Blood glucose management system using continuous glucose meter and method thereof

Abstract

A blood glucose management system using a continuous blood glucose measurement device and a method thereof are disclosed. The blood glucose management system of the embodiments can include a data collection part for collecting blood glucose data measured by a continuous blood glucose measurement device worn by a patient, a data analysis part for analyzing the collected blood glucose data to calculate a plurality of blood glucose variable values representing a blood glucose state of the patient, and a data providing part for providing the blood glucose variable values through a monitor panel. According to such a system, the blood glucose state of the patient can be thoroughly monitored and managed.

Description

Technical Field

[0001] This invention relates to a technique for monitoring and managing a patient's blood glucose using a continuous glucose monitor (CGM).

[0002] This patent is the result of a project supported by the Ministry of Trade, Industry and Energy of Korea (project management (professional) agency: Korea Institute for Industrial Technology Planning and Evaluation). The project's inherent number is 2410013887, the project number is 07892971, the research project name is Biotechnology Development (R&D), the research topic name is the development of a biosignal monitoring and real-time early warning system beacon for critically injured patients based on artificial intelligence (Biosiginal Emergency Advanced Critical Observation Network), the project implementing agency is ODN Co., Ltd, and the research period is from April 1, 2025 to December 31, 2025. Background Technology

[0003] Blood glucose, as one of the core biomarkers reflecting the body's metabolic state, plays a crucial role in the diagnosis, prognostic assessment, and monitoring of treatment for various metabolic diseases, including diabetes. In particular, abnormal blood glucose levels, such as hypoglycemia or hyperglycemia, can trigger acute or chronic complications. Therefore, stable blood glucose management is a shared challenge for both patients and healthcare professionals.

[0004] Recently, continuous glucose monitors (CGMs) have become widely used for real-time monitoring of patients' blood glucose levels. CGMs automatically measure blood glucose levels at predetermined intervals using sensors and transmit the data wirelessly to mobile devices or external devices, thus providing more detailed and continuous data compared to existing self-monitoring devices. However, several practical challenges remain in blood glucose management using these CGMs.

[0005] First, most blood glucose data provided by continuous glucose monitoring devices can only be viewed through the device manufacturer's proprietary application, resulting in low compatibility with third-party health management applications. Therefore, it is difficult to comprehensively manage or analyze blood glucose data within the daily health management platforms used by patients.

[0006] Secondly, continuous glucose monitoring devices are typically not integrated with older systems such as electronic medical records (EMR) in healthcare institutions, making it difficult to share meaningful blood glucose data collected by patients with healthcare professionals in real time. This not only hinders the clinical application of blood glucose data but can also delay decision-making by healthcare professionals. Furthermore, due to the closed nature, structural limitations, and complexity of older systems, it is practically difficult to integrate blood glucose data analysis and monitoring functions into these systems.

[0007] Therefore, in order to maximize the effectiveness of blood glucose management using continuous glucose meters, it is necessary to develop a system that can provide the function of analyzing and monitoring patients' blood glucose data and supports interoperability between continuous glucose meters and the legacy systems or health management platforms of medical institutions.

[0008] Existing technical documents

[0009] Patent documents

[0010] Korean Patent Publication No. 10-2024-0138379 (published on September 20, 2024) Summary of the Invention

[0011] Technical issues

[0012] The various technical objectives addressed by the embodiments of the present invention are to provide a system and method for effectively managing a patient's blood glucose using a continuous glucose monitor (CGM).

[0013] Specifically, the technical objective addressed by various embodiments of the present invention is to provide a system and method for accurately analyzing and monitoring collected patient blood glucose data using a continuous glucose meter.

[0014] Furthermore, another technical objective achieved through various embodiments of the present invention is to provide a system and method thereof capable of predicting the occurrence of abnormal blood glucose in patients at an early stage based on blood glucose data collected by a continuous glucose meter and supporting preventive management measures.

[0015] Furthermore, another technical objective addressed by various embodiments of the present invention is to provide a system and method thereof capable of supporting interoperability between a continuous glucose meter and legacy systems or health management platforms in medical institutions.

[0016] The technical objectives of this invention are not limited to those mentioned above. Those skilled in the art to which this invention pertains can clearly understand other technical objectives not mentioned through the following description.

[0017] Technical solution

[0018] To achieve the technical objectives described above, the blood glucose management system of various embodiments of the present invention includes: a data collection unit for collecting blood glucose data from a patient, the blood glucose data being measured by a continuous glucose meter worn by the patient; a data analysis unit for calculating multiple blood glucose variable values ​​representing the patient's blood glucose status by analyzing the collected blood glucose data; and a data providing unit for providing the blood glucose variable values ​​through a monitoring board. The blood glucose management system is linked to an existing system of a medical institution, and the blood glucose management system can independently perform processing based on the blood glucose data with the existing system.

[0019] In several embodiments, the legacy system may include an electronic medical record (EMR) system.

[0020] In several embodiments, the blood glucose management system provides at least a portion of the blood glucose variable values ​​to the legacy system and can receive the patient's medical information from the legacy system.

[0021] In several embodiments, the data collection unit collects blood glucose data from multiple patients, and the data providing unit can provide comprehensive blood glucose-related information from the patients through the monitoring board.

[0022] In several embodiments, the data collection unit may collect the blood glucose data from the manufacturer's server of the continuous glucose meter via an application programming interface (API).

[0023] In several embodiments, the blood glucose variable includes Time In Tight Range (TITR), which represents the proportion of time blood glucose is within a specified range. The upper limit of the specified range is set to a value between 130 mg / dL and 150 mg / dL, and the lower limit of the specified range may be set to a value between 60 mg / dL and 90 mg / dL.

[0024] In several embodiments, the blood glucose variable includes a Glycemia Risk Index (GRI), which is calculated based on a hypoglycemia risk index and a hyperglycemia risk index, wherein the hypoglycemia risk index may be assigned a weighted value higher than that of the hyperglycemia risk index.

[0025] In several embodiments, the data analysis unit obtains the patient's clinical information from the legacy system and analyzes the collected blood glucose data and clinical information together using a machine learning model to predict the risk of the patient developing abnormal blood glucose within a specific time period. The clinical information includes at least one of the patient's demographic information, diagnostic information, prescription information, medication history, and disease history. The blood glucose management system may also include a notification unit to provide notifications based on the predicted risk level.

[0026] In several embodiments, the machine learning model may include: an encoder that generates variable embeddings corresponding to the blood glucose variables; an interaction encoder that generates interaction embeddings corresponding to predetermined pairs of blood glucose variables based on the variable embeddings; a variable weighting estimator that estimates patient-specific variable weights for the blood glucose variables based on the clinical information; an interaction weighting estimator that estimates patient-specific interaction weights for the pairs of blood glucose variables based on the clinical information; an aggregator that generates a composite embedding by aggregating the variable embeddings reflecting the variable weights and the interaction embeddings reflecting the interaction weights; and a predictor that predicts the risk level based on the composite embeddings.

[0027] In several embodiments, when the predicted risk level is above a threshold, the notification unit retains the provision of the notification until a preset waiting timer expires. Before the waiting timer expires, if an event that increases the risk of the patient's abnormal blood glucose level is detected, the notification unit immediately provides the notification. Before the waiting timer expires, if an event that decreases the risk of the patient's abnormal blood glucose level is detected, the notification unit may reset the waiting timer and predict the risk level again using the machine learning model.

[0028] In several embodiments, the monitoring board may include: a first interface displaying the blood glucose variable value; a second interface displaying the patient's blood glucose-related information according to time periods; a third interface displaying the proportion of time the patient's blood glucose remains within the target blood glucose range; and a fourth interface displaying the patient's daily blood glucose profile.

[0029] In several embodiments, the blood glucose management system includes a backend-for-frontend (BFF) server, a Fast Healthcare Interoperability Resource (FHIR) relay server, a gateway server, a Fast Healthcare Interoperability Resource server, and an internal integration server. The data collection unit, the data analysis unit, and the data provision unit are implemented within the internal integration server. The Fast Healthcare Interoperability Resource server and the internal integration server are located within the medical institution's internal network along with the legacy system. The backend-for-frontend server transmits health data received from the patient's terminal to the Fast Healthcare Interoperability Resource relay server. The Fast Healthcare Interoperability Resource relay server converts the health data received from the backend-for-frontend server according to the Fast Healthcare Interoperability Resource protocol and transmits it to the gateway server. The gateway server transmits the converted health data to the Fast Healthcare Interoperability Resource server located in the internal network. The Fast Healthcare Interoperability Resource server can transmit the converted health data to the internal integration server.

[0030] To achieve the technical objectives described above, the blood glucose management method using a continuous glucose meter according to various embodiments of the present invention is executed by at least one processor of a blood glucose management system, comprising the following steps: collecting blood glucose data measured by a continuous glucose meter worn by a patient; calculating multiple blood glucose variable values ​​representing the patient's blood glucose status by analyzing the collected blood glucose data; and providing the blood glucose variable values ​​through a monitoring board, wherein the blood glucose management system is linked to an existing system of a medical institution, and the blood glucose management system can independently perform processing based on the blood glucose data with the existing system.

[0031] To achieve the technical objectives described above, the computer program of various embodiments of the present invention is combined with a computer processor and can be stored in a computer-readable recording medium to perform the following steps: collecting blood glucose data measured by a continuous glucose meter worn by a patient; calculating multiple blood glucose variable values ​​representing the patient's blood glucose status by analyzing the collected blood glucose data; and providing the blood glucose variable values ​​via a monitoring board.

[0032] The effects of the invention

[0033] According to various embodiments of the present invention, a continuous glucose monitor (CGM) collects patients' blood glucose data in real time or at set intervals. This data can be analyzed to calculate various blood glucose variable values ​​representing the patient's blood glucose status. These blood glucose variable values ​​can be provided to patients and / or healthcare personnel via a monitoring panel, resulting in comprehensive monitoring and management of the patient's blood glucose status. For example, it can effectively manage the blood glucose status of outpatients (e.g., patients recovering post-surgery, pre-operative inpatients, discharged patients, etc.).

[0034] Furthermore, the patient's blood glucose data is obtained through the application programming interface (API) provided by the manufacturer's server of the continuous glucose meter. Thus, even in environments where it is difficult to directly link with the continuous glucose meter, the patient's blood glucose data can be collected stably.

[0035] Furthermore, the blood glucose management system performs blood glucose data processing (e.g., analysis, monitoring, etc.) independently from the medical institution's legacy system. Therefore, without affecting the structure or performance of the corresponding legacy system, it can reflect (or integrate) the patient's blood glucose-related information to the legacy system.

[0036] Furthermore, by comprehensively analyzing patients' blood glucose-related information (e.g., blood glucose data, blood glucose variable values, etc.) and clinical information through machine learning models, the risk of developing abnormal blood glucose levels in corresponding patients can be accurately predicted. Moreover, if the risk level exceeds a threshold, notifications can be provided to patients and / or healthcare personnel to proactively prevent abnormal blood glucose levels from occurring.

[0037] Furthermore, the blood glucose management system integrates with continuous glucose monitors and existing systems in medical institutions, serving as a platform for patient health management, including blood glucose management, thereby enabling the provision of comprehensive health and medical services to patients.

[0038] The effects based on the technical concept of this invention are not limited to those mentioned above. Those skilled in the art to which this invention pertains can clearly understand other effects not mentioned through the following description. Attached Figure Description

[0039] Figure 1 and Figure 2 This is an illustrative diagram illustrating the working environment of the blood glucose management system for various embodiments of the present invention.

[0040] Figure 3 This is an illustrative block diagram used to explain the structure and operation of a blood glucose management system according to several embodiments of the present invention.

[0041] Figure 4 The illustration is provided to show examples of a blood glucose management system according to various embodiments of the present invention.

[0042] Figure 5 The following is an exemplary flowchart illustrating a blood glucose management method using a continuous glucose monitor (CGM) according to several embodiments of the present invention.

[0043] Figure 6 This is an illustrative diagram illustrating the screen structure of a monitoring board used to explain various embodiments of the present invention.

[0044] Figures 7a to 7e An example diagram illustrating a monitoring board.

[0045] Figure 8 An illustrative diagram showing the input and output data of a machine learning model according to several embodiments of the present invention.

[0046] Figure 9 The diagram illustrates the structure and operation of machine learning models in various embodiments of the present invention.

[0047] Figure 10 This is an illustrative diagram illustrating a model training method based on multi-task learning, which is used to explain several embodiments of the present invention.

[0048] Figure 11 An exemplary diagram is provided to illustrate a computing device capable of implementing various embodiments of the blood glucose management system of the present invention. Detailed Implementation

[0049] Hereinafter, several embodiments of the present invention will be described in detail with reference to the accompanying drawings. The advantages, features, and implementation methods of the present invention can be found by referring to the accompanying drawings. Figure 1 The detailed embodiments become clear. However, the technical concept of the present invention is not limited to the following embodiments and can be implemented through different methods. The following embodiments are only used to complete the technical concept of the present invention so that those skilled in the art can fully understand the scope of the present invention. The technical concept of the present invention is limited only to the scope of the claims.

[0050] In describing various embodiments of the present invention, detailed descriptions of known structures or functions will be omitted when it is determined that such descriptions may obscure the essence of the invention.

[0051] Unless otherwise defined, the terminology (including technical and scientific terms) used in the following embodiments is intended to be understood by one of ordinary skill in the art to which this invention pertains, and may vary depending on the intent, practice, or emergence of new technologies of one of ordinary skill in the art. The terminology used in this specification is for illustrative purposes only and does not limit the scope of the invention.

[0052] In the following embodiments, unless explicitly specified in the context as singular, the singular expressions used include plural meanings. Furthermore, unless explicitly specified in the context as plural, the plural expressions include singular meanings.

[0053] Furthermore, in the following embodiments, the terms “first,” “second,” “A,” “B,” “(a),” “(b),” etc., are used only to distinguish one structural element from another, and the terms do not limit the nature, importance, or order of the corresponding structural elements.

[0054] In the following embodiments, the structural elements described using terms such as "unit," "module," "block," and "device" and the functional blocks shown in the figures can be implemented in software, hardware, or a combination thereof. For example, software may include machine code, firmware, embedded code (or software), application software, or a combination thereof. Hardware may include, for example, electrical circuits, electronic circuits, processors, computers, integrated circuits, integrated circuit cores, manual devices, or a combination thereof. As more specific examples, "unit," "module," etc., may include structural elements such as software structural elements, object-oriented software structural elements, class structural elements, and task structural elements, as well as processes, functions, attributes, programs, subroutines, program code segments, drivers, firmware, microcode, circuits, data, databases, data structures, lists, arrays, and variables.

[0055] Hereinafter, several embodiments of the present invention will be described in detail with reference to the accompanying drawings.

[0056] Figure 1 This is an illustrative diagram illustrating the working environment of the blood glucose management system 10 according to various embodiments of the present invention.

[0057] like Figure 1 As shown, the blood glucose management system 10 of this embodiment is a computing device / system equipped with blood glucose management functions for one or more patients, and can work in conjunction with a continuous glucose monitor (CGM) 11 and a legacy system 13 in a medical institution. However, the scope of the present invention is not limited thereto. Figure 1The illustration shows the blood glucose management system 10 linked with a continuous blood glucose meter 11. However, in order to comprehensively manage the blood glucose of multiple patients, the blood glucose management system 10 can also be linked with multiple continuous blood glucose meters.

[0058] exist Figure 1 In the illustrated working environment, the blood glucose management system 10 collects blood glucose data from patients via a continuous glucose detector 11, and can analyze the data to perform blood glucose management functions for the corresponding patients (e.g., providing blood glucose-related information, notifying patients of blood glucose abnormalities, etc.). Here, blood glucose data, as a type of time-series data, can be understood as continuous blood glucose data.

[0059] Furthermore, the blood glucose management system 10 can provide the collected blood glucose data and / or its analysis results (e.g., at least a portion of the information indicating blood glucose status) to the legacy system 13 of the medical institution. In this case, the blood glucose management system 10 performs blood glucose data-based processing independently of the legacy system 13, and therefore, without affecting the structure or performance of the legacy system 13, it can reflect (or integrate) the patient's blood glucose-related information to the legacy system 13.

[0060] Furthermore, the blood glucose management system 10 obtains the patient's medical information (e.g., various clinical information) from the legacy system 13, and can analyze the corresponding patient's blood glucose data and medical information together to perform blood glucose management tailored to the patient. For example, the blood glucose management system 10 uses a machine learning model to predict the risk of the corresponding patient experiencing abnormal blood glucose levels within a specific time period (e.g., 24 hours, 6 hours, etc.) in advance, and can provide the patient with advance notifications for blood glucose management based on this. Such notifications can be provided through an application (e.g., a health management application, etc.) installed on the patient's terminal 12, but the scope of the invention is not limited thereto. The blood glucose management system 10 can also provide the aforementioned notifications to healthcare professionals.

[0061] Later, refer to Figure 3 The structure and operation of the blood glucose management system 10 are further explained in detail.

[0062] For reference, in addition to blood glucose management, the blood glucose management system 10 may also provide other health management services. In this case, the blood glucose management system 10 may also be named "health management system", etc.

[0063] A continuous glucose meter 11 is a device that continuously measures a patient's blood glucose status at predetermined intervals using a sensor attached to the patient and transmits the data to an external device (e.g., a patient terminal 12). The continuous glucose meter 11 can be used to monitor a patient's blood glucose status in real time or on a schedule. Unlike existing self-monitoring devices, its advantage is that it allows for continuous monitoring of blood glucose trends without the need for finger-prick blood sampling.

[0064] Patient terminal 12 refers to a patient-side device (i.e., a computing device) that can be linked with the blood glucose management system 10. A health management application (e.g., a blood glucose management application) linked with the blood glucose management system 10 can be installed on the patient terminal 12, allowing the patient to receive blood glucose management functions (e.g., receiving notifications of abnormal blood glucose levels). Furthermore, the patient terminal 12 can also transmit blood glucose data received from the continuous glucose meter 11 to the blood glucose management system 10.

[0065] For example, the patient terminal 12 can be a device in various forms, such as a smartphone, tablet, desktop computer, smartwatch, or other wearable device, but is not limited to these.

[0066] The legacy system 13 refers to the pre-built and operational information system of the medical institution. The legacy system 13 can obtain blood glucose-related information of patients from the blood glucose management system 10 (e.g., blood glucose data, blood glucose status information, etc.), and can also provide the blood glucose management system 10 with corresponding patient medical information.

[0067] For example, legacy system 13 could be an Electronic Medical Record (EMR), a Hospital Information System (HIS), an Order Communication System (OCS), a Picture Archiving and Communication System (PACS), a Laboratory Information System (LIS), etc., but is not limited to these. The illustrated systems are usually built on a closed architecture or non-standard interfaces, therefore, direct interaction with external systems is often limited.

[0068] In several embodiments, such as Figure 2 As shown, the blood glucose management system 10 can also work in conjunction with the manufacturer server 21 of the continuous glucose meter 11. The manufacturer server 21, acting as a server, manages the blood glucose data measured (or provided / generated) by the continuous glucose meter 11, for example, which may be implemented based on the cloud, but is not limited to this. Figure 2In the illustrated environment, the blood glucose management system 10 can collect blood glucose data from the continuous glucose meter 11 measured by the device via an application programming interface (API) provided by the manufacturer's server 21. In this case, blood glucose data from the patient can be collected stably even in environments where direct connection to the continuous glucose meter 11 is limited (e.g., the continuous glucose meter 11 is a product from another manufacturer).

[0069] like Figure 1 or Figure 2 As shown, the patient terminal 12, the blood glucose management system 10, and the manufacturer server 21 can communicate via a network. This network can be any type of wired / wireless network, such as a local area network (LAN), a wide area network (WAN), a mobile radio communication network, or a wireless broadband internet access network (Wibro).

[0070] Above, refer to Figure 1 and Figure 2 An exemplary operating environment for a blood glucose management system 10 according to several embodiments of the present invention has been described. Hereinafter, reference will be made to... Figure 3 The structure and operation of the blood glucose management system 10 are described.

[0071] Figure 3 This is an illustrative block diagram used to explain the structure and operation of a blood glucose management system 10 according to several embodiments of the present invention.

[0072] like Figure 3 As shown, the blood glucose management system 10 of this embodiment may include a data collection unit 31, a data analysis unit 32, a data providing unit 33, and a notification unit 34. However, Figure 3 Only structural elements relevant to embodiments of the present invention are shown. Therefore, those skilled in the art to which this invention pertains should understand that, in addition to... Figure 3 In addition to the structural elements 31, 32, 33, and 34 shown, other general-purpose structural elements may also be included (e.g., processors, memory, input / output devices, databases, etc.). Furthermore, Figure 3 The structural elements 31, 32, 33, and 34 of the blood glucose management system 10 shown represent functional components with different functions. In a real physical environment, multiple structural elements can also be implemented in a combined form, and a specific structural element can also be implemented in a form consisting of multiple separate sub-structural elements. The structural elements 31, 32, 33, and 34 of the blood glucose management system 10 will be described below.

[0073] The data collection unit 31 is a module used to collect patient blood glucose data. For example, the data collection unit 31 can collect blood glucose data through the continuous glucose meter 11 or from the manufacturer's server 21.

[0074] Next, the data analysis unit 32, as a module, is used to analyze the collected blood glucose data. The data analysis unit 32 can derive various information required for blood glucose management by analyzing the blood glucose data.

[0075] For example, the data analysis unit 32 can calculate various blood glucose variable values ​​representing a patient's blood glucose status by analyzing blood glucose data. These blood glucose variables may include, but are not limited to, time in range (TIR), time above range (TAR), time below range (BR), time in tight range (TITR), glycemia risk index (GRI), mean blood glucose, standard deviation, glucose variability (CV), and glucose management indicator (GMI). Please refer to Table 1 below for the meaning of the illustrated blood glucose variables and the calculation methods for their values.

[0076] Table 1

[0077]

[0078] For example, as a reference, the upper limit of the target range for time within the strict range is set to a value between approximately 130 mg / dL and 150 mg / dL, and the lower limit is set to a value between approximately 60 mg / dL and 90 mg / dL, preferably between approximately 70 mg / dL and 140 mg / dL. In this case, the potential glycemic risk for the patient can be effectively managed. However, the scope of the invention is not limited thereto, and different target glycemic ranges can be set for time within the target range and time within the strict range depending on the patient's type and characteristics.

[0079] Furthermore, the blood glucose risk index can be calculated based on hypoglycemia risk indicators and hyperglycemia risk indicators. For example, the blood glucose risk index can be calculated based on the weighted value of the two risk indicators, in which case the hypoglycemia risk indicator can be assigned a higher weighted value than the hyperglycemia risk indicator. This allows for a more effective assessment of hypoglycemia risk. The hypoglycemia risk indicator and the hyperglycemia risk indicator can be calculated based on the proportion of time a patient's blood glucose is below the target range and the proportion of time a patient's blood glucose is above the target range, respectively; however, the scope of this invention is not limited thereto. As a more specific example, the blood glucose risk index can be calculated according to the following mathematical formula 1.

[0080] Mathematical Formula 1

[0081]

[0082] In Formula 1, "VLow" and "Low" are the elements serving as indicators of hypoglycemia risk. "VLow" refers to the proportion of time with blood glucose levels below approximately 54 mg / dL, while "Low" refers to the proportion of time with blood glucose levels above approximately 54 mg / dL but below 70 mg / dL. Similarly, "VHigh" and "High" are the elements serving as indicators of hyperglycemia risk. "VHigh" refers to the proportion of time with blood glucose levels above approximately 250 mg / dL, while "High" refers to the proportion of time with blood glucose levels below approximately 250 mg / dL but above 180 mg / dL.

[0083] Furthermore, for example, the data analysis unit 32 analyzes blood glucose-related information and the patient's medical information (e.g., clinical information) together using a machine learning model, enabling early prediction of various information related to abnormal blood glucose levels. This will be discussed later. Figure 5 Please provide an explanation.

[0084] For example, a patient's medical information may include both clinical and non-clinical information. Clinical information may include patient type (e.g., pre-operative, post-operative, emergency room patient), demographic information, diagnostic information (e.g., diagnosis name), prescription information, examination results (e.g., blood test results, physical examination results), surgical information, medication history, medical history, blood glucose-related information, etc., but is not limited to these. Similarly, non-clinical information may include, for example, administrative information, insurance information, etc., but is not limited to these.

[0085] Furthermore, for example, the data analysis unit 32 can analyze blood glucose data to detect abnormalities in a patient's blood glucose levels in real time (e.g., severe blood glucose variability). The data analysis unit 32 can detect abnormalities using machine learning models or other methods.

[0086] Next, the data providing unit 33, as a module, is used to provide patients with blood glucose-related information (e.g., blood glucose data, blood glucose status information, etc.). For example, the data providing unit 33 can provide patients with various blood glucose-related information in real time or on a scheduled basis via a monitoring board; this will be discussed later. Figures 5 to 7e Please provide an explanation.

[0087] Subsequently, the notification unit 34, as a module, is used to provide notifications required for blood glucose management to patients and / or healthcare personnel. For example, when abnormal blood glucose information (e.g., risk level, etc.) predicted by a machine learning model meets preset conditions, the notification unit 34 can provide a notification via the patient terminal 12. The notification unit 34 can also provide the notification to healthcare personnel. Alternatively, when an abnormal blood glucose level is detected from the patient, the notification unit 34 can provide a notification to the patient and / or healthcare personnel. Later, refer to... Figure 5 The detailed instructions will provide specific methods for notification.

[0088] For reference, the term "notification" can also be used interchangeably with expressions such as "reminder," "alarm," and "warning," depending on the context.

[0089] Later, refer to Figure 5 Further details on the specific operation of the blood glucose management system 10.

[0090] The blood glucose management system 10 or its structural elements 31, 32, 33, and 34 can be implemented by at least one computing device. For example, all functions of the blood glucose management system 10 can be implemented by a single computing device, the first function of the blood glucose management system 10 can be implemented by a first computing device, and the second function can be implemented by a second computing device. Alternatively, specific functions of the blood glucose management system 10 can also be implemented by multiple computing devices.

[0091] The computing device may include any device with computing (or processing) capabilities, as will be discussed later. Figure 11 An example of the described device is given. A computing device, as an integration of various structural elements (e.g., memory, processor, etc.) interacting with each other, may also be referred to as a "computing system," depending on the context. Of course, the concept of the term "computing system" can also include an integration of multiple computing devices interacting with each other.

[0092] Figure 4 This is an example diagram illustrating an instance of the blood glucose management system 10.

[0093] like Figure 4 As shown, the blood glucose management system 10 may include a backend-for-frontend (BFF) server 41 serving the frontend, a rapid medical interoperability resource relay server 42, a gateway server 43, a rapid medical interoperability resource server 44, and an internal integration server 45. For example, Figure 3The multiple modules 31, 32, 33, and 34 shown can be implemented within the internal integrated server 45; however, the scope of the invention is not limited thereto. For example, the corresponding modules 31, 32, 33, and 34 can also be implemented separately in different servers (e.g., 41 and 45, etc.).

[0094] As shown in the figure, the Rapid Medical Interoperability Resource Server 44, the Internal Integration Server 45, and the Legacy System 13 can be located within the medical institution's internal network. For internal network security, the Gateway Server 43 can be located within a Demilitarized Zone (DMZ). The following is a brief description of each server.

[0095] The backend server 41 serving the front end refers to the server that provides interface functions for the patient terminal 12. For example, the backend server 41 serving the front end can receive and process requests from the health management application installed on the patient terminal 12 via an application programming interface (API), and can also transmit specific requests to the rapid medical interoperability resource relay server 42. Furthermore, the backend server 41 serving the front end receives the patient's health data (or requests containing such data) from the corresponding health management application and transmits it to the rapid medical interoperability resource relay server 42. The backend server 41 serving the front end can be understood as an application server that interacts with the health management application. As an example, health data may include blood glucose data, activity data (e.g., activity data related to blood glucose such as exercise, diet, sleep, medication use, and drug administration), but is not limited to these.

[0096] Next, the Rapid Medical Interoperability Resource Relay Server 42 refers to a server that performs processing based on the Rapid Medical Interoperability Resource Protocol. The Rapid Medical Interoperability Resource Relay Server 42 can internally act as a Rapid Medical Interoperability Resource Client. The Rapid Medical Interoperability Resource Relay Server 42 can transform health data (or requests, etc.) received from the backend server 41 serving the front end based on the Rapid Medical Interoperability Resource Protocol and transmit it to the gateway server 43. Of course, the Rapid Medical Interoperability Resource Relay Server 42 can also transmit data (or responses to requests, etc.) received from the gateway server 43 to the backend server 41 serving the front end.

[0097] Subsequently, gateway server 43 refers to a server that acts as a gateway, relaying communications with the outside world while maintaining the security of the internal network. For example, gateway server 43 may transmit health data (i.e., converted health data) (or conversion requests, etc.) received from rapid medical interoperability resource relay server 42 to rapid medical interoperability resource server 44 located on the internal network.

[0098] Then, the Rapid Healthcare Interoperability Resource Server 44 refers to a server that reads and processes health data (or requests, etc.) converted based on the Rapid Healthcare Interoperability Resource Protocol. For example, the Rapid Healthcare Interoperability Resource Server 44 may transmit health data (i.e., converted health data) (or conversion requests, etc.) received from the gateway server 43 to the internal integration server 45.

[0099] Finally, the internal integration server 45 refers to a server that comprehensively analyzes and manages patient health data and medical information / data. For example, the internal integration server 45 can receive patient medical information from the legacy system 13 and also provide patient health data to the legacy system 13. Furthermore, the internal integration server 45 receives patient health data from the Rapid Medical Interoperability Resource Server 44, or can retrieve relevant health data from the repository of the Rapid Medical Interoperability Resource Server 44. Additionally, the internal integration server 45 can provide patient health data and medical information to healthcare personnel through a monitoring panel (e.g., the monitoring panel can be provided through the user interface of the legacy system). The internal integration server 45 can comprehensively analyze and manage data from multiple patients.

[0100] Above, refer to Figure 4 An example of the blood glucose management system 10 is described below. Referring to the following... Figure 5 This describes the various methods that the blood glucose management system 10 can perform.

[0101] For ease of understanding, all steps / work of the method described below will be explained in the following sections. Figure 2 The examples shown are all performed by the blood glucose management system 10. Therefore, when the main body of a specific step / task is omitted, it can be understood that the corresponding step / task is performed by the blood glucose management system 10. However, in a real environment, some steps / tasks of the methods described later may also be performed by other computing devices. For example, a machine learning model for predicting blood glucose abnormalities (see reference...) Figure 8 Training (etc.) can also be performed by other computing devices.

[0102] For ease of explanation, the blood glucose management system 10 will be referred to as "the system" below.

[0103] Figure 5 This is an illustrative flowchart illustrating a blood glucose management method using a continuous glucose meter according to several embodiments of the present invention. However, this is merely an illustrative embodiment for achieving the objectives of the present invention, and some steps may be added or deleted as needed.

[0104] like Figure 5As shown, the blood glucose management method of the embodiment may begin in step S51, whereby blood glucose data of the patient is collected by a continuous glucose meter 11. For example, the system (e.g., data collection unit 31) 10 may collect the patient's blood glucose data in real time or on a scheduled basis via an application programming interface provided by the manufacturer's server 21.

[0105] In step S52, multiple blood glucose variable values ​​representing the blood glucose status of the corresponding patient are calculated by analyzing blood glucose data. For example, as shown in Table 1, the system (e.g., data analysis unit 32) 10 can calculate blood glucose variable values ​​by analyzing blood glucose data (i.e., raw measurement data).

[0106] It should be understood that the execution of steps S53 and S55 described later is not in any particular order.

[0107] In step S53, the risk of the patient developing abnormal blood glucose levels is predicted using a machine learning model (i.e., a trained model). For example, the system (e.g., data analysis unit 32) 10 uses a machine learning model (refer to...) Figure 8 This predicts the risk (i.e., probability) of an abnormal blood glucose level occurring within a specific time period (e.g., 24 hours) starting from the current point in time. As examples, abnormal blood glucose levels include, but are not limited to, hypoglycemia, hyperglycemia, severe blood glucose variability, abnormal blood glucose patterns, etc.

[0108] Depending on the circumstances, System 10 can also predict the time point when blood glucose abnormalities occur (i.e., the prediction time point), the type of blood glucose abnormality, the cause of the blood glucose abnormality, and the duration of the blood glucose abnormality.

[0109] The structure and training methods of machine learning models can be designed in various ways, which will be discussed later. Figures 8 to 10 Please provide an explanation.

[0110] In step S54, a notification is provided based on the predicted risk level. For example, when the predicted risk level meets a preset condition (e.g., above a threshold), the system (e.g., notification unit 34) 10 may send a notification message to the patient and / or medical staff (e.g., the medical staff responsible for treating the patient). For example, the notification message may be sent to the corresponding patient via a health management application installed on the patient terminal 12; however, the scope of the invention is not limited thereto.

[0111] For example, notification messages may include the type of blood glucose abnormality, the level of risk, the expected time of occurrence, recommended actions or blood glucose management guidelines, the cause of the blood glucose abnormality, and the expected duration of the blood glucose abnormality, but are not limited to these.

[0112] The specific method of providing notifications may vary depending on the implementation.

[0113] In several embodiments, system 10 may provide a notification immediately when the predicted risk level is above a threshold.

[0114] In several further embodiments, when the predicted risk level is above a threshold, system 10 can predict the duration of the abnormal blood glucose level using a machine learning model. Then, when the predicted duration is above a baseline value, system 10 can provide a notification immediately. Conversely, in other cases, system 10 can retain the notification.

[0115] In several other embodiments, when the predicted risk level exceeds a threshold, system 10 retains the notification until the waiting timer expires, allowing monitoring of the patient's blood glucose-related events (e.g., diet, exercise, sleep, medication administration, etc.). Furthermore, before the waiting timer expires, if an event increasing the risk of blood glucose (i.e., the risk of abnormal blood glucose) is detected, system 10 can immediately provide a notification. Conversely, if an event decreasing the risk of blood glucose is detected, system 10 can reset the waiting timer and re-predict the risk level using a machine learning model (i.e., re-predict based on re-collected blood glucose data, etc.). This approach not only reduces the inconvenience caused to patients by unnecessary frequent notifications but also helps address the problem of declining notification effectiveness. For reference, even for the same event, depending on the patient's condition, it can be determined as an event increasing the risk of blood glucose, or it can be determined as an event decreasing the risk of blood glucose (e.g., if the patient is fasting, exercise can be determined as an event increasing the risk of hypoglycemia). Furthermore, the end time of the waiting timer can be dynamically set. For example, the corresponding end time can be dynamically set based on the predicted risk level and the predicted time of occurrence (for example, if the predicted risk level is low and the time of occurrence is relatively long, a longer corresponding end time can be set).

[0116] In several additional embodiments, when the predicted risk level is above a threshold, system 10 can predict the time point of occurrence of blood glucose abnormalities using a machine learning model. Furthermore, when the difference between the current time point and the predicted time point is less than a baseline value, system 10 can immediately provide a notification. When the corresponding difference is above the baseline value, system 10 retains the notification until the waiting timer expires, allowing monitoring of the patient's blood glucose-related events (e.g., diet, exercise, sleep, medication administration, etc.). Subsequently, when an event that reduces the risk of blood glucose abnormalities (i.e., the risk of occurrence of blood glucose abnormalities) is detected (e.g., exercise, medication administration, etc.), system 10 can again predict the risk level using a machine learning model. Conversely, when an event that increases the risk of blood glucose abnormalities is detected (e.g., high-glycemic diet, failure to administer medication at a predetermined time, fasting exercise, etc.), system 10 can immediately provide a notification.

[0117] On the other hand, in several embodiments, system 10 can detect abnormal blood glucose levels in a patient in real time or on a scheduled basis by analyzing blood glucose data. The specific method for detecting abnormal blood glucose levels can be any method. Furthermore, when an abnormal blood glucose level is detected, system 10 can immediately send a notification message to the patient (and / or healthcare personnel). At this time, the notification message may include information related to the abnormal blood glucose level and information consistent with the patient's blood glucose management guidelines (e.g., insulin dosage adjustments, dietary improvement guidelines, exercise therapy, etc.).

[0118] In step S55, the blood glucose variable value is provided via a monitoring board. For example, the system (e.g., data providing unit 34) 10 can provide the value via a monitoring board (see reference). Figures 6 to 7e The system provides patients with blood glucose-related information (e.g., blood glucose data, blood glucose variable values, etc.). The system can also provide blood glucose-related information for multiple patients through corresponding monitoring boards.

[0119] Depending on the situation, the monitoring board may also be named "monitoring interface", "monitoring screen", "monitoring user interface (UI)", "monitoring graphical user interface (GUI)", "control panel", etc.

[0120] The monitoring board can be designed in various forms, and its specific form can vary depending on the implementation. The following is for ease of understanding, referring to... Figures 6 to 7e This section provides a design example of a monitoring board.

[0121] Figure 6 This is an illustrative diagram illustrating the screen structure of the monitoring board 61 in several embodiments of the present invention.

[0122] like Figure 6 As shown, the monitoring board 61 in the embodiment may include multiple interfaces (62 to 66 or areas).

[0123] The first interface 62 is for inputting search criteria. For example, search criteria can be time limits, patient names, blood glucose variables, ward numbers, etc., but are not limited to these. Figure 7a Examples of a first interface 62 and a second interface 63 are shown. The first interface 62 shows the case where period information is entered as a search criterion (see “Period Input Section”).

[0124] The second interface 63 to the fifth interface 66 can display information / data that matches the search criteria.

[0125] Next, the second interface 63 is used to display blood glucose variable values. That is, the second interface 63 can display various blood glucose variable values ​​that meet the search criteria. As shown in Table 1, Figure 7a This shows a display of blood glucose variable values.

[0126] Subsequently, the third interface 64 is used to display summary information related to active blood glucose (e.g., activity-based blood glucose changes). For example, the third interface 64 can display the patient's blood glucose-related information by time period or visualize the trend of blood glucose changes over time. In this case, blood glucose changes during periods of active patient activity (work hours, exercise time, etc.) and inactive periods (e.g., sleep time), as well as activity-based blood glucose change patterns, can be intuitively conveyed. For an example of the third interface 64, please refer to [link to example]. Figure 7b . Figure 7b The third interface 64 shows the blood glucose trend based on average blood glucose and the passage of time (refer to "blood glucose curve") displayed in terms of time periods.

[0127] Then, the fourth interface 65 is used to display the proportion of time a patient's blood glucose level remains within the target blood glucose range. The target blood glucose range can be set as a range of time within a strictly defined range; however, the scope of this invention is not limited to this. Different target blood glucose ranges can also be set according to the patient's type (e.g., pre-operative, post-operative recovery, emergency, etc.). For an example of the fourth interface 65, please refer to [link to example]. Figure 7c . Figure 7c The fourth interface 65 shows the proportion of time a patient's blood glucose levels fall within various ranges (e.g., target blood glucose range, hypoglycemia range, hyperglycemia range, etc.) according to the date.

[0128] Finally, the fifth interface 66 is used to display the patient's blood glucose profile. The blood glucose profile is a visual or numerical summary of blood glucose patterns based on blood glucose data measured within a specified period (e.g., one day, one week), including trends in blood glucose levels, highest / lowest blood glucose levels, blood glucose stability, and postprandial blood glucose response. For example, the fifth interface 66 can display the patient's daily blood glucose profile. Please refer to the relevant example. Figure 7d . Figure 7d The fifth interface 66 shows the patient's daily blood glucose profile in a curved visual format.

[0129] Figure 7e An example of a monitoring board (or a characteristic interface of a corresponding board) 71 for the comprehensive management of blood glucose in multiple patients is shown. Figure 7e The monitoring panel 71 visually provides changes in blood glucose levels in admitted patients and applies this information to the intensive care unit or to situations involving changing healthcare staff.

[0130] like Figure 7eAs shown, the monitoring panel 71 can highlight (“reference color”) and summarize the current blood glucose value, blood glucose status (e.g., hyperglycemia, hypoglycemia, normal blood glucose within the target range), and trends, allowing for a clear overview of blood glucose changes in multiple patients. Furthermore, by selecting the blood glucose summary information area 72 for a specific patient, the monitoring panel 71 can display detailed blood glucose information for that patient. For example, the detailed blood glucose information can be displayed by referring to… Figures 6 to 7d The information provided in at least one of the interfaces 62, 63, and 64 described herein is not limited to this. For example, detailed blood glucose information may include the proportion of time a patient's blood glucose is maintained within the target blood glucose range, information on active blood glucose, etc., and statistics for the day, week, two weeks, four weeks, etc., can be queried.

[0131] Above, refer to Figures 6 to 7e The monitoring board 61 of several embodiments of the present invention is described. The monitoring board 61 provides blood glucose-related information in real time or on a fixed schedule, which can be effectively used to monitor and manage the blood glucose status of patients, and can also be used to comprehensively manage the blood glucose status of multiple patients.

[0132] The following is for reference Figures 8 to 10 This document describes the instance structure and training method of a machine learning model 80 used to predict abnormal blood sugar levels.

[0133] Figure 8 An illustrative diagram is provided to show the input and output data of a machine learning model 80 according to several embodiments of the present invention.

[0134] like Figure 8 As shown, the machine learning model 80 receives clinical information (or medical information) such as patient demographics, diagnosis, prescriptions, medication history, and disease history, as well as blood glucose-related information. It can predict and output abnormal blood glucose information at an early stage. That is, the machine learning model 80 can predict information related to abnormal blood glucose by comprehensively analyzing the patient's clinical information and blood glucose-related information. In this way, when considering the clinical information of individual patients, predictions related to abnormal blood glucose can be performed in a more precise and personalized manner.

[0135] As mentioned above, blood glucose-related information may include blood glucose data (i.e., raw measurement data) and blood glucose status information composed of blood glucose variable values. However, for ease of understanding, the following explanation will focus on the case where blood glucose-related information is composed of "blood glucose variable" values.

[0136] For example, information about abnormal blood glucose levels may include the type of abnormality (e.g., hypoglycemia, hyperglycemia, severe blood glucose variability, abnormal blood glucose pattern, etc.), whether an abnormality has occurred, the expected time of occurrence, the duration of the abnormality, the reason for its occurrence, or a combination thereof, but is not limited to these.

[0137] The machine learning model 80 can be designed and implemented in various ways. For example, the machine learning model 80 can be designed and implemented based on various types of models / algorithms such as decision tree, random forest, eXtreme Gradient Boosting (XGBoost), Light Gradient Boosting Machine (LightGBM), Categorical Boosting (CatBoost), Support Vector Machine (SVM), logistic regression, Adaptive Boosting (AdaBoost), and neural network. Its specific type / form / structure can vary depending on the task of the machine learning model 80. Those skilled in the art to which this invention pertains are familiar with the basic structure and training methods of exemplary models; therefore, their description will be omitted.

[0138] The machine learning model 80 can be trained using datasets of normal individuals and patients in a supervised learning manner. Those skilled in the art to which this invention pertains are familiar with training based on supervised learning methods; therefore, its description will be omitted. For reference, in supervised learning, the details of the labels may vary depending on the task of the machine learning model 80.

[0139] Figure 9 This is an illustrative diagram used to explain the structure and operation of the machine learning model 80 in several embodiments of the present invention. Figure 9 The structure of a machine learning model 80 provided for performing individual patient-specific predictions is shown, and the corresponding model 80 is implemented based on a neural network.

[0140] like Figure 9 As shown, in various embodiments, the machine learning model 80 may include an encoder 91, a variable weighted estimator 92, an interactive weighted estimator 93, an interactive encoder 94, an aggregator 95, and a predictor 96.

[0141] Encoder 91 refers to a neural network module that encodes variable data 97 and generates embeddings. Variable data 97 can consist of various blood glucose variable values, or it can be equivalent to time-series data. For example, encoder 91 can receive each blood glucose variable value and generate its corresponding variable embedding (e.g., embedding vector).

[0142] Encoder 91 can be implemented as any type / morphology of neural network. For example, encoder 91 can be implemented based on a multi-layer perceptron (MLP) (or a fully-connected layer) or through other types of neural networks.

[0143] Next, the variable weighting estimator 92 is a module that estimates weighted values ​​for each blood glucose variable based on the patient's characteristic information 98-1. The characteristic information 98-1 may include the patient's clinical information and other characteristic information. Furthermore, the characteristic information 98-1 can be encoded into an appropriate form and input into the variable weighting estimator 92. For example, the variable weighting estimator 92 can estimate and output individual patient-specific weighted values ​​(i.e., variable weights) by analyzing the patient's clinical information 98-1. It should be understood that the variable weights refer to vectors used to reflect the varying degrees of influence of each blood glucose variable on blood glucose abnormalities from patient to patient.

[0144] Depending on the circumstances, the variable weighted value estimator 92 can receive the representative vector (e.g., center vector, weighted average vector, etc.) of the cluster to which the patient characteristic information 98-1 belongs, and this technique is also applicable to the interactive weighted value estimator 93.

[0145] The variable weighted value estimator 92 can be implemented in any form. For example, the variable weighted value estimator 92 can be implemented based on a neural network module such as a multiple perceptron (MLP), or it can be implemented by a module that performs attention operations.

[0146] Next, the interaction weighting estimator 93 is a module that estimates predetermined interaction weights for pairs of blood glucose variables based on the patient's characteristic information 98-1. For example, the interaction weighting estimator 93 can estimate and output individual patient-specific weights (i.e., interaction weights) for each pair of blood glucose variables by analyzing the patient's clinical information 98-1. It should be understood that the interaction weights are vectors used to reflect the degree of influence of the interaction between blood glucose variables on blood glucose abnormalities, which varies from patient to patient. The interaction weighting estimator 93 can be implemented in a similar manner to the variable weighting estimator 92.

[0147] Subsequently, interactive encoder 94 is a module for generating interactive embeddings corresponding to the blood glucose variable pairs respectively. For example, interactive encoder 94 can generate interactive embeddings corresponding to the predetermined blood glucose variable pairs based on the variable embeddings output by encoder 91.

[0148] The interactive encoder 94 can be implemented in any form. For example, the interactive encoder 94 can be implemented based on a neural network module such as a multiple perceptron (MLP), or it can be implemented as a module that performs operations such as averaging, concatenation, and element-wise multiplication on the embeddings of two variables.

[0149] Then, aggregator 95 refers to the module that aggregates variable embeddings and interaction embeddings reflecting weighted values ​​to generate a composite embedding. In other words, aggregator 95 can aggregate variable embeddings reflecting variable weighted values ​​and interaction embeddings reflecting interaction weighted values ​​to generate a composite embedding.

[0150] Aggregator 95 can be implemented in any form. For example, aggregation region 95 can be implemented based on neural network modules such as multiple perceptrons (MLP), or it can be implemented in other ways.

[0151] In several embodiments, as shown in the figure, aggregator 95 may also receive patient blood glucose-related event information 98-2. The blood glucose-related event information 98-2 may be encoded into an appropriate format and input into aggregator 95. In this case, predictor 96 may consider patient blood glucose-related events (e.g., exercise, sleep, diet, medication administration, etc.) to predict abnormal blood glucose information 99.

[0152] Finally, predictor 96 refers to a neural network module that predicts blood glucose abnormality-related information 99 based on comprehensive embedding. That is, predictor 96 can predict blood glucose abnormality information 99 of an individual patient by comprehensively analyzing various information included in the comprehensive embedding (e.g., blood glucose variable values ​​and various clinical information). For example, the predictor 96 can be implemented based on a multiple perceptron (MLP), but the scope of the invention is not limited thereto.

[0153] For example, predictor 96 can output the risk (i.e., probability) of a patient developing a blood glucose abnormality within (or after) a specific time / period (e.g., 6 hours, 12 hours, 1 day, 3 days, etc.) (e.g., outputting a confidence score related to whether or not a blood glucose abnormality occurs), or it can output the type of blood glucose abnormality (e.g., hypoglycemia, hyperglycemia, etc.), the cause of occurrence, the expected time of occurrence, the duration, etc. (e.g., outputting a confidence score according to the cause of occurrence).

[0154] On the other hand, in several embodiments, a machine learning model 80 can be constructed to further accurately predict abnormal blood glucose information using multi-task learning techniques. For this embodiment, see below. Figure 10 Please provide an explanation.

[0155] Figure 10The diagram illustrates the model training method based on multi-task learning, which is used to explain several embodiments of the present invention. Hereinafter, for the purpose of clearly explaining the present invention, "Predictor A," "Predictor B," etc., will be named "First Predictor," "Second Predictor," etc., and this naming convention also applies to other items using similar names.

[0156] like Figure 10 As shown, system 10 can perform multi-task learning using multiple predictors (e.g., 101, 102) connected to machine learning model 80. Figure 10 The illustration shows an aggregator 95 of a machine learning model 80 connected to two predictors 101 and 102; however, the scope of the invention is not limited thereto. For example, the aggregator 95 may also be connected to a third predictor (not shown). At least one of the predictors 101 and 102 may represent the execution of the target task. Figure 9 Predictor 96.

[0157] The first predictor 101 is a neural network module for predicting first blood glucose abnormality information (hereinafter referred to as "first information") 103, and the second predictor 102 is a neural network module for predicting second blood glucose abnormality information (hereinafter referred to as "second information") 104 that is different from the first information (refer to examples of the various types of information below). That is, the first predictor 101 and the second predictor 102 perform different tasks based on a comprehensive embedding (i.e., the task of predicting the first information and the task of predicting the second information).

[0158] The types and number of tasks used for multi-task learning can be defined and designed in various ways.

[0159] For example, machine learning model 80 can be trained using a first task to predict the risk of blood glucose abnormalities occurring within a specific time period and a second task to predict the cause of blood glucose abnormalities (i.e., the first information 103 is the risk of blood glucose abnormalities occurring within a specific time period, and the second information is the cause of blood glucose abnormalities). Specifically, the parameters of the first predictor 101, aggregator 95, encoders 91, 94, and / or weighted estimators 92, 93 are updated based on the prediction loss of the first task, and the parameters of the second predictor 102, aggregator 95, encoders 91, 94, and / or weighted estimators 92, 93 are updated based on the prediction loss of the second task.

[0160] As another example, the machine learning model 80 can be trained by a first task for predicting the risk of blood glucose abnormalities occurring in a first time period and a second task for predicting the risk of blood glucose abnormalities occurring in a second time period different from the first time period (i.e., the first information is the risk of blood glucose abnormalities occurring in the first time period, and the second information is the risk of blood glucose abnormalities occurring in the second time period). As a more specific example, the machine learning model 80 can be trained by performing a first task for predicting the risk of blood glucose abnormalities occurring within 24 hours through a first predictor 101 and a second task for predicting the risk of blood glucose abnormalities occurring within 48 hours through a second predictor 102.

[0161] As another example, machine learning model 80 can be trained using a first task for predicting the risk of blood glucose abnormalities occurring within a specific time period and a second task for predicting the time point of occurrence of blood glucose abnormalities (i.e., the first information is the risk of blood glucose abnormalities occurring within a specific time period, and the second information is the expected time point of occurrence of blood glucose abnormalities). More specifically, machine learning model 80 can be trained by performing a first task, predicting the risk of blood glucose abnormalities occurring within 24 hours, via a first predictor 101, and a second task, predicting the time point of occurrence of blood glucose abnormalities, via a second predictor 102, which is equivalent to a regression head.

[0162] As another example, the machine learning model 80 can be trained by a first task for predicting the risk of blood glucose abnormalities occurring within a specific time period and a second task for predicting the number of times (or duration, etc.) of blood glucose abnormalities occurring within a specific time period (i.e., the first information is the risk of blood glucose abnormalities occurring within a specific time period, and the second information is the number of times blood glucose abnormalities occur within a specific time period).

[0163] As another example, machine learning models can also be trained using tasks such as predicting the cumulative time of abnormal blood glucose within a specific time (or period), predicting blood glucose values ​​after a specific time, and predicting blood glucose-related events that affect blood glucose data.

[0164] As yet another example, machine learning model 80 can also be trained based on various combinations of the examples described.

[0165] If the machine learning model 80 is successfully trained, the system 10 can use it to predict abnormal blood glucose levels in patients and provide notifications. In this case, the abnormal blood glucose information predicted by the machine learning model 80 can be used to determine whether to provide a notification, the method of notification, and the content of the notification message. For example, the risk level of the abnormal blood glucose level, the expected time of occurrence, and the duration can be used to determine whether to provide a notification, the method of notification, and / or the blood glucose management guidelines included in the notification message.

[0166] In several embodiments, system 10 can derive a blood glucose management guideline suitable for the patient through machine learning model 80. Specifically, if the predicted risk of abnormal blood glucose is above a threshold, system 10 can modify the blood glucose-related event information 98-2 and repeatedly perform risk prediction. Then, system 10 can determine a blood glucose management guideline suitable for the patient based on the blood glucose-related event information 98-2 that minimizes the predicted risk. For example, if the blood glucose-related event is "exercise," system 10 can determine "exercise" as the corresponding patient's blood glucose management guideline if the predicted risk is minimized.

[0167] Above, refer to Figures 8 to 10 The machine learning model 80 of several embodiments of the present invention is described, including its input and output data, exemplary detailed structure, and training method.

[0168] Above, refer to Figures 5 to 10 This invention describes a blood glucose management method using a continuous glucose meter according to several embodiments. As described above, the continuous glucose meter 11 collects the patient's blood glucose data in real time or at set times, and analyzes this data to calculate various blood glucose variable values ​​representing the patient's blood glucose status. These blood glucose variable values ​​can be provided to the patient and / or healthcare personnel via a monitoring board 61, resulting in thorough monitoring and management of the patient's blood glucose status. For example, it can effectively manage the blood glucose status of outpatients (e.g., patients recovering from surgery, preoperative inpatients, discharged patients, etc.).

[0169] Furthermore, the patient's blood glucose data is obtained through the application programming interface (API) provided by the manufacturer's server 21 of the continuous glucose meter 11, thereby enabling stable collection of the patient's blood glucose data even in environments where it is difficult to directly link with the continuous glucose meter 11.

[0170] Furthermore, system 10 performs blood glucose data processing (e.g., analysis, monitoring, etc.) independently of the medical institution's legacy system 13. Therefore, without affecting the structure or performance of the corresponding legacy system 13, it can reflect (or integrate) the patient's blood glucose-related information to the legacy system 13.

[0171] Furthermore, by comprehensively analyzing patients' blood glucose-related information (e.g., blood glucose data, blood glucose variable values, etc.) and clinical information through machine learning model 80, the risk of developing abnormal blood glucose levels in corresponding patients can be accurately predicted. Moreover, if the risk level exceeds a threshold, notifications can be provided to patients and / or healthcare personnel to proactively prevent abnormal blood glucose levels from occurring.

[0172] The following is for reference Figure 11 An exemplary computing device 110 capable of implementing the system 10 will be described.

[0173] Figure 11A hardware structure diagram illustrating an example of a computing device.

[0174] like Figure 11 As shown, the computing device 110 may include one or more processors 111, a bus 113, a communication interface 114, a memory 112 for loading a computer program 116 executed by the processor 111, and a memory 115 for storing the computer program 116. However, Figure 11 Only structural elements relevant to embodiments of the present invention are shown. Therefore, those skilled in the art to which this invention pertains should understand that, in addition to... Figure 11 In addition to the structural elements shown, other general structural elements may also be included. That is, besides... Figure 11 In addition to the structural elements shown, the computing device 110 may also include various other structural elements. Furthermore, depending on the circumstances, the computing device 110 may be omitted. Figure 11 The structural elements shown are some of the structural elements. The following describes each structural element of the computing device 110.

[0175] Processor 111 controls the overall operation of various structures of computing device 110. Processor 111 may include at least one of a central processing unit (CPU), microprocessor (MPU), microcontroller (MCU), graphics processing unit (GPU), neural network processing unit (NPU), tensor processing unit (TPU), vision processing unit (VPU), accelerated processing unit (APU), or any type of processor known in the art to which this invention pertains. Furthermore, processor 111 can perform operations related to at least one application or program for performing a specific task / step / method. Computing device 110 may include more than one processor.

[0176] Next, memory 112 can store various data, instructions, and / or information. Memory 112 can load computer program 116 from memory 115 to perform specific tasks / steps / methods. Memory 112 can be volatile memory such as random access memory; however, the scope of the present invention is not limited thereto.

[0177] Subsequently, bus 113 can provide communication functions between the structural elements of computing device 110. Bus 113 can be of various forms, such as address bus, data bus, and control bus.

[0178] Then, the communication interface 114 can support wired and wireless communication of the computing device 110. To this end, the communication interface 114 may include communication modules known in the art to which this invention pertains (e.g., radio frequency modules that support wireless communication) and can support multiple communication methods.

[0179] Next, memory 115 may non-temporarily store one or more computer programs 116. Memory 115 may include read-only memory (ROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other non-volatile memory, hard disk, removable disk, or any form of computer-readable recording medium known in the art to which this invention pertains.

[0180] Subsequently, computer program 116 may include instructions that, when loaded into memory 112, cause processor 111 to perform specific work / steps / methods. That is, by executing the loaded instructions, processor 111 can perform specific work / steps / methods.

[0181] For example, computer program 116 may include instructions for performing the following tasks: collecting blood glucose data measured by a continuous glucose meter 11 worn by the patient; calculating multiple blood glucose variable values ​​representing the patient's blood glucose status by analyzing the collected blood glucose data; and providing the blood glucose variable values ​​via a monitoring board 61.

[0182] As another example, computer program 116 may include functions for performing the reference. Figures 1 to 10 The instructions described include at least a portion of the work / steps / methods.

[0183] As shown in the figure, the system 10 of various embodiments of the present invention can be implemented by the computing device 110.

[0184] On the other hand, in several embodiments, Figure 11 The computing device 110 shown refers to a virtual machine implemented based on cloud technology. For example, the computing device 110 can be a virtual machine running on one or more physical servers included in a server farm. In this case, Figure 11At least a portion of the processor 111, memory 112, and memory 115 shown may be virtual hardware, and the communication interface 114 may also be a virtual network component such as a virtual switch.

[0185] Above, refer to Figure 11 An exemplary computing device 110 is described, which can implement the system 10 of various embodiments of the present invention.

[0186] The following is for reference Figures 1 to 11 This document describes several embodiments of the present invention and their effects. The effects of the present invention are not limited to those mentioned above; those skilled in the art will clearly understand other effects not mentioned through the following description.

[0187] Furthermore, although multiple structural elements are combined into one or more in the above embodiments, this does not mean that the technical concept of the present invention is limited to this embodiment. That is, as long as it is within the scope of the purpose of the technical concept of the present invention, all structural elements can also be selectively combined into more than one for operation.

[0188] The technical concept of the present invention described above can be implemented by computer-readable code on a computer-readable recording medium. The computer program recorded on the computer-readable recording medium can be transmitted to other computing devices via networks such as the Internet and installed on the respective computing devices, thereby becoming usable by those devices.

[0189] Although multiple tasks are shown in a specific order in the accompanying drawings, it should not be construed as meaning that the tasks must be performed in the shown order or sequential order, or that all the shown tasks must be performed to obtain the desired result. In certain situations, multitasking and parallel processing may also be advantageous. While several embodiments of the invention have been described above with reference to the accompanying drawings, those skilled in the art should understand that the technical concept of the invention can also be implemented through other specific embodiments without changing its technical idea or essential features. Therefore, the embodiments described above are merely examples at all levels and should not be construed as limiting. The scope of protection of this invention should be interpreted based on the scope of the invention claims, and all technical ideas within the same scope fall within the scope of the invention claims as defined by the technical concept of this invention.

Claims

1. A blood glucose management system, characterized in that, include: A data collection unit for collecting blood glucose data from patients, the blood glucose data being measured by a continuous glucose meter worn by the patient; The data analysis department calculates multiple blood glucose variable values ​​representing the patient's blood glucose status by analyzing the collected blood glucose data; and The data provider provides the blood glucose variable values ​​through the monitoring board. The blood glucose management system is linked with the medical institution's existing system. The blood glucose management system performs processing based on the blood glucose data independently of the legacy system.

2. The blood glucose management system according to claim 1, characterized in that, The old system included an electronic medical record system.

3. The blood glucose management system according to claim 1, characterized in that, The blood glucose management system provides at least a portion of the blood glucose variable values ​​to the legacy system. Receive the patient's medical information from the old system.

4. The blood glucose management system according to claim 1, characterized in that, The data collection section collects blood glucose data from multiple patients. The data provider provides the patient's blood glucose-related information through the monitoring board.

5. The blood glucose management system according to claim 1, characterized in that, The data collection unit collects the blood glucose data from the manufacturer's server of the continuous glucose meter via an application programming interface.

6. The blood glucose management system according to claim 1, characterized in that, Blood glucose variables include time within a strict range. The "strict time range" refers to the proportion of time during which blood glucose levels are within a specified range. The upper limit of the specified range is set to a value between 130 mg / dL and 150 mg / dL. The lower limit of the specified range is set to a value between 60 mg / dL and 90 mg / dL.

7. The blood glucose management system according to claim 1, characterized in that, Blood glucose variables include blood glucose risk indices. The blood glucose risk index is calculated based on a hypoglycemia risk index and a hyperglycemia risk index, with the hypoglycemia risk index being assigned a higher weighting value than the hyperglycemia risk index.

8. The blood glucose management system according to claim 1, characterized in that, The data analysis unit obtains the patient's clinical information from the legacy system. The collected blood glucose data and clinical information are analyzed using a machine learning model to predict the patient's risk of developing abnormal blood glucose levels within a specific time period. The clinical information includes at least one of the patient's demographic information, diagnostic information, prescription information, medication history, and disease history. The blood glucose management system also includes a notification unit that provides notifications based on the predicted risk level.

9. The blood glucose management system according to claim 8, characterized in that, The machine learning model includes: The encoder generates variable embeddings corresponding to the blood glucose variables; An interactive encoder generates an interactive embedding corresponding to a predetermined blood glucose variable pair based on the variable embedding. A variable weighting estimator estimates patient-specific variable weights for the blood glucose variable based on the clinical information. An interactive weighted value estimator, based on the clinical information, estimates patient-specific interactive weighted values ​​for each pair of blood glucose variables; An aggregator generates a composite embedding by aggregating the variable embeddings reflecting the weighted values ​​of the said variables and the interaction embeddings reflecting the interaction weighted values; and The predictor predicts the risk level based on the comprehensive embedding.

10. The blood glucose management system according to claim 8, characterized in that, When the predicted risk level exceeds a threshold, the notification unit retains the provision of the notification until the preset waiting timer expires. Before the waiting timer expires, if an event is detected that increases the risk of the patient experiencing the abnormal blood glucose level, the notification unit immediately provides the notification. Before the waiting timer expires, if an event is detected that reduces the risk of the patient experiencing the abnormal blood glucose level, the notification unit resets the waiting timer and predicts the risk level again using the machine learning model.

11. The blood glucose management system according to claim 1, characterized in that, The monitoring board includes: The first interface displays the blood glucose variable value; The second interface displays the patient's blood glucose-related information according to time periods; The third interface displays the proportion of time the patient's blood glucose level remained within the target range; and The fourth interface displays the patient's daily blood glucose profile.

12. The blood glucose management system according to claim 1, characterized in that, The blood glucose management system includes a backend server serving the front end, a rapid medical interoperability resource relay server, a gateway server, a rapid medical interoperability resource server, and an internal integration server. The data collection unit, the data analysis unit, and the data provision unit are implemented within the internal integration server. The rapid medical interoperability resource server and the internal integration server are located together with the legacy system within the medical institution's internal network. The backend server serving the front end transmits health data received from the patient's terminal to the rapid medical interoperability resource relay server. The rapid medical interoperability resource relay server converts the health data received from the back-end server serving the front end according to the rapid medical interoperability resource protocol and transmits it to the gateway server. The gateway server transmits the converted health data to the rapid medical interoperability resource server located in the internal network. The rapid medical interoperability resource server transmits the converted health data to the internal integration server.

13. A blood glucose management method using a continuous glucose meter, executed by at least one processor of a blood glucose management system, characterized in that, Includes the following steps: Collect blood glucose data measured using a continuous glucose meter worn by the patient; The collected blood glucose data is analyzed to calculate multiple blood glucose variable values ​​representing the patient's blood glucose status; as well as The blood glucose variable values ​​are provided by the monitoring board. The blood glucose management system is linked with the medical institution's existing system. The blood glucose management system performs processing based on the blood glucose data independently of the legacy system.

14. The blood glucose management method using a continuous glucose meter according to claim 13, characterized in that, The steps for collecting the blood glucose data include the following: collecting the blood glucose data from the manufacturer's server of the continuous glucose meter via an application programming interface.

15. The blood glucose management method using a continuous glucose meter according to claim 13, characterized in that, It also includes the following steps: The patient's clinical information is obtained from the legacy system; The collected blood glucose data and clinical information are analyzed using a machine learning model to predict the patient's risk of developing abnormal blood glucose levels within a specific time period; and Notifications are provided based on the predicted level of risk. The clinical information includes at least one of the patient's demographic information, diagnostic information, prescription information, medication history, and disease history.

16. The blood glucose management method using a continuous glucose meter according to claim 13, characterized in that, The monitoring board includes: The first interface displays the blood glucose variable value; The second interface displays the patient's blood glucose-related information according to time periods; The third interface displays the proportion of time the patient's blood glucose level remained within the target range; and The fourth interface displays the patient's daily blood glucose profile.

Images