 and SCD 102, and the second distance 272 can be determined by the strength of a second signal between display device 120 and MDD 152.

[0441] In step 278, the site rotation application can be configured to calculate a third distance 274 between SCD 102 and MDD 152, where SCD 102 is located at a fixed position on the subject. Because the display device 120 is not fixed in one position, triangulation can be performed on the first and second signals to determine the third distance 274 between SCD 102 and MDD 152, which have fixed positions on the subject's body. The third distance 274 between MDD 152 and SCD 102 is calculated and can then be recorded in the site rotation application in the display device 120.

[0442] In step 282, the site rotation application can be configured to output a recommendation to move the MDD to a new injection site on the subject in response to determining that the calculated third distance is substantially similar to a previously calculated third distance. In one embodiment, if the calculated third distance 274 between SCD 102 and MDD 152 repeatedly exceeds a given threshold amount, the site rotation application can be configured to provide a guiding message prompting the user to inject insulin at the new site. The threshold amount can be once, alternatively twice, or alternatively three times. In one embodiment, the site rotation application can also be configured to provide the user with a list of acceptable injection sites to present recommendations for new injection areas. These selectable areas can include the arm, thigh, abdomen, and buttock.

[0443] Additional exemplary implementations

[0444] This document also describes other exemplary implementations related to drug delivery. These implementations will be described in the context of an insulin pen, although they are not limited thereto. In many implementations, the MDD 152 (e.g., an insulin pen) can communicate directly with the SCD 102, thus eliminating the need for an additional reader device. In some implementations, the reader device may also communicate with a sensor control device that communicates with the MDD 152. In these implementations, the reader device may also communicate with the MDD 152. In other implementations, the reader device may communicate with the MDD 152, which in turn communicates with the SCD 102, which does not communicate with the reader device. In other implementations, the SCD 102 communicates with the reader device, which in turn communicates with the MDD 152, but the MDD 152 does not communicate with the SCD 102. These different communication schemes can allow any type of information (e.g., analyte measurements, alarms, user information or settings, dosage instructions, etc.) to be transmitted from one device in the system to another via an intermediary device. Communication can occur via Bluetooth or Bluetooth Low Energy or other wireless protocols (such as NFC, RFID, Wi-Fi, etc.). Communication can be sustained (e.g., active Bluetooth pairing) or it can be intermittent (e.g., NFC near-field scanning).

[0445] In some implementations, for safety purposes, SCD 102 or display device 120 (such as reader device (RD)) can check whether MDD 152 maintains a wireless connection with SCD 102 or reader device to notify the user whether MDD 152 may be outside the immediate vicinity. If SCD 102 or RD detects a loss of connection, an alarm is generated at SCD 102 (if configured to generate an audible or visual alarm) or RD (or SCD 102 can notify RD to generate an alarm). Similar alarms can be generated when a problem or malfunction is detected, or when the drug supply is depleted (e.g., low cartridge). If SCD 102, RD, or MDD 152 detects a high analyte condition (e.g., hyperglycemia), SCD 102 or RD (through communication with MDD 152), or MDD 152 itself, can determine whether a dose was recently administered and suppress the generation of an alarm (at SCD 102, RD, or MDD 152).

[0446] Other notifications or alerts that can be generated for the user can be related to pen depletion, including estimating and / or reminding the user of enough remaining medication to be administered within a short period of time (e.g., 1 day). If it is determined that the next predetermined dose is greater than the remaining medication, a recommendation to use a new pen or load a new cartridge can be output. The MDD 152 can be configured to monitor medication dosage and generate an alarm or notification if the expected amount does not match the actual (sensed) amount. The MDD 152 can also be configured to monitor the duration for which the delivery button is pressed to ensure complete delivery and notify the user if incomplete delivery is suspected. If the MDD 152 or DGS 100 detects that the button has been pressed for too short a time, the MDD 152 or DGS 100 can assume that the user has ingested a dose lower than the recommended dose. The DGS 152 or MDD 152 can output a query to the user to verify whether the administered dose is an incomplete dose or an intentionally lower dose. If the MDD 152 or DGS 100 detects that the delivery button has been pressed for too long (e.g., longer than the time required to administer the recommended dose or dose indicated in the dosing guide), the MDD 152 or other system device (e.g., the DGA mounted on the display device 120) can notify or warn the user that a dose may have been administered too high. In one embodiment, the MDD 152 can be configured to output an audible notification (e.g., a beep), a tactile notification (e.g., a vibration or click), and / or a visual notification (e.g., an LED, light) once the dose has been reached, allowing the user to stop pressing the actuator or button. After the dose is administered, the SCD 102 or RD can be configured to send a disconnect time to the MDD 152. If the MDD 152 is disconnected or unable to communicate with the SCD 102 or RD 120 configured to provide dose guidance, the MDD 152 knows that further dose administration should be prevented or limited, at least during the disconnect period. Recommended feed data can be sent to the integrated pen based on the latest optimal estimate.

[0447] If the connection is lost and then re-established between MDD 152 and SCD 102 and / or RD120, the user can be prompted to input the amount of any dose administered during the period of connection loss. For embodiments where MDD 152 monitors the duration of button presses, an accelerometer or other pressure sensing mechanism may be included. Based on the press duration (or movement pattern, and / or pen orientation when the button is pressed) measured by the pressure sensing mechanism, MDD 152 can determine whether the button press is for initiation or dose administration purposes. MDD 152 can be configured to distinguish between different types of presses. For example, MDD 152 can assume a series of short pressures is for initiation purposes and therefore not for dose administration.

[0448] In one implementation, the DGS 100 (e.g., SCD 102, display device 120, or MDD 152) can be configured to detect the occurrence of dietary consumption and prompt the user to prepare medication if they have not already done so. The DGS 100 can update dosage guidance when collecting analyte data during or after meals and immediately before dose administration.

[0449] The MDD 152 may also include a temperature sensor, and the DGS 100 can adjust insulin dosage guidance based on temperature fluctuations (in cases of reduced insulin efficacy or insulin activity at IOB levels). The system can also be configured to generate a notification if the temperature exceeds a specified range.

[0450] The SCD 102 or RD120 can identify the MDD 152 via communication and determine whether the MDD 152 is the correct type for dosage administration. For example, if the MDD 152 contains rapid-acting insulin, and the dosage guidance involves long-acting insulin, the SCD 102 or RD120 can generate an alarm or notification, or cause the MDD 102 itself to generate an alarm or notification. The SCD 102 or RD120 can simultaneously support connections to multiple different MDDs 152, including multiple pens of the same and / or different types.

[0451] Any device in the DGS 100 (“system device”, such as SCD 102, RD120, or MDD 152) can estimate the amount of time before the next dose is needed and notify the user. The system device can use, for example, a GPS monitor in the MDD 152 (e.g., geo-tracking, geofencing) to track the location of the MDD 152 and can notify the user of the location of the MDD 152.

[0452] The MDD 152 can detect the application of a new needle and prompt the user to apply a new needle if no needle replacement is detected. The MDD 152 can be configured with a sensor to detect the presence of air or gas in the needle. The MDD 152 can be configured to detect whether the needle has been tapped (e.g., using a sensor or accelerometer) and prompt or request the user to tap the needle to remove air or gas. The durable pen can detect the presence of a reservoir or cartridge and prompt the user to enter the type of medication (e.g., rapid-acting insulin, long-acting insulin). The MDD 152 can be configured to detect whether the medication has been properly mixed (e.g., using a sensor or accelerometer) and prompt the user to mix or shake the MDD 152 to properly mix the medication.

[0453] MDD 152 can be configured to display the remaining time before the MDD 152 battery runs out or is depleted, and / or before the medication in the MDD 152 runs out or is depleted.

[0454] MDD 152 can be configured to lock users under specific circumstances, such as if the user is not authorized (login / password entry fails on the device), if the user's analyte level is too low or drops rapidly, if a large dose has recently been administered, if the insulin in MDD 152 is expired or too old, or has been exposed to excessively high temperatures, if the systemic state time has exceeded the previous dose (possibly in combination with glucose levels), and any combination thereof, or others.

[0455] The system device (e.g., the DGA on display device 120) can calculate the dose guidance and send it to MDD 152 and / or cloud 190 (e.g., a trusted server) in the event of connection loss or communication failure. MDD 152 can send an acknowledgment that the dose guidance has been received; if no acknowledgment is received, the system device can generate an alarm or notification to the user.

[0456] Before administering a dose, such as when the user picks up or otherwise activates the MDD 152, the MDD 152 (or other system devices communicating with the MDD 152) may prompt the user to verify that the last dose known to the system is correct (e.g., prompting the user to confirm, for example, that the last dose of 5 ml at 5 p.m. was correct and that no other doses were administered). This occurs if a different MDD 152 is used (e.g., a different pen or pump). If the user indicates that no other doses were ingested, the DGS 100 can continue. If the user indicates that another dose has been ingested (e.g., a dose unknown to the DGS 100), the system device may prompt the user to enter the amount, time, and type of medication, and determine a new dose instruction. The system device may also be configured to periodically prompt the user to determine if other pens or drug delivery devices are being used and request the user to integrate these pens or drug delivery devices into the system, for example, by establishing a Bluetooth pairing.

[0457] In various embodiments, a method is provided for parameterizing a patient's drug administration practice to configure dose guidance settings, the method comprising: classifying, by at least one processor, time-related data characterizing the patient's analyte and the drug doses received by the patient during the analysis period, wherein each of the drug doses is classified as a drug category; grouping, by at least one processor, each dose in one of a set of mealtime groups; generating, by at least one processor, dose parameters for the patient by at least partially applying data from each mealtime group to a model; and storing, by at least one processor, the dose parameters in a computer memory for configuring dose guidance settings.

[0458] In some embodiments, the classification step of the method further includes generating a feature matrix that associates a set of classification features with each dose. The method may also include classification features selected from the group comprising: the time of administration for each dose, time-filtered analyte values, the rate of change of the analyte value closest to the administration time, the area under the curve (AUC) representing the integral difference between the analyte value and the analyte value closest to the administration time over a period prior to the administration time, the right AUC representing the integral difference between the analyte value and the analyte value closest to the administration time over a period after the administration time, the time elapsed between administration times, the probability that a meal begins within a defined time interval prior to the administration time, the most likely time interval elapsed since the most recent meal, the probability that a meal begins within a defined time interval after the administration time, and the most likely time interval until the next meal. The method may also include estimating the time of each meal eaten by the patient during the analysis period. The method may further specify that estimating the time of each meal includes generating a feature matrix based on time-related analyte data, wherein the feature matrix associates a set of analyte data features with each distinct region classified as rising, preceding falling, and falling. The method may be further specified that estimating the time of each meal also includes using an algorithm to generate an estimated meal time based on a feature matrix. The method may be further specified that the set of analyte data features is selected from the following group: maximum analyte rate of change, maximum analyte acceleration, analyte value at the point of maximum analyte acceleration, duration of the region, height of the region, maximum deceleration, average rate of change within the region, and time of maximum analyte acceleration.

[0459] In some implementations, the mealtime groups include breakfast, lunch, and dinner. In some implementations, the method also includes grouping by cluster analysis.

[0460] In some implementations, the model used to fit the data pairs is selected from a linear model with zero slope, a linear model with non-zero slope, a piecewise model with a junction at a single point, or a nonlinear model that approximates a piecewise model.

[0461] In some implementations, the fitted data pair also includes minimizing the sum of squared residuals.

[0462] In some embodiments, the fitted data pairs further include evaluating each model using the Akaike Information Criterion (AIC) and selecting the number of models with the minimum AIC value. In some embodiments, the dosage parameters include a fixed dose of drug, analyte level, and a correction factor from the selected model for each group. In some embodiments, th...

Claims

1. A method for configuring dose guidance settings, the method comprising: The system comprises at least one processor receiving analyte data from an on-body unit and drug dosage data from a drug delivery device during an analysis period, wherein the on-body unit includes an analyte sensor and sensor electronics, the analyte sensor being configured to monitor analyte levels, the sensor electronics being coupled to the analyte sensor and configured to wirelessly transmit analyte data, and wherein the drug delivery device is configured to administer a drug dosage and wirelessly transmit drug dosage data. The at least one processor classifies the drug doses received during the analysis period based on the analyte data and the drug dose data, wherein each of the drug doses is classified as a drug category; The at least one processor groups the dosage of each class of drugs in one of a set of mealtime groups; The at least one processor generates the dosage parameters by applying the analyte data and the drug dosage data for each of the set of mealtime groups to the model, at least in part; and The at least one processor stores the dose parameters in the computer memory for configuring dose guidance settings.

2. The method according to claim 1, wherein, The classification also includes generating a feature matrix that associates a set of classification features with each drug dose.

3. The method according to claim 2, wherein, The classification features are selected from the following groups: drug time for each dose, time-filtered analyte value, rate of change of the analyte value closest to the drug time, area under the left curve (AUC) representing the integral difference between the analyte value and the analyte value closest to the drug time over a period before the drug time, right AUC representing the integral difference between the analyte value and the analyte value closest to the drug time over a period after the drug time, time elapsed between drug times, probability that a meal begins within a defined time interval before the drug time, most likely time interval elapsed since the most recent meal, probability that a meal begins within a defined time interval after the drug time, and most likely time interval until the next meal.

4. The method according to claim 2, wherein, The classification also includes estimating the time of each meal eaten during the analysis period.

5. The method according to claim 4, wherein, Estimating the time of each meal further includes: generating a feature matrix based on the analyte data and the drug dosage data, wherein the feature matrix associates a set of analyte data features with each distinct region classified as rising, falling, or declining.

6. The method according to claim 5, wherein, Estimating the time for each meal also includes using an algorithm to generate an estimated meal time based on the feature matrix.

7. The method according to claim 5, wherein, The set of analyte data features is selected from the following groups: maximum analyte change rate, maximum analyte acceleration, analyte value at the point of maximum analyte acceleration, duration of the region, height of the region, maximum deceleration, average rate of change in the region, and time of maximum analyte acceleration.

8. The method according to claim 1, wherein, The meal schedule includes: breakfast, lunch, and dinner.

9. The method according to claim 8, wherein, The grouping also includes grouping through cluster analysis.

10. The method according to claim 1, wherein, The analyte data includes indicators of glucose levels, and the drug dosage data includes insulin.

11. The method according to claim 1, wherein, Also includes: The fixed dose of medication for each corresponding meal group is output on the monitor.

12. The method according to claim 1, wherein, The categorized drug dosages include dietary dosage, pre-meal corrected dosage, post-meal corrected dosage, basal dosage, missed dietary dosage, or ambiguous dosage.

13. The method according to claim 1, wherein, Applying the analyte data and drug dosage data for each meal group to the model includes fitting data pairs to the model, wherein each data pair contains pre-meal glucose levels and corresponding dietary doses.

14. The method according to claim 13, wherein, The model used to fit the data pairs is selected from: a linear model with zero slope, a linear model with non-zero slope, a piecewise model with a junction at a single point, or a nonlinear model that approximates a piecewise model.

15. The method according to claim 13, wherein, Fitting data pairs also includes minimizing the sum of squared residuals.

16. The method of claim 14, wherein, The selection of the model to fit the data pair also includes: evaluating one or more models using the Akaike Information Criterion (AIC) and selecting the model with the minimum AIC value.

17. The method according to claim 16, wherein, The dosage parameters include: fixed dose of drug, analyte level, and correction factors from the selected model for each mealtime group.

18. The method according to claim 17, wherein, It also includes: combining data from multiple mealtime groups by the at least one processor to form a combined group, and selecting the best-fit model and correction factor for the combined group.

19. The method of claim 16, further comprising: The at least one processor compares the AIC value of the selected model with a threshold, and if the AIC value exceeds the threshold, requests user input.

20. A system for parameterizing patient medication dispensing practices to configure dosage guidance settings, the system comprising: A body unit, configured to be worn on the surface of the patient's skin, the body unit comprising: An analyte sensor is configured to contact the patient's interstitial fluid and monitor the patient's analyte levels; and Sensor electronics, coupled to the analyte sensor, and configured to wirelessly transmit the patient's analyte data; A drug delivery device is configured to administer a drug dose and wirelessly transmit the patient's drug dosage data. The display is configured to visually present information; and One or more processors coupled to the body unit, the drug delivery device, and the display, the one or more processors being coupled to a memory storing instructions that, when executed by the one or more processors, cause the system to: During the analysis period, the system receives analyte data from the patient in the body unit and drug dosage data from the patient in the drug delivery device. Based on the analyte data and the drug dosage data, the drug dosage received by the patient during the analysis period is classified. Group the dosage of each category of medication in one of the mealtime groups; Dosage parameters are generated, at least in part, by applying the analyte data and drug dosage data for each of the said mealtime groups to the model; and The dose parameters used to configure dose guidance settings are stored.

21. The system according to claim 20, wherein, The memory stores further instructions for classifying drug doses, at least in part, by generating a feature matrix that associates a set of classification features with each drug dose.

22. The system according to claim 21, wherein, The memory stores further instructions for the following operations: associating classification features from the group, wherein the classification features include: the time of drug administration for each dose, the time-filtered analyte value, the rate of change of the analyte value closest to the drug administration time, the area under the left curve (AUC) representing the integral difference between the analyte value and the analyte value closest to the drug administration time over a period before the drug administration time, the right AUC representing the integral difference between the analyte value and the analyte value closest to the drug administration time over a period after the drug administration time, the time elapsed between drug administration times, the probability that a meal begins within a defined time interval before the drug administration time, the most likely time interval elapsed since the most recent meal, the probability that a meal begins within a defined time interval after the drug administration time, and the most likely time interval until the next meal.

23. The system according to claim 21, wherein, The memory stores further instructions for classifying drug dosages, at least in part, by estimating the timing of each meal the patient ate during the analysis period.

24. The system according to claim 20, wherein, The memory stores further instructions for the following operation: grouping each of the classified drug doses into a mealtime group including breakfast, lunch, and dinner.

25. The system according to claim 24, wherein, The memory stores further instructions for grouping, at least in part, through cluster analysis.

26. The system according to claim 20, wherein, The memory stores the analyte data and the drug dosage data, the analyte data including indicators of glucose levels, and the drug dosage data including insulin.

27. The system according to claim 20, wherein, The memory stores further instructions for the following operation: outputting a fixed dose of medication for each corresponding meal group on the display.

28. The system according to claim 20, wherein, The categorized drug dosages include dietary dosage, pre-meal corrected dosage, post-meal corrected dosage, basal dosage, missed dietary dosage, or ambiguous dosage.

29. The system according to claim 20, wherein, The memory stores further instructions for the following operations: applying the analyte data and the drug dosage data for each meal group to the model, at least in part, by fitting data pairs to the model, wherein each data pair contains pre-meal glucose levels and corresponding dietary doses.

30. The system according to claim 29, wherein, The memory stores further instructions for the following operations: selecting a model for fitting data pairs from a linear model with zero slope, a linear model with non-zero slope, a piecewise model with a junction at a single point, or a nonlinear model that is approximately piecewise.

31. The system according to claim 29, wherein, The memory stores further instructions for fitting data pairs, at least in part, by minimizing the sum of squared residuals.

32. The system according to claim 30, wherein, The memory stores further instructions for the following operation: selecting the model for fitting the data pair by evaluating one or more models using the Akaike Information Criterion (AIC) and selecting the model with the smallest AIC value, at least in part.

33. The system according to claim 32, wherein, The memory stores the dosage parameters, which include: a fixed dose of drug, an analyte level, and a correction factor from a selected model for each mealtime group.

Images