Information processing system

Abstract

To predict with high accuracy, from few physical information of a user, a biological change amount due to a dietary habit improvement or an increase in momentum of the user.SOLUTION: An information processing system includes a control unit. The control unit inputs, to a first trained model having been trained using physical information of a plurality of attributes so as to estimate values of missing attributes from values of some attributes of the physical information, values of physical information of first attribute of a plurality of subjects for whom a biological change due to a dietary habit improvement or an increase in momentum is observed, and acquires the estimate value of physical information of a missing second attribute from the physical information of the first attribute of each subject from the first trained model. Meanwhile, the control unit conducts training for predicting a biological change amount of each subject due to a dietary habit improvement or an increase in momentum of each subject adopting the value of the first attribute and the estimate value of the second attribute as explanatory variables, and thereby generates a second trained model for predicting a biological change amount of a discretionary subject due to a dietary habit improvement or an increase in momentum of the discretionary subject.SELECTED DRAWING: Figure 4

Description

[Technical Field]

[0001] The present invention relates to an information processing system, an information processing method, and a program for predicting the amount of biological change (diet effect) due to a user's improved eating habits or increased exercise. [Background technology]

[0002] There have been systems for supporting users in improving their health. For example, Patent Document 1 below discloses a system in which a user is asked questions about their physical information, dietary information, lifestyle information, needs information, etc. using a medical questionnaire or questionnaire, and further a physical fitness test is conducted, and a health improvement program including two types of endurance training and muscle strength training that are individually suited to the user's physical condition, dietary status, and lifestyle status is created in accordance with the user's needs. The user performs the health improvement program and reports the amount of activity each time to a center, and the center analyzes and judges the amount of activity and provides advice based on the results. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Laid-Open No. 2004-54591 Summary of the Invention [Problem to be solved by the invention]

[0004] The technology in Patent Document 1 above allows various physical information about the user to be obtained through medical questionnaires and questionnaires, which can be used to create a health promotion program; however, if not much physical information is obtained from the user, an appropriate health promotion program cannot be created.

[0005] The present invention aims to provide an information processing system, an information processing method, and a program that can accurately predict the amount of biological change resulting from a user's improved diet or increased exercise based on the user's limited physical information. [Means for solving the problem]

[0006] An information processing system according to one aspect of the present invention includes a control unit. The control unit inputs values ​​of physical information of a first attribute of multiple subjects in whom biological changes due to an improved diet or increased physical activity have been observed into a first trained model, the first trained model having learned physical information of multiple attributes so as to estimate missing attribute values ​​from the values ​​of certain attributes of the physical information, and obtains estimated values ​​of physical information of a second attribute missing from the physical information of the first attribute of each subject from the first trained model. The control unit then generates a second trained model that predicts the amount of biological change of a given user due to the improved diet or increased physical activity of the given user by learning the values ​​of the first attribute and the estimated values ​​of the second attribute as explanatory variables.

[0007] An information processing system according to another aspect of the present invention includes a control unit. The control unit inputs values ​​of physical information of a first attribute of multiple subjects in which biological changes due to an improved dietary lifestyle or an increased amount of exercise have been observed into a first trained model, the first trained model having physical information of multiple attributes trained to estimate missing attribute values ​​from values ​​of certain attributes of the physical information, obtains estimated values ​​of physical information of a second attribute missing from the physical information of the first attribute of each of the subjects, and learns to predict an amount of biological change in each subject due to an improved dietary lifestyle or an increased amount of exercise using the values ​​of the first attribute and the estimated values ​​of the second attribute as explanatory variables. The control unit then inputs, into a second trained model generated to predict an amount of biological change in an arbitrary user due to an improved dietary lifestyle or an increased amount of exercise, the value of the first attribute received from the user terminal of the arbitrary user or a proxy user terminal of a proxy user acting on behalf of the user, and the value of the second attribute obtained by inputting the value of the first attribute into the first trained model as the explanatory variables. Then, the control unit obtains a predicted value of the biological change from the second trained model and transmits it to the user terminal or the proxy user terminal.

[0008] An information processing method according to another aspect of the present invention includes: inputting values ​​of the physical information of a first attribute of a plurality of subjects in whom biological changes due to an improvement in diet or an increase in exercise amount have been observed into a first trained model that has been trained to estimate missing attribute values ​​from values ​​of certain attributes of the physical information, and obtaining, from the first trained model, estimated values ​​of the physical information of a second attribute that is missing from the physical information of the first attribute of each of the subjects; generating a second trained model that predicts the amount of biological change of any user due to an improvement in the dietary habit or an increase in the amount of exercise by learning to predict the amount of biological change of each subject due to an improvement in the dietary habit or an increase in the amount of exercise of the subject using the value of the first attribute and the estimated value of the second attribute as explanatory variables.

[0009] An information processing method according to yet another aspect of the present invention includes: a first trained model is trained to estimate missing attribute values ​​from certain attribute values ​​of physical information, and the first trained model is input with values ​​of the physical information of a first attribute of multiple subjects in whom biological changes due to improved dietary habits or increased physical activity have been observed; the first trained model is then trained to estimate missing attribute values ​​from certain attribute values ​​of physical information, and an estimated value of the physical information of a second attribute that is missing from the physical information of the first attribute of each subject is obtained; and the first trained model is trained to predict the amount of biological change of each subject due to improved dietary habits or increased physical activity using the values ​​of the first attribute and the estimated values ​​of the second attribute as explanatory variables, and the second trained model is then trained to predict the amount of biological change of an arbitrary user due to improved dietary habits or increased physical activity, and the value of the first attribute received from the user terminal of the arbitrary user or the proxy user terminal of a proxy user acting on behalf of the user, and the value of the second attribute obtained by inputting the value of the first attribute into the first trained model are input as the explanatory variables; This includes obtaining a predicted value of the biological change from the second trained model and transmitting it to the user terminal or the proxy user terminal.

[0010] According to yet another aspect of the present invention, there is provided a program for executing the program on an information processing device, inputting values ​​of physical information of a first attribute of a plurality of subjects in whom biological changes due to improved diet or increased exercise have been observed into a first trained model that has been trained to estimate missing attribute values ​​from values ​​of certain attributes of the physical information, and acquiring, from the first trained model, estimated values ​​of physical information of a second attribute that is missing from the physical information of the first attribute of each of the subjects; and generating a second trained model that predicts the amount of biological change of any user due to an improvement in the user's diet or an increase in the amount of exercise by learning to predict the amount of biological change of the user due to an improvement in the user's diet or an increase in the amount of exercise of each subject using the value of the first attribute and the estimated value of the second attribute as explanatory variables.

[0011] According to yet another aspect of the present invention, there is provided a program for executing the program on an information processing device, a step of inputting values ​​of physical information of a first attribute of multiple subjects in which biological changes due to improved dietary habits or increased physical activity have been observed into a first trained model, the first trained model having physical information of multiple attributes trained to estimate missing attribute values ​​from certain attribute values ​​of physical information, obtaining estimated values ​​of physical information of a second attribute that is missing from the physical information of the first attribute of each of the subjects, and learning to predict the amount of biological change of each subject due to improved dietary habits or increased physical activity using the values ​​of the first attribute and the estimated values ​​of the second attribute as explanatory variables, and inputting the value of the first attribute received from the user terminal of any user or a proxy user terminal of a proxy user acting on behalf of the user and the value of the second attribute obtained by inputting the value of the first attribute into the first trained model as the explanatory variables; The system executes a step of obtaining a predicted value of the biological change from the second trained model and transmitting it to the user terminal or the proxy user terminal. [Effects of the Invention]

[0012] According to an information processing system of an embodiment of the present invention, it is possible to predict with high accuracy the amount of biological change resulting from an improvement in a user's diet or an increase in the amount of exercise, based on a small amount of physical information about the user. However, this effect does not limit the present invention. [Brief explanation of the drawings]

[0013] [Figure 1] 1 is a diagram showing the configuration of a diet effect prediction model generation system according to one embodiment of the present invention. [Figure 2] 1 is a diagram showing a hardware configuration of a diet information providing server according to an embodiment of the present invention. [Figure 3] 2 is a diagram showing the configuration of a database included in a diet information providing server according to one embodiment of the present invention. FIG. [Figure 4] 10 is a flowchart showing the flow of a process for generating a diet effect prediction model by a diet information providing server according to one embodiment of the present invention. [Figure 5] 10 is a flowchart showing details of a process for generating a diet effect prediction model by a diet information providing server according to one embodiment of the present invention. [Figure 6] 10 is a diagram illustrating the prediction accuracy of a diet effect prediction model by a diet information providing server according to one embodiment of the present invention. FIG. [Figure 7] 10 is a flowchart showing the flow of a diet effect prediction information providing process using the diet effect prediction model by a diet information providing server according to one embodiment of the present invention. [Figure 8] FIG. 10 is a diagram illustrating the prediction accuracy of a diet effect prediction model by a diet information providing server according to another embodiment of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0014] Hereinafter, an embodiment of the present invention will be described with reference to the drawings.

[0015] [System Configuration] As shown in FIG. 1, this system includes a diet information providing server 100 on the Internet 50 and a plurality of user terminals 200.

[0016] The diet information providing server 100 provides a diet method suggestion service (diet effect prediction information providing service) for users of the user terminal 200. Specifically, the diet information providing server 100 is connected to a plurality of user terminals 200 via the Internet 50, and predicts the diet effect when the user improves their eating habits based on the user's physical information and answers to a questionnaire, and transmits the predicted information to the user terminal 200. In this embodiment, information on the weight loss rate is provided as a specific example of the diet effect.

[0017] Here, "improving eating habits" means, for example, performing at least one of the following, and the diet information providing server 100 can set a definition of improving eating habits by appropriately combining the following. Eat less (set a daily calorie intake) Reduce snacking or eat healthier snacks Reduce the amount of carbohydrates (staple foods) you consume Reduce fat intake Reduce sugar in drinks Increase your intake of fruits and vegetables Eat meals at regular times Don't eat late at night (e.g. after 8pm) Limit the time during the day when you can eat (for example, 10 hours) Stop eating when you are 80% full Replace at least one meal a day with a menu designed specifically for dieting (for example, a menu that reduces fat and increases protein, replaces fat with omega-3, and includes carbohydrates with dietary fiber).

[0018] The user terminals 200 (200A, 200B, 200C...) are terminals used by users, such as smartphones, mobile phones, tablet PCs (Personal Computers), notebook PCs, desktop PCs, etc. The user terminals 200 transmit the user's physical information and answers to questionnaires to the diet information providing server 100, receive information indicating the diet effect, and display it on the screen using a browser or the like. An application corresponding to a diet prediction effect information providing service may also be installed in the user terminals 200, and the user terminals 200 may access the diet information providing server 100 using the application to display the diet effect prediction information.

[0019] When providing the diet effect prediction information, the diet information providing server 100 generates a diet effect prediction model that predicts the diet effect using the physical information and answer information to the questionnaire as explanatory variables. In this case, when the diet information providing server 100 receives the physical information and answer information to the questionnaire from the user terminal 200, the diet information providing server 100 inputs the physical information and answer information to a trained model (first trained model) 10 that can estimate the value of a missing attribute (second attribute) of the physical information from the value of a certain attribute (first attribute) of the physical information, acquires the physical information of the missing attribute from the trained model 10, and generates the diet effect prediction model (second trained model) using the acquired physical information and the physical information and answer information provided by the user as explanatory variables. Details of the generation process of the diet effect prediction model will be described later.

[0020] The first attribute value and the second attribute value may each be a single attribute value or a set of multiple attribute values. The trained model 10 may be, for example, a model capable of estimating a joint probability distribution (e.g., HI-VAE (https: / / arxiv.org / pdf / 1807.03653.pdf) or Tab Transformer (https: / / arxiv.org / pdf / 2012.06678.pdf)), but is not limited thereto. Here, a "missing attribute" includes at least the following cases: "when the value of the attribute does not exist," "when the value of the attribute is missing," "when the value of the attribute is not measured," "when the value of the attribute is not obtained," "when the value of the attribute is not input and the value of the attribute is flagged as missing," "when the value of the attribute is inaccessible," or "when the value of the attribute is lost."

[0021] Here, the (first and second) attributes may be basic information such as height, weight, sex, and age, as well as information indicating the state of a person's body, such as information on general blood, liver function, lipids, metabolic system, blood pressure, etc., which are test items in a health checkup; scores, numerical values, and classification information obtained by measuring all or part of the body; and scores and numerical classification information obtained from a medical interview or questions.For example, information about your physical condition may include your physique / body composition (body fat, muscle mass, visceral fat, etc.), blood test values ​​(triglycerides, total cholesterol, HDL cholesterol, LDL cholesterol, blood sugar, HbA1c, ALT, red blood cell count, aspartate aminotransferase, alanine aminotransferase, gamma-glutamyltranspeptidase, alkaline phosphatase), various hormones (cortisol, triiodothyronine, thyroxine, dehydroepiandrosterone, testosterone, estradiol, progesterone, follicle-stimulating hormone, luteinizing hormone, prolactin, etc.), vascular function (arteriosclerosis index, lower limb arterial stenosis / occlusion index, etc.), metabolic function (blood glucose, insulin, glucagon, GLP-1, GIP), cognitive function (functional test (Cognitrax (registered trademark)), blood D or L-amino acids), motor function (grip strength, gross motor ability, walking function (speed, stride length, pitch), etc.), immune-related index (lymphoid subset analysis, NK cell activity, cytokine analysis, etc.), disease information (presence or absence of disease, medical treatment, medication, etc.), Activity level (calories burned per day, steps taken per day), sleep-related indicators (score from medical questionnaire), productivity (score from questionnaire), diet / nutritional status (calories burned, PFC balance, etc.), menstrual-related indicators (stage and disability level determined by medical interview and questions), menopausal-related indicators (stage and disability level determined by questions), hair (diameter, degree of waviness), skin condition (skin blood flow rate, skin glycation level, ceramide amount), body odor (by gas chromatography) All kinds of physical attributes are included, such as information on: quantification of odor components, Oriental medical constitution classification (clustering by questionnaire), biomicrobiota information (quantity and composition ratio of bacteria in the oral cavity, intestines, scalp, and skin), progression of hair thinning and alopecia (score by photographic assessment), DNA and RNA information, overactive bladder / urinary incontinence (score by questionnaire), lifestyle questionnaires (BDHQ (short self-administered dietary history questionnaire), international standardized physical activity questionnaire, eating behavior questionnaire, etc.), health questionnaires (35-item dietary questionnaire, WHO-5 mental health status questionnaire, brief occupational stress questionnaire, Chalder fatigue scale, Athens insomnia scale), personality questionnaires (Big-five questionnaire, etc.).

[0022] For example, by inputting the attribute values ​​of age, gender, height, and weight into the trained model 10 as a set of physical information of a first attribute, estimated values ​​of information of each attribute of visceral fat cross-sectional area, triglyceride, blood glucose level, and stress level can be obtained as a set of physical information of a second attribute that is missing from the set of physical information of the first attribute.

[0023] [Hardware configuration of diet information server] As shown in FIG. 2, the diet information providing server 100 includes a CPU (Central Processing Unit) 11, a ROM (Read Only Memory) 12, a RAM (Random Access Memory) 13, an input / output interface 15, and a bus 14 connecting these components to one another.

[0024] The CPU 11 accesses the RAM 13 etc. as needed, and performs various arithmetic processing while controlling all blocks of the diet information providing server 100. Multiple CPUs 11 may be provided depending on the processing. The ROM 12 is a non-volatile memory in which firmware such as the OS, programs, and various parameters to be executed by the CPU 11 are permanently stored. The RAM 13 is used as a working area for the CPU 11, and temporarily stores the OS, various applications currently being executed, and various data currently being processed.

[0025] The input / output interface 15 is connected to a display unit 16, an operation reception unit 17, a storage unit 18, a communication unit 19, and the like.

[0026] The display unit 16 is a display device that uses, for example, an LCD (Liquid Crystal Display), an OLED (Organic ElectroLuminescence Display), a CRT (Cathode Ray Tube), or the like.

[0027] The operation reception unit 17 is, for example, a pointing device such as a mouse, a keyboard, a touch panel, or other input device. When the operation reception unit 17 is a touch panel, the touch panel can be integrated with the display unit 16.

[0028] The storage unit 18 is a non-volatile memory such as a hard disk drive (HDD), a flash memory (solid state drive (SSD)), or other solid-state memory. The storage unit 18 stores the OS, various applications, and various data.

[0029] As will be described later, particularly in this embodiment, the memory unit 18 has programs such as applications necessary for the process of generating the diet effect prediction model described later and the process of providing diet effect prediction information using the prediction model, as well as a user physical information database, a questionnaire information database, an estimated physical information database, and an effect prediction information database.

[0030] The communication unit 19 is, for example, a NIC (Network Interface Card) for Ethernet or various modules for wireless communication such as wireless LAN, and is responsible for communication processing with the user terminal 200.

[0031] Although not shown, the basic hardware configuration of the user terminal 200 is also substantially the same as the hardware configuration of the diet information providing server 100 described above.

[0032] [Database configuration of diet information server]

[0033] 3, the diet information providing server 100 has, in the storage unit 18, a user physical information database 31, a questionnaire information database 32, an estimated physical information database 33, and an effect prediction information database 34. Note that each of these databases may be stored in a storage device or server externally connected to the diet information providing server 100, rather than in the storage unit 18.

[0034] The user physical information database 31 stores physical information received from the user terminal 200, such as the user's age, sex, height, weight, body composition, visceral fat area, waist size, blood pressure, heart rate, and body temperature, in association with general information such as the user's name, a user ID for identifying the user, occupation, address, and email address.

[0035] The diet information providing server 100 may receive the physical information not directly from the user terminal 200 but via another organization such as a health checkup institution or a sports gym.

[0036] The questionnaire information database 32 stores questionnaire information for users and information on responses to the questionnaire information received from the user terminal 200. Specifically, the questionnaires are about the user's lifestyle, health status, or personality. Questionnaires about lifestyle include the BDHQ (Brief Self-Administered Diet History Questionnaire), the International Standardized Physical Activity Questionnaire, and the Eating Behavior Questionnaire. Questionnaires about health status include the 35-item Diet Questionnaire, the WHO-5 Mental Health Questionnaire, the Occupational Stress Brief Questionnaire, the Chalder Fatigue Scale, the Athens Insomnia Scale, and the Self-Efficacy Questionnaire. Questionnaires about personality include the Big-five questionnaire.

[0037] The estimated physical information database 33 inputs the physical information (physical information of the first attribute) and the response information received from the user terminal 200 into a trained model 10 capable of estimating the physical information, and stores the physical information (physical information of the second attribute) missing from the physical information of the first attribute acquired from the trained model 10. There may be multiple types of physical information of the first attribute and physical information of the second attribute.

[0038] The effect prediction information database 34 stores diet effect prediction information (weight loss rate information) in association with the user ID, which is obtained by inputting the physical information and response information of the first attribute received from the user terminal 200 and the physical information of the second attribute obtained from the trained model 10 into the diet effect prediction model.

[0039] These databases are used by mutual reference as necessary in the diet effect prediction model generation process and diet effect prediction information provision process using the prediction model by the diet information providing server 100 described later.

[0040] [Operation of diet information server] Next, the operation of the diet information providing server 100 configured as above will be described. The operation is performed by the cooperation of hardware such as the CPU 11 and communication unit 19 of the diet information providing server 100 and software stored in the storage unit 18. For convenience, in the following description, the CPU 11 is the subject of the operation.

[0041] (Diet effect prediction model generation process) First, we will explain the process of generating a diet effect prediction model by the diet information providing server 100. As a premise of this process, it is assumed that a clinical trial of a predetermined dietary improvement (for example, daily calorie intake restriction) is conducted on a plurality of subject users for, for example, several weeks, and subject users who have observed a diet effect (weight loss) as a result are identified.

[0042] 4, the CPU 11 of the diet information providing server 100 first receives (a set of) physical information of a first attribute and answer information to a questionnaire from the user terminals 200 of a plurality of subject users who have been observed to lose weight due to the dietary improvement (step 41). The information is received, for example, via a physical information input screen having input fields and a questionnaire (questionnaire) screen having answer fields, which are displayed on the user terminal 200. Upon receiving the physical information of the first attribute and the answer information, the CPU 11 may transmit an input request for the respective pieces of information to the user terminals 200 of a plurality of subject users who have been confirmed to lose weight in the clinical trial.

[0043] Next, CPU 11 inputs the received physical information of the first attribute and the answer information to the questionnaire into the trained model 10 (step 42). The physical information and answer information may be input each time they are received from user terminal 200, or a predetermined number of pieces of physical information and answer information may be input all at once when the physical information and answer information for a predetermined number of subject users have been received.

[0044] Next, the CPU 11 acquires an estimated value of the physical information of the second attribute from the trained model 10 (step 43).

[0045] Then, the CPU 11 generates a diet effect prediction model that estimates the diet effect (weight loss rate) of any user due to an improvement in their eating habits, using the physical information of the first attribute, the physical information of the second attribute, and the answer information as explanatory variables (step 44).

[0046] FIG. 5 is a flowchart showing the details of the process for generating a diet effect prediction model in step 44 above.

[0047] As shown in the figure, first, the CPU 11 performs preprocessing, i.e., normalization processing (step 51), on the (set of) physical information of the second attribute of each subject user acquired from the trained model 10, with the minimum value set to 0 and the maximum value set to 1. The physical information of the second attribute is received as data of, for example, about 2000 types of features.

[0048] Next, CPU 11 divides the set of physical information of the second attribute after the normalization into training (teacher) data and verification data (step 52).

[0049] Next, the CPU 11 selects feature quantities (parameters) to be used for prediction from the large number of feature quantities of the training data (for example, selects 20 from the 2000 quantities) (step 53). For this selection, for example, Lasso regression can be used. Lasso regression is a regression model in which an L1 regularization term is added to the objective function. By adding the regularization term, the values ​​of some regression coefficients become 0, allowing for automatic feature selection.

[0050] Then, the CPU 11 performs multiple regression analysis using the selected feature data as training data to construct a prediction model for the weight loss rate resulting from dietary improvement (step 54).

[0051] The CPU 11 then inputs the verification data into the constructed prediction model to predict the weight loss rate (step 55) and verifies the prediction accuracy (step 56). The CPU 11 repeatedly executes the above process by changing the training data and verification data among the data of multiple subjects until a predetermined or higher prediction accuracy is achieved, and constructs a final prediction model for predicting the weight loss rate of any user due to the user's improvement in their dietary habits.

[0052] FIG. 6 is a diagram illustrating the prediction accuracy of the diet effect prediction model.

[0053] Figure 1A shows the prediction accuracy of a model for predicting weight loss rates due to dietary improvements constructed without using the trained model 10, using age, sex, weight, and height as the first attribute of physical information and responses to the Obesity Standardized Questionnaire (five factors: appetite, health orientation, mealtime, dietary restrictions, and exercise) as explanatory variables. Figure 1B shows the prediction accuracy of a model for predicting weight loss rates constructed by inputting the first attribute of physical information and responses to the questions into the trained model 10 as described above to obtain second attribute of physical information. Figure 1C also shows the prediction accuracy of a model constructed by adding the WHO-5 Mental Health Status Questionnaire as a questionnaire, obtaining responses to the questionnaire from the user terminal 200, and inputting the responses to the first attribute of physical information and the Obesity Standardized Questionnaire into the trained model 10 to obtain second attribute of physical information. In this example, the target users were 20 men with a BMI of 23 kg / m or higher.

[0054] The WHO-5 Mental Health Questionnaire asks participants to answer the following five items (1) to (5) on a five-point scale, indicating how often they have experienced each of the following in the past two weeks: "Never (0 points)," "Very rarely (1 point)," "Less than half the time (2 points)," "More than half the time (3 points)," "Almost always (4 points)," and "Always (5 points)." (1) I was in a cheerful and happy mood. (2) I felt calm and relaxed. (3) I was motivated and active. (4) I had a good night's rest and woke up feeling refreshed. (5) There were many interesting things in my daily life.

[0055] As shown in Figure 1(A), when the trained model 10 was not used, the prediction accuracy (correlation coefficient r between the true value and the predicted value) was 0.52 for Spearman and 0.50 for Pearson. On the other hand, as shown in Figure 1(B), by acquiring the physical information of the second attribute using the trained model 10, the prediction accuracy was 0.79 for Spearman and 0.64 for Pearson, which was significantly improved compared to the case of Figure 1(A).

[0056] Furthermore, as shown in Figure 1(C), by increasing the number of types of questionnaires input into the trained model 10, the prediction accuracy was further improved to 0.92 for Spearman and 0.86 for Pearson.

[0057] In this way, by using the trained model 10 to amplify the attributes of physical information that can be used as explanatory variables, it is possible to construct a predictive model that can accurately predict the weight loss rate due to a user's improved dietary habits, even from limited physical information (non-invasive information such as gender, age, height, and weight) and response information of the user.

[0058] (Diet effect prediction information provision process) Next, a description will be given of a process for providing diet effect prediction information to the user terminal 200 using the diet effect prediction model constructed above. Fig. 7 is a flowchart showing the flow of the process for providing the diet effect prediction information.

[0059] As shown in the figure, first, the CPU 11 determines whether or not a diet effect (weight loss rate) prediction request has been received from the user terminal 200 (step 71). The diet effect prediction request includes the physical information of the first attribute and the answer information to the questionnaire, and is transmitted via an application installed in the user terminal 200, for example.

[0060] If it is determined that the prediction request has been received (Yes in step 71), the CPU 11 inputs the physical information and questionnaire information of the first attribute into the trained model 10 (step 72).

[0061] Next, the CPU 11 acquires an estimated value of the physical information of the second attribute from the trained model 10 (step 73).

[0062] Next, the CPU 11 inputs the first attribute physical information, the second attribute physical information, and the questionnaire information into the diet effect prediction model that has been created (step 74).

[0063] Next, the CPU 11 obtains an estimated value of the weight loss rate from the diet effect prediction model (step 75).

[0064] The CPU 11 then transmits the acquired estimated value information of the weight loss rate to the user terminal (step 76). The estimated value information may be displayed, for example, on the application as a percentage of the weight loss rate when the user improves their eating habits for a predetermined period of time, together with a message encouraging the user to go on a diet.

[0065] As described above, according to this embodiment, the diet information providing server 100 can accurately predict the weight loss rate of a user due to an improvement in the user's eating habits from the user's limited physical information.

[0066] [Variations] Although the embodiments of the present invention have been described above, the present invention is not limited to the above-described embodiments, and it goes without saying that various modifications can be made within the scope of the gist of the present invention.

[0067] In the above-described embodiment, the diet information providing server 100 uses the trained model 10 to generate a prediction model that predicts a weight loss rate when a user improves their eating habits. Alternatively, the diet information providing server 100 may similarly generate a prediction model that predicts a weight loss rate of a user due to an increase in the user's amount of exercise. In this case, too, a clinical trial of a predetermined increase in the amount of exercise is conducted on multiple subject users for, for example, several weeks, and subject users who have experienced weight loss as a result are identified, and the first attribute of physical information and answer information to a questionnaire are received from the subject users.

[0068] Here, "increasing the amount of exercise" refers to, for example, performing at least one of the following, and the diet information providing server 100 can set the definition of increasing the amount of exercise by appropriately combining the following. Engage in a set number of minutes (e.g., at least 150-300 minutes) of moderate-intensity (3.0-6.0 METs) physical activity per week (e.g., brisk walking, dancing, mowing the lawn). Engage in vigorous-intensity (6.0 METs or greater) physical activity (e.g., running, swimming, stair climbing) for a set period of time (e.g., at least 75-150 minutes) per week. · Perform prescribed strength training (e.g., 15 minutes of multiple strength training such as squats) a prescribed number of times per week (e.g., 3 times)

[0069] In addition to the above, "increasing the amount of exercise" also includes increasing the intensity and / or duration of daily physical activity or exercise.

[0070] Figure 8 is a diagram illustrating the prediction accuracy of a model predicting a diet effect due to increased exercise volume by the diet information providing server 100. Figure 8(A) shows the prediction accuracy when a model predicting a weight loss rate due to increased exercise volume is constructed without using the trained model 10, using the values ​​of age, sex, weight, and height as the first attribute of physical information and the answer information to the WHO-5 mental health status questionnaire as the explanatory variables. Figure 8(B) shows the prediction accuracy when the first attribute of physical information and the answer information to the question information are input into the trained model 10 to obtain the second attribute of physical information, and then the model predicting a weight loss rate is constructed.

[0071] As shown in Figure 1(A), when the trained model 10 was not used, the prediction accuracy (correlation coefficient r between the true value and the predicted value) was 0.31 for Spearman and 0.37 for Pearson. On the other hand, as shown in Figure 1(B), by acquiring the physical information of the second attribute using the trained model 10, the prediction accuracy was 0.51 for Spearman and 0.50 for Pearson, which was significantly improved compared to the case of Figure 1(A).

[0072] In this way, by amplifying the attributes of physical information using the trained model 10, it is possible to construct a predictive model that can accurately predict the rate of weight loss due to increased exercise volume of a user from limited physical information and response information of the user.

[0073] In the above-described embodiment and modified example, the diet information providing server 100 has constructed a prediction model for predicting a weight loss rate as a biological change of a user due to the user's dietary improvement or increased exercise amount. However, the biological change to be predicted is not limited to the weight loss rate.

[0074] The diet information providing server 100 can construct a prediction model to predict, for example, biological changes such as weight loss, visceral fat area reduction rate, changes in skeletal muscle (quantity and quality), changes in blood pressure, changes in blood components (cholesterol, fasting blood glucose level or HbA1c), changes in stress levels, and changes in sleep quality (proportion of non-REM sleep and REM sleep), in the same manner as in the above-mentioned embodiments.

[0075] In the above-described embodiment, the diet information providing server 100 inputs the answer information to the questionnaire along with the physical information of the first attribute into the trained model 10. However, inputting the answer information to the questionnaire is not required. Furthermore, the input physical information of the first attribute is not limited to the above-described age, sex, height, and weight, and various other physical information can be input as the physical information of the first attribute, thereby allowing any other physical information of the second attribute to be acquired from the trained model 10.

[0076] In the above-described embodiment, the diet information providing server 100 receives a diet effect prediction request from the user terminal 200 and transmits information regarding the predicted value of the diet effect according to the diet effect prediction model to the user terminal 200. However, the diet information providing server 100 may also receive a diet effect prediction request from a proxy user terminal of a proxy user acting as a proxy for the user of the user terminal 200 and transmit information regarding the predicted value of the diet effect to the proxy user terminal. Here, the proxy user terminal is, for example, a terminal of a business operator (such as a medical checkup business operator or a gym trainer) that acts between the user and the diet method proposal system, but is not limited to these. Furthermore, even when generating the diet effect prediction model, the diet information providing server 100 may receive the first attribute of physical information and the answer information to the questionnaire from a proxy subject terminal, such as a terminal of a test person in a clinical trial, instead of receiving the first attribute of physical information and the answer information to the questionnaire from the user terminal 200 of the subject user.

[0077] In the above embodiment, only one diet information providing server 100 is shown, but the processes executed by the diet information providing server 100 may be distributed and executed by multiple servers. For example, the process of generating the diet effect prediction model shown in Fig. 4 and the process of providing diet effect prediction information using the prediction model shown in Fig. 7 may be executed by separate servers.

[0078] Among the inventions described in the claims of this application, the invention described as an "information processing method" is one in which each step is automatically performed by at least one device such as a computer through software-based information processing, and is not performed by a human using a device such as a computer. In other words, the "information processing method" is an information processing method using computer software, and is not a method in which a human operates a computing tool called a computer. [Explanation of symbols]

[0079] 10…Trained model 11...CPU 18...Storage section 19…Communications Department 31...User physical information database 32...Questionnaire information database 33...Estimated physical information database 34...Effectiveness prediction information database 100...Diet information server 200...User terminal

Claims

1. inputting values ​​of the physical information of a first attribute of a plurality of subjects in whom biological changes due to an improvement in dietary habits or an increase in the amount of exercise have been observed into a first trained model that has been trained to estimate missing attribute values ​​from the values ​​of certain attributes of the physical information, and obtaining, from the first trained model, estimated values ​​of the physical information of a second attribute that is missing from the physical information of the first attribute of each of the subjects; generating a second trained model that predicts an amount of biological change of an arbitrary user due to an improvement in the dietary habit or an increase in the amount of exercise by performing learning to predict an amount of biological change of each of the subjects due to an improvement in the dietary habit or an increase in the amount of exercise of the arbitrary user using the value of the first attribute and the estimated value of the second attribute as explanatory variables; Control unit An information processing system comprising:

2. The control unit obtaining estimates of physical information of a plurality of different second attributes from the first trained model; and generating the second trained model by selecting, from the estimated values ​​of the plurality of second attributes, at least one estimated value having a high prediction accuracy of the biological change as the explanatory variable to be used for the prediction. The information processing system according to claim 1 .

3. The control unit inputting at least the values ​​of age, sex, height, and weight of each subject into the first trained model as values ​​of physical information of the first attribute; Generate the second trained model that predicts the weight loss rate of each subject as the biological change amount. The information processing system according to claim 1 .

4. inputting at least the values ​​of age, sex, height, and weight of each subject into the first trained model as values ​​of physical information of the first attribute; Generate the second trained model that predicts the visceral fat area reduction rate of each subject as the biological change amount. The information processing system according to claim 1 .

5. The control unit Response information of each subject to a questionnaire regarding the lifestyle habits or health condition of each subject is received from the subject terminal of each subject or a surrogate subject terminal of a surrogate subject acting on behalf of each subject, and the physical information of the first attribute and the response information are input into the first trained model.

5. The information processing system according to claim 3 or 4.

6. The control unit inputs the answer information of each of the subjects to the questionnaire regarding the mental health state of each of the subjects into the first trained model, and generates the second trained model that predicts the biological change due to the improvement in dietary habits or the increase in the amount of exercise. The information processing system according to claim 5 .

7. The control unit inputs, as explanatory variables, the value of the first attribute received from the user terminal of the arbitrary user or the proxy user terminal of a proxy user acting on behalf of the user, and the value of the second attribute obtained by inputting the value of the first attribute into the first trained model, obtains a predicted value of the biological change from the second trained model, and transmits the predicted value to the user terminal or the proxy user terminal. The information processing system according to claim 1 .

8. a first trained model is trained to estimate missing attribute values ​​from the values ​​of certain attributes of physical information, and the first trained model is input with values ​​of the physical information of a first attribute of multiple subjects in whom biological changes due to improved dietary habits or increased physical activity have been observed; an estimated value of the physical information of a second attribute that is missing from the physical information of the first attribute of each subject is obtained from the first trained model; and a second trained model is generated to predict the amount of biological change of each subject due to improved dietary habits or increased physical activity using the values ​​of the first attribute and the estimated values ​​of the second attribute as explanatory variables, and the second trained model is generated to predict the amount of biological change of an arbitrary user due to improved dietary habits or increased physical activity, and the value of the first attribute received from the user terminal of the arbitrary user or a proxy user terminal of a proxy user acting on behalf of the user, and the value of the second attribute obtained by inputting the value of the first attribute into the first trained model are input as the explanatory variables; Obtaining a predicted value of the biological change from the second trained model and transmitting it to the user terminal or the proxy user terminal; Control Unit An information processing system comprising:

9. inputting values ​​of the physical information of a first attribute of a plurality of subjects in whom biological changes due to an improvement in dietary habits or an increase in the amount of exercise have been observed into a first trained model that has been trained to estimate missing attribute values ​​from the values ​​of certain attributes of the physical information, and obtaining, from the first trained model, estimated values ​​of the physical information of a second attribute that is missing from the physical information of the first attribute of each of the subjects; generating a second trained model that predicts an amount of biological change of an arbitrary user due to an improvement in the dietary habit or an increase in the amount of exercise by performing learning to predict an amount of biological change of each of the subjects due to an improvement in the dietary habit or an increase in the amount of exercise of the arbitrary user using the value of the first attribute and the estimated value of the second attribute as explanatory variables; An information processing method executed by an information processing device.

10. a first trained model is trained to estimate missing attribute values ​​from the values ​​of certain attributes of physical information, and the first trained model is input with values ​​of the physical information of a first attribute of multiple subjects in whom biological changes due to improved dietary habits or increased physical activity have been observed; an estimated value of the physical information of a second attribute that is missing from the physical information of the first attribute of each subject is obtained from the first trained model; and a second trained model is generated to predict the amount of biological change of each subject due to improved dietary habits or increased physical activity using the values ​​of the first attribute and the estimated values ​​of the second attribute as explanatory variables, and the second trained model is generated to predict the amount of biological change of an arbitrary user due to improved dietary habits or increased physical activity, and the value of the first attribute received from the user terminal of the arbitrary user or a proxy user terminal of a proxy user acting on behalf of the user, and the value of the second attribute obtained by inputting the value of the first attribute into the first trained model are input as the explanatory variables; Obtaining a predicted value of the biological change from the second trained model and transmitting the predicted value to the user terminal or the proxy user terminal. Information processing methods.

11. In the information processing device, inputting values ​​of physical information of a first attribute of a plurality of subjects in whom biological changes due to improved diet or increased exercise have been observed into a first trained model that has been trained to estimate missing attribute values ​​from values ​​of certain attributes of the physical information, and acquiring, from the first trained model, estimated values ​​of physical information of a second attribute that is missing from the physical information of the first attribute of each of the subjects; generating a second trained model that predicts an amount of biological change of an arbitrary user due to an improvement in the dietary habit or an increase in the amount of exercise by performing learning to predict an amount of biological change of each of the subjects due to an improvement in the dietary habit or an increase in the amount of exercise of the arbitrary user using the value of the first attribute and the estimated value of the second attribute as explanatory variables; A program that executes the following.

12. a step of inputting values ​​of physical information of a first attribute of multiple subjects in whom biological changes due to improved dietary habits or increased physical activity have been observed into a first trained model, the first trained model having physical information of multiple attributes trained to estimate missing attribute values ​​from certain attribute values ​​of physical information, obtaining estimated values ​​of physical information of a second attribute that is missing from the physical information of the first attribute of each of the subjects, and learning to predict the amount of biological change of each subject due to the improved dietary habits or increased physical activity of each of the subjects using the values ​​of the first attribute and the estimated values ​​of the second attribute as explanatory variables, and inputting the value of the first attribute received from the user terminal of the user or a proxy user terminal of a proxy user acting on behalf of the user and the value of the second attribute obtained by inputting the value of the first attribute into the first trained model as the explanatory variables; obtaining a predicted value of the biological change from the second trained model and transmitting the predicted value to the user terminal or the proxy user terminal; A program that executes the following.

Images