Blood glucose risk prediction and management system using blood glucose and personalized data

Abstract

The present invention relates to a blood glucose risk prediction and management system using blood glucose and personalized data. The blood glucose risk prediction and management system using blood glucose and personalized data includes: a user terminal which receives measurement data from a blood glucose measurement device and receives correlation factors as input; and a blood glucose management server that determines search conditions including two or more of a blood glucose measurement time, a personal state, and the correlation factors, and searches for a standard pattern among a plurality of standard patterns, compares the measurement data with the searched standard pattern to determine whether the user health state is abnormal, and if the user health state is determined to be abnormal, collects additional information and analyzes the user health state to determine whether the user health state is abnormal, and transmits an alarm for reporting the occurrence of an abnormal state to the user terminal.

Description

Blood Sugar Risk Prediction and Management System Using Blood Sugar and Personalized Data

[0001] The present invention relates to a blood sugar risk prediction and management system using blood sugar and personalized data.

[0002] Patients with type 2 diabetes regularly visit the hospital every three months to have their blood sugar and glycated hemoglobin levels measured. Glycated hemoglobin represents the average blood sugar level over the past two to three months and is used as an indicator of good blood sugar control. However, average levels make it difficult to assess changes in blood sugar levels due to factors such as meals, fasting, and medications, making them ineffective in preventing and managing hypoglycemia and hyperglycemia.

[0003] In particular, it's crucial for diabetes patients to manage their blood sugar levels stably to prevent complications. To achieve this, continuous glucose monitoring (CGM) devices, which automatically measure blood sugar levels periodically, are used, and the Time-In-Range (TIR) ​​is used to determine how well blood sugar levels remain within target range. However, CGMs require permanent wear of the sensor, which can be a significant financial burden for patients not covered by health insurance.

[0004] The embodiment of the present invention defines a pattern according to individual characteristics to analyze abnormalities and blood sugar change trends in specific situations, thereby expecting effects similar to those of CGM use, and provides services such as suggesting appropriate health management guidelines for each situation, guiding hospital treatment, and notifying the attending physician.

[0005] According to one aspect of the present invention, a blood sugar risk prediction and management system using blood sugar and personalized data is provided. The blood sugar risk prediction and management system using blood sugar and personalized data may include a blood sugar management server that receives measurement data from a blood sugar measurement device, determines a search condition including a user terminal receiving a correlation factor, a blood sugar measurement time, a personal condition, and two or more of the correlation factors to search for a standard pattern from a plurality of standard patterns, compares the measurement data with the searched standard pattern to determine whether the user's health condition is abnormal, and if the determination result is abnormal, collects additional information to analyze the user's health condition to determine whether the condition is abnormal, and transmits an alarm to notify the user terminal of the occurrence of the abnormal condition. Here, the standard pattern represents a blood sugar change over time classified by a combination of a plurality of conditions from clinically collected measurement data, and the plurality of conditions are combinations of the blood sugar measurement time, the correlation factor, and the personal condition, and the correlation factor includes a meal time and a food type, and the personal condition may include age, gender, and BMI.

[0006] In one embodiment, the blood sugar management server may include a standard pattern storage unit in which the plurality of standard patterns are stored, a first abnormality detection unit that determines a search condition including two or more of the blood sugar measurement time, the personal condition, and the correlation factor to search for a standard pattern from the plurality of standard patterns, and compares the measurement data with the searched standard pattern to determine whether the user's health condition is abnormal, an additional information collection unit that outputs a questionnaire question through a chatbot dialogue window executed on the user terminal and extracts symptoms from user discourse input in response thereto, a second abnormality detection unit that compares the collected symptoms with predicted symptoms registered in relation to the user to determine whether the user's health condition is abnormal, and a situation analysis and notification unit that transmits the alarm to the user terminal.

[0007] In one embodiment, the first abnormal state detection unit collects the blood sugar measurement time, the correlation factor, the personal condition, and the measurement data, and determines whether the measurement data is a fasting blood sugar level or a postprandial blood sugar level based on the blood sugar measurement time, and if the measurement data is a fasting blood sugar level, determines whether the measurement data was measured at dawn, and if the measurement data is a fasting blood sugar level measured at dawn, searches for a standard pattern with fasting blood sugar level, a dawn range, and personal condition, and if the measurement data is a fasting blood sugar level measured at a time other than dawn, searches for a standard pattern with fasting blood sugar level and personal condition, and if the measurement data is a postprandial blood sugar level, searches for a standard pattern with postprandial blood sugar level, a correlation factor, and personal condition, and analyzes the user's health condition with the searched standard pattern.

[0008] In one embodiment, the measurement data may be time series measurement data collected from before the analysis time point for the same user.

[0009] In one embodiment, the second abnormal condition detection unit can determine the occurrence of the predicted symptom and the non-predicted symptom.

[0010] In one embodiment, the additional information acquisition unit may include a chatbot module that receives a user discourse including information related to a health condition and generates a questionnaire question to check whether the user has a similar symptom, a user discourse analysis module that analyzes the user discourse to classify it as one of positive, negative, and neutral, and extracts symptoms from the user discourse classified as positive and negative and registers them in an additional information list, wherein positive means that the user has the symptom, negative means that the user does not have the symptom, and neutral means that it is impossible to determine whether the user has the symptom, and a symptom generation module that learns about diseases and symptoms related thereto and constructs symptoms by vector embedding them in space to form clusters by disease, and provides similar symptoms to symptoms registered in the additional information list based on similarity.

[0011] In one embodiment, the system further includes a medical staff terminal that receives notification of the abnormal condition and inputs an opinion based on the notified abnormal condition, and can transmit the alarm to the user terminal based on the opinion received from the medical staff terminal.

[0012] According to another aspect of the present invention, a method for predicting and managing blood sugar risk using blood sugar and personalized data is provided. The method for predicting and managing blood sugar risk using blood sugar and personalized data may include the steps of collecting blood sugar measurement timing, personal status, correlation factors, and measurement data; determining a search condition including at least two of the blood sugar measurement timing, the personal status, and the correlation factors to search for a standard pattern among a plurality of standard patterns; comparing the measurement data with the searched standard patterns to determine whether the user's health condition is abnormal; if the determination result is abnormal, collecting additional information to analyze the user's health condition to determine whether the user's health condition is abnormal; and transmitting an alarm to notify a user terminal of the occurrence of the abnormal condition. Here, the standard pattern represents a blood sugar change over time obtained by classifying clinically collected measurement data by a combination of a plurality of conditions, and the plurality of conditions are a combination of the blood sugar measurement timing, the correlation factors, and the personal status. The correlation factors include a meal time and a food type, and the personal status may include age, gender, and BMI.

[0013] In one embodiment, if the determination result is an abnormal state, the step of collecting additional information and analyzing the user's health state to determine whether the state is abnormal may include the steps of outputting a questionnaire question through a chatbot dialogue window running on the user terminal, receiving a user discourse as a response to the questionnaire question, classifying the user discourse as one of positive, negative, and neutral, wherein positive is a case where the user has a symptom, negative is a case where the user does not have a symptom, and neutral is a case where it is impossible to determine whether the user has a symptom, extracting a symptom from the user discourse classified as positive or negative and registering it in an additional information list, and comparing the symptom registered in the additional information list with a predicted symptom registered in association with the user to determine whether the user's health state is abnormal.

[0014] In one embodiment, a method for predicting and managing blood sugar risk using blood sugar and personalized data may further include a step of notifying a medical staff terminal of the abnormal condition.

[0015] According to an embodiment of the present invention, by defining a pattern according to individual characteristics, the presence or absence of abnormalities and blood sugar change trends in specific situations can be analyzed, thereby expecting an effect similar to that of CGM use, and services such as suggesting appropriate health management guidelines for each situation, guiding hospital treatment, and notifying the attending physician can be provided.

[0016] The present invention is described below with reference to embodiments illustrated in the accompanying drawings. To facilitate understanding, identical components are assigned the same reference numerals throughout the accompanying drawings. The components depicted in the accompanying drawings are merely exemplary embodiments implemented to illustrate the present invention and are not intended to limit the scope of the present invention. In particular, the accompanying drawings may slightly exaggerate some of the elements depicted in the drawings to facilitate understanding of the invention.

[0017] FIG. 1 schematically illustrates the physical components of a blood sugar management server for implementing a blood sugar risk prediction and management system using blood sugar and personalized data according to an embodiment of the present invention.

[0018] FIG. 2 schematically illustrates the physical components of a user terminal for implementing a blood sugar risk prediction and management system using blood sugar and personalized data according to an embodiment of the present invention.

[0019] FIG. 3 is a drawing for exemplarily explaining a standard pattern generation process according to an embodiment of the present invention.

[0020] Figure 4 is a drawing exemplarily showing a standard pattern according to an embodiment of the present invention.

[0021] FIG. 5 is a diagram illustrating a blood sugar risk prediction and management system using blood sugar and personalized data according to an embodiment of the present invention.

[0022] FIG. 6 is a diagram exemplifying a blood sugar management method executed by a blood sugar risk prediction and management system using blood sugar and personalized data according to an embodiment of the present invention.

[0023] Figure 7 is a drawing exemplarily showing an analysis process using a standard pattern according to an embodiment of the present invention.

[0024] FIG. 8 is a drawing exemplarily showing an additional information acquisition unit that acquires additional information according to an embodiment of the present invention.

[0025] FIG. 9 is a diagram exemplarily illustrating a process of extracting additional information from user discourse according to an embodiment of the present invention.

[0026] The present invention is susceptible to various modifications and embodiments. Specific embodiments are illustrated in the drawings and described in detail. However, this is not intended to limit the present invention to specific embodiments, but rather to encompass all modifications, equivalents, and alternatives falling within the spirit and technical scope of the present invention. In particular, the functions, features, and embodiments described below with reference to the accompanying drawings may be implemented alone or in combination with other embodiments. Therefore, it should be noted that the scope of the present invention is not limited to the forms illustrated in the accompanying drawings.

[0027] Throughout the attached drawings, identical or similar elements are referenced using the same drawing reference numerals. However, for convenience of explanation and understanding, the drawings are somewhat exaggerated.

[0028] FIG. 1 schematically illustrates the physical components of a blood sugar management server for implementing a blood sugar risk prediction and management system using blood sugar and personalized data according to an embodiment of the present invention.

[0029] Figure 1 schematically illustrates the physical configuration of a blood sugar management server (100). The blood sugar management server (100) is comprised of one or more servers, and each server is equipped with one or more processors (110) and memory (120) for executing programs suited to their respective purposes. Additionally, the blood sugar management server (100) may include an internal or external storage device, and the storage device may record and reproduce information electrically, magnetically, or optically. Meanwhile, the blood sugar management server (100) may communicate with other servers within the same system or with servers within other systems via a communication network.

[0030] In this specification, for convenience of explanation, the entity that processes each step or task is expressed as a system, but in reality, all processing can be performed by a processor (110) and memory (120) that execute commands by a program installed in the blood sugar management server (100).

[0031] A blood sugar management server (100) having a basic configuration illustrated in FIG. 1 includes a processor (110), a memory (120), a data storage device (130), an input / output device (140), and a network device (150).

[0032] The processor (110) has data processing capabilities and controls the overall operation of the user terminal. Specifically, the processor (110) controls components to perform the basic functions of the blood sugar management server (100) and controls components to execute programs. To this end, the processor (110) cooperates with the installed operating system to execute the programs.

[0033] Depending on the type of server, the memory (120) may include volatile memory and non-volatile memory. The memory (120) is controlled for operation by a memory controller (not shown). Volatile memory is, for example, RAM (Random Access Memory), and non-volatile memory is, for example, ROM (Read Only Memory), Flash memory, etc. The memory (120) stores one or more files constituting an operating system, middleware, API, etc. necessary for performing basic functions of the user terminal, a program executed by the processor (110), etc., and stores data necessary during execution of the program.

[0034] The blood sugar management server (100) may additionally include a data storage device (130). The data storage device (130) may be removable or fixed, and may be, for example, a magnetic disk, an optical disk, etc., but is not limited thereto. The data storage device (130) may store a program for implementing a real-time skin analysis method using facial photographs, data for executing the program, a database, etc.

[0035] The blood sugar management server (100) may additionally include an input / output device (140). Input devices may include, but are not limited to, a keyboard, a mouse, a touchpad, etc. Output devices may include, but are not limited to, a display, a speaker, a printer, etc.

[0036] The blood sugar management server (100) may additionally include a network device (150). The network device (150) enables the server (100) to transmit and receive data with an external device via a communication network. The data transmitted and received by the blood sugar management server (100) may include, but is not limited to, computer-related commands, data structures, program modules, etc. The network device (150) converts data output from the blood sugar management server (100) into electrical / optical signals or wireless signals according to a predetermined communication protocol and transmits the converted data, and transmits the received signals to the processor (110).

[0037] FIG. 2 schematically illustrates the physical components of a user terminal for implementing a blood sugar risk prediction and management system using blood sugar and personalized data according to an embodiment of the present invention.

[0038] The user terminal (200) illustrated in FIG. 2 is a portable communication terminal, for example, a smart phone, a tablet, a laptop, etc., and may include a processor (210), a wireless communication module (220), a display (230), a display controller (235), a memory (270), a memory controller (275), an input / output device (260), an auxiliary communication module (225), a camera (240), and a power source (250). For reference, the components of the user terminal (200) illustrated in FIG. 2 are only examples, and addition or omission of one or more components is possible. Therefore, it should be understood that embodiments according to the present invention are not necessarily implementable only in the user terminal (200) having the components illustrated in FIG. 2.

[0039] The user terminal (200) communicates with an external device via a wireless communication module (220), an auxiliary communication module (225), and / or a connector. The external communication device includes another user terminal (200), a computer, or a server capable of communicating via a communication network. The communication network should be interpreted to include not only wired and wireless networks such as the Internet or wireless LAN, but also cables such as USB that connect two or more devices, or short-range communication networks such as Bluetooth.

[0040] The processor (210) is a processor (210) equipped with a data operation function and controls the overall operation of the user terminal (200). Specifically, the processor (210) controls the components so that the user terminal (200) performs basic functions regardless of the presence or absence of user input, and controls the components so that commands according to user input are executed. To this end, the processor (210) cooperates with the installed operating system to execute a computer program. Here, the computer program includes not only a program selectively executed by the user or an external device, but also a program executed regardless of the user's selection to perform the basic functions of the terminal. Specifically, the processor (210) processes voice / sound / data input in the form of an electrical signal through a wireless communication module (220) and / or an auxiliary communication module (225). The processed voice / sound / data is output through a display (230) or an input / output device (260). The user terminal (200) displays one or more computer programs that can be selected by the user and outputs a user interface on the display (230) that executes the selected computer programs so that the user can use them. Through the user interface, the user can select and execute the computer programs and perform necessary tasks.

[0041] The wireless communication module (220) transmits or receives a wireless signal according to one or more communication methods through an antenna (not shown) under the control of the processor (210). Here, the wireless communication module (220) can communicate with a base station according to a mobile communication standard such as W-CDMA, LTE, etc., for example. The wireless communication module (220) converts an electrical signal output from the processor (210) into a wireless signal and transmits it to an external device through a communication network, and converts a wireless signal received through the communication network into an electrical signal and inputs it to the processor (210). The wireless signals transmitted and received by the wireless communication module (220) may be, for example, voice calls, data, text / multimedia messages, etc.

[0042] The auxiliary communication module (225) transmits or receives wired or wireless signals according to one or more communication methods under the control of the processor (210). The auxiliary communication module (225) can support various wired and wireless communication methods. For example, it can support not only short-range wireless communication methods such as wireless LAN, and short-range wireless communication methods such as Bluetooth, Zigbee, Wifi-Direct, and NFC, but also wired communication methods such as LAN and USB.

[0043] The user terminal (200) includes a display (230) and a display controller (235). The display (230) may be implemented as, for example, an LCD, an OLED, etc. Meanwhile, there may be one or more displays (230), and for example, if there are two displays (230), each display (230) may be placed on the front and back of the user terminal (200). The display controller (235), under the control of the processor (210), processes an electric signal corresponding to a screen output from the processor (210) and controls the display (230) so that the electric signal is output to the display (230). Meanwhile, the display (230) may be a touch screen that can receive a touch input from a user. In this case, the touch screen can operate as an input device. If the display (230) is a touch screen, the display controller (235) can generate coordinates corresponding to a user's touch detected on the display (230). The generated coordinates are input to the processor (210). Here, if the touch is on more than one point at the same time, there can be more than one coordinate, and if the user drags, there can be more than two coordinates.

[0044] The camera (240) includes one or more cameras that capture at least one of still images and moving images under the control of the processor (210). The still images and / or moving images captured by the camera (240) are stored in the memory (270). When there are two or more cameras, one of the plurality of cameras may be placed on the front of the user terminal (200), and the rest may be placed on the rear of the user terminal (200).

[0045] The power source (250) may include one or more built-in batteries or may be a commercial power source (250) located externally to the user terminal (200), and supplies power to components of the user terminal (200) under the control of the processor (210). Meanwhile, when power is supplied through the commercial power source (250), the user terminal (200) may be connected to the commercial power source (250) located externally through a connector.

[0046] The input / output device (260) may include, for example, buttons, microphones, speakers, vibration motors, connectors, etc. Meanwhile, the input / output device (260) illustrated in FIG. 2 is an example of devices that are practically commonly provided in various types of user terminals (200), and it should be noted that various input / output devices (260) may be optionally provided or omitted.

[0047] The button is located on the outer surface of the user terminal (200) and is used by the user to input a command to select and / or execute a function of the user terminal (200). The button may include, for example, an on / off button, a home button, a volume button, etc. When the user presses any button, the button outputs an electrical signal to the processor (210).

[0048] A microphone receives voice or sound and outputs a corresponding electrical signal.

[0049] A speaker converts an input electrical signal into voice or sound and outputs it. The speaker converts an electrical signal received from an external device or from a processor (210) into voice or sound. Here, the processor (210) can output a corresponding sound through the speaker when a user selects or executes a specific function of the user terminal (200). For example, when playing a video, a voice file included in the video can be output through the speaker.

[0050] A haptic device converts an input electrical signal into a haptic motion and outputs it. A representative haptic device is a vibration motor. The vibration motor outputs a vibration corresponding to the input electrical signal. The haptic device converts an electrical signal received from an external device or from a processor (210) into a haptic motion. Here, the processor (210) can output a corresponding haptic motion through the haptic device when a user selects or executes a specific function of the user terminal (200). For example, when a user presses a button, a vibration can be output each time the button is pressed.

[0051] A connector is an interface that connects a user terminal (200) to an external device. In the simplest example, the connector may be a power connector for connecting the user terminal (200) to an external power source (250) or a speaker / earphone jack for connecting to a speaker. Meanwhile, in the case of a serial high-speed interface such as USB, data and power can be supplied to the user terminal (200) simultaneously through a cable connected to the connector, and the user terminal (200) can transmit data to an external device. In addition, in the case of a wired LAN connector, for example, the user terminal (200) can transmit and receive data with an external device connected to the Internet through a cable connected to the connector.

[0052] The user terminal (200) includes a memory controller (275) and a memory (270). The memory (270) includes a volatile memory (270) and a non-volatile memory (270). The volatile memory (270) is, for example, a RAM (Random Access Memory), and the non-volatile memory (270) is, for example, a ROM (Read Only Memory), a Flash memory, etc. The memory (270) stores one or more files constituting an operating system, middleware, API, etc. necessary for performing the basic functions of the user terminal (200), a computer program executed by the processor (210), etc., and stores data necessary during the execution of the computer program. In addition, the memory (270) also stores information necessary for configuring a user interface, for example, an image of a computer program, properties of a computer program, a processing routine, etc. The memory controller (275) controls the operation of the memory (270). For example, the memory controller (275) controls the memory (270) to store an electrical signal corresponding to data output by the processor (210), and controls the memory (270) to output data requested by the processor (210).

[0053] The user terminal (200) having the exemplary configuration described above will be described with reference to the blood sugar risk prediction and management system using blood sugar and personalized data according to an embodiment of the present invention below. Even if the physical components illustrated in FIG. 2 are not explicitly cited, it should not be construed that the blood sugar risk prediction and management system using blood sugar and personalized data according to an embodiment of the present invention is implemented and executed solely through data processing without a physical configuration.

[0054] FIG. 3 is a drawing for exemplarily explaining a standard pattern generation process according to an embodiment of the present invention, and FIG. 4 is a drawing for exemplarily showing a standard pattern according to an embodiment of the present invention.

[0055] A standard pattern represents blood glucose changes over time based on a combination of multiple conditions. The multiple conditions may include one or a combination of the following: blood glucose measurement timing, correlated factors, individual status, and symptoms. The blood glucose measurement timing includes the time of measurement and whether or not a meal was taken (fasting or postprandial). Correlated factors include, for example, the timing of the meal (breakfast, lunch, dinner) and the type of food. Individual status includes age, sex, and BMI (Body Mass Index). Symptoms may include symptoms caused by blood glucose changes and symptoms of blood glucose-related diseases. BMI can be calculated from weight and height.

[0056] Standard patterns can be generated based on clinically collected measurement data (hereinafter referred to as clinical data). In "Chinese diabetes datasets for data-driven machine learning" (Nature Scientific Data, 2023), clinical data were obtained by measuring blood glucose levels of 12 type 1 diabetes patients and 100 type 2 diabetes patients using continuous glucose monitoring (CGM) at 15-minute intervals for 14 days.

[0057] Patients are categorized by individual status, namely age, gender, and BMI. Age is categorized as [0-50, 51-60, 61-70, 70+], gender as [male, female], and BMI as [0-25, 25-30, 30+]. In other words, patients are assigned to one of 24 categories based on individual status, allowing for more detailed analysis by demographic group.

[0058] Clinical data is related to the timing of blood glucose measurement, individual status, and correlated factors. Fasting blood glucose is measured after a certain period of time without food intake. Postprandial blood glucose is measured after a certain period of time after a meal. Meal times can be breakfast, lunch, or dinner, and food types can be converted into characteristic codes representing the main components of the food: carbohydrates, protein, vegetables and fruits, dairy products, and fat / oil / sugar. For example, if a patient's meal contains "Egg 68g, Coarse grain steamed bread 129g, Boiled vegetables 119g," the food types would be converted to "Protein=1," "Carbohydrate=1," and "Vegetables and fruits=1," since "Egg" is protein, "Coarse grain steamed bread" is carbohydrate, and "Boiled vegetables" is vegetable.

[0059] After obtaining clinical data on fasting and postprandial blood glucose levels, each patient's clinical data is classified into one of 24 categories based on their individual status. After classification, the average of all blood glucose values ​​within that category is calculated to create a standard pattern. This classification allows for a comprehensive review of blood glucose patterns and dietary responses across various demographic groups, helping to identify potential variations in diabetes management and treatment effectiveness.

[0060] - Creating a standard fasting blood sugar pattern

[0061] When creating a standard fasting blood glucose pattern, essential information is individual status. For example, if you find clinical data for a diabetic patient with the following characteristics: age=20; gender=male; BMI=21, this patient falls into the categories of "age (0-50), gender (male), BMI (0-25)". When collecting data from this many patients, the average value of the data can be used as the standard pattern for that category, generating a total of 24 standard fasting blood glucose patterns.

[0062] Figures 4 (a) and (b) are standard patterns of fasting blood sugar levels. (a) is the standard pattern of normal fasting blood sugar levels. If the patient's personal condition and measurement data are age = 20; gender = male; BMI = 21; current blood sugar level = 90; and blood sugar level type = 'fasting blood sugar level', it is considered normal. (b) is the standard pattern of the dawn phenomenon. If the patient's personal condition and measurement data are age = 55; gender = male; BMI = 23; current blood sugar level = 100; and blood sugar level type = 'fasting blood sugar level', it is considered possible for the dawn phenomenon to occur. The measurement time is from 12:00 to 6:00 a.m. and until the moment of breakfast, and blood sugar measurement data is displayed every 15 minutes. The dawn phenomenon occurs when the fasting blood sugar level is 140 mg / dL or higher between 3:00 a.m. and 8:00 a.m., and if the blood sugar level is very high, excessive thirst, frequent urination, dry mouth, nausea, vomiting, confusion, and rapid breathing may occur. When the possibility of a dawn phenomenon occurring is predicted, an alarm is sent to the user.

[0063] - Creating a standard postprandial blood sugar pattern

[0064] Each patient's postprandial blood glucose level is calculated. Observations are conducted at 15-minute intervals for two hours immediately after a meal. Each patient's clinical data is separated into breakfast, lunch, and dinner, allowing for a more focused analysis of postprandial blood glucose changes. Clinical data are collated based on individual status and correlated factors, and the average value of the clinical data is used as the standard pattern for that category. A total of 455 standard patterns can be generated: 183 standard patterns for breakfast, 155 standard patterns for lunch, and 147 standard patterns for dinner. Figure 4 (c) and (d) illustrate standard patterns for postprandial blood glucose levels. (c) is a standard pattern for general postprandial blood glucose levels. If the patient's individual status and measurement data are age = 20; gender = male; BMI = 20; current blood glucose value = 90; and blood glucose type = 'postprandial blood glucose level,' the patient is considered normal. (d) is a standard pattern for sugar crash levels. If the patient's individual status and measurement data are age = 20; gender = female; BMI = 20; If your current blood sugar level is 70 and your blood sugar type is "postprandial blood sugar," a sugar crash is considered likely. A sugar crash occurs when your blood sugar level drops below 70 mg / dL for up to four hours after a meal. Sudden drops in blood sugar levels can cause fatigue, weakness, irritability, shaking, dizziness, and difficulty concentrating. If a sugar crash is predicted, an alert is sent to the user.

[0065] FIG. 5 is a diagram illustrating a blood sugar risk prediction and management system using blood sugar and personalized data according to an embodiment of the present invention.

[0066] Referring to FIG. 5, a blood sugar risk prediction and management system using blood sugar and personalized data includes a blood sugar management server (100), and the blood sugar management server (100) is connected to user terminals (200) of multiple users requiring blood sugar management and medical staff terminals (202) of medical staff managing users' blood sugar through a communication network.

[0067] The user terminal (200) can transmit measurement data received from a blood sugar measurement device (201) that measures blood sugar, one or a combination of correlation factors and symptoms input by the user, to the blood sugar management server (100). The user regularly visits a hospital to check blood sugar, glycated hemoglobin, weight, height, etc., and usually checks fasting blood sugar and postprandial blood sugar using the blood sugar measurement device (201). Meanwhile, the user terminal (200) can display content and / or alarms transmitted by the blood sugar management server (100) to prompt the user to take immediate action.

[0068] The blood sugar management server (100) provides blood sugar management services to users based on measurement data. The blood sugar management server (100) selects an appropriate standard pattern from among a plurality of standard patterns based on one or a combination of blood sugar measurement timing, correlation factors, personal condition, and symptoms, performs a primary analysis to detect abnormal conditions by comparing the selected standard pattern with the measurement data, collects additional information through an additional information acquisition unit (104), and performs a secondary analysis to detect abnormal conditions. The analysis results can then be provided to medical staff.

[0069] The blood sugar management server (100) may include a standard pattern storage unit (101), a first abnormal condition detection unit (102), a second abnormal condition detection unit (103), an additional information acquisition unit (104), and a situation analysis and notification unit (105).

[0070] The standard pattern storage unit (101) stores a plurality of standard patterns. The standard patterns are generated based on clinically collected clinical data.

[0071] The first abnormal condition detection unit (102) selects one of a plurality of standard patterns based on the blood glucose measurement time, correlation factors, and individual conditions, and detects an abnormal condition based on the selected standard pattern and measurement data. If the collected measurement data does not conform to the standard pattern, the first abnormal condition detection unit (102) can notify the second abnormal condition detection unit (103) and / or the situation analysis and notification unit (105) of the occurrence of the abnormal condition.

[0072] When an abnormal condition occurs, the second abnormal condition detection unit (103) collects additional information necessary to analyze the user's health condition through the additional information acquisition unit (104), and detects the abnormal condition based on the collected additional information. When the additional information is collected, the second abnormal condition detection unit (103) determines whether a predicted symptom registered in a medical database related to the user has occurred, and can additionally determine whether a symptom other than the predicted symptom (hereinafter, "non-predicted symptom") has occurred. A predicted symptom is a symptom that may occur when a diabetes-related disease that the user has worsens, and when a predicted symptom has occurred, the user's health condition is in an abnormal state or may progress to an abnormal state. A non-predicted symptom can be used to determine whether complications other than the current diabetes-related disease have occurred.

[0073] Additional information primarily includes symptoms necessary to assess a user's health status, and may also include life events that impact the user's health status and / or correlated factors that the user may have missed. Life events may include drinking, smoking, medication, and stress. Like symptoms, life events are also used to assess a user's health status, and are therefore collectively referred to as "symptoms" below.

[0074] The situation analysis and notification unit (105) notifies the user terminal (200) and / or the medical staff terminal (202) when an abnormal condition is detected based on the analysis results of the first abnormal condition detection unit (102) and the second abnormal condition detection unit (103), and manages the analysis results collected over a certain period of time.

[0075] FIG. 6 is a diagram exemplifying a blood sugar management method executed by a blood sugar risk prediction and management system using blood sugar and personalized data according to an embodiment of the present invention.

[0076] In S10, the blood sugar management server (100) analyzes the user's health status based on the blood sugar measurement time, correlation factors, personal condition, and measurement data. The first abnormal condition detection unit (102) selects one of a plurality of standard patterns based on the blood sugar measurement time, correlation factors, and personal condition, and detects the abnormal condition based on the selected standard pattern and measurement data. The standard pattern includes changes in fasting blood sugar and postprandial blood sugar defined according to the personal condition. Additionally or optionally, the standard pattern may also include changes in blood sugar that occur when, for example, sugar crush or the dawn phenomenon occurs.

[0077] When an abnormal condition is detected, the first abnormal condition detection unit (102) transmits a first alarm to the user terminal (200) through the situation analysis and notification unit (105). The first alarm provides the user with nutritional and exercise treatment methods, etc. to induce the user to take care of his or her health, or to take immediate action. When the blood sugar management server (100) detects an abnormal condition in the first analysis, it performs a second analysis. Meanwhile, the first abnormal condition detection unit (102) can additionally apply the treatment guideline criteria defined in the medical database to detect an abnormal condition based on measurement data other than blood sugar, for example, blood pressure.

[0078] In S20, the blood sugar management server (100) collects additional information through the additional information acquisition unit (104) to perform a secondary analysis of the user's health status. The second abnormal state detection unit (103) collects additional information through an interactive questionnaire conducted by the chatbot module (310). The collection of additional information is described below with reference to FIGS. 8 and 9. The second abnormal state detection unit (103) can detect an abnormal state by analyzing the user's health status based on symptoms. Meanwhile, the user's health status can be analyzed using a standard pattern reselected based on the additional information collected by the second abnormal state detection unit (103) to detect an abnormal state. An abnormal state is a case where an emergency situation has occurred or is likely to occur for the user. In addition, the second abnormal state detection unit (103) can additionally apply the treatment guideline criteria defined in the medical database to detect an abnormal state based on measurement data other than blood sugar, for example, blood pressure.

[0079] When an abnormal condition is detected, the second abnormal condition detection unit (103) transmits a second alarm to the user terminal (200) through the situation analysis and notification unit (105). The second alarm prompts the user to take immediate action.

[0080] In S30, the blood sugar management server (100) notifies the medical staff terminal (202) of an abnormal condition. The abnormal condition notification may include measurement data and collected additional information. The collected additional information includes symptoms registered in the additional information list.

[0081] In S40, the blood sugar management server (100) takes action based on the medical staff's opinion. Upon receiving an abnormal condition notification, the medical staff can retrieve measurement data and additional information regarding the user in whom the abnormal condition has been detected via the medical staff terminal (202) and determine whether an emergency situation exists. The measurement data may include measurement data collected in the past since the abnormal condition occurred. If an emergency situation is determined, the medical staff or the blood sugar management server (100) may make a phone call and / or send a third alarm to the user to check the user's current condition and / or request a visit to the hospital. If it is determined that the situation is not an emergency or that the user can take appropriate action, a third alarm is sent to prompt the user to take immediate action.

[0082] Figure 7 is a drawing exemplarily showing an analysis process using a standard pattern according to an embodiment of the present invention.

[0083] In S11, the first abnormal condition detection unit (102) of the blood sugar analysis server (100) collects blood sugar measurement timing, correlation factors, individual status, and measurement data. The individual status is associated with the user terminal (200) and stored in the blood sugar analysis server (100) or an app server (not shown) that is linked to the blood sugar analysis server (100), and the blood sugar measurement timing, correlation factors, and measurement data are provided by the user terminal (200).

[0084] In S12, the first abnormal condition detection unit (102) determines whether the measured data is a fasting blood sugar level by checking the blood sugar measurement time. Additionally, the first abnormal condition detection unit (102) can determine whether the measured data is a fasting blood sugar level by referring to a correlation factor.

[0085] In S13, if the measurement data is a fasting blood sugar level, the first abnormal condition detection unit (102) determines whether the measurement time falls within the early morning range.

[0086] In S14, if the measurement data is a fasting blood sugar level measured at dawn, the first abnormal condition detection unit (102) determines the fasting blood sugar level, the dawn range, and the individual condition as the first search conditions for finding a standard pattern.

[0087] In S15, if the measurement data is a fasting blood sugar level measured at a time other than dawn, the first abnormal condition detection unit (102) determines the fasting blood sugar level and personal condition as the second search conditions for finding a standard pattern.

[0088] In S16, the first abnormal condition detection unit (102) checks the correlation factor to confirm the type of food consumed by the user.

[0089] In S17, the first abnormal condition detection unit (102) determines postprandial blood glucose, correlation factors, and individual status as third search conditions for finding a standard pattern. The food type can be input by the user through the user terminal (200) or collected through the chatbot module (310).

[0090] In S18, the first abnormal state detection unit (102) searches for a standard pattern corresponding to any one of the first to third search conditions determined from the standard pattern storage unit (101).

[0091] In S19, the first abnormal state detection unit (102) analyzes the user's health status using the searched standard pattern. For example, the first abnormal state detection unit (102) can determine whether the measurement data falls within the normal range allowed by the searched standard pattern. Here, the measurement data may be time series measurement data collected for the same user from before the analysis time. If the measurement data falls outside the normal range, it may be determined to be an abnormal state and a first alarm may be transmitted. Meanwhile, additionally, if the measurement data falls outside the normal range and the search condition includes fasting blood sugar, the first abnormal state detection unit (102) may analyze it using the dawn phenomenon standard pattern, and if the measurement data falls outside the normal range and includes postprandial blood sugar, the first abnormal state detection unit (102) may analyze it using the sugar crash standard pattern.

[0092] Fig. 8 is a drawing exemplarily showing an additional information acquisition unit according to an embodiment of the present invention.

[0093] The additional information acquisition unit (104) includes a chatbot module (310), a user discourse analysis module (320), a single word extraction module (330), a compound word extraction module (340), and a symptom generation module (350), and may additionally include a clustering module (360).

[0094] The chatbot module (310) provides a chatbot dialogue window in which the user can input user discourse that explicitly includes additional information or is necessary to infer additional information. The user can directly input user discourse including additional information (e.g., “my hand is numb,” “I can’t see well,” etc.) or answer questionnaire questions (e.g., “yes,” “no,” etc.) into the chatbot dialogue window running on the user terminal (200). For example, when the chatbot dialogue window is first run, the user can input user discourse including symptoms, and the chatbot module (310) can generate questionnaire questions based on the analysis results of the user discourse. From then on, the user can input simple answers or input symptoms that are not included in the questionnaire questions. The chatbot module (310) transmits the user discourse entered into the chatbot dialogue window to the user discourse analysis module (320) and displays questionnaire questions including symptoms generated by the symptom generation module (350) in the chatbot dialogue window.

[0095] The user discourse analysis module (320) can analyze user discourse and classify it into three types: positive, negative, and neutral. Positive refers to a case where the user has a symptom or is contextually interpreted as being sick, negative refers to a case where the user does not have a symptom or is contextually interpreted as being healthy, and neutral refers to a case where no symptom can be identified from the user discourse. User discourse classified as positive includes symptoms that the user has (hereinafter, symptomatic), whereas user discourse classified as negative includes symptoms that the user does not have (hereinafter, asymptomatic). Symptoms can be extracted by single word extraction or compound word extraction.

[0096] Representative natural language processing models include Bidirectional Encoder Representations from Transformers (BERT), Generative Pre-trained Transformer (GPT), and Embeddings from Language Model (ELMo). The user discourse analysis module (320) can be implemented by training BERT, which is capable of bidirectional learning, to understand the context of user discourse. The user discourse analysis module (320) was trained using the Fine Tuning technique using BERT, which is a pre-trained model.

[0097] The user discourse analysis module (320) can extract additional information from user discourse classified as positive and / or negative using the single word extraction module (330) and the compound word extraction module (340). A symptom can be composed of one or more words or syllables. Here, the symptom can be a term registered in the disease database (370). The single word extraction module (330) extracts a single-word symptom from the user discourse using a keyword extraction method, and the compound word extraction module (340) can extract a symptom of two or more words from the user discourse using a symptom chunking rule. The chunking rule is a method of dividing a sentence into several types of parts of speech and creating phrases using the divided parts of speech. The symptom chunking rule can extract a symptom of two or more words by applying the chunking rule to the user discourse.

[0098] Additionally or optionally, the compound word extraction module (340) can extract additional information composed of two or more nouns, as well as additional information composed of a noun and a verb or an adjective from the user's discourse. The user can express the additional information in the form of an adjective or a combination of a verb and a noun, such as "my throat hurts" or "my lips are blue." The compound word extraction module (340) can generate one or more phrases composed of a noun and a verb or an adjective and convert them into symptoms. The extracted symptoms are registered in the additional information list.

[0099] When a new symptom is registered in the additional information list, the symptom generation module (350) outputs one or more symptoms (hereinafter, "similar symptoms") that are highly similar to the registered symptom. Similar symptoms may be output in order of similarity or together with their associated similarities. The symptom generation module (350) may output a set number of similar symptoms, and among the output similar symptoms, similar symptoms that are not included in the additional information list may be provided to the chatbot module (310).

[0100] The symptom generation module (350) is generated by vector-embedding one or more symptoms related to an arbitrary symptom into a two-dimensional or more space. The similarity is calculated using the cosine similarity method to measure the distance to other symptoms located around the input symptom, and a higher similarity indicates a closer distance to the input symptom. For example, hypoglycemia is accompanied by symptoms such as lack of energy, tremors all over the body, tremors, paleness, palpitations, decreased consciousness, fatigue, hunger, headache, anxiety, convulsions, dizziness, excitement, sweating, and shock. This means that each symptom is similar to each other and has the same context when viewed from the perspective of hypoglycemia. Therefore, the symptom generation module (350) learns by understanding the symptoms of each disease as a single context. Since the symptom generation module (350) learns by understanding the symptoms of a single disease as having the same context, the symptoms form clusters by disease according to the learning characteristics of the symptom generation module (350). Clusters of symptoms possess unique characteristics, and these characteristics encode information about the disease. That is, symptoms that are close together are more likely to be symptoms of the same disease.

[0101] The clustering module (360) determines whether the symptoms included in the additional information list form a cluster. Since a cluster is a state in which a large number of symptoms are densely packed in a small area, clustering can be determined based on density rather than area. For example, hypoglycemia is accompanied by symptoms such as lethargy, tremors, hand tremors, palpitations, decreased consciousness, fatigue, hunger, headache, anxiety, convulsions, dizziness, excitement, sweating, and shock. From the perspective of diabetes as a disease, each symptom is similar to another and shares a common context. In other words, symptoms that are close to each other in terms of distance are more likely to be diabetes-related symptoms. Cluster formation can be determined using a baseline similarity. The baseline similarity can be calculated from the average similarity of m diseases, and the similarity between two symptoms can be calculated using the cosine similarity method. If the average similarity of the collected symptoms is greater than the standard similarity, it is determined that a cluster has been formed, and if it is less than the standard similarity, it is determined that a cluster has not been formed.

[0102] The clustering module (360) calculates an average similarity for multiple symptoms included in the additional information list and compares it with a reference similarity. If the average similarity is lower than or equal to the reference similarity, the chatbot module (310) can display questions including similar symptoms through a chatbot dialogue window to conduct additional questionnaires. If the average similarity is lower than or equal to the reference similarity, the number of symptoms included in the additional information list can be compared with the reference number of symptoms. Alternatively, the clustering module (360) can first compare the number of symptoms and then compare the average similarity.

[0103] The clustering module (360) can search for one or more predicted diseases related to diabetes from a disease database using multiple symptoms included in the additional information list, and determine the predicted diseases in order of highest matching rate. For example, if there are a total of 7 symptoms included in the additional information list, and keywords related to the first to third predicted diseases are 10, 12, and 14, respectively, and all symptoms in the additional information list are included in the first to third predicted diseases, the matching rates of the first to third predicted diseases are 70% (i.e., 100 X 7 / 10), 58% (i.e., 100 X 7 / 12), and 50% (i.e., 100 X 7 / 14), respectively. If the standard matching rate is 70%, only the first predicted disease can be selected, and related information can be provided to the medical staff terminal (202) and / or the user terminal (200). As another example, assuming that there are a total of six symptoms included in the additional information list and that all other conditions remain the same, the concordance rates for the first, second, and third predicted diseases are 60%, 50%, and 43%, respectively. In this case, the clustering module (360) can select similar symptoms excluding symptoms associated with the first predicted disease, i.e., symptoms included in the additional information list, and provide them to the chatbot module (310).

[0104] FIG. 9 is a diagram exemplarily illustrating a process of extracting additional information from user discourse according to an embodiment of the present invention.

[0105] The chatbot dialogue window (311) is an area where a user inputs a response for a questionnaire. The chatbot module (310) can display a discourse (311a) indicating that the automatic questionnaire is ready to proceed in the chatbot dialogue window (311). When the automatic questionnaire starts, the user can input a user discourse (311b) including additional information into the chatbot dialogue window (311). The input user discourse (311b) is transmitted to the user discourse analysis module (320). The user discourse analysis module (320) analyzes the context of the user discourse (311b) and classifies it as positive (S50), and the simple word extraction module (330) can extract 'numbness' from the user discourse (311b) (S51). A list of additional information is created for each user when the automatic questionnaire is first executed, and can be updated when additional questionnaires are conducted. 'Numbness' is first registered in the list of additional information (S52).

[0106] Symptoms can be classified as symptomatic or asymptomatic and entered in the additional information list. Symptoms classified as unconfirmed and entered in the additional information list may be changed to symptomatic or asymptomatic during the additional interview process.

[0107] When the list of additional information is updated, a process is performed to determine whether to proceed with additional questionnaires. Since 'numbness' is registered in the list of additional information in S52, the first to nth similar symptoms associated with 'numbness' are provided by the symptom generation module (350).

[0108] The similar symptoms provided can be used for additional questioning by the chatbot module (310).

[0109] If there is one or more unidentified similar symptoms, the chatbot module (310) generates a question including the unidentified similar symptoms and displays it in the chatbot dialogue window (311). If there are two or more unidentified similar symptoms, the similar symptoms may be displayed sequentially starting from the similar symptoms with a relatively high degree of similarity. The chatbot module (310) displays a question (311c) including the first similar symptom, 'sweating', which has the highest similarity to 'numbness' among n similar symptoms, in the chatbot dialogue window (311). In response, the user inputs "I don't know." The input user discourse (311d) is transmitted to the user discourse analysis module (320). The user discourse analysis module (320) analyzes the context of the user discourse (311d) and classifies it as neutral (S53, S54).

[0110] In one embodiment, if a user discourse regarding a question including a similar symptom is classified as neutral, the chatbot module (310) can change the similar symptom into an expression that is easy for the user to understand and ask the question again. In the illustrated drawing, if the user discourse (311d) regarding 'sweating' is classified as neutral, the chatbot module (310) can display a question (311e) including an expression that is easy for the user to understand the meaning of 'sweating' in the chatbot dialogue window (311). In response, the user inputs "Yes, I sweat a lot." The input user discourse (311f) is transmitted to the user discourse analysis module (320). The user discourse analysis module (320) analyzes the context and classifies it as positive (S55). The compound word extraction module (340) can extract 'sweating' as a symptom from the user discourse (111d) (S56). The extracted 'sweating' can be added to the additional information list as a symptom (S57).

[0111] In another embodiment, if a user's discourse regarding a question including a first similar symptom is classified as neutral, the chatbot module (310) may ask a question about a second similar symptom with a lower similarity than the first similar symptom. The first similar symptom may be classified as unconfirmed in the additional information list.

[0112] Again, referring to FIG. 4, when the first similar symptom is classified as a symptom and the list of additional information is updated, a process of determining whether to proceed with additional questionnaire is performed. Since there are two symptoms classified as symptoms up to S57, this may be the case where a cluster is not formed or the number of symptoms is less than the standard number. Accordingly, the chatbot module (310) generates a question (311g) including the similar symptom with the highest similarity among the n-1 unidentified similar symptoms and displays it in the chatbot dialogue window (311). In response, the user inputs “yes.” The input user discourse (311h) is transmitted to the user discourse analysis module (320). The user discourse analysis module (320) analyzes the context of the user discourse (311h) and classifies it as positive (S58). If the user discourse for the question including the second similar symptom is classified as positive, the second similar symptom can be classified as a symptom (S59).

[0113] When the second similar symptom is classified as a symptom and the list of additional information is updated, a process is performed to determine whether to proceed with additional questionnaire. Since there are three symptoms classified as symptoms up to S59, this may be the case where a cluster is not formed or the number of symptoms is less than the standard number. Therefore, the chatbot module (310) generates a question (311i) including the third similar symptom with the highest similarity among the unidentified similar symptoms and displays it in the chatbot dialogue window (311). In response, the user enters “No.” The input user discourse (311j) is transmitted to the user discourse analysis module (320). The user discourse analysis module (320) analyzes the context of the user discourse (311j) and classifies it as negative (S60). If the user discourse in response to the question including the similar symptom is classified as negative, the third similar symptom may be classified as asymptomatic (S61).

[0114] The foregoing description of the present invention is for illustrative purposes only, and those skilled in the art will readily appreciate that the present invention can be readily modified into other specific forms without altering the technical spirit or essential characteristics of the present invention. Therefore, it should be understood that the embodiments described above are exemplary in all respects and not restrictive. In particular, the features of the present invention described with reference to the drawings are not limited to the structures depicted in the specific drawings, and may be implemented independently or in combination with other features.

[0115] The scope of the present invention is indicated by the claims described below rather than the detailed description above, and all changes or modifications derived from the meaning and scope of the claims and their equivalent concepts should be interpreted as being included in the scope of the present invention.

Claims

1. A user terminal that receives measurement data from a blood sugar measurement device and inputs correlation factors; A blood sugar management server that determines search conditions including a blood sugar measurement time, a personal condition, and two or more of the above correlation factors to search for a standard pattern from a plurality of standard patterns, compares the measurement data with the searched standard pattern to determine whether the user's health condition is abnormal, and if the determination result is abnormal, collects additional information to analyze the user's health condition to determine whether it is abnormal, and transmits an alarm to notify the user terminal of the occurrence of the abnormal condition. The above standard pattern represents the blood sugar change over time classified by a combination of multiple conditions based on clinically collected measurement data, and the multiple conditions are a combination of the blood sugar measurement time, the correlation factor, and the individual condition. The above correlation factors include meal timing and food type. The above personal status is a blood sugar risk prediction and management system using blood sugar and personalized data including age, gender, and BMI.

2. In claim 1, the blood sugar management server, A standard pattern storage unit in which the above plurality of standard patterns are stored; A first abnormal state detection unit that determines a search condition including two or more of the above blood sugar measurement time, the above personal condition, and the above correlation factor to search for a standard pattern from a plurality of standard patterns, and compares the measurement data with the searched standard pattern to determine whether the user's health condition is abnormal; An additional information collection unit that outputs questionnaire questions through a chatbot dialogue window running on the user terminal and extracts symptoms from user discourse entered in response to the questionnaire; A second abnormal condition detection unit that compares the collected symptoms with the predicted symptoms registered in relation to the user and determines whether the user's health condition is abnormal; and A blood sugar risk prediction and management system using blood sugar and personalized data, including a situation analysis and notification unit that transmits the alarm to the user terminal.

3. In claim 2, the first abnormal state detection unit, Collect the above blood sugar measurement time, the above correlation factors, the above personal status and the above measurement data, Based on the above blood sugar measurement time, distinguish whether the measurement data is a fasting blood sugar level or a postprandial blood sugar level. If the above measurement data is a fasting blood sugar level, determine whether the above measurement data was measured at dawn. If the above measurement data is a fasting blood sugar level measured at dawn, search for a standard pattern by fasting blood sugar level, dawn range, and personal condition. If the above measurement data is a fasting blood sugar level measured at a time other than dawn, search for a standard pattern by fasting blood sugar level and individual condition. If the above measurement data is a postprandial blood sugar level, a standard pattern is searched for postprandial blood sugar level, correlation factors, and individual status. A blood sugar risk prediction and management system that uses blood sugar and personalized data to analyze the user's health status using searched standard patterns.

4. In claim 2, the measurement data is time series measurement data collected from before the analysis time for the same user, a blood sugar risk prediction and management system using blood sugar and personalized data.

5. In claim 2, the second abnormal condition detection unit is a blood sugar risk prediction and management system using blood sugar and personalized data that determines the occurrence of the predicted symptom and the unpredicted symptom.

6. In claim 2, the additional information acquisition unit, A chatbot module that receives user discourse including information related to health status and generates questionnaire questions to determine whether the user has similar symptoms; A user discourse analysis module that analyzes the user discourse above and classifies it into one of positive, negative, and neutral, and extracts symptoms from the user discourse classified as positive and negative and registers them in the additional information list, where positive means that the user has the symptom, negative means that the user does not have the symptom, and neutral means that it cannot be determined whether the user has the symptom; and A blood sugar risk prediction and management system using blood sugar and personalized data, comprising a symptom generation module that learns about diseases and their associated symptoms, and clusters symptoms by disease and vector embedding them in space, and provides similar symptoms to symptoms registered in the additional information list based on similarity.

7. In claim 1, a medical staff terminal is further included that is notified of the abnormal condition and inputs an opinion based on the notified abnormal condition. A blood sugar risk prediction and management system using blood sugar and personalized data, which transmits the alarm to the user terminal based on the opinion received from the medical staff terminal.

8. A method for predicting and managing blood sugar risk using blood sugar and personalized data, Step of collecting blood sugar measurement time, individual condition, correlation factors and measurement data; A step of searching for a standard pattern from a plurality of standard patterns by determining a search condition including two or more of the above blood sugar measurement time, the above personal condition, and the above correlation factor; A step of comparing the above measurement data with the searched standard pattern to determine whether the user's health condition is abnormal; If the judgment result is abnormal, a step of collecting additional information and analyzing the user's health status to determine whether it is abnormal; Including a step of transmitting an alarm to notify the user terminal of an abnormal condition, The above standard pattern represents the blood sugar change over time classified by a combination of multiple conditions based on clinically collected measurement data, and the multiple conditions are a combination of the blood sugar measurement time, the correlation factor, and the individual condition. The above correlation factors include meal timing and food type. The above personal status is a method for predicting and managing blood sugar risk using blood sugar and personalized data including age, gender and BMI.

9. In claim 8, if the above judgment result is an abnormal state, the step of collecting additional information and analyzing the user's health status to determine whether it is an abnormal state is, A step of outputting a questionnaire through a chatbot dialogue window running on the user terminal; A step of receiving user discourse in response to the above questionnaire; A step of classifying the above user discourse as either positive, negative, or neutral, where positive means that the user has the symptom, negative means that the user does not have the symptom, and neutral means that it cannot be determined whether the user has the symptom; A step of extracting symptoms from user discourse classified as positive and negative and registering them in a list of additional information; and A method for predicting and managing blood sugar risk using blood sugar and personalized data, comprising a step of comparing symptoms registered in the above-mentioned additional information list with predicted symptoms registered in relation to the user and determining whether the user's health condition is abnormal.

10. In claim 8, A method for predicting and managing blood sugar risk using blood sugar and personalized data, further comprising a step of notifying the medical staff terminal of the abnormal condition.

Images