WO2024117258A1 - Blood sugar level model creating method and predicting method for blood sugar level transition - Google Patents

Abstract

[Problem] To provide a method for appropriately obtaining evaluation for a meal to be taken or for a taken meal, even without health examination results of a user. [Solution] The present invention involves a meal score acquisition method including: an image input step for inputting, to a computer, an image relating to a meal; a meal analysis step in which the computer analyzes the image and obtains meal analysis information; a blood sugar level transition information acquisition step in which the computer obtains, by using a transition pattern model of blood sugar level of a subject and the meal analysis information, transition information of blood sugar level of the subject after the subject has taken the meal; and a meal score acquisition step for obtaining a meal score for evaluating the meal on the basis of the transition information of the blood sugar level of the subject, which was obtained in the blood sugar level transition information acquisition step.

Description

How to create a blood glucose model and how to predict blood glucose trends

This invention relates to a method for creating a blood glucose model and a method for predicting blood glucose trends using the blood glucose model.

　Patent No. 6659049 describes a blood glucose prediction method. This method includes an acquisition step in which a computer acquires the measured blood glucose level of a user, the measured HbA1c level of the user, and the results of the user's health check; a classification step in which the computer determines whether the user is normal, borderline, or diabetic based on the measured blood glucose level, HbA1c level, and the results of the health check; a prediction step in which the computer uses the result of the classification and the measured fasting blood glucose level of the user at a past time to predict the user's future fasting blood glucose level; and a risk determination step in which the computer generates a standard normal distribution using the variability in the change in fasting blood glucose level within a specified period of time, divides the confidence interval of the generated standard normal distribution by a division line set at the fasting blood glucose level, which is the threshold value for determining diabetes, calculates an area ratio using the area of the confidence interval divided by the division line and the area of the confidence interval, and multiplies the area ratio by the reliability of the confidence interval to calculate the risk of developing diabetes.

The above method cannot predict blood glucose levels without information such as the user's health check results and fasting blood glucose measurements, and therefore cannot properly evaluate the food consumed.

Patent No. 6659049

This invention provides a method for appropriately obtaining an evaluation of meals that are planned to be consumed or meals that have been consumed, even if the user does not have health check results, etc.

An invention relates to a computer-based method for obtaining a meal score, which is a score for evaluating a meal. The method evaluates the meal using images of the meal.

The first method for obtaining a meal score includes a step of inputting blood glucose information of a subject and a step of obtaining a meal score. The meal score is an evaluation value for evaluating a certain meal. The meal score may be an evaluation value for evaluating a certain meal from the perspective of diabetes treatment or prevention.

The subject's blood glucose information input step is a step of inputting subject's blood glucose information, which is the blood glucose or glucose value of a subject who has taken a certain meal, or transition information of the subject's blood glucose information, into the computer. The transition information of the subject's blood glucose information may be time-series transition information of the subject's blood glucose level, or time-series transition information of the subject's glucose level. This step may include a step of inputting the subject's real-time blood glucose or glucose value measured using a sensor into the computer. Examples of the sensor may be a low-invasive blood glucose sensor, a non-invasive blood glucose sensor, a low-invasive glucose measurement sensor, and a non-invasive glucose measurement sensor. The sensor may be a sensor partly embedded in the body, a wristwatch type sensor, or a type attached to the body surface. The blood glucose or glucose value may not be the blood glucose value or the glucose value itself, but may be a value related to the blood glucose or glucose value. In this specification, the glucose value may also be simply expressed as the blood glucose value.
When subject blood glucose information, which is the blood glucose level or glucose level of a subject who has consumed a certain meal, is input to the computer, the computer may further include a blood glucose level etc. transition information acquisition step of obtaining transition information of the subject's blood glucose level after consuming a certain meal using the input subject blood glucose information. The transition information of the subject's blood glucose level (estimated value) can be obtained by inputting the subject's blood glucose information into a trained model (transition pattern model) described later, for example.

The transition pattern model may be obtained by a process including a blood glucose level input process and a blood glucose level transition pattern model acquisition process. The blood glucose level input process is a process in which a computer receives multiple blood glucose or glucose values after a subject has eaten a meal. These blood glucose levels, etc. are preferably input to the computer at a predetermined timing for a predetermined period after eating a meal. When these values are input for a predetermined period, one transition information is obtained. By inputting multiple such data as teacher data and performing machine learning (deep learning), a transition model of the subject's blood glucose level, etc. can be obtained.
The blood glucose level trend pattern model acquisition process is a process in which a computer obtains a blood glucose level trend pattern model of the subject through machine learning using the subject's postprandial blood glucose level or glucose value inputted in the blood glucose level input process.

The meal score acquisition process is a process in which a computer uses the subject's blood glucose information or information on trends in the subject's blood glucose information to obtain a meal score, which is an evaluation value for evaluating a certain meal.
The meal score acquisition process may include a process of obtaining a meal score by inputting the subject's blood glucose information, or information on trends in the subject's blood glucose information, into a trained model for obtaining a meal score.
Such a trained model may be constructed by machine learning or deep learning using, for example, information about a plurality of individuals (age, sex, medical history, diabetic state), meals, blood glucose levels, etc., transition information on blood glucose levels, etc., and meal scores as teacher data. By using such a trained model, the transition of blood glucose levels can be obtained by inputting blood glucose levels, etc., and meal scores can be obtained. In addition, this trained model may be constructed by machine learning or deep learning using one or more of past meals, past blood glucose levels, etc., and transitions of blood glucose levels, etc., and meal scores as teacher data for the subject. By using such a trained model, the transition of blood glucose levels can be obtained by inputting blood glucose levels, etc., and transition information on blood glucose levels, etc., and meal scores can be obtained.

The meal score acquisition process may obtain a meal score using either or both of the sum of the changes in the subject's blood glucose information in the transition information of the subject's blood glucose information. For example, the computer may create a graph of the transition information of blood glucose levels and glucose levels over time (the horizontal axis is time and the vertical axis is these values), and calculate the area (sum) of the part of the graph where the vertical axis is equal to or greater than the reference value. The computer may then store a meal score corresponding to the sum in the memory unit, and use the obtained sum to read out the meal score and calculate the meal score. The sum may be an integral value or an approximate value. An example of the approximate value may be the area of a triangle with the blood glucose level, etc. of the reference value (when fasting or when a meal is started) and the maximum value as vertices, and the graph area to which the reference value or threshold value is assigned as the base. The amount of change may also be, for example, the difference between the blood glucose level, etc. of the reference value (when fasting or when a meal is started) and the maximum value. The computer may then store a meal score in association with the difference between the reference value and the maximum value in the memory unit, and read out the meal score from the memory unit using the difference between the reference value and the maximum value. The above describes a case where the difference between a reference value and a maximum value is used as the amount of change. The amount of change is not limited to this example, and various change values can be used. Examples of such amounts of change include the reference value and the maximum value, as well as the time at which the maximum value is reached and a local maximum value.

The above describes a case where the subject's blood glucose level, etc. is input to a computer. The subject's blood glucose level, etc. may or may not be input to a computer.

The second meal score acquisition method includes an image input process, a meal analysis process, and a meal score acquisition process. This method may further include a blood glucose level and other trend information acquisition process. This method may further include the subject's blood glucose information input process described above.

The image input step is a step in which an image relating to a certain meal is input to a computer.
The diet analysis step is a step in which a computer analyzes the image and obtains diet analysis information.
The meal score acquisition process is a process of obtaining a meal score based on the meal analysis information. This process may obtain a meal score by constructing a trained model using the meal analysis information and the meal score as teacher data and inputting the meal analysis information into the trained model. The blood glucose level, etc. transition information acquisition process is a process in which a computer uses a transition pattern model of the subject's blood glucose level or glucose level and the meal analysis information to obtain transition information of the subject's blood glucose information, which is transition information of the subject's blood glucose information, which is the blood glucose level or glucose level of the subject after ingesting a certain meal. In this case, the meal score acquisition process may be a process of obtaining a meal score based on the transition information of the subject's blood glucose information obtained in the blood glucose level, etc. transition information acquisition process.

It is preferable that the meal score obtained as described above is output as appropriate. The next invention relates to a method for outputting a meal score. An example of this output method is to output an image related to a certain meal and a meal score so that the image related to the certain meal and the meal score are displayed on a single screen. For example, this output method can be achieved by having a computer read out an image related to a certain meal and a meal score stored in a memory unit and outputting them to an output unit.

The third invention relates to a method for displaying transition information of a subject's blood glucose level, etc. This invention may be combined with the first or second invention.
This invention provides a method for obtaining a representative value line indicating a representative value, a first area band indicating an area included with a first probability, a second area band indicating an area included with a second probability that is higher than the first probability, and a target value line indicating a target blood glucose value or target glucose value, for a blood glucose value or glucose value at a predetermined time after a meal, using a plurality of subject blood glucose information obtained on different days;
The method includes a step of outputting the median line, the first area band, the second area band and the target value line so that the representative value line, the first area band, the second area band and the target value line are displayed on a single screen.

It is preferable that the median line is output in white or gray, the first range band in a blue color, the second range band in a lighter blue color than the first range band, and the target value line in a yellow or red color.

The fourth invention relates to a method for calculating an insulin dosage using a computer.
The method includes a differential blood glucose calculation step and an insulin dosage calculation step.
The differential blood glucose value calculation step is a step in which a computer calculates a differential blood glucose value between a subject's predicted blood glucose value, which is the blood glucose value or glucose value of the subject calculated using a transition pattern model of the subject's blood glucose value or glucose value, and a target blood glucose value, which is a target blood glucose value or glucose value. The insulin dosage calculation step is a step in which a computer calculates an insulin dosage for the subject using the differential blood glucose value calculated in the differential blood glucose value calculation step.

An example of an insulin dosage calculation process is a process in which information about the insulin preparation to be administered to the subject and the differential blood glucose level are input into a blood glucose or glucose level transition pattern model for each insulin preparation, and the insulin dosage for the subject is calculated.

This specification also describes programs for causing a computer to execute the various methods described above, as well as non-transitory information recording media that store such programs and can be read by a computer.

The above-mentioned invention allows the user to appropriately evaluate meals that are planned to be consumed or have been consumed without using the user's health check results, etc.

FIG. 1 is a flowchart illustrating a method for obtaining a meal score. FIG. 2 is a conceptual diagram showing an example of blood glucose levels measured in real time over a certain period of time and the resulting transition pattern of blood glucose levels. FIG. 3 is a conceptual diagram showing an example of transition information of a subject's blood glucose level. FIG. 4 is a conceptual diagram showing an example of determining the insulin dosage. FIG. 5 shows an example of displaying a date, an image of a meal, and a meal score. FIG. 6 is a conceptual diagram for explaining meals, blood glucose level trends, and meal scores. FIG. 7 shows an example of a graph showing changes in estimated blood glucose levels on a given day together with meal scores. FIG. 8 is a diagram showing an example of displaying transition information of the subject blood glucose level and the like. FIG. 9 shows a sample example of a display screen. FIG. 10 shows an example of a weekly menu along with the meal score for each menu.

Below, the embodiments for carrying out the present invention will be explained with reference to the drawings. The present invention is not limited to the embodiments described below, but also includes appropriate modifications of the embodiments below within the scope that would be obvious to a person skilled in the art.

In the following explanation, blood glucose level will be used as an example. However, meal scores and insulin dosages can be estimated using not only blood glucose levels, but also values related to blood glucose levels, such as blood glucose levels and glucose values. For this reason, this specification also discloses cases where blood glucose level is interpreted as glucose value or a value related to blood glucose levels other than glucose value.

In this invention, basically, a computer performs various processes. The computer has an input unit, an output unit, a control unit, a calculation unit, and a storage unit, and each element is connected by a bus or the like so as to be able to send and receive information. For example, the storage unit may store a program or various information. When specific information is input from the input unit, the control unit reads out a control program stored in the storage unit. The control unit then reads out the information stored in the storage unit as appropriate and transmits it to the calculation unit. The control unit also transmits the input information as appropriate to the calculation unit. The calculation unit performs calculation processing using the various information received and stores it in the storage unit. The control unit reads out the calculation results stored in the storage unit and outputs them from the output unit. In this way, various processes and each step are performed. The various parts and means perform these various processes. The computer may have a processor, and the processor may realize various functions and steps. The computer may have a memory for storing information and a circuit for performing processing. The computer may be standalone. The computer may have some of its functions distributed to a server and a terminal. In this case, it is preferable that the server and terminals are able to send and receive information via a network such as the Internet or an intranet.

The elements of the computer that implement each process described below are the individual parts and means, and the elements that implement each process can be read as individual parts or means. A computer can convert various information into digital information and perform various calculation processes using the digital information. For example, various calculation processes can be performed by converting numerical information or image information into binary data, storing it in a memory unit, reading out the binary data, and having the calculation unit calculate it. Machine learning can improve the accuracy of a trained model by inputting a large amount of training data, inputting a large amount of training data with correct answers, or providing feedback. An example of feedback is when an obtained result is an abnormal value, and inputting that fact into the trained model.

An invention relates to a method using a computer for obtaining a meal score. The meal score is a score for evaluating a certain meal. The meal score may be an index that evaluates the fluctuation in blood glucose level within a certain time (e.g., 3 hours) after ingesting a certain meal. In this case, the smaller the fluctuation in blood glucose level, the higher the evaluation.

The meal score may be an evaluation value that evaluates a certain meal from the viewpoint of diabetes treatment or prevention. The evaluation value may be a numerical value, or may be a shade or a type of color. For example, the meal score may be such that the redder the meal, the worse the meal score. In an example of a meal score, when a doctor sets a treatment index (= treatment goal, ΔHbA1c) to be targeted, the blood glucose transition for each meal according to the treatment goal is expressed as a score. Specific values may be displayed, for example, from 1 to 10 points, and the standard score may be a Meal Score of 6 points. In this case, the meal score is intended to change the patient's behavior. By making the patient aware that if they achieve 6 points or more, they can achieve the goal treatment index, the patient will be more likely to take a positive approach to treatment. Such a meal score may be displayed on a display unit or may be printed, etc., as appropriate. The better the meal score, the higher the value, or vice versa. In this specification, an example will be explained in which a higher meal score is given to a more favorable blood glucose level, etc.

FIG. 1 is a flow chart for explaining a method for obtaining a meal score. The first method for obtaining a meal score that will be explained involves evaluating a meal using an image of the meal. As shown in FIG. 1, this method includes an image input process (S101), a meal analysis process (S102), a blood glucose level transition information acquisition process (S103), and a meal score acquisition process (S104). Hereinafter, each process in this specification may be performed by the corresponding parts or means that are elements of a computer.

The image input process (S101) is a process in which an image of a certain meal is input to a computer. The image of a certain meal may be, for example, a photograph or video of the meal taken by the subject using a mobile terminal. The subject transmits the captured image to a system having the above-mentioned computer. The system (for example, an image input unit or image input means of the system) then receives the captured image and stores it in memory. In this way, the image of the meal in the system is input. At this time, various information about the subject (user) may be input to the system.

The image of a certain meal may be an image selected by the subject from images related to a meal menu stored in the memory unit. In this case, it is possible to predict the meal score not only for meals that the subject plans to eat or has actually eaten, but also for meals to be eaten. For example, when considering what to eat for breakfast, lunch, or dinner, if a meal image (e.g., an image of a steak) is selected from the memory unit along with a menu, the image of the steak will be input to the computer's processing unit. The subject will then be able to obtain a predicted meal score if he or she eats the steak.

For example, various information about the user is stored in the user information management section of the user's mobile terminal or server. Examples of information about the user include the user's identification number, name, age, sex, allergies, medication information (particularly information about insulin medications), and disease information. A model of the user's blood glucose level transition pattern may be stored as information about the user. When user information is managed by the system, the user information may be stored in association with the user's identification number (user ID).

The meal analysis process (S102) is a process in which a computer analyzes an image and obtains meal analysis information. The meal analysis information is information about a meal based on an image of a certain meal. The meal information may include information about the dish, ingredients, calories (energy), and sugar content. The meal information may also include information about whether the meal is a staple food, main dish, side dish, or main soup. The meal information may also include the time when the image of the meal was taken, or the time when the meal was consumed. In this process, for example, the system may read an image stored in a dish storage section of the memory, and perform pattern matching between the read image and an image of a certain meal, thereby analyzing the dish and ingredients and obtaining various information.

　Obtaining dietary analysis information without performing the image input step (S101) is a different invention from the above described in this specification. For example, a user reads out a menu item they plan to eat (e.g., steak) from the memory unit. Then, dietary analysis information is stored in the memory unit in association with the menu item. In this case, dietary analysis information may be obtained by reading it out from the memory unit.

The dish storage unit is an element for storing a plurality of dishes and images of the plurality of dishes. The storage unit of a computer functions as the dish storage unit. The information on the plurality of dishes may include, for example, the names of the staple food, main dish, side dish, and main soup (and also drinks, desserts, and fruits), their respective nutrients, sugar content, and energy. This information may be stored, for example, in association with the identification number (ID) of each dish. Then, by specifying the identification number, the above-mentioned information on the name, nutrients, sugar content, and energy of the dish can be read out. Note that, although fruits and drinks are not cooked, they may be included as one of the "dishes" if they form part of a meal. For example, this system stores images of each dish in association with the above-mentioned dish identification number in the storage unit. Then, the image of the dish can be read out using the dish identification number. The dish storage unit 3 may store information other than the above (for example, information on food classification) in association with the above-mentioned identification number. Examples of food categories are fruits, meat, fish, green vegetables, root vegetables, dairy products, seaweed, legumes, grains, and alcohol. Allergy-related information may also be stored in association with the above identification number. For example, for a dish that contains sesame, information about sesame allergies may be stored in the memory unit in association with the identification number.

The blood glucose level trend information acquisition process (S103) is a process in which the computer uses the blood glucose level trend pattern model of the subject and the meal analysis information to obtain information on the trend of the subject's blood glucose level after the subject has eaten a certain meal. At this time, information on the subject's intake of insulin, the time of insulin intake, the insulin dose, and/or the type of insulin may also be input to the computer.

A model of the transition pattern of a subject's blood glucose level can be obtained, for example, as follows.
That is, an example of a method for obtaining a blood glucose level transition pattern model of a subject includes a blood glucose level input step and a blood glucose level transition pattern model acquisition step.

The blood glucose input process is a process in which a computer receives multiple blood glucose values after a subject has eaten a meal. For example, the subject wears a sensor for measuring blood glucose levels. When the subject eats a meal, the subject's mobile device linked to the sensor or the system that receives the sensing information determines that the subject has eaten a meal from the change in the subject's blood glucose level. After that, the measured blood glucose level is stored in the memory of the mobile device or the memory of the system at predetermined intervals (for example, every minute). In this way, the computer can receive and store multiple blood glucose values after the subject has eaten a meal. At this time, if the subject sends an image of the meal they have eaten to the system, the system can store the change in blood glucose level along with the meal analysis information. Note that even if an image of the meal they have eaten is not sent to the system, if the subject's blood glucose level data over a long period of time (for example, 12 hours or more) is used, it can be determined that the subject ate a meal (including snacks) during a time period when blood glucose levels fluctuate greatly. For example, blood glucose levels measured in real time for a certain period of time or more are stored in memory. Then, the system reads out the blood glucose level and analyzes the fluctuations in blood glucose level. By analyzing the fluctuations in blood glucose level, the system can determine the time when blood glucose levels become high. The system then analyzes the time periods when blood glucose fluctuations were above a certain level as the time when a meal was consumed and the time after the meal. The system can then store the pattern of blood glucose fluctuations after a meal in memory.

The glucose level trend pattern model is a model that shows the pattern of changes in blood glucose levels after a meal or the ingestion of sugar, or after the administration of insulin. There are no particular limitations on the glucose level trend pattern model, so long as it is capable of analyzing blood glucose level fluctuation data and obtaining a blood glucose level trend pattern model. An example of a glucose level trend pattern model acquisition process is a process in which a computer obtains a blood glucose level trend pattern model of a subject through machine learning using the subject's postprandial blood glucose level input in the blood glucose level input process. Also, as explained above, blood glucose level fluctuation patterns for a certain subject may be stored in multiple memories, and the blood glucose level fluctuation patterns may be read out as appropriate to obtain a glucose level trend pattern model using artificial intelligence or machine learning.

Figure 2 is a conceptual diagram showing an example of blood glucose level (blood glucose level transition) measured in real time for a certain period of time or more, and the blood glucose level transition pattern obtained from it. An example of the unit of blood glucose level is mg/dl. As shown in the left diagram of Figure 2, when the blood glucose level of a subject is measured for a certain period of time or more, information (graph) showing the transition of blood glucose level can be obtained. By combining this graph and the insulin intake information of the subject, it is possible to obtain blood glucose level transition patterns such as blood glucose level transition information when insulin is taken, blood glucose level transition information when a meal is taken after insulin is taken, blood glucose level transition information after a meal, blood glucose level transition information after breakfast, blood glucose level transition information after lunch, blood glucose level transition information after a snack, and blood glucose level transition information after dinner. The center of Figure 2 is an example of a blood glucose level transition pattern obtained in this way. The obtained blood glucose level transition pattern is appropriately stored in memory. A blood glucose level transition pattern model can be obtained by collecting a large number of blood glucose level transition patterns. At this time, if information on meals and medication that give blood glucose level transition patterns is also stored, a blood glucose level transition pattern model corresponding to meals and medication can be obtained. The obtained glucose level trend pattern model can be stored in memory as appropriate. The right part of Figure 2 shows an example of predicting the glucose level trend of a subject using the glucose level trend pattern model.

Figure 3 is a conceptual diagram showing an example of the blood glucose level trend information (graph) of a subject. As explained above, the blood glucose level trend information acquisition process is a process in which a computer uses a blood glucose level trend pattern model of the subject and dietary analysis information to obtain blood glucose level trend information of the subject after the subject ingests a certain meal. Dietary analysis information includes, for example, information on the sugar content of a meal that a certain subject has eaten. The computer reads the information on the sugar content from memory and has the processor perform a calculation using the information on the sugar content to apply it to the blood glucose level trend pattern model of the subject, thereby obtaining blood glucose level trend information of the subject after the subject ingests a certain meal. The computer appropriately stores the blood glucose level trend information of the subject after the subject ingests a certain meal obtained in this way in memory.

The meal score acquisition process (S104) is a meal score acquisition process for obtaining a meal score, which is an evaluation value for evaluating a certain meal, based on the blood glucose level trend information of the subject obtained in the blood glucose level trend information acquisition process. The meal score acquisition process may be capable of obtaining a meal score based on the blood glucose level trend information of the subject. For example, the computer may read the blood glucose level trend information of the subject from memory, obtain a meal score by pattern matching the read blood glucose level trend information of the subject, and store it in memory. Also, for example, the computer may obtain the area enclosed by the blood glucose level trend graph (blood glucose level fluctuation area) from the blood glucose level trend information of the subject (blood glucose level trend graph with the horizontal axis being elapsed time and the vertical axis being blood glucose level), and obtain a meal score from the obtained blood glucose level fluctuation area using the range of the blood glucose level fluctuation area stored in the memory unit and information regarding the value of the meal score. Also, for example, the computer may obtain the maximum change in blood glucose level before and after a meal from the subject's blood glucose level transition information (reading the maximum blood glucose level after the meal from the memory, reading the blood glucose level before the meal from the memory, and having the processor perform a calculation to subtract the blood glucose level before the meal from the maximum blood glucose level after the meal to obtain the maximum change in blood glucose level), and store it in the memory as appropriate. The computer may then obtain a meal score from the obtained maximum change in blood glucose level using information about the range of the maximum change in blood glucose level and the meal score value stored in the storage unit. Table 1 is a table showing the relationship between the meal score value and the range of the blood glucose fluctuation area or the maximum blood fluctuation value associated therewith. In this way, the meal score can be obtained using either the sum or the amount of change in blood glucose level or both in the subject's blood glucose level transition information. The meal score may be obtained using the range of the blood glucose fluctuation area and the maximum blood fluctuation value. In this case, a score may be given for each range of the blood glucose fluctuation area and the range of the maximum blood fluctuation value, and the scores multiplied by a predetermined coefficient may be added together to obtain an evaluation score, and the meal score may be obtained according to the range of the evaluation score. This calculation may be performed using a computer.

 

 

An invention relates to a computer-based method for obtaining a meal score, which is different from the above method. This method includes a blood glucose level input step, a blood glucose level transition information acquisition step, and a meal score acquisition step.

The blood glucose input step is a step in which a computer receives a plurality of blood glucose levels of a subject after ingesting a meal. An example of a blood glucose input step is a step in which a real-time blood glucose level of a subject measured using a sensor is input into the computer.
The blood glucose level trend information acquisition process is a process in which a computer obtains blood glucose level trend information of a subject after ingesting a certain meal based on the subject's multiple blood glucose levels. An example of the blood glucose level trend information is the graph shown in FIG. 3. The blood glucose level trend information is preferably information on blood glucose levels measured in real time. However, the blood glucose level trend information may also be information including blood glucose levels at 12 or more points within 3 hours after the meal. The meal score acquisition process is a process for obtaining a meal score for evaluating a certain meal based on the blood glucose level trend information of the subject obtained in the blood glucose level trend information acquisition process. This meal score acquisition process may be performed in the same manner as described above.

One invention relates to a method for calculating insulin dosage using a computer. This method includes a step of calculating a blood glucose difference and a step of calculating an insulin dosage.

The blood glucose difference calculation step is a step in which a computer calculates a blood glucose difference, which is the difference between an estimated blood glucose level, which is a blood glucose level calculated using a blood glucose level transition pattern model, and a target blood glucose level, which is a target blood glucose level.
The memory stores a target blood glucose level in association with the subject. An example of the target blood glucose level is a blood glucose level in a fasting state. The computer stores a blood glucose level transition pattern model of the subject in association with the subject. The blood glucose level transition pattern model may be obtained, for example, from multiple blood glucose level transition patterns for the amount of sugar ingested by the subject using artificial intelligence or machine learning. Information about the subject and information about meals are input to the computer. Then, based on the input information about the subject, the computer reads out the blood glucose level transition pattern model of the subject from the memory, and obtains an estimated blood glucose level using information about meals (information for obtaining a blood glucose level included in the information). An example of the estimated blood glucose level is an estimated blood glucose level that is an estimated value of the subject's blood glucose level in a fasting state. The estimated blood glucose level obtained by the computer is appropriately stored in the memory. The computer reads out the target blood glucose level of the subject from the memory using the information about the subject. The computer reads out the estimated blood glucose level and the target blood glucose level from the memory, and has the processor obtain the difference between them to obtain the differential blood glucose level. The computer appropriately stores the obtained differential blood glucose level in the memory. In this way, the computer can obtain the differential blood glucose level.

The insulin dosage calculation process is a process in which the computer calculates the insulin dosage for the subject using the differential blood glucose value obtained in the differential blood glucose value calculation process. The insulin dosage calculation process may perform any processing as long as the computer can calculate the insulin dosage for the subject using the differential blood glucose value obtained in the differential blood glucose value calculation process.

An example of an insulin dosage calculation process uses a blood glucose level transition pattern model of a subject when insulin is administered to the subject. The blood glucose level transition pattern model of a subject when insulin is administered can be obtained in the same manner as the blood glucose level transition pattern model of a subject after eating a meal. For example, a computer obtains multiple blood glucose level transition patterns when a subject administers a predetermined unit of insulin and eats a meal, and stores them in memory. The computer can then use artificial intelligence or machine learning to obtain a blood glucose level transition pattern model of the subject when insulin is administered. The blood glucose difference value is the blood glucose level value that should be lowered for the subject. The computer can then read out the blood glucose difference value from the memory and use the blood glucose level transition pattern model of the subject and the read out blood glucose difference value to calculate the amount of insulin to be administered to the subject. When the computer stores the blood glucose level transition pattern model of the subject when insulin is administered, the blood glucose level change amount per unit of insulin can be visualized. Then, by inputting an image of the meal into the system before eating, blood glucose level transition information when insulin is not administered can be obtained. Then, by combining the information on blood glucose level trends when a given amount of insulin is administered with the information on blood glucose level trends when no insulin is administered, the optimal insulin dosage for that meal can be determined.

Figure 4 is a conceptual diagram showing an example of determining an insulin dosage. In the example shown in Figure 4, based on a meal image, it is estimated that the maximum blood glucose level will be 150 mg/dl if the meal is consumed. The estimated maximum blood glucose level is stored in memory. There is a blood glucose level trend pattern model that matches the blood glucose level trend pattern when the meal is consumed, and the maximum blood glucose level was 130 mg/dl when the meal was consumed after 2 units of insulin were administered. The subject's target blood glucose level is 90 mg/dl. The computer reads this information from memory, calculates 6 units as the amount of insulin required to consume a certain meal, stores this in memory, and outputs it to the subject's mobile device. The subject then knows that the amount of insulin is 6 units, and is able to consume the required amount of insulin.

The insulin dosage calculation step may calculate the insulin dosage for the subject by applying information about the insulin preparation administered to the subject and the differential blood glucose value to a blood glucose level trend pattern model for each insulin preparation. In this case, the memory stores the blood glucose level trend pattern model for each insulin preparation for the subject. Information about the subject as well as information about the insulin preparation is input to the system. The system then reads out the blood glucose level trend pattern model for each insulin preparation using the input information about the subject and information about the insulin preparation. The system can calculate the insulin dosage for the subject using the differential blood glucose value and the read blood glucose level trend pattern model.

One invention relates to a program for causing a computer to execute any of the above-mentioned methods. Another invention relates to a non-transitory information recording medium that can be read by a computer that stores the above-mentioned program. Examples of non-transitory information recording media are CD-ROMs, DVDs, and USB memories.

Diabetes Treatment Application A diabetes treatment application was created and installed on a user's mobile device. First, the user's HbA1c value was measured. The obtained HbA1c measurement value was input to the system and stored in memory. A doctor set a target HbA1c value for three months from now. The doctor input the target HbA1c value to the system. The system's memory stored the HbA1c measurement value and the target HbA1c value in association with the user's ID. The system has a blood glucose level transition pattern model, and the insulin dosage was calculated using the HbA1c measurement value, the target HbA1c value, and the blood glucose level transition pattern model. The blood glucose level transition pattern model may be, for example, a blood glucose level transition pattern model of the user stored in association with the user ID, or may be a general one. However, the insulin dosage may be calculated using the differential blood glucose value.

　Before injecting insulin before a meal, the user takes a photo of the meal. The photo of the meal is then sent to the system. The system analyzes the meal based on the photo of the meal and obtains information on the user's blood glucose level trends over the three hours after the meal. The system then calculates a meal score based on this information on blood glucose levels.

The system stores the calculated meal score in memory along with the date and time of the meal (or the date and whether it was breakfast, lunch, snack, or dinner). At this time, the system may store this information along with a photo (image) of the meal. The system can then output the daily change in meal score.

Figure 5 shows an example of displaying a date, meal images, and meal score. As shown in Figure 5, in this example, the date, and images of breakfast, lunch, and dinner are displayed on the mobile device of a user who has installed the application. Furthermore, in this example, the meal score is displayed on top of the images of breakfast, lunch, and dinner. In this way, displaying breakfast, lunch, and dinner together with the meal score has the effect of discouraging the consumption of meals other than breakfast, lunch, and dinner (e.g. snacks).

In the example of FIG. 5, the meal score is displayed as a numerical value. For example, the meal score may be expressed using the color of the frame showing the meal. For example, the more red the color, the worse the meal score may be, and the more blue the color, the better (healthier) the meal score may be. It is preferable that the meal score obtained in this manner is output as appropriate. The computer may output an image of a certain meal and the meal score so that the image of the certain meal and the meal score are displayed on a single screen. For example, this output method can be achieved by the computer reading out the image of a certain meal and the meal score stored in the memory unit and outputting them to the output unit.

Figure 6 is a conceptual diagram for explaining meals, blood glucose level trends, and meal scores. As shown in Figure 6, the user launches the application and takes a picture of the meal they plan to eat using their mobile device. The application then sends the photo of the planned meal to the server. The server receives information about the user along with the photo of the meal. The server analyzes the received image of the meal and reads out a trend pattern model based on the received information about the user. The server uses this information to predict the trend of the user's blood glucose level. The center diagram in Figure 6 is a graph showing the predicted trend of blood glucose level. The server calculates a meal score from the calculated blood glucose level trend. The server stores the calculated meal score in memory and sends it to the user's terminal. The user's terminal can then obtain the meal score along with the photo image. If the user does not like the meal score, he or she can modify the meal content and send a revised photo of the planned meal to the server. The user can then obtain a meal score for the revised meal. In this way, the user can eat an appropriate meal.

Figure 7 shows an example of a graph showing changes in estimated blood glucose levels for a given day along with meal scores. The server estimates the trends in a user's blood glucose levels based on the received meal photos. The server can then estimate the trends in a user's blood glucose levels over the course of a day. The server stores a program for calculating a daily evaluation value based on the trends in blood glucose levels, and based on that program, obtains a daily evaluation value using the trends in blood glucose levels. As shown in Figure 7, by displaying the daily trends in blood glucose levels, meal score, and daily evaluation value on the user's device, it is possible to increase the user's motivation to control their blood glucose levels.

By looking at their daily meal score and images of the associated meals, users can visually understand what meals are good for their blood sugar levels and what meals are not. They can also graph the changes in their daily meal score. In this way, by recording daily blood sugar level trends and meals, users can select meals that will keep their post-meal blood sugar levels within the target range, enabling stable blood sugar control. The application also provides motivation to control blood sugar levels by visualizing blood sugar level progress information.

FIG. 8 is a diagram showing an example of displaying trend information of a target blood glucose level, etc. The vertical axis shows blood glucose level, and the horizontal axis shows time. In this example, the meal score is not shown, but the meal score may be displayed together. The range of blood glucose or glucose levels after ingesting breakfast, lunch, or dinner may be graphed. Also, the trend of blood glucose levels at night may be displayed. In this invention, a computer uses multiple subject blood glucose information acquired on different days to obtain a representative value line indicating a representative value (e.g., a median value), a first area band indicating an area included with a first probability, a second area band indicating an area included with a second probability that is a wider probability than the first probability, and a target value line indicating a target blood glucose level or target glucose value for the blood glucose level or glucose value at a predetermined time after ingesting a meal. Then, the computer outputs the median line, the first area band, the second area band, and the target value line so that the representative value line, the first area band, the second area band, and the target value line are displayed on one screen. In this way, trend information of a target blood glucose level, etc., such as that shown in FIG. 8, for example, can be displayed. In the above example, a representative value line showing a representative value (e.g., a median value), a first area band showing an area included with a first probability, a second area band showing an area included with a second probability that is a higher probability than the first probability, and a target value line showing a target blood glucose value or target glucose value are obtained. This specification also describes obtaining and displaying one or more of these.

Each output may be output on the screen of a terminal (client) held by the user (subject), may be output on a medium such as paper, or may be output on the screen of a terminal (client) held by a medical institution such as a doctor or a supporter such as a nutritionist.

The target blood glucose transition is, for example, a curve that represents the blood glucose transition for each meal according to the treatment goal when the doctor sets the treatment index (= treatment goal, ΔHbA1c). On the doctor screen/subject screen, for example, a graph is drawn with a yellow line. The target blood glucose transition is calculated, for example, as follows, but may be obtained by machine learning. Calculate the reference fasting blood glucose level (H1). This value can be obtained, for example, as the median value of the fasting blood glucose level (H) from the start of the test (day 1) to two weeks later (day 14). The set value may be input into the computer as appropriate. The following processes may be performed automatically by the computer.
Next, the estimated HbA1c (I) is calculated. This value can be obtained, for example, by substituting the average value of the blood glucose level (A) from the start of the test (day 1) to two weeks later (day 14) into a linear equation and calculating. Using this value, the target HbA1c (J) and the target average blood glucose level (C) are calculated. Next, the target fasting blood glucose level (K) is set. This value is set by the doctor as a constant, which is the ideal fasting blood glucose level in clinical practice, and may be input into the computer as appropriate. Then, the computer may read the value stored in the memory unit and use it for various calculations. It is specified with reference to the standard fasting blood glucose level (H1) from the start of the test (day 1) to two weeks later (day 14) and the target HbA1c. The doctor specifies the approximate target fasting blood glucose level for each HbA1c. This often complies with the guidelines for diabetes treatment, so it is sufficient to select these values stored in the computer. Then, these values are input into the computer. Next, the computer calculates the target blood glucose level (G) using the target average blood glucose level (C) and the target fasting blood glucose level (K). The target blood glucose transition is plotted in a graph. In this way, a graph like that shown in FIG. 8 can be obtained.

The example in Figure 8 shows a graph that makes it easy to see the patient's actual blood glucose trend, based on three hours of blood glucose data from the timestamp of the patient app recording. For example, it is displayed with the median (white line), 25-75% band (blue), 10-90% band (light blue), less than 10% and over 90% (dotted line), hypoglycemia reference line (red line), and target blood glucose trend (yellow line).

The median line (white line) (M) is a line that connects the median values of blood glucose trends from the start of a meal (0 minutes) to 180 minutes later. For example, it can be expressed from the day the insulin dosage is changed to the day before the next insulin dosage change.

The 25-75% band (blue line) (N) connects the lines that represent the 25% and 75% bands of blood glucose trends from the start of the meal (0 minutes) to 180 minutes later.

The 10-90% band (light blue line) (O) connects the lines that represent the 10% and 90% bands of blood sugar trends from the start of the meal (0 minutes) to 180 minutes later.

It is preferable that the median line is output in white or gray, the first range in a bluish color (e.g., blue, navy blue, purple), the second range in a lighter bluish color than the first range (e.g., light blue, light purple), and the target value line in a yellowish or reddish color (e.g., yellow or red). As a result of experiments using multiple colors, it was found that people who are interested in blood sugar levels, such as diabetics, tend to feel more mentally stable the more bluish areas there are. For this reason, the above color scheme was able to obtain a higher evaluation.

Figure 9 shows a sample example of the display screen. In this example, blood glucose level transition information is estimated from the meal image using a transition model and drawn in blue. The input meal image is also displayed together with the meal score. In this example, the blood glucose level transition is displayed as a blue curve. Survey results showed that displaying the blood glucose level transition curve in blue makes it easier to recognize diabetes.

FIG. 10 shows an example of a display of a week's menus along with the meal scores for each menu. In this example, meal images corresponding to the week's menus are read from the menus stored in the memory unit, and the meal score for when the subject eats meals based on each menu is calculated and displayed on the screen. By displaying seven days' worth of data, a variety of meals can be covered. Also, for example, when a nutritionist is deciding on a meal menu for a subject (e.g. a patient), if they select a menu for each subject, they can obtain a meal score for each meal for that subject, making it easier to give dietary advice to each subject. Also, by displaying images and meal scores for each meal eaten in the past week, an overview of the week can be made.

The present invention can be used in the fields of information-related industries and medical equipment.

Claims (14)

a subject's blood glucose information input step of inputting subject's blood glucose information, which is a blood glucose level or glucose value of a subject who has taken a certain meal, or time-varying information of the subject's blood glucose information into a computer;
A meal score acquisition step in which the computer obtains a meal score, which is an evaluation value for evaluating the certain meal, using the subject's blood glucose information or the transition information of the subject's blood glucose information;
How to obtain Meal Score including: The method for obtaining a meal score according to claim 1,
The method, wherein the subject's blood glucose information input step is a step in which the subject's real-time blood glucose or glucose value measured using a sensor is input into the computer. 2. The method of claim 1 ,
The method includes a step of obtaining the meal score by inputting the subject's blood glucose information or information on trends in the subject's blood glucose information into a trained model for obtaining a meal score. 2. The method of claim 1 ,
The method further includes a blood glucose level transition information acquisition step in which the computer uses the subject's blood glucose information to obtain transition information of the subject's blood glucose level after ingesting the certain meal,
The method, wherein the meal score acquisition process is a process for obtaining the meal score based on the transition information of the blood glucose information. The method for obtaining a meal score according to claim 4,
The meal score acquisition step includes:
A method comprising the step of obtaining the meal score using either the sum of the changes in the subject's blood glucose information in the transition information of the subject's blood glucose information or the changes in the subject's blood glucose information, or both. 2. The method of claim 1 ,
an image input step of inputting an image relating to a certain meal into the computer;
a diet analysis step in which the computer analyzes the image and obtains diet analysis information;
The computer further includes a blood glucose level transition information acquisition step of acquiring transition information of the subject's blood glucose information, which is transition information of the subject's blood glucose level or glucose level after ingesting the certain meal, using a transition pattern model of the subject's blood glucose level or glucose level and the meal analysis information,
The meal score acquisition process is a process of obtaining the meal score based on the trend information of the subject's blood glucose information obtained in the blood glucose level trend information acquisition process. The method for obtaining a meal score according to claim 6,
The transition pattern model is
A blood glucose level input process in which the computer receives a plurality of blood glucose levels or glucose levels after the subject has eaten a meal;
The method is obtained by a process including a blood glucose level trend pattern model acquisition process in which the computer obtains a blood glucose level trend pattern model of the subject through machine learning using the subject's postprandial blood glucose level or glucose value inputted in the blood glucose level etc. input process. Obtaining a meal score using the meal score acquisition method according to claim 6;
The method includes a step of outputting the image of the certain meal and the meal score so that the image of the certain meal and the meal score are displayed on one screen.
How to output meal score. 2. The method of claim 1 ,
The computer obtains, using a plurality of the subject's blood glucose information obtained on different days, a representative value line indicating a representative value, a first area band indicating an area included with a first probability, a second area band indicating an area included with a second probability that is a higher probability than the first probability, and a target value line indicating a target blood glucose value or target glucose value, for the blood glucose value or glucose value at a predetermined time after ingestion of a meal;
outputting the median line, the first area band, the second area band and the target value line so that the representative value line, the first area band, the second area band and the target value line are displayed on a single screen. 10. The method of claim 9,
The median line is white or gray,
The first zone is blue-based colors,
The second zone is a lighter blue color than the first zone.
The method wherein the target line is output as a yellowish or reddish color. 1. A method for calculating an insulin dosage using a computer, comprising:
The computer,
A differential blood glucose value calculation step of calculating a differential blood glucose value between a subject's predicted blood glucose value, which is a blood glucose value or glucose value of the subject calculated using a transition pattern model of the blood glucose value or glucose value of the subject, and a target blood glucose value, which is a target blood glucose value or glucose value;
an insulin dosage calculation step in which the computer calculates an insulin dosage for the subject using the differential blood glucose value obtained in the differential blood glucose value calculation step. 12. The method of claim 11,
The insulin dosage calculation step is a step of inputting information about the insulin preparation to be administered to the subject and the differential blood glucose value into a blood glucose or glucose value trend pattern model for each insulin preparation, and calculating the insulin dosage for the subject. a subject's blood glucose information input step of inputting subject's blood glucose information, which is a blood glucose level or glucose value of a subject who has taken a certain meal, or time-varying information of the subject's blood glucose information into a computer;
A meal score acquisition step in which the computer obtains a meal score, which is an evaluation value for evaluating the certain meal, using the subject's blood glucose information or the transition information of the subject's blood glucose information;
A program for carrying out a method for obtaining a meal score, including: A non-transitory information recording medium that can be read by a computer and stores the program according to claim 13.

Images