Computer program, information processing apparatus, and method

Abstract

To provide a computer program, an information processing apparatus, and a method used for estimating a target user's state of health on the basis of the history of behaviors executed by the target user, and having at least partially improved performance.SOLUTION: The method includes the steps of: acquiring target behavior data for identifying the history of behaviors executed by a target user; and inputting the target behavior data to an estimation model generated by executing supervised learning, to cause the estimation model to output, in target state data, of the target user, for identifying the state of exercises, the state related to a lifestyle-related disease, or the state of insufficiency of physical activity, the target state data including at least one of data for identifying the body weight of the target user, data for identifying the BMI of the target user, data for identifying the number of steps of the target user, data for identifying the presence or absence of smoking of the target user, and data for identifying a metabolic syndrome diagnostic reference value of the target user.SELECTED DRAWING: Figure 12

Description

[Technical Field]

[0001] The technology disclosed in the present application is executed by a user who is the subject of estimation (target user). By inputting data that identifies the history of user behavior into the learning model, A computer program that outputs data for identifying health conditions from the learning model. , and an information processing device and method. [Background technology]

[0002] The health status of a target user is estimated based on data describing the user's purchasing history. A method for constructing an estimation model is disclosed in Japanese Patent No. 6916367 (Patent Document 1). This patent document is incorporated herein by reference in its entirety. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Patent No. 6916367 Summary of the Invention [Problem to be solved by the invention]

[0004] The health status of the target user based on the history of actions taken by the target user recently There is a need to provide a method with improved performance for estimating do.

[0005] Therefore, the technology disclosed in the present application is based on the history of actions performed by the target user. and at least partially improves the health status of the target user based on the The present invention provides a computer program, an information processing device, and a method having improved performance. [Means for solving the problem]

[0006] A computer program according to one embodiment is "executed by at least one processor." By doing so, target behavior data that identifies a history of behaviors performed by a target user is acquired. The target behavior data is applied to the estimation model generated by performing supervised learning. The input of the target sleep state data identifies the sleep state of the target user. and target characteristic data for identifying the sleep characteristics of the target user, the sleep rhythm of the target user, target rhythm data identifying the target user's sleep time; and , target category data identifying a sleep category to which the target user belongs, and outputting target sleep state data including at least one of the plurality of sleep states from the estimation model. "It can function at least one processor."

[0007] An information processing device according to one embodiment includes at least one processor, a processor, target behavior data identifying a history of behaviors performed by the target user; and then applying the target behavior data to the generated estimation model by performing supervised learning. By inputting the data, target sleep state data that identifies the sleep state of the target user is obtained. and target characteristic data identifying the sleep characteristics of the target user. target rhythm data for identifying the rhythm of the target user; target time data for identifying the sleeping time of the target user; and target category data identifying a sleep category to which the target user belongs. and outputting target sleep state data including at least one of the sleep states from the estimation model. It can be achieved.

[0008] In one embodiment, the method includes: A method executed by one processor, the at least one processor By executing the above command, a target line that identifies the history of actions performed by the target user is generated. The first is to acquire dynamic data and then perform supervised learning to generate an estimated model. A target that identifies a sleep state of the target user by inputting the target behavior data. sleep state data, subject characteristic data identifying the sleep characteristics of the subject user; Target rhythm data for identifying a user's sleep rhythm, identifying the target user's sleep time Target time data and a target category that identifies a sleep category to which the target user belongs and outputting target sleep state data from the estimation model, the target sleep state data including at least one of and "causing the processing to be performed."

[0009] Another embodiment of the method is a method for executing computer-readable instructions. a method executed by at least one processor, the method comprising: By executing the above command, the teacher data of each group is obtained from one sample user corresponding to the group. The sample behavior data that identifies the history of behaviors performed by the user and the corresponding user and sample sleep state data identifying sleep states of the sample users. An acquisition step of acquiring training data, and a learning step of inputting the plurality of sets of training data into a learning model to learn the data. By doing so, target behavior data that identifies the history of behaviors performed by the target user is input. and target sleep state data identifying a sleep state of the target user, target characteristic data for identifying the sleep characteristics of the target user; target time data for identifying the target user's sleeping time; and and target category data identifying a sleep category to which the user belongs. target sleep state data, the estimation model configured such that: and generating the signal.

[0010] According to another embodiment, the computer program is The target behavior data identifying the history of behaviors performed by the target user is generated. acquiring and applying the subject behavior data to an estimation model generated by performing supervised learning; By inputting the target menopausal state data, the target user's menopausal state is identified. and subject overall condition data identifying the overall menopausal condition of the subject user. target first condition data identifying a first symptom state of the target user; and subject second condition data identifying a menopausal condition. and causing the at least one processor to output data from the estimation model. It can be made to function.

[0011] An information processing device according to another aspect includes at least one processor, Another processor generates target behavior data identifying a history of behaviors performed by the target user. The target behavior data is acquired, and the target behavior data is applied to an estimation model generated by performing supervised learning. A target menopausal symptom indicator that identifies the menopausal symptom state of the target user by inputting the data. and condition data, the condition data including target overall condition data identifying the overall condition of menopausal symptoms of the target user. target first condition data identifying a first symptom state of the target user; and (b) a second condition data identifying the second symptom state of the subject. The method may be configured to output menopausal status data from the estimation model.

[0012] Another embodiment of the method is a method for executing computer-readable instructions. a method executed by at least one processor, the method comprising: and executing said instructions to identify a history of actions performed by said target user. Obtaining behavioral data and generating an inference model by performing supervised learning By inputting the target behavior data into the target menopausal status data for identifying the overall status of menopausal symptoms of the target user; subject first condition data identifying a first symptom state of the subject user; and target second condition data identifying a second symptom condition of the target user. and outputting target menopausal status data, including at least one, from the estimation model. It is possible.

[0013] Another embodiment of the method is a method for executing computer-readable instructions. a method executed by at least one processor, the method comprising: By executing the above command, the teacher data of each group is obtained from one sample user corresponding to the group. The sample behavior data that identifies the history of behaviors performed by the user and the corresponding user and sample menopausal status data identifying a menopausal symptom status of a sample user of the An acquisition step of acquiring a plurality of sets of training data, and inputting the plurality of sets of training data into a learning model. By learning from the target behavior data, the target behavior data is generated to identify the history of behaviors performed by the target user. and inputting target menopausal status data that identifies the menopausal symptom status of the target user. subject overall condition data identifying the overall condition of menopausal symptoms of the subject user; target first condition data identifying a state of a first symptom of the target user, and a second condition data identifying a state of the target user and subject second condition data identifying a menopausal condition. and a generation step of generating an estimation model configured to output the data. can be done. [Brief explanation of the drawings]

[0014] [Figure 1] FIG. 1 is a block diagram showing an example of the configuration of a communication system according to an embodiment. [Figure 2] FIG. 2 is a block diagram showing an example of a hardware configuration of the server device 10 shown in FIG. [Figure 3] FIG. 3 is a flow diagram showing an example of an operation executed by the communication system 1 shown in FIG. [Figure 4] FIG. 4 is a diagram showing an example of the Athens Insomnia Scale. [Figure 5A] FIG. 5A is a diagram showing an example of the Pittsburgh Sleep Quality Index. [Figure 5B] FIG. 5B is a diagram showing an example of the Pittsburgh Sleep Quality Index. [Figure 5C] FIG. 5C shows an example of the Pittsburgh Sleep Quality Index. [Figure 6A] FIG. 6A is a diagram showing an example of a three-dimensional sleep scale. [Figure 6B] FIG. 6B is a diagram showing an example of a three-dimensional sleep scale. [Figure 7] FIG. 7 is a diagram showing an example of an insomnia severity questionnaire. [Figure 8A] FIG. 8A shows an example of the Munich Chronotype Questionnaire. [Figure 8B] FIG. 8B shows an example of the Munich Chronotype Questionnaire. [Figure 9] FIG. 9 is a diagram showing an example of sleep category data used in the communication system 1 shown in FIG. [Figure 10] FIG. 10 is a diagram showing another example of sleep category data used in the communication system 1 shown in FIG. [Figure 11] FIG. 11 is a diagram conceptually illustrating an example of a search table used in the communication system 1 shown in FIG. [Figure 12] FIG. 12 is a flow diagram showing another example of the operation executed by the communication system 1 shown in FIG. [Figure 13] FIG. 13 is a diagram showing an example of the Simplified Menopausal Index (SMI). [Figure 14] FIG. 14 is a diagram conceptually showing another example of a search table used in the communication system 1 shown in FIG. [Figure 15] FIG. 15 is a diagram schematically showing the flow of data for predicting the health of a customer using the health prediction system according to the embodiment. [Figure 16] FIG. 16 is a diagram showing a specific example of the functional configuration of the integrated health prediction program 100 according to the embodiment. [Figure 17] FIG. 17 is a diagram illustrating a specific example of purchase data according to the embodiment. [Figure 18] FIG. 18 is a diagram illustrating a specific example of purchase data according to the embodiment. [Figure 19] FIG. 19 is a diagram showing a specific example of a table for realizing proposals to customers according to output values ​​of a prediction model. [Figure 20] FIG. 20 is a diagram showing a specific example of a table for realizing proposals to customers according to output values ​​of a prediction model. [Figure 21]FIG. 21 is a diagram showing the processing flow of the integrated health prediction program according to the embodiment. [Figure 22] FIG. 22 is a diagram illustrating a specific example of the hardware configuration of an information processing device that executes an integrated health prediction program according to an embodiment. [Figure 23] FIG. 23 is a diagram showing a specific example of the functional configuration of a learning system including a model learning program according to an embodiment. [Figure 24] FIG. 24 is a diagram showing a specific example of a product category. [Figure 25] FIG. 25 is a diagram showing a specific example of the relationship between menopausal symptoms and health issues found from items purchased by subjects in the H group regarding menopausal symptoms. [Figure 26] FIG. 26 is a diagram showing specific examples of product categories for women who belong to the H group and have menopausal symptoms. [Figure 27] FIG. 27 is a diagram showing a processing flow of the model learning program according to the embodiment. [Figure 28] FIG. 28 is a diagram illustrating a specific example of the hardware configuration of an information processing device that executes the model learning program according to the embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0015] Various embodiments of the present invention will be described below with reference to the accompanying drawings. Components throughout the drawings are given the same reference numerals. Please note that some elements may be omitted in other drawings for clarity of illustration. Furthermore, the accompanying drawings are not necessarily drawn to scale. Please be careful.

[0016] The various systems, methods, and devices described herein may be used in any manner. Indeed, the present disclosure is not to be construed as limiting the scope of the invention as disclosed. Each of the various embodiments, combinations of the various embodiments with each other, and Any novel features and combinations of the various embodiments of the present invention are intended to be included in the present invention. The present invention is directed to various systems, methods, and apparatuses described herein. refers to a particular aspect, a particular feature, or a combination of such particular aspects and features. The products and methods described herein may be used in combination with one or more Nor does it require that any particular effect be present or problem be solved. Also, various features or aspects of the various embodiments described herein, Or, some of such features or aspects may be used in combination with each other.

[0017] The operations of some of the various methods disclosed herein may be conveniently summarized. However, descriptions in this manner may be inconsistent with the specific order specified below. Unless otherwise required by the text, this includes re-arranging the order of the actions described above. It should be understood that, for example, actions listed in a sequential order may, in some cases, be reordered. Furthermore, for the sake of simplicity, the accompanying drawings may be The various features and methods described herein may be used in conjunction with other features and methods. It does not show the various ways in which

[0018] Any theory of operation, scientific principles or other information presented herein in connection with the devices or methods of the present disclosure may be used without departing from the spirit and scope of the present invention. Other theoretical descriptions are provided for the purpose of better understanding and are not intended to limit the scope of the technology. The apparatus and methods of the appended claims are not intended to The present invention is not limited to apparatus and methods that operate in a manner described by any theory of operation.

[0019] Any of the various methods disclosed herein may be implemented using a computer readable or more media, using a plurality of computer-executable instructions stored thereon. The one or more media may be implemented and executed in a computer. , for example, at least one optical media disk, a plurality of volatile memory components, or a plurality of Non-transitory computer-readable storage, such as non-volatile memory components Here, the plurality of volatile memory components may be, for example, DRAM or SR. The non-volatile memory components include, for example, hard drives and hard disk drives. Furthermore, the computer may perform calculations, for example, in the market, including smartphones and other mobile devices with hardware that performs Includes any available computer.

[0020] Such computer-executable programs for implementing the techniques disclosed herein may be implemented using the Any of a number of possible instructions may be used during the implementation of the various embodiments disclosed herein. together with any data generated or used in one or more computers. a non-transitory computer-readable storage medium (e.g., a non-transitory computer-readable storage medium) Such computer executable instructions may be stored in a memory (e.g., a memory device). It can be part of a separate software application or can be integrated into a web browser or or other software applications (such as remote computing applications). Part of a software application accessed or downloaded via the Such software may be, for example, any suitable software available on the market. on a single local computer (as a process running on any suitable computer) or in a network using one or more network computers environment (e.g., Internet, wide area network, local area network) networks, client-server networks (such as cloud computing networks), or or other such networks).

[0021] For clarity, we will refer to a specific selected version of the various software-based implementations. Only various aspects are described, and other details that are well known in the art are omitted. For example, the techniques disclosed herein may be implemented in a specific computer language or program. For example, the technology disclosed in this specification is not limited to C, C++, Java ( trademark), or software written in any other suitable programming language. Similarly, the techniques disclosed herein may be implemented by a particular computer or The present invention is not limited to any particular type of computer or hardware. The specific details of A are well known and need not be described in detail herein.

[0022] Furthermore, various such software-based embodiments (e.g., as described herein) may be implemented using the A computer program for causing a computer to perform any of the various methods disclosed in Any of the instructions (including instructions executable by the computer) may be transmitted to the uploader by suitable communication means. It can be loaded, downloaded, or accessed remotely. Such suitable communication means are, for example, the Internet, the World Wide Web, an intranet, software applications, cables (including fiber optic cables), magnetic communications , electromagnetic communications (including RF communications, microwave communications, and infrared communications), electronic communications, or other This includes such means of communication.

[0023] I.Chapter 1 1. Overview In the technology disclosed in the present application, in brief, at least one information processing device Target actions that identify the history of actions performed by a user who is the target of a specific action (target user) By acquiring data and inputting this target behavior data into the estimation model, The data (health condition data) that identifies the health condition of the user is output from the estimation model. It is possible.

[0024] In addition, at least one information processing device is At least one target product and / or at least one target product corresponding to the health condition data of the target user Furthermore, at least one target service can be determined. The device then provides at least one target service and / or at least one It is also possible to output information identifying the target service.

[0025] 2.Communication System Configuration Such a technique can be implemented using a communication system such as the one shown in FIG. 1 is a block diagram showing an example of the configuration of a communication system according to an embodiment.

[0026] As shown in FIG. 1, a communication system 1 according to an embodiment includes, for example, a At least one server device (information processing device) 10 and at least one The system may include one or more terminal devices (information processing devices) 20. Although two server devices 10A and 10B are illustrated as one server device 10, any Any number of server devices 10 may be used. Similarly, FIG. 1 illustrates at least one terminal device. Although two terminal devices 20A and 20B are shown as an example of the server 20, any number of server devices may be used. 10 can be used.

[0027] Each server device 10 is connected to at least one other server device 10 and / or at least It is possible to connect to one terminal device 20 via the communication network 2. at least one other terminal device 20 and / or at least one server device 10 can be connected to via a communication network 2.

[0028] Each server device 10 may be any information processing device. For example, personal computers, workstations, supercomputers and mainframes Each terminal device 20 may include, but is not limited to, any information. Such information processing devices may be, for example, mobile phones, smartphones, Tablets, personal computers, workstations and mobile information terminals, etc. These may include, but are not limited to:

[0029] In one example, at least one server device 10 estimates the health status of any user. Services ("Estimation Services"), and / or at least An operating company that operates a service that proposes at least one target product, etc. ("Proposal Service") For convenience, the information processing device may be an information processing device managed by the estimation service and / or may use the term "estimated services, etc." to refer to proposed services.

[0030] In one example, at least one server device 10 and / or at least one terminal device 20 is provided to corporate users and / or individual users who receive estimated services, etc. from the operating companies. It can be an information processing device that is managed by the information processing device.

[0031] The operation of acquiring the target behavior data described above is performed by at least one server device 10 and / or Alternatively, it may be executed by at least one terminal device 20.

[0032] The above-mentioned estimation model is implemented by at least one server device 10 and / or at least one The estimation model thus generated can be generated by at least one terminal device 20. It can be held by one server device 10 and / or at least one terminal device 20. The estimation model thus stored is transmitted to at least one server device 10 and / or It can be used by at least one terminal device 20.

[0033] The operation of outputting the health condition data from the estimation model is performed by at least one server. It may be executed by the device 10 and / or at least one terminal device 20.

[0034] Determining at least one eligible product and / or at least one eligible service as described above The operation is performed by at least one server device 10 and / or at least one terminal device 2. 0. At least one of the target services and / or at least one of the target services The operation of outputting the data identifying the target services is performed by at least one server device 10. , and / or may be executed by at least one terminal device 20.

[0035] In addition, communication network 2 includes mobile phone networks, wireless networks, fixed telephone networks, the Internet, Intranet, Local Area Network (LAN), Wide Area Network (W AN) and / or Ethernet networks The wireless network may include, for example, Bluetooth (registered trademark) standard), WiFi (such as IEEE 802.11a / b / n), WiMax, Cellular This includes, but is not limited to, RF connections via radar, satellite, laser, infrared, etc. This can be done.

[0036] 3. Hardware configuration of each information processing device Next, the hardware configuration of each information processing device (server device 10, terminal device 20) is An example will be described.

[0037] (1) Hardware Configuration of Server Device 10 FIG. 2 is a block diagram showing an example of a hardware configuration of the server device 10 shown in FIG. (Note that in FIG. 2, the reference numerals in parentheses refer to terminal devices 20, as will be described later.) are listed consecutively).

[0038] As shown in FIG. 2, the server device 10 includes a central processing unit 11, a main memory device 12, and an input / output a power interface device 13, an input device 14, an auxiliary storage device 15, and an output device 16. These devices are connected to each other by a data bus and / or a control bus. It is being done.

[0039] The central processing unit 11 is called the "CPU" and executes instructions and It is possible to perform operations on data and store the results of the operations in the main memory device 12. Furthermore, the central processing unit 11 communicates with the input device 1 via the input / output interface unit 13. 4, auxiliary storage device 15, output device 16, etc. The server device 10 It is possible to include one or more such central processing units 11 .

[0040] The main memory device 12 is referred to as "memory", and is connected to the input device 14, the auxiliary memory device 15, and the communication network. 2 (terminal device 20, etc.) via the input / output interface device 13. The main memory 1 can store the data, the operation results of the central processing unit 11, and the 2 is volatile memory (e.g., registers, cache, random access memory (RAM) )), non-volatile memory (e.g., read-only memory (ROM), EEPROM, flash memory, flash memory), and storage (e.g., hard disk drive (HDD), computer hard disks (SSDs), magnetic tape, optical media, etc. These may include, but are not limited to, computer-readable media. As will be understood, the term "computer-readable medium" includes modulated The transmission medium is not a fixed data signal, i.e., a transient signal, but rather memory and storage. The data storage medium may include a storage medium.

[0041] The auxiliary storage device 15 is a storage device having a larger capacity than the main storage device 12. The storage device 15 stores instructions and data (computer program instructions) that make up a particular application. These instructions are stored in the memory and controlled by the central processing unit 11. Commands and data (computer programs) are input via an input / output interface device 13. The auxiliary storage device 15 can be a magnetic disk device and / or Or, optical disk devices, etc. may be included without being limited to these.

[0042] The specific application here refers to the operating system, various software, Applications (including web browsers), which run to provide estimation services, etc. dedicated applications that are run to receive the provision of estimation services, etc. This may include, but is not limited to, applications, etc.

[0043] The input device 14 is a device for inputting data from the outside, and may be a keyboard, a touch panel, Including, but not limited to, buttons, mouse and / or sensors (microphone, camera) It is possible.

[0044] The output device 16 may include a display device, a touch panel, and / or a printer device. This can include, but is not limited to:

[0045] In such a hardware configuration, the central processing unit 11 stores the data in the auxiliary storage device 15. The instructions and data (computer programs) that make up the specific application stored in the The instructions and data are sequentially loaded into the main memory 12, and the loaded instructions and data can be operated. As a result, the central processing unit 11 can access the output device via the input / output interface unit 13. 16, or via the input / output interface device 13 and the communication network 2, Various information ( It is possible to send and receive data.

[0046] In this way, the server device 10 executes the specific application that has been installed. By carrying out this service, you will be able to provide the Estimated Services, etc. and / or receive the Estimated Services, etc. 3 and 12, etc.) Additionally or alternatively, the server device 10 may execute a browser. By accessing other server devices 10 via the / or actions related to receiving the provision of estimated services, etc. (see Figures 3 and 12, etc.) It is also possible to perform various operations (such as those described below) using the

[0047] The server device 10 may be configured to perform the following functions in place of or in addition to the central processing unit 11: One or more microprocessors and / or graphics processing units It may also include a graphics processing unit (GPU).

[0048] (2) Hardware Configuration of Terminal Device 20 As shown in parentheses in FIG. 2, the terminal device 20 is substantially the same as the server device 10. As shown in FIG. 2, the terminal device 20 may have a central processing unit (CPU) device 21, a main memory device 22, an input / output interface device 23, an input device 24, The device may include an auxiliary storage device 25 and an output device 26. These devices may exchange data. Each of these devices is connected by a bus and / or a control bus. As explained in relation to

[0049] A specific application stored in the auxiliary storage device 25 and executed by the central processing unit 21 The application includes the operating system, various software applications (web browsers, (including the use of third-party applications), dedicated applications executed to receive the provision of estimation services, etc. In addition, specific applications may require It may also include dedicated applications that run to provide specific services, etc.

[0050] In such a hardware configuration, the central processing unit 21 stores the data in the auxiliary storage device 25. The instructions and data (computer programs) that make up the specific application stored in the The instructions and data are sequentially loaded into the main memory 22, and the loaded instructions and data can be operated. As a result, the central processing unit 21 can access the output device via the input / output interface unit 23. 26, or via the input / output interface device 23 and the communication network 2, Various information is exchanged between other devices (for example, the server device 10 and / or other terminal devices 20). (Data) can be sent and received.

[0051] In this way, the terminal device 20 executes the specific application that has been installed. By doing so, you will receive the Estimated Services, etc. and / or provide the Estimated Services, etc. 3 and 12, etc.) related to the Additionally or alternatively, the server device 10 may execute a browser. By accessing other server devices 10 via the or actions related to receiving the provision of the estimated service, etc. (see Figures 3 and 12, etc.) It is also possible to perform various operations (such as those described below).

[0052] The terminal device 20 may be used in place of the central processing unit 21 or in addition to the central processing unit 21. or higher microprocessor and / or graphics processing unit It can also include a graphics processing unit (GPU).

[0053] 4. Executed by communication system 2 to provide sleep state data regarding the state of sleep. Actions that can be performed Next, the communication system 2 described above is used to provide sleep state data relating to the state of sleep. A specific example of the operations performed by the system will be further described with reference to FIG. 10 is a flowchart showing an example of an operation executed by the communication system 1 shown in FIG.

[0054] (1) Step 1000 First, in step (hereinafter referred to as "ST") 1000, an information processing device, e.g. For example, a server device 10A managed by a company that operates an estimation service or the like may generate an estimation model. It is possible to acquire and store multiple sets of training data used to generate the

[0055] The training data for each group of multiple training data sets is a sample of one person corresponding to this group. data identifying a history of actions performed by a user ("Sample Behavior Data"); data identifying the sleep state of one sample user corresponding to the set of For example, the first set of training data may include the first set of training data. Sample actions identify the history of actions performed by one sample user (user A) and sample sleep state data identifying a sleep state of Person A. The second set of training data was run by one sample user (Mr. B) corresponding to the second set. Sample behavioral data that identifies the history of user B's behavior and sample behavioral data that identifies user B's sleep state. and pull sleep state data.

[0056] (1A) Sample behavioral data Each sample behavior data is a data set of at least one of a plurality of predetermined behaviors. The data may be data identifying a predetermined number of (one or more) actions. Each has at least a partial positive or negative effect on the sleep state of the person who performed the action. Such predetermined actions may be, for example, the following: Examples may include, but are not limited to: (Examples of behaviors that can at least partially positively influence the state of human sleep) - Participated in aerobic exercise for at least one hour at least two days per week. I woke up at the same time every day. Limit caffeinated drinks to two or fewer drinks per day. The working hours per day were less than 8 hours. Purchased 10 or more coffees (this behavior is, for example, the product category to which coffee belongs). (These products can be identified by data including the product and its purchase quantity.) (Examples of behaviors that may at least partially adversely affect a person's sleep state) Participated in aerobic exercise for at least one hour on less than one day per week. · I woke up at a different time every day. Drinking three or more caffeinated beverages per day. -Working hours were 13 hours or more per day. - Purchased 5 or more bottles of milk (this behavior is related to the product category to which milk belongs and its (These items may be identified by data including the quantity purchased).

[0057] Each sample behavior data is a set of at least one of the predetermined multiple behaviors exemplified in this way. The data may be data identifying each of at least one behavior.

[0058] In one example, the products purchased by the user and / or the services used by the user may be Since the above-mentioned pre-treatment may tend to have a positive or negative effect on the user's sleep state, The defined actions are the products purchased by the sample users (or the categories to which the products belong). services used by sample users (or the categories to which those services belong) This type of behavior can include, for example, shopping at retail stores (drug stores, supermarkets, etc.). POS (Point of Sales) owned by retailers (e.g., convenience stores) It can be generated using data.

[0059] In one example, each of the plurality of predetermined actions may be provided with unique identification data (alphabetical, numeric, or Each sample behavioral data is assigned such a classification. For example, if you use POS data, you can use multiple samples. Each piece of user behavior data is a JICFS classification (classification code), product code, and Product category, purchase amount, purchase quantity, number of purchases, gender, age, and / or purchase date, etc. These may include, but are not limited to:

[0060] An information processing device, such as a server managed by a company that operates an estimation service The device 10A may collect such sample behavioral data, for example, by one of the following methods: It can be obtained by at least one method. The information processing device communicates with at least one other server device 10 (e.g., Another server device 10 that stores POS data installed in a drug store or the like, or The POS data is received from another server device (10) that can access the POS data. The information processing device sends a predetermined questionnaire (executed by a sample user) via the communication network 2. A small number of individuals may receive an email or web page containing their responses to a questionnaire about their behavior. It is received from at least one terminal device 20 and / or at least one server device 10. The information processing device collects responses from multiple users to the predetermined questionnaire. The content of the reply is received from at least one server device 10 via the communication network 2. The information processing device collects responses from multiple users to the predetermined questionnaire. The response content is received via a recording medium (USB memory, DVD-ROM, etc.).

[0061] (1B) Sample sleep state data The sample sleep state data included in each set of training data is one of the following examples: may include, but is not limited to, at least one of: Data identifying the sleep characteristics of one sample user corresponding to the set ("sample "Characteristic Data") Data identifying the sleep rhythm of one sample user corresponding to the set ("sample" "Lurism Data") Data identifying the sleep time of one sample user corresponding to the set ("sample time" Interval data) Identify the sleep category to which the sample user (whose sleep) corresponds to the set belongs. Data ("Sample Sleep Category Data")

[0062] First, the sample property data indicates whether the sample users have good sleep quality (e.g., It can be data that identifies whether something is good (e.g., how bad) or bad (e.g., how bad) In one example, the sample property data may include an overall assessment of the sample user's sleep quality. In this case, the sleep quality of the sample user may be, for example, For example, the data may identify a score calculated by any of the following methods.

[0063] Sample users' responses to the questions listed in the Athens Insomnia Scale (AIS) (see Figure 4) A score calculated based on the responses of the user using the method described in the Athens Insomnia Scale, or The modified score obtained by performing any calculation on this score (addition, subtraction, multiplication, division, etc.) Questions listed in the Pittsburgh Sleep Quality Index (PSQI) (see Figures 5A to 5C) Based on sample user responses to the items listed in the Pittsburgh Sleep Quality Index, The score calculated by the method, or the calculation of this score by any method (addition, subtraction, multiplication, division, etc.) The corrected score obtained by - The questionnaire items listed in the 3-dimensional sleep scale (3DSS) (see Figures 6A and 6B) Based on the responses of sample users, the data was calculated using the method described in the 3D sleep scale. The score obtained by performing any calculation (addition, subtraction, multiplication, division, etc.) on this score. Corrected score Sample questions from the Insomnia Severity Inventory (ISI) (see Figure 7) A score calculated based on the user's responses using the method described in the insomnia severity questionnaire Or, a modified score obtained by performing any operation on this score (addition, subtraction, multiplication, division, etc.) a - Sample users' responses to questions on any other questionnaire or scale A score calculated based on the content and in the manner described in the questionnaire or scale, or Corrected score obtained by calculating the score using any method

[0064] Second, the sample rhythm data is of good quality (for example, It can be data that identifies whether something is good (e.g., how bad) or bad (e.g., how bad) In one example, the sample rhythm data may be a sleep rhythm assessment of the sample user. It may be data that identifies the core.

[0065] In this case, the sleep rhythm of the sample user is, for example, the sample user's weekday sleep rhythm. The social networking factor is the absolute value of the difference between the median time of sleep during the day and the median time of sleep during the weekend. Data that identifies social jet lag (or a value calculated using any method) For example, the sleeping hours of a certain sample user on weekdays are from midnight to If it is 6:00 AM, then the central time is 3:00 AM and the sample user is on a holiday. If the sleep period is from 3:00 AM to 11:00 AM, the median time is 7:00 AM. Therefore, the social jet lag is 4 hours. The smaller (or larger) the sample user's sleep rhythm is, the better (or worse) It can be said that.

[0066] The sample user's sleeping hours on weekdays and weekends are, for example, The results of the questionnaire on the Multimodal Communication Test (MCTQ) (see Figures 8A and 8B) In another example, the sample user The information can be obtained from the website via email, a web page, etc.

[0067] Third, the sample time data shows the sleep time (sleep time per day) of the sample user. The sleep duration of a sample user may be, in one example, Munich Chrono The results of the questionnaire (MCTQ) (see Figures 8A and 8B) It can be extracted from the responses of sample users, or in another example, from the sample users. The information can be obtained from the website via email, a web page, etc.

[0068] Fourth, the sample sleep category data is the sleep category to which the sample user (sleep) belongs. As an example, the sample sleep category data may be: The sample user's sample property data and the sample user's sample rhythm data Identify the category to which the sample user (sleep) belongs, determined based on the data. In this case, the sleep category may be determined by the following method, for example: obtain.

[0069] FIG. 9 is an example of sleep category data used in the communication system 1 shown in FIG. As shown in FIG. 9, one of the vertical and horizontal axes (here, the horizontal axis) is a graph showing the relationship between the sample The scores identified by the property data are arranged on the other of the vertical and horizontal axes (here, the vertical axis) On one axis, values ​​identified by sample rhythm data can be arranged. At least one (here, one) threshold is set for the score identified by the property data. A value, i.e., a threshold A1 (6 points) can be set on the other axis. For the values ​​identified by the data, there is at least one (here one) threshold, i.e. A threshold B1 (1 hour) can be set. There are four different sleep categories ( Here, sleep categories 10, 11, 20, 21) can be formed. Category 10 is the category where sleep disturbances are most likely to be evident. Sleep category 21 is the category most likely to have the least sleep disturbances. Sleep categories 11 and 20 are positioned between sleep categories 10 and 21. The sleep category data is divided into these four sleep categories. The data may be data that identifies any of the sleep categories.

[0070] FIG. 10 shows the sleep category data used in the communication system 1 shown in FIG. In the example shown in FIG. 10, one of the axes is a graph showing the relationship between sample property data and the average value of the sample property data. For example, two thresholds are set for the scores to be discriminated, namely, threshold A1 (6 points) and threshold A2 (8 points) can be set. The other axis can be identified by the sample rhythm data. For example, two thresholds, i.e., threshold B1 (1 hour) and threshold B2 (3 hours), are set for the value. ) can be set. As a result, the quality of sleep can be determined by the thresholds A1, A2 and B1, B2. Nine different sleep categories (herein referred to as "sleep categories") are divided into categories based on quality and sleep rhythm. sleep categories 100, 101, 102, 200, 201, 202, 300, 301 , 302). In this example, the sleep category 300 is a category that represents sleep disorders. The category where harm is most likely to be evident is sleep category 202. The remaining sleep categories are those most likely to be experiencing sleep disturbances. It can be said that this category is positioned between sleep categories 300 and 202. Category data is based on one of these nine sleep categories. It may be identifying data.

[0071] The examples shown in Figures 9 and 10 are merely examples shown for convenience, and may differ depending on the sample property data. Any number of thresholds can be set for the score to be discriminated, and the discriminator can discriminate based on the sample rhythm data. Any number of thresholds may be set for the values ​​to be classified. By setting more thresholds for each rhythm data, the quality of sleep and Generate more distinct sleep categories divided in terms of sleep rhythms It is possible.

[0072] An information processing device, such as a server managed by a company that operates an estimation service The device 10A (and / or the terminal device 20) generates such sample The behavioral data and the sample sleep state data are combined to generate a A set of training data can be generated and stored. The information processing device identifies a certain sample user from the acquired plurality of sample behavior data. extracting sample behavioral data with associated identification data, and A sample with identification data that identifies the same sample user from the behavioral data. The sleep state data is extracted and combined to form a set of The training data can be generated and saved. The same process can be performed for each of multiple sample users. By executing the above, the information processing device generates and stores multiple sets of training data. It is possible.

[0073] (2) Step 1002 Returning to FIG. 3, next, in ST1002, an information processing device, for example, an estimation service The server device 10A (and / or the terminal device 20) managed by a company that operates a service, etc. Using multiple sets of training data acquired by the ST1000, a learning model is created through machine learning. By further training, an estimation model can be generated. ,any well-known machine learning method, e.g., gradient boosted trees (e.g., LightGBM), These may include, but are not limited to, neural networks or linear regression.

[0074] The information processing device sequentially inputs a plurality of sets of training data into the learning model. Here, as described above, each set of training data is Sample behavior data of a sample user corresponding to the pair and the sample user corresponding to the pair The information processing device may include the sample sleep state data of each set of training data. The sample behavior data included in the training data of the set is used as explanatory variables, and the sample behavior data included in the training data of the set is used as explanatory variables. Sample sleep state data can be used as the response variable.

[0075] (2A) Machine Learning Examples For example, in the case of using gradient boosting trees, the information processing device uses a set of training data. When the sample behavior data (explanatory variables) included in the data are input into the training data, From the sleep state data (objective variable) and the estimated value (i.e., the value output from the learning model), Decision trees can be created and added to the learning model to improve the calculated objective function. The information processing device performs this process for the number of decision trees determined by the hyperparameters. For the second and subsequent decision trees, the target variable and the previously created In this way, information can be learned by comparing the difference between the estimated value and the value estimated by the decision tree. The processing unit can determine the weights of the branches and leaves of each decision tree: In gradient boosting trees, each decision tree uses explanatory variables from the training data as input data. The value of the leaf node reached by following each branch of the decision tree is the output data (objective variable). Furthermore, the error between this data (objective variable) and the objective variable in the training data is small. The method for each branch of the decision tree and the value of each leaf node can be optimized so that:

[0076] The information processing device may include, for example, a neural network including an input layer, an intermediate layer, and an output layer. In this case, the information processing device inputs explanatory variables from the training data into the input layer. The data output from the output layer (objective variable) and the objective variable in the training data are To reduce the error of Each parameter (weight) can be optimized (learned).

[0077] In the case where linear regression is used, the information processing device improves the error between the objective variable and the estimated value. The parameters of the linear regression model can be learned as follows. It is divided into an input layer and an output layer, and the explanatory variables in the training data are used as input data for the input layer. The data (objective variable) output from the output layer by inputting The least squares method is used to find the error between the objective variable and the linear regression model. The parameters can be optimized.

[0078] (2B) First sets of training data included in the sets of training data In one example, the information processing device The estimation model can be generated (updated) using the first plurality of sets of training data. The training data for each group is the sample behavior data of the sample user corresponding to that group, and sample sleep state data for a sample user corresponding to the set. Here, the sample sleep state data in the training data of each group is the sample corresponding to that group. sample property data of a user and sample rhythm data of the sample user. It is possible.

[0079] In this case, the information processing device performs the following on each set of teacher data included in the first plurality of sets of teacher data: Specifically, the information processing device performs preprocessing for each set of training data. 9 and 10 using the sample property data and sample rhythm data included in the data. By performing the above-described process with reference to the sample sleep category data, the set is This allows the information processing device to perform preprocessing such as generating the data. First, multiple sets of training data after preprocessing (each set of training data is a sample corresponding to that set) Sample behavioral data of a user and sample sleep category data of the sample user. , including ) can be input to the learning model to generate (update) an estimation model. As will be described later, such an estimation model uses the actions performed by the target user as explanatory variables. and receiving a sleep pattern associated with the target user in response to inputting target behavior data identifying a history of the target user's sleep patterns. At least the target sleep category data for identifying the category can be output. It becomes like this.

[0080] (2C) Second multiple sets of training data included in multiple sets of training data In one example, the information processing device The estimation model can be generated (updated) using the second set of training data. The training data for each group is the sample behavior data of the sample user corresponding to that group, and sample sleep state data for a sample user corresponding to the set. Here, the sample sleep state data in the training data of each group is the sample corresponding to that group. Sample sleep category data for the user may be included.

[0081] In this case, the information processing device may store a second plurality of sets of training data (each set of training data is stored in the Sample behavioral data of a sample user corresponding to the sample and sample sleep of the sample user The categorical data (including the above) is input to the learning model to generate (update) an estimation model. As will be described later, such an estimation model uses the target user's In response to inputting subject behavior data identifying a history of behaviors performed by the subject, At least outputting target sleep category data that identifies the sleep category to which the user belongs You will be able to do this.

[0082] (2D) Third multiple sets of training data included in multiple sets of training data In one example, the information processing device The estimation model can be generated (updated) using the third set of training data. The training data for each group is the sample behavior data of the sample user corresponding to that group, and sample sleep state data for a sample user corresponding to the set. Here, the sample sleep state data in the training data of each group is the sample corresponding to that group. sample property data of a user and sample rhythm data of the sample user. It is possible.

[0083] In this case, the information processing device may store a third plurality of sets of training data (each set of training data is stored in the set) Sample behavior data of the sample user corresponding to the sample and sample properties of the sample user The data (including the sample user's sample rhythm data) is input to the learning model. Such an estimation model can be generated (updated) by inputting the As explained above, we use the target behavior data, which identifies the history of actions performed by the target user, as an explanatory variable. target characteristic data identifying a sleep characteristic of the target user in response to inputting the sleep characteristic data; and outputting at least target rhythm data for identifying the sleep rhythm of the target user. You will be able to do this.

[0084] (2E) Fourth set of training data included in the set of training data In one example, the information processing device may include a fourth set of training data included in the sets of training data. The estimation model can be generated (updated) using the above. The training data for each group is the sample behavior data of the sample user corresponding to that group, and sample sleep state data for a sample user corresponding to the set. Here, the sample sleep state data in the training data of each group is the sample corresponding to that group. It may include sample property data of the user.

[0085] In this case, the information processing device may store a fourth plurality of sets of training data (each set of training data is stored in the Sample behavior data of the sample user corresponding to the sample and sample properties of the sample user The data (including the data) can be input to the learning model to generate (update) an estimation model. As will be described later, such an estimation model uses the results of the target user's execution as explanatory variables. and determining a sleep state of the target user in response to inputting target behavior data identifying a history of the target user's sleep state. At least the target property data for identifying the property of sleep can be output.

[0086] (2F) The fifth set of training data included in the set of training data In one example, the information processing device may include: The estimation model can be generated (updated) using the above. The training data for each group is the sample behavior data of the sample user corresponding to that group, and sample sleep state data for a sample user corresponding to the set. Here, the sample sleep state data in the training data of each group is the sample corresponding to that group. It may contain sample rhythm data of the user.

[0087] In this case, the information processing device may generate a fifth plurality of sets of training data (each set of training data is The sample behavior data of the sample user corresponding to the sample user and the sample list of the sample user The training model can be used to generate (update) an estimation model. As will be described later, such an estimation model uses the results of the experiment performed by the target user as explanatory variables. In response to inputting target behavior data identifying a history of behaviors performed by the target user, It is now possible to at least output target rhythm data for identifying sleep rhythms. do.

[0088] (2G) The sixth set of training data included in the set of training data In one example, the information processing device may include: The estimation model can be generated (updated) using the sixth set of training data. The training data for each group is the sample behavior data of the sample user corresponding to that group, and sample sleep state data for a sample user corresponding to the set. Here, the sample sleep state data in the training data of each group is the sample corresponding to that group. It may contain user sample time data.

[0089] In this case, the information processing device may generate a sixth plurality of sets of training data (each set of training data is The sample behavior data of the sample user corresponding to the sample time of the sample user The data (including the data) can be input to the learning model to generate (update) an estimation model. As will be described later, such an estimation model uses the results of the target user's execution as explanatory variables. and determining a sleep state of the target user in response to inputting target behavior data identifying a history of the target user's sleep state. At least the target time data for identifying the sleeping time can be output.

[0090] The generation (update) of such estimation models is managed by the company that operates the estimation service. The server device 10A (and / or the terminal device 20) is replaced by the server device 10A (and / or the terminal device 20), or the server device 10A is provided with an estimation service. The server device 10A (and / or the terminal device 20) is managed by a company that operates the The server device 10 (and / or the terminal device 20) is managed by the same company. The estimation model generated (updated) by the other server device 10 (terminal device 20) is , may be transmitted to any server device 10, including server device 10A, or It can be accessed and used by any server device 10, including server device 10A.

[0091] The names "1st" to "6th" immediately before "multiple sets of training data" are , "multiple sets of training data" with a certain name among these names, and This is used for convenience to distinguish it from the "multiple sets of training data" which has a different name. However, this does not limit the order or content of the information.

[0092] (3) Step 1004 Referring again to FIG. 3, next, in ST1004, an information processing device, for example, A server device 10A (and / or a terminal device) managed by a company that operates an estimation service, etc. 20) is, for example, a specific application (an application for receiving estimated services, etc.). A terminal device 20 (or a terminal device 21) that has accessed this server device 10A by executing a from another server device 10A, and / or by running a browser to this server device 10A. The terminal device 20 that has accessed the system receives the information about the behavior performed by the target user to be estimated. The target behavior data that identifies the history can be obtained. Data identifying at least one action(s) performed by a user could be.

[0093] The acquisition of such target behavioral data is managed by the company that operates the estimation service, etc. Instead of the server device 10A (and / or the terminal device 20), or to operate an estimation service, etc. The server device 10A (and / or the terminal device 20) managed by the company is also managed by the other company. may be executed by other server devices 10 (and / or terminal devices 20) managed by the same company. .

[0094] (4) Step 1006 Next, in ST1006, an information processing device, for example, a device that operates an estimation service, etc. The server device 10A (and / or the terminal device 20) managed by a company is The target behavior data acquired in step ST1002 and the estimation model generated in step ST1002 can be used to obtain target sleep state data that identifies a sleep state of the target user. Specifically, the server device 10A converts the target behavior data as explanatory variables into an estimation model. By inputting the data into the estimation model, the target sleep state data is output as the objective variable. It can be done.

[0095] The target sleep state data may include at least one of the following data: . Data identifying the sleep characteristics of the target user (“Target Characteristic Data”) Data identifying the target user's sleep rhythm ("Target Rhythm Data"); Data identifying the target user's sleeping hours ("target time data") Data to identify the sleep category to which the target user (or their sleep) belongs ("target sleep category" (Reader Data)

[0096] The estimation model determines which data to output as the target sleep state data. Which data was used as sample sleep state data when generating (updating) the model? This may depend on the

[0097] For example, when generating (updating) an estimation model, the first complex data is used as sample sleep state data. When several sets of training data are used (see above "4(2)(2B)"), the estimated model At least target sleep category data may be output as target sleep state data. can.

[0098] For example, when generating (updating) an estimation model, the second complex data is used as sample sleep state data. When several sets of training data are used (see above "4(2)(2C)"), the estimated model At least target sleep category data may be output as target sleep state data. can.

[0099] For example, when generating (updating) an estimation model, the third complex data is used as sample sleep state data. When several sets of training data are used (see above "4(2)(2D)"), the estimated model is , and output at least target property data and target rhythm data as target sleep state data. It is possible.

[0100] For example, when generating (updating) an estimation model, the fourth complex data is used as sample sleep state data. When several sets of training data are used (see above "4(2)(2E)"), the estimated model At least the target property data can be output as the target sleep state data.

[0101] For example, when generating (updating) an estimation model, the fifth complex data is used as sample sleep state data. When several sets of training data are used (see above "4(2)(2F)"), the estimated model At least the target rhythm data can be output as the target sleep state data.

[0102] For example, when generating (updating) an estimation model, the sixth complex data is used as sample sleep state data. When several sets of training data are used (see above "4(2)(2G)"), the estimated model At least the target time data can be output as the target sleep state data.

[0103] For example, a first plurality of sets of training data are used as sample sleep state data to generate an estimation model. (4(2)(2B)) and then the sample sleep state The estimation model is generated (updated) using the sixth set of training data as the state data. In this case (see above "4(2)(2G)"), the estimation model uses the target sleep state data , target sleep category data and target time data can be output at least. Furthermore, after this, estimation is performed using a fifth set of training data as sample sleep state data. If a fixed model is generated (updated) (see above "4(2)(2F)"), the estimated model The target sleep state data includes target sleep category data, target time data, and target At least the rhythm data can be output.

[0104] The estimation model can be used in one of the following ways: It is possible. The server device 10A acquires and stores the estimation model generated in ST1004. The target behavior data is input to this estimation model, and the target behavior data output from this estimation model is Obtain sleep state data. The server device 10A acquires and stores the estimation model generated in ST1004. The server device 10 communicates with an external device (any other server device 10 or any terminal device 20) By transmitting target behavior data via the communication network 2, the target behavior data is used as explanatory variables. The server device 1 then outputs the target sleep state data to the estimation model. The target sleep state data output by this estimation model is transmitted from the external device. Received via network 2.

[0105] The acquisition of such target sleep state data is managed by the company that operates the estimation service. The server device 10A (and / or the terminal device 20) is replaced by the server device 10A (and / or the terminal device 20), or the server device 10A is provided with an estimation service. The server device 10A (and / or the terminal device 20) is managed by a company that operates the The server device 10 (and / or the terminal device 20) is managed by the same company. obtain.

[0106] (5) Step 1008 Next, in ST1008, the information processing device, for example, a device that operates an estimation service, etc. The server device 10A (and / or the terminal device 20) managed by the company Using the target sleep state data acquired in the At least one eligible product / service can be determined. The product / target service may be used to determine the current (estimated) sleep state of the target user identified by the target sleep state data. The sleep state may be appropriate for the sleep state.

[0107] Specifically, in one example, the server device 10A first receives target sleep state data and a plurality of products. And / or preparing and storing a search table that stores a plurality of services in association with each other. Next, the server device 10A receives the target sleep state data acquired in ST1006. By entering data into the lookup table, at least one target is retrieved from this lookup table. Collect data identifying the product and / or at least one Offering ("Proposal Data"); You can gain.

[0108] FIG. 11 is a conceptual diagram showing an example of a search table used in the communication system 1 shown in FIG. The lookup table illustrated in FIG. 11 is a diagram illustrating a sleep state that may be included in the target sleep state data. Each data (target property data, target rhythm data, target time data and target sleep category) At least one target product and / or at least one target product to be proposed is associated with each of the data. Each of them remembers one target service.

[0109] In this example, for the sake of simplicity, we will use the target property data, target rhythm data, and target time data. The data and target sleep category data correspond to three mutually different levels. Additionally, at least one eligible product and / or at least one eligible service to be proposed Taking the target property data as an example, the target property data indicates Class V. 1 (class with good sleep characteristics for the target user), class V2 (class with good sleep characteristics for the target user) Class V1 (class in which the target user's sleep quality is poor) and Class V2 (class in which the target user's sleep quality is poor) and at least one target product and / or at least one target service to be proposed. A service may be assigned.

[0110] Classes W1 to W3 are also in this order, and are used to determine whether the target user's sleep rhythm is "good" or "normal." Classes X1 to X3 can also be used to indicate the target user's The sleep duration may be indicated as "long," "normal," or "short."

[0111] Classes Y1 to Y3 are also, in this order, sleep categories to which the target user (sleep) belongs. can indicate "good", "standard", or "bad". In one example, class Y1 is 9 corresponds to "Sleep Category 21" in the same figure, and Class Y2 corresponds to "Sleep Category 22" in the same figure. Class Y3 corresponds to "Sleep Category 20" or "Sleep Category 11" in the same figure. It can also be thought of as corresponding to "Category 10."

[0112] All of the products P1 to P11 illustrated in FIG. 11 are designed to improve the user's sleep quality. The services S1 to S8 illustrated in FIG. 11 are all suitable for These products include any service suitable for improving the sleep quality of the user. This may include, but is not limited to, household goods, medicines, sports equipment, electrical appliances, etc. In addition, these services are limited to hospitals, clinics, sports gyms, etc. It can be included without

[0113] For example, the server device 10A detects whether the target property data in the target sleep state data is Class V. In the case of 3, product P4 and service S1 are the target product and target service, respectively. In addition, the server device 10A can determine the target sleep state data by, for example, If the data contains target time data (indicating class X2) and target rhythm data (indicating class W3), In this case, product P8 is determined as the target product and services S1, S3, and S4 are determined as the target services. It can be determined.

[0114] For each data included in the target sleep state data, the total number of classes used, The total number of products and / or services assigned to a class may be determined arbitrarily.

[0115] In another example, the server device 10A may generate a Using another estimation model developed, at least one target product and / or at least one The other estimation model can determine the target service based on the target sleep state data. The data indicating each class of each data included in the data are used as explanatory variables, and at least one target Data indicating the product and / or at least one target service is used as a target variable, and It can be generated (updated) by performing learning.

[0116] The machine learning for training the separate estimation model is, for example, As mentioned, any well-known machine learning method, such as gradient boosting trees (e.g., Lightning tGBM), neural networks or linear regression, etc. obtain.

[0117] The server device 10A inputs the target sleep state data into the different estimation model. , from the other estimation model, at least one target product and / or at least one target Data identifying the service ("suggestion data") may be output.

[0118] The method for using the other estimation model is one of the following examples: It is possible to do this. The server device 10A acquires and stores the generated different estimation model, and The target sleep state data is input to another estimation model, and the proposal output from the estimation model is Get the data. The server device 10A is an external device that acquires and stores the generated separate estimation model. (any other server device 10 or any terminal device 20) via the communication network 2 By transmitting the target sleep state data, the target sleep state data is used as an explanatory variable. The server device 10A then outputs the proposed data to the input different estimation model. The proposed data output by the different estimation model is transmitted from the external device via the communication network 2. and receive it.

[0119] The acquisition of such proposal data is conducted through a service managed by a company that operates an estimation service, etc. Instead of the server device 10A (and / or the terminal device 20), or a company that operates an estimation service, etc. The server device 10A (and / or the terminal device 20) managed by the company is also managed by another company. The above-mentioned process may be executed by another server device 10 (and / or terminal device 20) managed by the other server device 10.

[0120] (6) Step 1010 Referring again to FIG. 3, next, an information processing device, for example, a device that operates an estimation service, etc. The server device 10A (and / or the terminal device 20) managed by a company is The proposal data acquired in the At least one terminal device 20 (or another server device 10) is connected to the communication network 2. The "individual user" here refers to the "target user" It may include.

[0121] Proposal data may be provided via web pages, email, chat, electronic files, or recording media. (USB memory, CD-ROM, etc.), and / or by any means, including paper media, At least one terminal device 20 (other servers) managed by a business user and / or a personal user The server device 10 may be provided with the above-mentioned information.

[0122] The provision of such proposed data is a service managed by the company that operates the estimation service, etc. Instead of the server device 10A (and / or the terminal device 20), or a company that operates an estimation service, etc. The server device 10A (and / or the terminal device 20) managed by the company is also managed by another company. The above-mentioned process may be executed by another server device 10 (and / or terminal device 20) managed by the other server device 10.

[0123] 5. Variations In the various examples given above, sample property data, sample rhythm data, sample time data, etc. The estimation model is generated and used using the data and sample sleep category data as the objective variables. In another example, sample property data, sample rhythm data, At least one of sample time data and sample sleep category data (maximum 3) It is also possible to generate and use an estimation model by using the above-mentioned variables as explanatory variables instead of the objective variable. It is possible.

[0124] Also, in the various examples described above, the various steps (e.g., ST1000) illustrated in FIG. ~ST1010) is managed mainly by companies that operate estimation services, etc. The case where the server device 10A (and / or the terminal device 20) is used has been described. However, The various steps illustrated in FIG. 3 may be performed by any of a plurality of server devices 10 and / or a plurality of terminals. It is also possible to have the server devices 20 share the execution of the process. In one example, the terminal device 20 is provided with a service managed by a company that operates an estimation service or the like. In another example, the server device 10A (and / or the terminal device 20) may be included. A server device 10A (and / or terminal device 2) managed by a company that operates a fixed service, etc. 0) is not included.

[0125] In the above "4(4) step 1006", , target property data, target rhythm data, target time data, and target sleep category data In another embodiment, the target sleep state data is the data set defined in "4(4)" above. Instead of the data shown in step 1006, at least one of the following data is used: may include: Data that "uniquely" identifies the sleep characteristics of the target user ("target characteristic data"); Data that "uniquely" identifies the target user's sleep rhythm ("target rhythm data") Data that "uniquely" identifies the target user's sleep time ("target time data") Data that "uniquely" identifies the sleep category to which the target user (their sleep) belongs ("target" Sleep Category Data

[0126] Here, "uniquely identifying" the "target property data" means that the "target property data" is uniquely identified by This means that the sleep characteristics of the target user can be identified without relying on other data. That is, multiple data items work together to identify the sleep characteristics of the target user. In such cases, these multiple data sets "uniquely identify" the sleep characteristics of the target user. For example, if the first data indicates that the target user's sleep quality is "poor," and a flag "0", and the second data includes data indicating that the quality of sleep of the target user is The third data includes data indicating "standard" and a flag "0", Consider a case where the data includes data indicating that the quality of sleep is "high quality" and a flag "1." In this case, the first data and the second data alone cannot determine the sleep characteristics of the target user. By checking the third data that has a flag "1", For the first time, it is possible to determine whether a target user's sleep quality is "high quality." Therefore, all of these data are data that "uniquely" identify the sleep quality of the target user. isn't it.

[0127] Each of the "target rhythm data," "target time data," and "target sleep category data" The same can be said for the term "uniquely identify" in the above definition of Cut.

[0128] As described above, according to the technology disclosed in the present application, By inputting target behavior data that identifies the history of at least one behavior into the estimation model, From this estimation model, the current (predicted) sleep state of the target user can be identified. It is possible to acquire data on the sleeping state of the elephant. and / or a small number of target services that are suitable for the target sleep state data thus obtained. At least one target product and / or at least one target service is provided to the target user. Suggestions can be made to any user, including the target user, to improve the situation.

[0129] This allows the target user to generate a list of users based on the history of at least one action performed by the target user. used to estimate the sleep state of a target user, and at least partially improve performance. It is possible to provide a computer program, an information processing device, and a method.

[0130] 6. By communication system 2 to provide menopausal status data regarding the status of menopausal symptoms. Actions performed by Next, the above-mentioned communication system is used to provide menopausal status data regarding the status of menopausal symptoms. A specific example of the operations performed by the system 2 will be further described with reference to FIG. 12 is a flow diagram showing another example of the operation performed by the communication system 1 shown in FIG. do.

[0131] (1) Step 1100 First, in ST1100, an information processing device, for example, a device that operates an estimation service, etc. A server device 10A managed by a company that owns the server device 10A stores multiple data used to generate an estimation model. Several sets of training data can be acquired and stored.

[0132] The training data for each group of multiple training data sets is a sample of one person corresponding to this group. data identifying a history of actions performed by a user ("Sample Behavior Data"); Data identifying the menopausal symptom status of one sample user corresponding to the set of For example, the first set of training data may include the first set of training data. A sample user (V) is identified as a sample of the history of actions performed by the sample user. pull behavior data, sample menopausal status data for identifying the state of menopausal symptoms of Ms. V, and The second set of training data may include one sample user (W Sample behavioral data identifying the history of actions performed by Ms. W and Ms. W's menopause and sample menopausal status data identifying a menopausal condition.

[0133] (1A) Sample behavioral data Each sample behavior data is a data set of at least one of a plurality of predetermined behaviors. The data may be data identifying a predetermined number of (one or more) actions. Each has at least a partial positive effect on the menopausal condition of the person who performs the action. The predetermined actions may be actions that may have a negative impact. Examples of such predetermined actions include: Examples may include, but are not limited to, the following: (Examples of actions that may at least partially positively influence menopausal symptoms in humans) - Participated in aerobic exercise for at least one hour at least two days per week. I woke up at the same time every day. I set aside time each week to relax. Purchased 10 or more cans of beer (this behavior is, for example, a category to which beer belongs). (These items can be identified by data including the product and its purchase quantity.) (Examples of behaviors that may at least partially negatively impact a person's menopausal state) Participated in aerobic exercise for at least one hour on less than one day per week. · I woke up at a different time every day. I worked without a single day off in a week. - Purchased 5 or more bottles of green juice (this behavior is, for example, the product category to which green juice belongs and its (These items may be identified by data including the quantity purchased).

[0134] Each sample behavior data is a set of at least one of the predetermined multiple behaviors exemplified in this way. The data may be data identifying each of at least one behavior.

[0135] In one example, the products purchased by the user and / or the services used by the user may be The above may tend to have a positive or negative effect on the user's menopausal symptoms. The predetermined actions are based on the products purchased by the sample users (or the products to which the products belong). Category), services used by sample users (or the category to which the services belong) This type of behavior can also include, for example, retail stores (drug stores, supermarkets, etc.). POS (Point of Sale) owned by retailers, convenience stores, etc. es) data.

[0136] In one example, each of the plurality of predetermined actions may be provided with unique identification data (alphabetical, numeric, or Each sample behavioral data is assigned such a classification. For example, if you use POS data, you can use multiple samples. Each piece of user behavior data is a JICFS classification (classification code), product code, and Limit the product category, purchase amount, purchase quantity, number of purchases, and / or purchase date, etc. It can be included without

[0137] An information processing device, such as a server managed by a company that operates an estimation service The device 10A may collect such sample behavioral data, for example, by one of the following methods: It can be obtained by at least one method. The information processing device communicates with at least one other server device 10 (e.g., Another server device 10 that stores POS data installed in a drug store or the like, or The POS data is received from another server device (10) that can access the POS data. The information processing device sends a predetermined questionnaire (executed by a sample user) via the communication network 2. A small number of individuals may receive an email or web page containing their responses to a questionnaire about their behavior. It is received from at least one terminal device 20 and / or at least one server device 10. The information processing device collects responses from multiple users to the predetermined questionnaire. The content of the reply is received from at least one server device 10 via the communication network 2. The information processing device collects responses from multiple users to the predetermined questionnaire. The response content is received via a recording medium (USB memory, DVD-ROM, etc.).

[0138] (1B) Sample Menopausal Status Data The sample menopausal state data included in each set of training data is as follows: The present invention may include, but is not limited to, at least one of the following: Data that identifies the overall state of one sample user corresponding to the set ("sample" ("Comprehensive Status Data") Data identifying the first symptom state of one sample user corresponding to the set ("sample"). Pull first state data) Data identifying the second symptom state of one sample user corresponding to the set ("sample"). Pull second state data)

[0139] First, the sample overall condition data indicates the menopausal condition of the sample user. Data that identifies whether something is good (e.g., how good) or bad (e.g., how bad) In one example, the sample overall status data may include the sample user's menopausal symptoms. In this case, the data may be a score that identifies the overall evaluation of the state of the sample. The state of the user's menopausal symptoms is calculated using, for example, a score calculated by one of the following methods: It can be data that identifies the

[0140] Sample users of the questionnaire listed in the Simplified Menopause Index (SMI) (see Figure 13) The score is calculated based on the user's responses using the method described in the SMI, or A modified score obtained by performing any calculation (addition, subtraction, multiplication, division, etc.) on the core -Menopause Rating Scale Based on the sample users' responses to the questionnaire, the results of the menopausal symptom assessment scale were The score calculated in this way, or the score can be calculated in any way (addition, subtraction, multiplication, division, etc.) and the corrected score obtained by Questions listed in the Kupperman index Based on the sample user's responses to the Kupperman Menopause Index, The score calculated by the method, or by performing any calculation (addition, subtraction, multiplication, division, etc.) on this score. Corrected score obtained from ·Quality as listed in the Menopausal Symptom Assessment (Green Climacteric Scale) Based on the sample users' responses to the questions, the method described in the menopausal symptom evaluation The score calculated by the above method or by performing any calculation (addition, subtraction, multiplication, division, etc.) on this score. The corrected score obtained ·PSST(Premenstrual Symptoms Screening To Based on the responses of sample users to the questions listed in the The score calculated using the method described above, or the score calculated using any method (addition, subtraction, multiplication, division, etc.) ) and the corrected score obtained by - Listed in the WHQ (Women's Health Questionnaire) Calculated according to the method described in WHQ based on the responses of sample users to the questionnaire. The score obtained by performing any operation (addition, subtraction, multiplication, division, etc.) on this score. Corrected score obtained -Responses to questions written on the VAS (Visual Analogue Scale) The score was calculated based on the responses of the sample users using the method described in the VAS. Or, a modified score obtained by performing any calculation (addition, subtraction, multiplication, division, etc.) on this score. ·HFRDI(Hot Flash Related Daily Interferen Based on the responses of sample users to the questions listed on the ce Scale , the score calculated by the method described in the HFRDI, or the score calculated by any method. Corrected score obtained by arithmetic (addition, subtraction, multiplication, division, etc.) Questions listed in HFCS (Hot Flash Composite Score) Calculated using the method described in HFCS based on the responses of sample users to the questions. The score obtained by performing any calculation (addition, subtraction, multiplication, division, etc.) on this score. Corrected score ·MENQOL(Menopause-Specific Quality Of Li Based on the responses of sample users to the questions listed in MENQO fe), The score calculated using the method described in L, or this score calculated using any method (addition, subtraction, multiplication, etc.) Corrected score obtained by dividing Sample users' responses to questions listed on the Japanese Women's Menopausal Symptom Assessment Table Based on the content, the score was calculated using the method described in the Japanese Women's Menopausal Symptom Assessment Table. Or, a modified score obtained by performing any calculation (addition, subtraction, multiplication, division, etc.) on this score. - Sample users' responses to questions on any other questionnaire or scale A score calculated based on the content and in the manner described in the questionnaire or scale, or Corrected score obtained by calculating the score using any method

[0141] Second, the sample first condition data may include a symptom of the sample user among a plurality of menopausal symptoms. The state of one symptom is judged to be good (e.g., how good) or bad (e.g., how bad). In one example, the sample first state data may be data that identifies the sample. The data may be data identifying a score that assesses the status of the user for one of the symptoms.

[0142] The above multiple menopausal symptoms, for example, in SMI, are "hot face" and "easy sweating." This could be multiple menopausal symptoms listed in the SMI, such as "feeling cold in the lower back, hands and feet." In addition, the above multiple menopausal symptoms can be measured using any of the indices listed above in addition to SMI. The above-mentioned symptom may be a plurality of menopausal symptoms described in the above. The symptom may be any one selected from a number of menopausal symptoms.

[0143] Taking SMI as an example, the sample first state data is, for example, in SMI (see FIG. 13) It can be extracted from the answers given by the sample users to the questions written in the questionnaire. In this case, the sample first condition data may be a symptom such as "shortness of breath, palpitations" For example, a score of 0, 4, 8, or 12 (see FIG. 13) In another example, the sample first state data may be data identifying the sample. It can be obtained from the user via email, a web page, or the like.

[0144] Third, the sample second condition data is used to distinguish the sample user from multiple menopausal symptoms. Is the condition of one of the symptoms good (e.g., how good) or bad (e.g., how bad)? In one example, the sample second state data may be data that identifies the sample. The data may identify a score that evaluates the status of the user for one of the other symptoms.

[0145] The above multiple menopausal symptoms, for example, in SMI, are "hot face" and "easy sweating." This could be multiple menopausal symptoms listed in the SMI, such as "feeling cold in the lower back, hands and feet." In addition, the above multiple menopausal symptoms can be measured using any of the indices listed above in addition to SMI. The other symptom may be a plurality of menopausal symptoms described in the above. It may be any one symptom selected from a plurality of menopausal symptoms, but it is not limited to the one symptom. Symptoms may be different from the condition.

[0146] Taking SMI as an example, the sample second state data is, for example, It can be extracted from the answers given by the sample users to the questions written in the questionnaire. In this case, the sample second state data may be, for example, the state of another symptom, "easily tired." For example, the data for identifying a score of 0, 2, 4, or 7 (see FIG. 13) In another example, the sample second state data may be data obtained from the sample user. It may be obtained via email, a web page, or the like.

[0147] An information processing device, such as a server managed by a company that operates an estimation service The device 10A (and / or the terminal device 20) generates such sample Combining the behavioral data with the sample menopausal status data, , a set of training data can be generated and stored. To achieve this, for example, The information processing device identifies a certain sample user from the acquired plurality of sample behavior data. Extract sample behavior data with identification data for distinguishing between the two, and A sample with identification data that identifies the same sample user from the initial state data. Extract menopausal status data and combine them to create a A set of training data can be generated and saved. The same process can be repeated for multiple sample users. By executing each of the above, the information processing device generates and stores multiple sets of training data. It can be preserved.

[0148] (2) Step 1102 Returning to FIG. 3, next, in ST1102, an information processing device, for example, an estimation service The server device 10A (and / or the terminal device 20) managed by a company that operates a service, etc. Using multiple sets of training data acquired by the ST1100, a learning model is created through machine learning. By further training, an estimation model can be generated. ,any well-known machine learning method, e.g., gradient boosted trees (e.g., LightGBM), These may include, but are not limited to, neural networks or linear regression.

[0149] (2A) First sets of training data included in the sets of training data In one example, the information processing device The estimation model can be generated (updated) using the first plurality of sets of training data. The training data for each group is the sample behavior data of the sample user corresponding to that group, and sample menopausal status data for a sample user corresponding to the set. Here, the sample menopausal state data in the training data for each group is the sample corresponding to that group. It may contain sample aggregate status data for pull users.

[0150] In this case, the information processing device may include: a first plurality of sets of training data (each set of training data is Sample behavior data of the sample user corresponding to the sample and the overall sample of the sample user The state data (including the state data) can be input to the learning model to generate (update) the estimation model. As will be described later, such an estimation model uses the target user's preferences as explanatory variables. a target user in response to inputting target behavior data identifying a history of performed behaviors; The overall status data that identifies the overall status of the target can be output at least. becomes.

[0151] In one embodiment, the sample menopausal state data in each set of training data is The overall condition of menopausal symptoms of one sample user corresponding to the score is identified. Consider the case where they are different.

[0152] In this case, the information processing device performs the following on each set of teacher data included in the first plurality of sets of teacher data: Specifically, the information processing device performs preprocessing for each set of training data. For the data, the score identified by the sample overall condition data corresponding to this set is Convert the data into one of several categories using at least one threshold. For example, the information processing device may perform preprocessing by setting one threshold (first threshold) When using the score (value), the score identified by the sample overall condition data corresponding to each set is Depending on whether the score is above or below the first threshold, the score is divided into two categories. can be converted into one of two categories ("High" and "Low"). Alternatively, for example, the information processing device may set two thresholds (a first threshold and a second threshold) When using a smaller second threshold, the sample overall state data corresponding to each group is used to identify the The score is determined based on whether the score is equal to or greater than the first threshold, or equal to or greater than the second threshold and less than the first threshold. There are three categories ("High", "Middle") depending on whether the The data can be converted into one of the following categories: "High" and "Low"

[0153] In this way, the information processing device performs the processing for each set of teacher data included in the first plurality of sets of teacher data. For the data, the overall condition of menopausal symptoms of one sample user corresponding to this set is calculated as " The sample overall state data that identifies the "corresponding score" is used to identify one sample corresponding to this pair. A sample summary of the user's overall condition of menopausal symptoms to identify the corresponding category It is possible to perform preprocessing, such as converting the data into state data.

[0154] As a result, the information processing device generates the first plurality of sets of preprocessed teacher data (each set of teacher data The data is the sample behavior data of the sample user corresponding to the pair and the The sample overall state data (including the above) is input to the learning model to generate (update) an estimation model. ) can be done.

[0155] As will be described later, such an estimation model uses the results of the target user's execution as explanatory variables. and receiving a target behavioral data entry identifying a history of the target user's menopausal behavior. At least one subject overall condition data identifying categories corresponding to the overall condition of the patient's symptoms is collected. It will also be possible to output

[0156] (2B) Second multiple sets of training data included in multiple sets of training data In one example, the information processing device The estimation model can be generated (updated) using the second set of training data. The training data for each group is the sample behavior data of the sample user corresponding to that group, and sample menopausal status data for a sample user corresponding to the set. Here, the sample menopausal state data in the training data for each group is the sample corresponding to that group. It may include sample first state data for the pull user.

[0157] In this case, the information processing device may store a second plurality of sets of training data (each set of training data is stored in the The sample behavior data of the sample user corresponding to the sample number 1 of the sample user The state data (including the state data) can be input to the learning model to generate (update) the estimation model. As will be described later, such an estimation model uses the target user's preferences as explanatory variables. a target user in response to inputting target behavior data identifying a history of performed behaviors; The first symptom of the condition can be identified by outputting at least the first condition data. It becomes like this.

[0158] In one embodiment, the sample first state data in each set of training data is Identify the score corresponding to the first symptom state of one sample user. Consider a certain case.

[0159] In this case, the information processing device performs the following on each set of teacher data included in the second plurality of sets of teacher data: Specifically, the information processing device performs preprocessing for each set of training data. For the data, the score identified by the sample first state data corresponding to this set is Convert the data into one of several categories using at least one threshold. For example, the information processing device may perform preprocessing by setting one threshold (first threshold) When using the value), the score identified by the sample first state data corresponding to each set Depending on whether the score is above or below the first threshold, the score is divided into two categories. can be converted into one of two categories ("High" and "Low"). Alternatively, for example, the information processing device may set two thresholds (a first threshold and a second threshold) When a small second threshold is used, the first state data corresponding to each set is used to identify the first state data. The score is determined based on whether the score is equal to or greater than the first threshold, or equal to or greater than the second threshold and less than the first threshold. There are three categories ("High", "Middle") depending on whether the The data can be converted into one of the following categories: "High" and "Low"

[0160] In this way, the information processing device For the data, the first symptom state of one sample user corresponding to this set is The sample first state data that identifies the "core" is Convert sample first condition data into a category that identifies the first symptom condition , the preprocessing can be performed.

[0161] As a result, the information processing device generates the second plurality of sets of pre-processed teacher data (each set of teacher data The data is the sample behavior data of the sample user corresponding to the pair and the The sample first state data is input to the learning model to generate (update) an estimation model. ) can be done.

[0162] As will be described later, such an estimation model uses the results of the target user's execution as explanatory variables. a first action of the target user in response to inputting target behavior data identifying a history of the target user's actions; Outputting at least target first condition data that identifies a category corresponding to the symptom state. You will be able to do this.

[0163] Furthermore, in addition to the pre-processing described above (for convenience, referred to as "pre-processing A"), the information processing device It is also possible to perform further preprocessing (for convenience, referred to as "preprocessing B"). Specifically, first, The information processing device extracts another plurality of sets of teacher data from the second plurality of sets of teacher data. The "preprocessing B" can be performed by adding the following two sets of training data. Each set of training data has sample first state data that identifies scores equal to or greater than a threshold.

[0164] For example, if the first symptom is "easy sweating" in SMI (see Figure 13), When this is done, the training data of each group included in the above separate multiple sets of training data is set to a threshold value of 5. The sample first condition data may identify a score of 10 or above (6 or 10). The training data of each group included in the above separate multiple sets of training data has a stronger first symptom (here The training data includes only sample users with the following characteristics: do.

[0165] Furthermore, the information processing device may use the extracted different sets of teacher data as "new After this, the information processing device can generate a "new" second set of training data. The above-described pre-processing A can be performed on the second plurality of sets of training data. In preprocessing A, at least one threshold to be used is set to the same value as in the above example where preprocessing B is not performed. Next, the information processing device performs the " A new second set of training data is input to the learning model to generate (update) an estimation model. It is possible.

[0166] Such a "new" second set of training data is a sample with a stronger first symptom. Sample behavioral data and sample first state data for only the first user. Therefore, the estimation results learned using such "new" second multiple sets of training data are The fixed model classifies target users who have engaged in behavior that may lead to a strong first symptom as a strong first symptom. Users with symptoms can be more easily identified and, furthermore, It is also possible to precisely estimate the level of intensity of the user's first symptom.

[0167] (2C) Third sets of training data included in the sets of training data In one example, the information processing device The estimation model can be generated (updated) using the third set of training data. The training data for each group is the sample behavior data of the sample user corresponding to that group, and sample menopausal status data for a sample user corresponding to the set. Here, the sample menopausal state data in the training data for each group is the sample corresponding to that group. It may include sample second state data for the pull user.

[0168] In this case, the information processing device may store a third plurality of sets of training data (each set of training data is stored in the set) The sample behavior data of the sample user corresponding to the sample user and the sample second data of the sample user The state data (including the state data) can be input to the learning model to generate (update) the estimation model. As will be described later, such an estimation model uses the target user's preferences as explanatory variables. a target user in response to inputting target behavior data identifying a history of performed behaviors; The second symptom state data is at least capable of being output. It becomes like this.

[0169] In one embodiment, the sample second state data in each set of training data is Identify the score corresponding to the second symptom state of one corresponding sample user. Consider a certain case.

[0170] In this case, the information processing device performs the following on each set of teacher data included in the third plurality of sets of teacher data: Specifically, the information processing device performs preprocessing for each set of training data. For the data, the score identified by the sample second state data corresponding to this set is Convert the data into one of several categories using at least one threshold. For example, the information processing device may perform preprocessing by setting one threshold (first threshold) When using the value), the score identified by the sample second state data corresponding to each set Depending on whether the score is above or below the first threshold, the score is divided into two categories. can be converted into one of two categories ("High" and "Low"). Alternatively, for example, the information processing device may set two thresholds (a first threshold and a second threshold) When a small second threshold is used, the second state data corresponding to each set is used to identify the second state data. The score is determined based on whether the score is equal to or greater than the first threshold, or equal to or greater than the second threshold and less than the first threshold. There are three categories ("High", "Middle") depending on whether the The data can be converted into one of the following categories: "High" and "Low"

[0171] In this way, the information processing device can For the data, the state of the second symptom of one sample user corresponding to this set is The sample second state data that identifies the "core" is Convert sample second condition data into data that identifies the category corresponding to the second symptom condition , the preprocessing can be performed.

[0172] As a result, the information processing device generates a third plurality of sets of pre-processed teacher data (each set of teacher data The data is the sample behavior data of the sample user corresponding to the pair and the The sample first state data is input to the learning model to generate (update) an estimation model. ) can be done.

[0173] As will be described later, such an estimation model uses the results of the target user's execution as explanatory variables. a second behavior of the target user in response to inputting target behavior data identifying a history of the target user's behavior; Outputting at least target second condition data that identifies a category corresponding to the symptom state. You will be able to do this.

[0174] Furthermore, in addition to the pre-processing described above (for convenience, referred to as "pre-processing A"), the information processing device It is also possible to perform further preprocessing (for convenience, referred to as "preprocessing B"). Specifically, first, The information processing device extracts another plurality of sets of teacher data from the third plurality of sets of teacher data. The "preprocessing B" can be performed by adding the following two sets of training data. Each set of training data has sample second state data that identifies scores equal to or greater than a threshold.

[0175] For example, if the second symptom is "easily tired" in SMI (see Figure 13), In this case, the training data of each group contained in the above separate multiple sets of training data is equal to or greater than the threshold value of 3. The sample second state data may be a sample second state data that identifies the score (7 or 4) of the above. The training data of each group included in the multiple sets of training data is the one with stronger second symptoms (here, "fatigue"). It contains training data on only sample users who have the characteristics "prone to being affected" (e.g., "easily affected").

[0176] Furthermore, the information processing device may use the extracted different sets of teacher data as "new After this, the information processing device can generate a "new" third set of training data. The above-described pre-processing A can be performed on the third plurality of sets of training data. In preprocessing A, at least one threshold to be used is set to the same value as in the above example where preprocessing B is not performed. Next, the information processing device performs the " New third sets of training data are input to the learning model to generate (update) the estimation model. It is possible.

[0177] Such a "new" third set of training data is used to identify samples with stronger second symptoms. Sample behavioral data and sample second state data for only users Therefore, the estimation results learned using such a "new" third set of training data are The fixed model classifies target users who have engaged in behavior that may lead to strong secondary symptoms as strong secondary symptoms. Users with symptoms can be more easily identified and, furthermore, It is also possible to precisely estimate the level of intensity of the user's secondary symptoms.

[0178] The generation (update) of the estimation model described above is managed by the company that operates the estimation service. Instead of the server device 10A (and / or the terminal device 20) that is provided, or together with the server device 10A (and / or the terminal device 20) managed by the operating company, Executed by another server device 10 (and / or terminal device 20) managed by another company The estimation model generated (updated) by the other server device 10 (terminal device 20) may be used. may be transmitted to any server device 10, including server device 10A, or , can be accessed and used by any server device 10, including server device 10A.

[0179] The names "first" to "third" immediately before "multiple sets of training data" are , "multiple sets of training data" with a certain name among these names, and This is used for convenience to distinguish it from the "multiple sets of training data" which has a different name. However, this does not limit the order or content of the information.

[0180] (3) Step 1104 Referring again to FIG. 12, next, in ST1104, the information processing device, for example, The server device 10A (and / or the terminal device) managed by the company that operates the estimation service, etc. The device 20) is, for example, a specific application (an application for receiving the provision of an estimated service, etc.). The terminal device 20 ( or another server device 10), and / or by running a browser on this server device 10A The behavior performed by the target user to be estimated from the terminal device 20 that accessed the The target behavior data can be obtained to identify the history of the target behavior. Data identifying at least one action(s) performed by the user It could be.

[0181] The acquisition of such target behavioral data is managed by the company that operates the estimation service, etc. Instead of the server device 10A (and / or the terminal device 20), or to operate an estimation service, etc. The server device 10A (and / or the terminal device 20) managed by the company is also managed by the other company. may be executed by other server devices 10 (and / or terminal devices 20) managed by the same company. .

[0182] (4) Step 1106 Next, in ST1106, the information processing device, for example, a device that operates an estimation service, etc. The server device 10A (and / or the terminal device 20) managed by a company is The target behavior data acquired in step ST1102 and the estimation model generated in step ST1102 and acquiring target menopausal state data for identifying the state of menopausal symptoms of the target user using the above method. Specifically, the server device 10A uses the target behavior data as an explanatory variable to estimate the By inputting the data into the estimation model, the target menopausal status data as the objective variable can be extracted from the estimation model. The data can be output.

[0183] The target menopausal status data may include at least one of the following data: do. Data identifying the overall state of menopausal symptoms of the target user ("target overall state data") ) Data identifying the subject user's first symptom state ("Subject First State Data") Data identifying the subject user's second symptom state ("Subject Second State Data")

[0184] The estimation model determines which data to output as the target menopausal status data. When generating (updating) the model, which data is used as sample menopausal status data? This may depend on how the

[0185] For example, when generating (updating) an estimation model, the first When multiple sets of training data are used (see above "6(2)(2A)"), the estimation model can output at least the subject's overall condition data as the subject's menopausal condition data. Cut.

[0186] For example, when generating (updating) an estimation model, the second When multiple sets of training data are used (see above "6(2)(2B)"), the estimation model can output at least the first target state data as the target menopausal state data. Cut.

[0187] For example, when generating (updating) an estimation model, the third When multiple sets of training data are used (see above "6(2)(2C)"), the estimation model can output at least the target second state data as the target menopausal state data. Cut.

[0188] For example, a first plurality of sets of training data are used as sample menopausal state data to generate an estimation model. The model is generated (updated) (see above "6(2)(2A)"), and then the sample The estimation model is generated (updated) using the second set of training data as the initial state data. (Section 6(2)(2B) above), the estimated model is As a result, at least the target overall status data and the target first status data can be output. After this, a third set of training data is used as sample menopausal state data. If an estimation model is generated (updated) using the above (section 6(2)(2C)), the estimation The fixed model uses the target menopausal state data as the target overall state data, the target first state data, and and the target second status data can be output.

[0189] The estimation model can be used in one of the following ways: It is possible. The server device 10A acquires and stores the estimation model generated in ST1104. The target behavior data is input to this estimation model, and the target behavior data output from this estimation model is Obtain menopausal status data. The server device 10A acquires and stores the estimation model generated in ST1104. The server device 10 communicates with an external device (any other server device 10 or any terminal device 20) By transmitting target behavior data via the communication network 2, the target behavior data is used as explanatory variables. The server device then outputs the target menopausal state data to the estimation model. 10A transmits the target menopausal state data output by this estimation model from the external device. and receives it via communication network 2.

[0190] The acquisition of such target menopausal status data is managed by the company that operates the estimation service, etc. Instead of the server device 10A (and / or the terminal device 20) that is provided, or together with the server device 10A (and / or the terminal device 20) managed by the operating company, Executed by another server device 10 (and / or terminal device 20) managed by another company It is possible.

[0191] (5) Step 1108 Next, in ST1108, the information processing device, for example, a device that operates an estimation service, etc. The server device 10A (and / or the terminal device 20) managed by a company is Using the target menopausal state data acquired in At least one target product / one target service can be determined. The target product / target service is the current ( It may be suitable for the condition of presumed (menopausal symptoms).

[0192] Specifically, in one example, the server device 10A first receives the target menopausal state data and a plurality of commodities. A search table is prepared and stored to store the product and / or multiple services in association with the product and / or multiple services. Next, the server device 10A receives the target menopausal condition information acquired in ST1106. By entering the state data into the lookup table, at least one Data identifying the Target Product and / or at least one Target Service ("Proposal Data"); can be obtained.

[0193] FIG. 14 is a schematic diagram illustrating another example of a search table used in the communication system 1 shown in FIG. The search table illustrated in FIG. 14 is a conceptual diagram of the menopausal state data included in the target menopausal state data. Each data that can be acquired (each of the target overall status data, target first status data, and target second status data) At least one target product and / or at least one pair to be proposed in association with each other Remember the elephant service.

[0194] In this example, for the sake of simplicity, the target overall status data, the target first status data, and the target The proposed items are associated with three different levels indicated by each of the second state data of the subject. At least one eligible product and / or at least one eligible service may be assigned. Taking the target comprehensive condition data as an example, the target comprehensive condition data indicates Class V1 (target Class V2 (the overall condition of the user's menopausal symptoms is good), Class V3 (the overall condition of the target user's menopausal symptoms is good) Class V2 (the overall condition of the menopausal symptoms of the target user is standard) and Class V3 (the overall condition of the menopausal symptoms of the target user is standard) At least one target product to be proposed for each of the following classes (classes with poor overall condition) and / or at least one target service may be assigned.

[0195] Classes W1 to W3 are also in this order, with the target user's condition of the first symptom being "good" and "standard". Classes X1 to X3 can also be used to indicate the target user's The condition of the second symptom of the patient may be indicated as "good," "normal," or "poor."

[0196] The products illustrated in FIG. 14 (references consisting of "P" and numbers such as P10 and P20) All of the products (marked with the symbol) are suitable for improving the menopausal symptoms of users. It indicates any product and the service (S10, S20, etc.) shown in FIG. and numerical references) are all designed to provide information on the user's menopausal symptoms. These products include food, daily necessities, medicines, etc. This may include, but is not limited to, sports equipment, electrical appliances, etc. These services include, but are not limited to, hospitals, clinics, gyms, etc. This can be done.

[0197] The server device 10A may, for example, determine whether the target overall condition data in the target menopausal condition data is a When class V3 is shown, product P40 and service S10 are the target product and target service, respectively. In addition, the server device 10A can determine, for example, the target menopausal condition The state data is the object second state data (indicating class X2) and the object first state data (indicating class W3). If the status data is included, the product P80 is the target product, and services S10, S30, and S 40 can be determined as the target service.

[0198] For each data included in the target menopausal status data, the total number of classes used, The total number of products and / or services allocated to each class may be determined arbitrarily.

[0199] In another example, the server device 10A may generate a Using another estimation model developed, at least one target product and / or at least one The other estimation model can also determine the target service for the target menopausal status data. The data indicating each class of each data included in the data are used as explanatory variables, and at least one pair The target product and / or at least one target service are used as the objective variable. It can be generated (updated) by performing ant learning.

[0200] The machine learning for training the other estimation model can be performed, for example, by using any Well-known machine learning techniques, such as gradient boosting trees (e.g., LightGBM), neural networks, These may include, but are not limited to, neural networks or linear regression.

[0201] The server device 10A inputs the target menopausal state data into the other estimation model. and from the separate estimation model, at least one subject product and / or at least one counterpart The service may output data identifying the service ("suggestion data").

[0202] The method for using the other estimation model is one of the following examples: It is possible to do this. The server device 10A acquires and stores the generated different estimation model, and The target menopausal state data is input to another estimation model, and the proposal output from the other estimation model is Obtain proposal data. The server device 10A is an external device that acquires and stores the generated separate estimation model. (any other server device 10 or any terminal device 20) via the communication network 2 By sending the target menopausal status data, the target menopausal status data is used as an explanatory variable. Then, the server device 10A outputs the proposed data to the other estimation model. The proposed data output by the other estimation model is transmitted from the external device through the communication network 2. Receive via.

[0203] The acquisition of such proposal data is conducted through a service managed by a company that operates an estimation service, etc. Instead of the server device 10A (and / or the terminal device 20), or a company that operates an estimation service, etc. The server device 10A (and / or the terminal device 20) managed by the company is also managed by another company. The above-mentioned process may be executed by another server device 10 (and / or terminal device 20) managed by the other server device 10.

[0204] (6) Step 1110 Referring again to FIG. 12, next, an information processing device, for example, an information processing device that operates an estimation service, etc. The server device 10A (and / or the terminal device 20) managed by the company The proposal data obtained in 8 is managed by corporate users and / or individual users. At least one terminal device 20 (which may be another server device 10) is connected to a communication network. 2. The "individual user" here refers to the "target user" may include:

[0205] Proposal data may be provided via web pages, email, chat, electronic files, or recording media. (USB memory, CD-ROM, etc.), and / or by any means, including paper media, At least one terminal device 20 (other servers) managed by a business user and / or a personal user The server device 10 may be provided with the above-mentioned information.

[0206] The provision of such proposed data is a service managed by the company that operates the estimation service, etc. Instead of the server device 10A (and / or the terminal device 20), or a company that operates an estimation service, etc. The server device 10A (and / or the terminal device 20) managed by the company is also managed by another company. The above-mentioned process may be executed by another server device 10 (and / or terminal device 20) managed by the other server device 10.

[0207] 7. Variations In the various examples described above, sample total status data, sample first status data, and sample This explains how to generate and use an estimation model using the second state data as the objective variable. In another example, sample total status data, sample first status data, and sample second status data are At least one (maximum two) of the state data is used as an explanatory variable rather than a target variable. It is also possible to generate and use estimation models.

[0208] In addition, in the various examples described above, various steps (e.g., ST110) illustrated in FIG. 0 to ST1110) are mainly managed by companies that operate estimation services, etc. In the above description, the server device 10A (and / or the terminal device 20) is configured as a 12 may be performed on any of a plurality of server devices 10 and / or a plurality of It is also possible to have the terminal device 20 share and execute the process. In one example, the device 10 and / or the plurality of terminal devices 20 are provided with a service by a company that operates an estimation service or the like. In another example, the server device 10A (and / or the terminal device 20) managed by the The server device 10A (and / or the terminal device) managed by the company that operates the estimation service, etc. position 20) is not included.

[0209] As described above, according to the technology disclosed in the present application, By inputting target behavior data that identifies the history of at least one behavior into the estimation model, From this estimated model, the current (predicted) menopausal symptom status of the target user can be identified. Furthermore, it is possible to acquire data on the menopausal state of a plurality of subjects prepared in advance. The targeted menopausal status data obtained in this manner for the product and / or multiple targeted services. At least one Eligible Product and / or at least one Eligible Service suitable for the Eligible User In order to improve menopausal symptoms of users, it can be proposed to any user, including the target user. can.

[0210] This allows the target user to generate a list of users based on the history of at least one action performed by the target user. Used to estimate the menopausal symptom status of the target user and at least partially improved It is possible to provide a computer program, an information processing device, and a method having the above-mentioned performance. .

[0211] 8. Estimation models for predicting various health conditions The technology disclosed in the present application is capable of predicting the above-mentioned sleep state and menopausal symptom state. It can also be applied to predict various health conditions, such as: be. Exercise status Immune status Hydration status Nutritional status

[0212] (1) Exercise status The estimation model for predicting exercise status is also used to predict conditions related to lifestyle-related diseases. To generate such an estimation model, sample behavioral data, which are explanatory variables, are used. The data may be at least one of the following examples of data: - Products purchased by sample users for a certain period (e.g., one year) (or the products themselves) data that identifies the category to which the - The service (or its contents) used by the sample user for a certain period (e.g., one year) Data identifying the category to which the service belongs Data to identify whether the sample user is obese or not -Weight of sample users over a certain period (for example, from their 30s to the present) Data that identifies the amount of change Data identifying the amount of exercise (e.g., number of steps) of the sample user

[0213] In addition, the data for identifying symptoms, which are the objective variables, are the following multiple data (measurements) may be at least one of: Data identifying the weight of the sample user Data identifying the sample user's BMI (Body Mass Index) Data identifying the number of steps taken by the sample user Data identifying whether the sample user smokes or not Data to identify the diagnostic reference values ​​for metabolic syndrome for sample users

[0214] The diagnostic criteria for metabolic syndrome are based on the visceral fat accumulation and / or or waist circumference and at least two of the following values: nothing. Sample users with hypertriglyceridemia and / or low HDL cholesterol Defined threshold for diagnosis Sample user's systolic and / or diastolic blood pressure - Thresholds established for the diagnosis of fasting hyperglycemia for sample users

[0215] The information processing device performs machine learning using such explanatory variables and objective variables. By using this method, it is possible to generate an estimation model that predicts the state of movement. The operations performed by the communication system 2 in relation to the generation and use of the above-mentioned menopausal The operations described in paragraphs "6" and "7" above in relation to the aspect of predicting the state of the In the above items "6" and "7" (particularly in Figs. 12 and 13), In the explanation given above (see 14), the estimated model for predicting menopausal symptom status The explanatory variables and target variables used are the explanatory variables described in Section 8(1) of this Generate an estimation model to predict the state of movement by replacing the number of subjects with the objective variable. The operations performed by the communication system 2 in relation to the communication method and the like will be understood by those skilled in the art. Let's do it.

[0216] As described above, according to the technology disclosed in the present application, By inputting target behavior data that identifies the history of at least one behavior into the estimation model, From this estimated model, we can identify the current (predicted) movement state of the target user. Furthermore, it is possible to acquire target product status data. The target service is selected from among a plurality of target services, and at least one of the target services is selected as a target service suitable for the target state data thus obtained. One target product and / or at least one target service is provided based on the target user's exercise status. For improvement, suggestions can be made to any user, including the target user.

[0217] This allows the target user to generate a list of users based on the history of at least one action performed by the target user. used to estimate the target user's movement state, and at least partially improve performance. It is possible to provide a computer program, an information processing device, and a method.

[0218] (2) Immune status The estimated model for predicting immune status is based on the severity of symptoms in people with weakened immune systems. To generate such an estimation model, the explanatory variables, The sample behavior data is at least one of the following types of data: It is possible. - Products purchased by sample users for a certain period (e.g., one year) (or the products themselves) data that identifies the category to which the - The service (or its contents) used by the sample user for a certain period (e.g., one year) Data identifying the category to which the service belongs

[0219] In addition, the data for identifying symptoms, which are the objective variables, are the following multiple data (measurements) may be at least one of: Data to identify the sample user's susceptibility to colds Data identifying the frequency of colds in sample users Data identifying the SIgA concentration of the sample user Data identifying the sample user's allergy symptoms Data identifying the oral environment and / or immune status of the sample user Data identifying the sleep state of the sample user Data identifying the sample user's exercise status Data to identify the stress state of the sample user

[0220] The data for identifying such symptoms was collected from sample users through questionnaires, etc. The symptoms may be subjective symptoms of the user, or may be acquired from the reception history of a sample user at a medical institution. The data for identifying such symptoms may be a plurality of (preferably three or more) The term "category" may identify any of the following categories:

[0221] The information processing device performs machine learning using such explanatory variables and objective variables. By using this method, it is possible to generate a prediction model for predicting the immune status. The operations performed by the communication system 2 in relation to the generation and use of the above-mentioned menopausal The operations described in paragraphs "6" and "7" above in relation to the aspect of predicting the state of the In the above items "6" and "7" (particularly in Figs. 12 and 13), In the explanation given above (see 14), the estimated model for predicting menopausal symptom status The explanatory variables and target variables used are the explanatory variables described in Section 8(2) of this Generate an inferential model to predict immune status by replacing the number and objective variables The operations performed by the communication system 2 in relation to the communication method and the like will be understood by those skilled in the art. Let's do it.

[0222] As described above, according to the technology disclosed in the present application, By inputting target behavior data that identifies the history of at least one behavior into the estimation model, From this estimated model, we can identify the current (predicted) immune status of the target user. Furthermore, it is possible to acquire target product status data. The target service is selected from among a plurality of target services, and at least one of the target services is selected as a target service suitable for the target state data thus obtained. One Target Product and / or at least one Target Service is provided based on the immune status of the Target User. For improvement, suggestions can be made to any user, including the target user.

[0223] This allows the target user to generate a list of users based on the history of at least one action performed by the target user. used to estimate the immune status of a target user, and at least partially improve performance. It is possible to provide a computer program, an information processing device, and a method.

[0224] (3) Hydration status The estimated model for predicting hydration status is based on the symptoms of people with weakened immune systems. To generate such an estimation model, explanatory variables are used to predict the degree of The sample behavioral data to be used is at least one of the multiple data examples shown below. It could be. - Products purchased by sample users for a certain period (e.g., one year) (or the products themselves) data that identifies the category to which the - The service (or its contents) used by the sample user for a certain period (e.g., one year) Data identifying the category to which the service belongs

[0225] In addition, the data for identifying symptoms, which are the objective variables, are the following multiple data (measurements) may be at least one of: -Data identifying the sample user's serum Na level Data identifying the sample user's urine osmolality Data identifying the sample user's urine specific gravity -Data identifying the urine color of the sample user Data identifying the sample user's BUN / creatinine ratio Data identifying sample user dehydration rating scale Data identifying subjective symptoms associated with dehydration in sample users

[0226] These data are obtained from the medical examination or medical checkup received by the sample user. The data is generated from clinical test values ​​or from sample user responses to health questionnaires. The data identifying the subjective symptoms of the sample user due to lack of water can be The degree of subjective symptoms can be generated from the responses of sample users to the questionnaire. Serum Na level, urine osmolality, urine specific gravity, urine color, BUN / creatinine ratio, water intake and / or This can be determined from the amount of physical activity, etc.

[0227] The information processing device performs machine learning using such explanatory variables and objective variables. This allows us to generate an estimation model that predicts the state of hydration. The operations performed by the communication system 2 in relation to the generation and use of the model are The method described in "6" and "7" above in relation to the aspect of predicting the state of a terminal symptom The operation may be substantially the same as that in the above items "6" and "7" (particularly in FIG. 12 and In the description given above (see Figure 14), a hypothesis model for predicting menopausal symptom status was developed. The explanatory variables and target variables used in the analysis are the same as those described in Section 8(3). By replacing the objective variables with the objective variables, we developed an estimation model for predicting hydration status. The operations performed by the communication system 2 in relation to the generation and use of the It will be understood.

[0228] As described above, according to the technology disclosed in the present application, By inputting target behavior data that identifies the history of at least one behavior into the estimation model, From this estimation model, the current (predicted) hydration status of the target user can be identified. Furthermore, it is possible to acquire target condition data for a plurality of target products and and / or one or more target services, the least of which is suitable for the target status data thus obtained. and one Target Product and / or at least one Target Service to promote hydration for the Target User. In order to improve the condition of the user, the user can make a suggestion to any user, including the target user.

[0229] This allows the target user to generate a list of users based on the history of at least one action performed by the target user. used to estimate the hydration status of a target user, and at least partially improve It is possible to provide a computer program, an information processing device, and a method having the above-mentioned functions.

[0230] (4) Nutritional status The estimated model for predicting nutritional status is based on the severity of symptoms in people who are not receiving nutritional support. To generate such an estimation model, the sample The user behavior data may be at least one of the following types of data: do. - Products purchased by sample users for a certain period (e.g., one year) (or the products themselves) data that identifies the category to which the - The service (or its contents) used by the sample user for a certain period (e.g., one year) Data identifying the category to which the service belongs

[0231] In addition, the data for identifying symptoms, which are the objective variables, are the following multiple data (measurements) may be at least one of: Data to identify whether or not the sample user has any subjective symptoms associated with nutritional deficiency Data identifying the dietary diversity score (DVS) of sample users Sample User Simplified Nutritional Appetite Scale (SNAQ) Data identifying the Personal Appetite Questionnaire Data identifying the sample user's BMI (Body Mass Index)

[0232] The data for identifying such symptoms was collected from sample users through questionnaires, etc. The symptoms may be subjective symptoms of the user, or may be acquired from the reception history of a sample user at a medical institution. The data for identifying such symptoms may be a plurality of (preferably three or more) The term "category" may identify any of the following categories:

[0233] The information processing device performs machine learning using such explanatory variables and objective variables. This makes it possible to generate a prediction model for predicting nutritional status. The operations performed by the communication system 2 in relation to the generation and use of the above-mentioned menopausal The operations described in paragraphs "6" and "7" above in relation to the aspect of predicting the state of the In the above items "6" and "7" (particularly Figs. 12 and 13), In the explanation given above (see 14), the estimated model for predicting menopausal symptom status The explanatory variables and target variables used are the explanatory variables described in Section 8(4) of this Generate an estimation model to predict nutritional status by replacing the number and objective variables The operations performed by the communication system 2 in relation to the communication method and the like will be understood by those skilled in the art. Let's do it.

[0234] As described above, according to the technology disclosed in the present application, By inputting target behavior data that identifies the history of at least one behavior into the estimation model, From this estimated model, we can identify the current (predicted) nutritional status of the target user. Furthermore, it is possible to acquire target product status data. The target service is selected from among a plurality of target services, and at least one of the target services is selected as a target service suitable for the target state data thus obtained. One Target Product and / or at least one Target Service is used to evaluate the nutritional status of a Target User. For improvement, suggestions can be made to any user, including the target user.

[0235] This allows the target user to generate a list of users based on the history of at least one action performed by the target user. used to estimate the nutritional status of a target user and achieve at least partially improved performance It is possible to provide a computer program, an information processing device, and a method.

[0236] 9. Various Aspects The computer program according to the first aspect is "executed by at least one processor." target behavior data that identifies a history of actions performed by a target user and then applying the target behavior data to the generated estimation model by performing supervised learning. By inputting the data, target sleep state data that identifies the sleep state of the target user is obtained. and target characteristic data identifying the sleep characteristics of the target user. target rhythm data for identifying the rhythm of the target user; target time data for identifying the sleeping time of the target user; and target category data identifying a sleep category to which the target user belongs. and outputting target sleep state data including at least one of the sleep states from the estimation model. "to cause said at least one processor to function." The computer program according to the second aspect is the same as the computer program according to the first aspect, except that "each set of teacher data" is The data identifies a history of actions performed by one sample user corresponding to the set. The sample behavior data and the sample data that identify the sleep state of one sample user corresponding to the set are By inputting multiple sets of training data, including the sleep state data, into the learning model, and obtaining the estimation model generated by the at least one processor. It can be made to function. The computer program according to the third aspect is the same as the computer program according to the first aspect, except that "each set of teacher data" is The data identifies a history of actions performed by one sample user corresponding to the set. The sample behavior data and the sample data that identify the sleep state of one sample user corresponding to the set are By inputting multiple sets of training data, including the sleep state data, into the learning model, and the at least one processor is connected to the generated estimation model via a communication line. "It allows one processor to function at the same time." The computer program according to the fourth aspect is the same as the first aspect in that "each set of teacher data" is The data identifies a history of actions performed by one sample user corresponding to the set. The sample behavior data and the sample data that identify the sleep state of one sample user corresponding to the set are and a plurality of sets of teacher data including the sleep state data are acquired, and the plurality of sets of teacher data are The estimation model is generated by inputting the data to a learning model and learning the data. "It can function at least one processor." The computer program according to the fifth aspect is the same as the computer program according to the fourth aspect, The plurality of sets of teacher data includes a first plurality of sets of teacher data, and the first plurality of sets of teacher data The teacher data of each group included in the above is used as the sample sleep state data. and sample characteristic data for identifying the sleep characteristics of one sample user corresponding to the set. and sample rhythm data identifying a corresponding sleep rhythm of one sample user. and for each set of teacher data included in the first plurality of sets of teacher data acquired, sample property data for identifying the sleep property of one sample user corresponding to the set; Based on the sample rhythm data, the sleep rhythm of the corresponding sample user is identified. and a sample that identifies a sleep category to which one sample user corresponding to the set belongs. The preprocessing of generating sleep category data is performed, and the teacher data for each group is assigned to that group. The sample behavior data of one corresponding sample user and one sample corresponding to the pair the sample sleep category data of the first user; A plurality of sets of training data are input to the learning model to generate the estimation model, and the estimation model and identifying a sleep category to which the target user belongs as the target sleep state data from the data. and outputting target sleep category data. "It can be made to work." A computer program according to a sixth aspect of the present invention is a computer program according to any one of the second to fourth aspects of the present invention. In either case, "the plurality of sets of teacher data includes a second plurality of sets of teacher data, and the second Each set of teacher data included in the plurality of sets of teacher data is used as the sample sleep state data. , a sample characteristic that identifies the sleep characteristic of at least one sample user corresponding to the set The quality data and the sample rhythm that identifies the sleep rhythm of one sample user corresponding to the set are The sleep category to which the sample user corresponding to the set belongs is determined based on the sleep data. and sample sleep category data identifying a sleep category from the estimation model. As the target sleep state data, a target sleep category that identifies the sleep category to which the target user belongs is and causing said at least one processor to output category data. It is possible. In the seventh aspect, the computer program according to the fifth aspect or the sixth aspect In the embodiment, "one sample corresponding to the set" is a sample that identifies the sleep characteristics of the user. The quality data were obtained using the Athens Insomnia Scale, Pittsburgh Sleep Quality Index, 3D Sleep Scale, or The score calculated based on the sample user's responses to the sleepiness severity questionnaire. and a sample list showing the sleep rhythm of one sample user corresponding to the set. The rhythm data is the median time of the sample user's weekday sleep time and the sleep time on weekends. The social jet lag is the absolute value of the difference from the central time zone. The sleep category to which the sample user belongs is the one on either the horizontal or vertical axis. At least one threshold value arranged in association with the score, and one of the horizontal axis and the vertical axis At least one of the axes is arranged in correspondence with the social jet lag. The threshold value may be any one of a number of categories divided by A computer program according to an eighth aspect of the present invention is a computer program according to any one of the second to fourth aspects of the present invention. In any one of the above, "the plurality of sets of teacher data includes a third plurality of sets of teacher data, and the third Each set of teacher data included in the plurality of sets of teacher data is used as the sample sleep state data. , a sample characteristic that identifies the sleep characteristic of at least one sample user corresponding to the set The quality data and a sample list that identifies the sleep rhythm of one sample user corresponding to the set. and calculating from the estimation model the target sleep state data, Target characteristic data for identifying the sleep characteristics of a user and a sleep rhythm of the target user and outputting target rhythm data corresponding to the target rhythm data. It is possible to "make it so." A computer program according to a ninth aspect is a computer program according to any one of the second to fourth aspects. In any one of the above, "the plurality of sets of teacher data includes a fourth plurality of sets of teacher data, and the fourth Each set of teacher data included in the plurality of sets of teacher data is used as the sample sleep state data. , a sample characteristic that identifies the sleep characteristic of at least one sample user corresponding to the set and from the estimation model, the target sleep state data includes quality data of the target user. and outputting target characteristic data that identifies the sleep characteristics of the at least one processor. This allows the processor to function. A computer program according to a tenth aspect of the present invention is a computer program according to any one of the second to fourth aspects of the present invention. In any one of the above, "the plurality of sets of teacher data includes a fifth plurality of sets of teacher data, and the fifth Each set of teacher data included in the plurality of sets of teacher data is used as the sample sleep state data. and generating a sample that identifies the sleep rhythm of at least one sample user corresponding to the set. and from the estimation model, the target sleep state data includes the target rhythm data. and outputting target rhythm data that identifies the user's sleep rhythm. "It can also function as a single processor." A computer program according to an eleventh aspect of the present invention is a computer program according to any one of the second to fourth aspects of the present invention. In any one of the above, "the plurality of sets of teacher data includes a sixth plurality of sets of teacher data, and the sixth Each set of teacher data included in the plurality of sets of teacher data is used as the sample sleep state data. and generating a sample time that identifies the sleep time of at least one sample user corresponding to the set. and from the estimation model, the target sleep state data is obtained from the target user. and outputting target time data identifying the sleep period of the at least one processor. "Make the processor work." A computer program according to a twelfth aspect is a computer program according to any one of the first to eleventh aspects. "Among the multiple products and / or services, the target sleep state data Determine at least one eligible product and / or at least one eligible service corresponding to the data. The at least one processor may be caused to function so as to "determine the A computer program according to a thirteenth aspect is the same as the "supervised" computer program according to the twelfth aspect. The target sleep state data is input to another estimation model generated by performing learning. By doing so, the at least one target product and / or the at least one target product can be estimated from the other estimation model. and outputting proposal data identifying at least one target service. "It can make one processor work." A computer program according to a fourteenth aspect of the present invention is a computer program according to the twelfth aspect of the present invention. a search table that associates status data with the plurality of products and / or the plurality of services; By inputting the target sleep state data into the lookup table, the at least one obtaining proposal data identifying the at least one eligible product and / or the at least one eligible service; "The at least one processor may be caused to function so as to: In a computer program according to a fifteenth aspect, the computer program according to the first aspect to the first aspect In any one of the four aspects, "the at least one processor is a central processing unit (CP U), microprocessor and / or graphics processing unit (GP U). The information processing device according to the sixteenth aspect includes at least one processor, At least one processor identifies a history of actions performed by a target user. The target row is then applied to the generated estimated model by performing supervised learning. A target sleep state data is generated by inputting the motion data to identify the sleep state of the target user. target characteristic data identifying a sleep characteristic of the target user; Target rhythm data for identifying the sleep rhythm, target time data for identifying the sleep time of the target user and target category data identifying a sleep category to which the target user belongs. outputting target sleep state data including at least one of the above from the estimation model; It may be configured so that An information processing device according to a seventeenth aspect is the same as the information processing device according to the sixteenth aspect, except that It can be a device. In the information processing device according to the eighteenth aspect, the information processing device according to the seventeenth aspect is A processor may be a central processing unit (CPU), a microprocessor, and / or It may include a graphics processing unit (GPU). The method according to the 19th aspect is a method for "executing computer-readable instructions by at least one a method executed by at least one processor, the method comprising: executes the instructions to identify a history of actions performed by the target user. The steps include acquiring target behavior data and performing supervised learning to generate an estimated model. and inputting the target behavior data into the device to identify the sleep state of the target user. target sleep state data, the target sleep characteristic data identifying the sleep characteristics of the target user; target rhythm data for identifying the sleep rhythm of the target user; and target time data for identifying the sleep category to which the target user belongs. and (c) extracting target sleep state data from the estimation model. and outputting the data from the data. In the method according to the twentieth aspect, the method according to the nineteenth aspect, "the at least one The processor may be a central processing unit (CPU), a microprocessor, and / or a graphics The processor may include a graphics processing unit (GPU). The method according to the 21st aspect is a method for "executing computer-readable instructions by at least one a method executed by at least one processor, the method comprising: However, by executing the above command, the teacher data of each group is sample behavior data identifying a history of behaviors performed by a user and a corresponding sample sleep state data identifying a sleep state of a sample user; an acquisition step of acquiring sets of training data; and a learning step of inputting the sets of training data into a learning model. target behavior data that identifies a history of behaviors performed by a target user by learning the target behavior data; target sleep state data for identifying a sleep state of the target user by inputting the corresponding target characteristic data for identifying the sleep characteristics of the target user, target rhythm data for identifying the target user's sleeping time, target time data for identifying the target user's sleeping time, and target category data identifying a sleep category to which the user belongs; generating an estimation model configured to output target sleep state data including: and "the steps of: The method according to the 22nd aspect is the same as the method according to the 21st aspect, wherein "the acquiring step includes: The method includes a step of acquiring a first plurality of sets of training data included in the training data, The teacher data of each group included in the teacher data of the group is a small number of the sample sleep state data. A sample characteristic data set that identifies the sleep characteristics of at least one sample user corresponding to the set. and a sample rhythm that identifies the sleep rhythm of one sample user corresponding to the set. data, and further, the generating step includes: For each group of training data included in Sample property data for identifying the quality and the corresponding sleep rhythm of one sample user Based on the sample rhythm data that identifies the set, one sample user A preliminary step is to generate sample sleep category data to identify the sleep category to which the sample belongs. and executing a process in which the training data of each set is the training data of one sample user corresponding to the set. The sample behavior data and the sample sleep category of one sample user corresponding to the set and the first plurality of sets of training data after the preprocessing, including the training data, are used as the learning model. and generating the estimation model by inputting the estimated data into a rule. The method according to the 23rd aspect is the same as the method according to the 21st aspect, wherein "the acquiring step includes: a step of acquiring a second plurality of sets of training data included in the training data, The teacher data of each group included in the teacher data of the group is a small number of the sample sleep state data. A sample characteristic data set that identifies the sleep characteristics of at least one sample user corresponding to the set. and sample rhythm data that identifies the sleep rhythm of one sample user corresponding to the set. The sleep category to which one sample user corresponding to the set belongs is determined based on the data. and sample sleep category data for identifying the sleep state of the subject. and inputting the above-mentioned data into the learning model to generate the estimation model. . In the method according to the 24th aspect, in the 23rd aspect, "one corresponding to the set" is The sample characteristics data that identify the sleep characteristics of a sample user are the Athens Insomnia Scale, Pi The above-mentioned samples for the Robertsburg Sleep Questionnaire, the Three-Dimensional Sleep Scale, or the Insomnia Severity Questionnaire A score calculated based on the answers of the pull users is displayed. The sample rhythm data identifying the sleep rhythm of the sample user is The absolute value of the difference between the median sleep time on weekdays and the median sleep time on weekends is and a sleep category to which one sample user corresponding to the set belongs, which shows a partial jet lag. The goal is to display a small number of points arranged in correspondence with the score on one of the horizontal and vertical axes. At least one threshold and the other of the horizontal axis and the vertical axis At least one threshold value corresponding to the jet lag; and It can be any of the categories. The method according to the 25th aspect is the method according to any one of the 21st to 24th aspects. "The target behavior data that identifies the history of behaviors performed by the target user is used as the estimation model." The sleep state of the target user is identified from the estimation model by inputting the sleep state information into the target sleep state information. and outputting the sleep state data. In the method according to the 26th aspect, any one of the 21st to 25th aspects is used. In the above, "the at least one processor is a central processing unit (CPU), a microprocessor, processor and / or graphics processing unit (GPU) . The computer program according to the 27th aspect is a program for executing, by at least one processor, Target behavior data that, when executed, identifies a history of actions performed by a target user. The target behavior data is acquired, and the target behavior data is applied to an estimation model generated by performing supervised learning. A target menopausal symptom indicator that identifies the menopausal symptom state of the target user by inputting the data. and condition data, the condition data including target overall condition data identifying the overall condition of menopausal symptoms of the target user. target first condition data identifying a first symptom state of the target user; and subject second condition data identifying two symptom states. the at least one processor to output state data from the estimation model. "It can be made to work." A computer program according to a 28th aspect of the present invention is the same as the computer program according to the 27th aspect of the present invention, except that "each group of classes" is The supervisor data identifies a history of actions performed by one sample user corresponding to the set. The sample behavior data and the menopausal symptom status of one sample user corresponding to the set are identified. and sample menopausal state data to be classified into two groups. and obtaining the estimated model generated by performing the at least one process. This allows the processor to function. A computer program according to a 29th aspect of the present invention is the same as the computer program according to the 27th aspect of the present invention, except that "each group of lessons The supervisor data identifies a history of actions performed by one sample user corresponding to the set. The sample behavior data and the menopausal symptom status of one sample user corresponding to the set are identified. and sample menopausal state data to be classified into two groups. The estimation model generated by the above-mentioned method is connected to the computer via a communication line. "causing the at least one processor to function." A computer program according to a 30th aspect of the present invention is the same as the computer program according to the 27th aspect of the present invention, except that "each group of classes" is The supervisor data identifies a history of actions performed by one sample user corresponding to the set. The sample behavior data and the menopausal symptom status of one sample user corresponding to the set are identified. and sample menopausal state data for classification, and generating the estimation model by inputting training data into a learning model and allowing it to learn; "to cause said at least one processor to function in such a manner." The computer program according to the 31st aspect is the computer program according to the 30th aspect. The plurality of sets of teacher data includes a first plurality of sets of teacher data, and the first plurality of sets of teacher data Each set of training data included in the data is, as the sample menopausal state data, at least , a sample corresponding to the set, a sample identifying the overall state of menopausal symptoms of the user and from the estimation model, the target menopausal state data includes the following: outputting target overall condition data identifying the overall menopausal condition of the target user; "to cause said at least one processor to function in such a manner." A computer program according to a 32nd aspect of the present invention is the computer program according to the 31st aspect of the present invention, Each set of training data included in the plurality of sets of training data is the sample menopausal state data. Then, the overall state of menopausal symptoms of at least one sample user corresponding to the set is calculated. the first plurality of sets of teacher data including sample overall condition data identifying corresponding scores; For each group of training data included in the data, the menopausal period of one sample user corresponding to the group is calculated. The sample overall status data corresponding to the set is used to identify a score corresponding to the overall status of the symptom. Identify the category that corresponds to the overall state of menopausal symptoms of one sample user and converting the first sample into sample overall state data. A plurality of sets of training data are input to the learning model to generate the estimation model, and the estimation model The target menopausal state data is a comprehensive state of menopausal symptoms of the target user. outputting the object overall status data that identifies the category corresponding to the at least "It allows one processor to function at the same time." In the computer program according to the 33rd aspect, in the 32nd aspect, The score corresponding to the overall condition of menopausal symptoms of one sample user corresponding to the set is , Simplified Menopause Index (SMI), Menopause Rating Scale (Menopause Rating Scale) Scale), Kupperman index, menopause Harm assessment (Green Climacteric Scale), PSST (Premens WHQ(Women's Symptoms Screening Tool), WHQ(Women's s Health Questionnaire), VAS (Visual Analog ue Scale), HFRDI(Hot Flash Related Daily In terference Scale), HFCS (Hot Flash Composite e Score), MENQOL (Menopause-Specific Qualit y Of Life) or the sample user responses to the Japanese Women's Menopausal Symptom Assessment Table It can be a score calculated based on content. The computer program according to the 34th aspect is the computer program according to the 30th aspect. The plurality of sets of teacher data includes a second plurality of sets of teacher data, and the second plurality of sets of teacher data Each set of training data included in the data is, as the sample menopausal state data, at least a sample first condition data set that identifies the first symptom condition of one sample user corresponding to the set; data of the target user as the target menopausal state data from the estimation model; and outputting target first condition data identifying the state of the first symptom of the at least one of the first symptoms. "It can make one processor work." A computer program according to a 35th aspect of the present invention is the computer program according to the 34th aspect of the present invention, Each set of training data included in the plurality of sets of training data is the sample menopausal state data. Then, at least one sample user's first symptom state corresponding to the set is identified. The second plurality of sets of training data includes sample first state data for identifying the core. For each set of training data, the first symptom state of one sample user corresponding to the set is calculated. The sample first state data identifying the corresponding score is then sent to one sample user corresponding to the set. Convert sample first condition data into a category that corresponds to the user's first symptom state. and performing a preprocessing of the second plurality of sets of training data after the preprocessing. a model to generate the estimation model, and the target menopausal state data is extracted from the estimation model. a target first symptom category that identifies a category corresponding to the first symptom category of the target user as the data; and causing the at least one processor to output the state data. can. A computer program according to a 36th aspect of the present invention is the computer program according to the 30th aspect of the present invention. The plurality of sets of teacher data includes a third plurality of sets of teacher data, and the third plurality of sets of teacher data Each set of training data included in the data is, as the sample menopausal state data, at least a sample second condition data set that identifies a second symptom condition of one sample user corresponding to the set; data of the target user as the target menopausal state data from the estimation model; and outputting target first condition data that identifies the state of the second symptom of the at least one of the first condition data and the second symptom data. "It can make one processor work." A computer program according to a 37th aspect of the present invention is the same as that according to the 36th aspect of the present invention, except that "the third Each set of training data included in the plurality of sets of training data is the sample menopausal state data. Then, at least one sample user corresponding to the set has a status corresponding to the second symptom. The third plurality of sets of training data includes sample second state data for identifying the core. For each set of training data, the state of the second symptom of one sample user corresponding to the set is calculated. The sample second state data identifying the corresponding score is then sent to one sample user corresponding to the set. Convert the sample second condition data into a category that corresponds to the user's second symptom state. and performing a preprocessing of the third plurality of sets of training data after the preprocessing. a model to generate the estimation model, and the target menopausal state data is extracted from the estimation model. a target second condition category that identifies a category corresponding to a second symptom state of the target user as the data; and causing the at least one processor to output the state data. can. A computer program according to a 38th aspect of the present invention is a computer program according to any one of the 27th to 37th aspects of the present invention. In any of the above cases, "among multiple products and / or multiple services, the target menopausal symptoms" At least one target product and / or at least one target service corresponding to the behavior data "The at least one processor may be caused to function so as to determine: A computer program according to a 39th aspect is the same as the computer program according to the 38th aspect, except that The target menopausal state data is input to another estimation model generated by executing learning. By doing so, from the other estimation model, the at least one target product and / or the outputting proposal data identifying at least one target service; "It can also function as a single processor." The computer program according to the 40th aspect is the computer program according to the 38th aspect, a search table that associates the initial status data with the plurality of products and / or the plurality of services; By inputting the target menopausal state data into the search box, the search table is searched for the at least one of the target menopausal state data. and providing proposal data identifying at least one target product and / or at least one target service. The at least one processor may be configured to "obtain the at least one signal." In the computer program according to the 41st aspect, the computer program according to the 27th aspect to the 30th aspect In any of the 40 aspects, "the at least one processor is a central processing unit (C PU), microprocessor and / or graphics processing unit (G PU). The information processing device according to the 42nd aspect "includes at least one processor, At least one processor identifies a history of actions performed by a target user. The target row is then applied to the generated estimated model by performing supervised learning. a target menopausal user to identify the state of menopausal symptoms of the target user by inputting the target user's movement data; and a target overall menopausal status data identifying the overall menopausal status of the target user. condition data, target first condition data identifying a first symptom condition of the target user, and and target second condition data identifying a second symptom state of the user. The method may be configured to output subject menopausal status data from the estimation model. An information processing device according to a 43rd aspect is the information processing device according to the 42nd aspect, It can be a device. In the information processing device according to the 44th aspect, the information processing device according to the 42nd aspect or the 43rd aspect "The at least one processor is a central processing unit (CPU), a microprocessor, The processor may include a graphics processor and / or a graphics processing unit (GPU). The method according to the 45th aspect is "at least one computer-readable instruction executing a method executed by at least one processor, the method comprising: executes the instructions to identify a history of actions performed by the target user. The steps include acquiring target behavior data and performing supervised learning to generate an estimated model. By inputting the target behavior data into the database, the state of menopausal symptoms of the target user can be determined. The target menopausal state data to be identified is used to identify the overall state of menopausal symptoms of the target user. target overall condition data identifying the condition of the first symptom of the target user; and target second condition data identifying a second symptom condition of the target user. outputting target menopausal state data including at least one from the estimation model. You can do this. In the method according to the 46th aspect, the method according to the 45th aspect, "the at least one The processor may be a central processing unit (CPU), a microprocessor, and / or a graphics The processor may include a graphics processing unit (GPU). The method according to the 47th aspect is a method for performing a computer-readable instruction. a method executed by at least one processor, the method comprising: However, by executing the above command, the teacher data of each group is sample behavior data identifying a history of behaviors performed by a user and a corresponding sample menopausal status data identifying the menopausal symptom status of one sample user; An acquisition step of acquiring multiple sets of training data, including: inputting the multiple sets of training data into a learning model; By training the system, the target user can identify the history of actions performed by the target user. and inputting the dynamic data to generate target menopausal status data that identifies the menopausal symptom status of the target user. subject overall status data identifying the overall menopausal status of the subject user; target first condition data identifying a first symptom state of the target user, and and subject second condition data identifying two symptom states. and generating an estimation model configured to output the state data. It is possible. A method according to a 48th aspect is the method according to the 47th aspect, wherein "the acquiring step includes: The method includes a step of acquiring a first plurality of sets of training data included in the training data, Each group of training data included in the group of training data is, as the sample menopausal state data, At least one sample user corresponding to the set has a general condition of menopausal symptoms. and the generating step includes sample overall condition data that identifies a score to be obtained. For each set of teacher data included in the first plurality of sets of teacher data, A sample that identifies a score corresponding to the overall state of menopausal symptoms for one sample user The comprehensive state data is converted into the comprehensive state of menopausal symptoms of one sample user corresponding to the set. Perform preprocessing by converting the sample overall condition data into the corresponding category-identifying data. and each set of training data is a sample of one sample user corresponding to the set. behavioral data and the sample overall state data of one sample user corresponding to the set; The first plurality of sets of preprocessed training data, including generating the estimation model. The method according to the 49th aspect is the method according to the 47th aspect, wherein "the acquiring step The step of acquiring a first plurality of sets of training data included in the training data of the second plurality of sets of training data Each group of training data included in the group of training data is, as the sample menopausal state data, At least one sample user corresponding to the set has a score corresponding to the first symptom state. and the generating step further includes sample first state data identifying the acquired second state data. For each group of training data included in the multiple sets of training data, one sample corresponding to the group is The sample first condition data identifying a score corresponding to the first symptom condition of the user is A sample user's first symptom state is identified as a category corresponding to the pair. The first stage of preprocessing is to convert the training data into sample first state data. , the sample behavior data of one sample user corresponding to the set, and the sample first state data of a sample user; and a step of inputting the plurality of sets of training data into the learning model to generate the estimation model; It can include. A method according to a 50th aspect is the method according to the 47th aspect, wherein "the acquiring step includes: a step of acquiring a third plurality of sets of teacher data included in the teacher data, Each group of training data included in the group of training data is, as the sample menopausal state data, At least one sample user corresponding to the set has a score corresponding to the second symptom state. and further comprising: sample second state data identifying the acquired third state data; For each group of training data included in the multiple sets of training data, one sample corresponding to the group is The sample second condition data identifying a score corresponding to the second symptom condition of the user is A sample user's second symptom state is identified as a category corresponding to the pair. The preprocessing stage converts each set of training data into sample second state data. , the sample behavior data of one sample user corresponding to the set, and the sample second state data of a sample user; and a step of inputting the plurality of sets of training data into the learning model to generate the estimation model; It can include. In the method according to the 51st aspect, the method according to the 47th aspect to the 50th aspect "The at least one processor may be a central processing unit (CPU), a microprocessor, and / or may include a graphics processing unit (GPU).

[0237] II.Chapter 2 The technology explained in Chapter 1 is combined with the technology explained below in Chapter 2. It is possible to use it in this way.

[0238] For example, the estimation model for estimating the sleep state described in Chapter 1 can be used as follows: In the technology, it can be used as the sleep state prediction model 110b. The estimation model for estimating the state of menopausal symptoms described in Chapter 1 is implemented using the technology described below. In this case, it can be used as a women's health prediction model 100a. In this paper, we will introduce the estimation model for predicting exercise status and immune status, which are explained in Chapter 1. A prediction model for hydration status, a prediction model for nutritional status, and a prediction model for hydration status. In the technology described below, the lifestyle-related disease prediction model 110c and the immune A state prediction model 110e, a hydration status prediction model 110d, and a nutritional status prediction model 11 It can be used as 0f.

[0239] Without being limited to this, at least one feature (or part) included in the technology described in Chapter 1 At least one feature included in the technology described in Chapter 2) It can also be used in the technology described in Chapter 1.

[0240] 1. Overview 15 to 28 are diagrams for explaining the embodiment. A health prediction system that predicts the state (also called health symptoms) of a person and a health prediction system that predicts the state (also called health symptoms) of a person. and learning systems that generate predictive models to forecast customer health status. We will then explain the overview, functional configuration, processing flow, etc.

[0241] 2. Health Prediction System 2.1 Overview of Health Prediction System Processing First, referring to FIG. 15, the general process of the health prediction system 1Z according to the embodiment will be described. FIG. 15 shows a health prediction system 1Z according to an embodiment of the present invention. 1 is a schematic diagram showing the flow of data for making health predictions.

[0242] Health Prediction System 1Z is a system for predicting an individual's health status based on their purchasing behavior. This health prediction system 1Z can predict the health status of individual customers based on their medical checkup data. Without any personal health data, the predictions are based on purchasing data, which is the customer's purchasing behavior. The estimated health condition can be used to improve the health of the customer, for example.

[0243] Here, an integrated health prediction program 1 predicts the health status of an individual based on their purchasing behavior. 00Z includes multiple types of trained predictive models that predict health status based on individual purchasing behavior. In the example of Figure 15, the integrated health prediction program 100Z integrates six prediction models. However, it is not limited to six, and it can be five or less, or seven It may be more than that.

[0244] The integrated health prediction program 100Z according to this embodiment is a program for predicting women's health. Health prediction model 110a, sleep state prediction model 110b for predicting sleep state, lifestyle-related disease Lifestyle-related disease prediction model 110c predicts the hydration status, and hydration status prediction model 11 predicts the hydration status. 0d, immune status prediction model 110e, nutritional status prediction model 110f, The six prediction models of the measurement model 110f (hereinafter collectively referred to as "prediction models 110Z") Each predictive model included in the Integrated Health Prediction Program 100Z is based on the customer's purchase history. From the purchasing data 150Z showing women's health status (SMI related to menopausal symptoms, etc.) / sleep Values ​​indicating condition / lifestyle-related disease / hydration status (hydration status) / immune status / nutritional status (specific disease The value indicating the possibility and / or severity of the disease) is used as the customer's health prediction value 160Z. In the example of Figure 15, the values ​​indicating the six health conditions are shown in percentage. Depending on the type of disease, treatment may be targeted to men only, women only, or both men and women. These prediction models 110Z can be described as functions. The prediction model is constructed by machine learning using training data including various health conditions. The construction method will be described later.

[0245] More specifically, the integrated health prediction program 100Z receives customer purchase data 150Z. Purchase data 150Z is data on the purchase history of customers whose health is to be predicted. , training data for each prediction model 110Z included in the integrated health prediction program 100Z The data corresponds to the explanatory variables of each prediction model. Enter purchase data 150Z into 110Z in parallel and calculate customer health prediction value 160Z. In this embodiment, only the purchase data 150Z is input to the prediction model 110Z. However, customer attribute information such as the customer's gender and age may also be input.

[0246] In addition, the Integrated Health Prediction Program 100Z provides the customer with the following information according to the calculated health prediction value 160Z. Various suggestions regarding health improvement may be made to the customer. It points out to customers categories with high health prediction values ​​and identifies (potential) symptoms that customers are unaware of. Alternatively, it may be possible to propose measures that can be taken while the symptoms are still mild. The proposed method will be described later with reference to FIG.

[0247] 2.2 Structure of the Integrated Health Prediction Program The functional configuration of the integrated health prediction program 100Z will be described below with reference to FIG. 16. FIG. 16 is a diagram showing the functional configuration of the integrated health prediction program 100Z. The program 100Z includes an input unit 120Z, a product category determination unit 125Z, and a prediction model 1 10Z, an output unit 130Z, and a proposal unit 135Z. Z can be implemented as a single program, or it can operate in cooperation with other programs by inputting and outputting data. It may be implemented as multiple programs.

[0248] The input unit 120Z receives (acquires) the input of the purchasing data 150Z. The integrated health prediction program 100Z is executed by an information processing device (computer) built in. HDD (Hard Disk Drive) or SSD (Solid State Drive) The purchasing data 150Z may be read from a storage medium such as a LIBRARY or the like. External devices connected via N (Local Area Network) or the Internet, etc. The purchasing data 150Z may be input from an information processing device or the like in the department.

[0249] Here, referring to FIG. 17, the input unit 120Z receives the purchase data 150Z. A specific example will be described. FIG. 17 is a diagram showing a specific example of purchase data 150Z. Kaidata 150Z uses panel data that accumulates customer purchasing history provided by research companies. The panel data is collected when a product is purchased by a customer. This data is acquired by scanning the smartphone or rental car. When a product is scanned using the provided barcode reader, the product and the pre-set The data is stored in association with the customer's attributes (office worker / student / housewife, etc.) and information on where they purchased the product. This allows us to understand who purchased what, how much, and at what store. It is grasped.

[0250] The example of purchase data 150Z in FIG. 17 shows the purchases made by a customer over a certain period (e.g., one year). Each product purchased by a customer is shown in the table below. In the JICFS (JAN Item Code File Service) classification, In other words, each product category has at least one product. The purchase data 150Z contains, for each customer ID, the product to which the product purchased by the customer belongs. Regarding the product category, those included in the major categories of "food" and "daily necessities" in the JICFS classification 150a for each of the approximately 500 subcategories (e.g., soft drinks, mouthwash, etc.) The actual data 150b is numerical data such as the purchase amount, purchase quantity, and purchase frequency during the period. The data is entered accordingly.

[0251] In this embodiment, the purchase data 150Z includes purchases of food and / or daily necessities. Purchase history is limited to those that are presumed to be related to the customer's health condition. For example, other products such as medicines, or various services such as massage and acupuncture treatments It is also possible to include the purchase history of services, etc. in the purchase data 150Z. 50Z can be identified by payment information managed by the retailer, or by credit card information, for example. It is also possible to identify the person using payment information from their credit card or payment information to financial institutions. .

[0252] The product category determination unit 125Z receives the purchase data 150Z from the input unit 120Z. From the purchase data 150Z to be input to each forecast model 110Z, Determine the classification code (JICFS classification) and product category to input, and then enter the obtained classification code. Numerical data such as purchase amount, purchase quantity, and purchase frequency for each product and product category. The actual data 150b (only a part of which may be used) is input to each prediction model 110Z. Input in parallel.

[0253] In addition, the classification code (product category) of the purchase data 150Z to be input to the prediction model 110Z (-) can be determined for each prediction model 110Z. For details on how to determine the product category to be used, see 3. below regarding the creation of the prediction model 110Z. This will be discussed later in Section 4.

[0254] In addition, the purchase data 150Z input from the input unit 120Z can be narrowed down by product category. It is also possible to input the actual data 15-b directly into the forecast model 110Z without In this case, the product category determination unit 125Z is not necessary.

[0255] The prediction model 110Z is based on the classification code and product selected by the product category determination unit 125Z. Performance data, which is numerical data such as purchase amount, purchase quantity, and purchase frequency for a category The data 150b is input, and the women's health status (menopausal symptoms), sleep status (sleep disturbances), ), lifestyle-related diseases (lack of physical activity), hydration status (degree of deficiency / hydration status), immune status ( In the example of Figure 15, the health prediction values ​​of the immune system (immunocompromised level) and nutritional status (nutritional deficiency level) are calculated. For the customer with user ID 001, each prediction model 110Z achieved 90% accuracy for menopausal symptoms and 90% accuracy for sleep. 21% of respondents had poor blood circulation, 40% had insufficient physical activity, 76% had poor fluid intake, 72% had a weakened immune system, and malnutrition The calculated value is 11%.

[0256] As mentioned above, the prediction model 110Z is based on the purchase data 150Z, which is the purchase history of each customer. , and the training data that corresponds to the health status of those customers are used to determine the former as an explanatory variable and the latter as It can be constructed by machine learning using the objective variable. The construction method will be described in section 3 below.

[0257] The output unit 130Z outputs the values ​​calculated by each prediction model 110Z as a health prediction for the customer. The value 160Z is the value of the proposal unit 135Z, the built-in storage medium, or the like connected via a network. Alternatively, the output unit 130Z may display the calculated results on a display device. In the example of FIG. 15, the output unit 140Z may display the calculated health prediction value 160Z. The values ​​of customer ID, menopause, sleep disturbance, etc. are arranged in order in the row direction, and this is the customer data By repeating this for each ID, the health prediction value 160Z for multiple customers is output. The health prediction value 160Z not only predicts the actual health condition of the customer, but also predicts future events. It may also include health conditions (potential health conditions) that may be predicted to occur.

[0258] The proposal unit 135Z proposes food, supplies, and services according to the predicted health prediction value 160Z for each customer. There are various ways to propose this, for example, various types of products used by retail store clerks. It may be displayed on the device or proposed directly to the customer via a messenger service, etc. The contents may be sent to the customer or displayed on a web page.

[0259] 2.3 Examples of various proposals and processing to customers 18 to 20, the integrated health prediction program 100Z will be described below. We will explain the various proposals we will make.

[0260] FIG. 18 shows the food / goods that the suggestion unit 135Z suggests in accordance with the customer's health prediction value 160Z. For example, if the menopausal symptoms are predicted to be 90%, For a woman (customer ID=001), the proposed item 135Z is sold at drug store 41Z. We will propose products that will improve menopausal symptoms. For the man (customer ID=002), the proposed product 135Z was sold at drug store 41Z. As another example, we propose a highly functional nutritional food that has the potential to predict menopause and sleep disturbances. For 79% and 89% of women (customer ID=003), respectively, the proposal unit 135Z In addition, the proposal department 135Z will recommend a consultation to customers who are relatively inactive. suggests going to the 45Z gym and recommends that customers whose physical inactivity level is lower than a certain level will propose a medical insurance product that addresses lifestyle-related diseases, offered by insurance company 47Z.

[0261] In this way, the proposal unit 135Z of the integrated health prediction program 100Z uses the purchase data 150 Z, by obtaining prediction results from each of the multiple prediction models 110Z, a suitable solution for the customer can be provided. In order to realize such a proposal, The foods, supplies, and / or services associated with each health symptom that is the output of the predictive model. A list of these should be prepared in advance.

[0262] Figure 19 shows Table 50Z for realizing suggestions regarding menopausal symptoms and the data to be referenced. Databases 52Z, 54Z and 56Z are shown.

[0263] Table 50Z shows the menopause prediction model (Women's Health Prediction Model 110 in Figure 16). Class 50a, where the output range of a) is less than 40%, Class 50b, where the output range is 40% or more but less than 80%, and There are three classes: Class 50a, Class 50c, and Class 50b, which are 80% or more. The product database 52Z for preventing menopausal symptoms from Rug Store is referenced. Base 52Z has products X1 and X2 in product category P1, and product X in product category P2. 3 is described. The suggestion unit 135Z refers to the product database 52Z and selects one of them. may suggest some or all of the categories and / or product names to applicable customers. If the output value of the prediction model falls under class 50b, the suggestion unit 135Z will Refer to the product database 54Z to be improved, and select products Y1 and Y2 in product category Q1. If the output value of the prediction model corresponds to 50c, the suggestion unit 135Z proposes the current location. and / or menopause outpatient medical institution data using pre-registered customer residence information Based on the base 56Z, medical institutions are recommended. This means that the medical institutions are recommended based on the customer's health prediction results. This corresponds to a referral to a medical institution's client for preventive care and / or treatment.

[0264] Another example will be described with reference to Fig. 20. Fig. 20 shows a lifestyle-related disease prediction model 110c 6 shows a table 60Z for realizing proposals according to the output value of the

[0265] Table 60Z shows the output value of the lifestyle-related disease prediction model 110c, in other words, the physical activity Class 60a, where the range of deficiency is less than 20%, Class 60b, where the range of deficiency is 20% or more but less than 80%, and There are three classes: Class 60a, Class 60c, and Class 60b. Physical activity is considered to be a fulfilling activity, and in order to encourage continued physical activity, nutritional guidance (1) and , a monthly membership plan for E1 Sports Gym is proposed. Nutritional guidance (1) is, for example, , dietary suggestions for consuming less than 10 grams of fat and more than 30 grams of protein In class 60b, physical activity is not considered to be insufficient, and nutritional guidance (2) and regular To encourage physical activity, the E2 Fitness Club offers a pay-as-you-go plan. Nutritional guidance (2) is, for example, less than 20 grams of fat and more than 20 grams of protein per meal. This is a dietary suggestion for getting high-quality nutrients. In class 60c, physical activity is insufficient. Considering the possibility of developing lifestyle-related diseases in the future, we provide nutritional guidance (3) and E3 The medical insurance products proposed by the life insurance company are proposed. Nutritional guidance (3) is, for example, It contains less than 20 grams of fat, more than 20 grams of protein, and less than 20 grams of carbohydrates. This is a dietary suggestion for taking in the necessary nutrients. Nutritional guidance, sports gyms, fitness clubs, and The proposal of insurance products falls under the category of product or service provision. Regarding the output value of the above, the proposing unit 135Z also refers to the table 60Z to obtain the output value It is possible to make proposals to customers that correspond to their needs.

[0266] In addition to the examples in Figures 19 and 20, the following data were collected for each of sleep status, immune status, hydration status, and nutritional status: By providing similar tables and databases as needed, the proposal section 135Z For each customer, the health prediction value 160Z, which is the output value of each prediction model 110Z, is provided to the customer. Proposals can be made.

[0267] It should be noted that the integrated health prediction program 100Z itself does not necessarily include the proposal part 135Z. For example, the output unit 130Z of the integrated health prediction program 100Z may With the customer's permission, a health plan may be provided to a business other than the business that operates Program 100Z. The program of the other business operator transmits the measurement value 160Z, and the program of the other business operator transmits the program related to the proposal section 135Z. It is also possible to implement the program and make proposals such as those mentioned above to customers. Health prediction program 100Z outputs health prediction value 160Z, which is the output value of prediction model 110Z. The information may be sent to an insurance company so that the insurance company can offer insurance products to the customer.

[0268] 2.4 Processing flow The processing flow of the integrated health prediction program 100Z will be described below with reference to FIG. FIG. 21 shows a customer's health prediction process executed by the integrated health prediction program 100Z. 10 is a flowchart showing the procedure of the process.

[0269] The input section 120Z of the integrated health prediction program 100Z is a The input unit 120Z acquires purchase data 150Z indicating the purchase history of the user (S701). For example, when purchasing goods at a retail store, the customer's membership card is presented at the POS (Point of Sale) ) system, a database linking the customer to their purchase history is created. With the customer's consent, purchasing data 150Z is extracted from such a database. By doing so, the input unit 120Z inputs the purchasing data 150 into the integrated health prediction program 100Z. As described above, such purchase data 150Z can be input, for example, Includes JICFS classification and product categories.

[0270] The product category determination unit 125Z determines each prediction model 110Z from the purchasing data 150Z. Determine the JICFS classification (classification code) and product category to be input (S703) As mentioned above, the 110Z prediction model does not limit the product categories used for prediction. If the entire purchase data 150Z is used, the process of S703 is not necessary.

[0271] Next, each prediction model 110Z uses the classification code determined by the product category determination unit 125Z. The health condition is predicted using the product category purchase data 150Z (S705). The output unit 130Z outputs a health prediction value 160Z indicating the health condition predicted by the prediction model 110Z. The proposing unit 135Z provides the customer with a proposal based on the health predicted value 160Z (S 707) Specifically, printing on receipts, displaying on screens, and The health prediction value 160Z or the proposal unit 135Z provides a health improvement proposal to the customer in the form of a notification or the like. This is what I can think of.

[0272] 2.5 Hardware Configuration Referring to FIG. 22, the information processing device 8 that executes the integrated health prediction program 100Z A specific example of the hardware configuration of the integrated health prediction program 100Z will be described. 8 is a hardware configuration diagram of an information processing device 800Z on which the information processing device 800Z is executed. 0Z uses the integrated health prediction program 100Z to provide information on customers' menopausal symptoms, sleep status, etc. The information processing device 800Z is a commonly available computer system. systems, such as desktop PCs (Personal Computers), notebooks The computer system may be a PC, a tablet PC, a server computer, etc. The system may be installed at the business premises of the operator of the Health Prediction System 1Z, or may be installed on a cloud service. In the latter case, the business must have a system that can communicate with the cloud service. It would be good if a PC or other device capable of handling the data was provided.

[0273] The information processing device 800Z includes a control unit 810Z and an input interface (I / F) unit 820. Z, a storage device 830Z, and an output I / F unit 840Z.

[0274] The control unit 810Z includes a CPU (Central Processing Unit). (not shown), ROM (Read Only Memory, not shown), RAM (Random Access Memory, not shown) The control unit 810Z may include a memory device 830Z. The integrated health prediction program 100Z stored in the In addition to the functions of a general computer, the device 800Z has the functions related to the above-mentioned health prediction. The arithmetic circuit included in the control unit 810Z does not have to be a CPU. It may be realized by various processors such as MPU and GPU, or may be realized by a single processor. The present invention may be realized by a plurality of processors instead of a single processor.

[0275] The input I / F unit 820Z receives purchase data 150Z relating to a customer whose health condition is to be predicted. As mentioned above, the input I / F unit 820Z receives data from a network such as the Internet or a LAN. Purchase data 150Z is obtained from an information processing device such as an external server connected via a network. The input I / F unit 820Z can receive, for example, Ethernet (registered trademark). communication terminal, USB (registered trademark) terminal, IEEE802.11, 4G, or 5G standards This can be realized by a communication circuit that performs communication in accordance with the above.

[0276] The storage device 830Z is a computer program required to operate the information processing device 800Z. The storage device 830Z is a storage medium for storing programs and data. The memory device 830Z may be an SSD, which is a semiconductor memory device. The control unit 810Z may be provided with a temporary storage element configured by a RAM such as AM. The storage device 830Z may function as an integrated health prediction program 100Z. 19 and 20 are table 50Z and table 60Z, respectively. Stores Lu 831Z.

[0277] The 110Z predictive model is incorporated as part of the 100Z integrated health prediction program. Alternatively, the data may be provided as separate data from the integrated health prediction program 100Z. The prediction model 110Z is stored in the storage device 830Z as a table or a group of parameters. This may be done.

[0278] In addition, the proposal table 831Z displays the predicted results indicating the level of health, for example, H layer / M layer. / L layer, 90% / 50% / 20%, and corresponding products and services to be proposed according to each level It can be considered as an attachment.

[0279] The output I / F unit 840Z is connected to various output devices provided outside the information processing device 800Z. For example, the output I / F unit 840Z is a communication circuit and / or a communication terminal to which the receiver is connected. It can be a USB terminal that outputs print data indicating the content to be printed on the integrated printer P. Health Prediction Program 100Z calculates prediction results regarding the customer's health condition, and as a result, Or send a proposal to the customer based on the result to the receipt printer P and print it on the receipt. It is possible.

[0280] The output I / F unit 840Z may be a video output terminal connected to the display D. The integrated health prediction program 100Z calculates prediction results for the customer's health condition and The results or a proposal to the customer based on the results are displayed on Display D. For example, a pharmacist can check the information displayed on the screen and provide the appropriate information according to the customer's symptoms. One or more products can be proposed.

[0281] In another embodiment, the output I / F unit 840Z is connected to a communication network N or the like to transmit data. The output I / F unit 840Z may be a communication terminal or a communication circuit capable of communication. In the case of a communication terminal or a communication circuit, the input I / F unit 820Z and the output I / F unit 840Z The hardware may be the same. The output I / F unit 840Z is connected to the and provide the customer with a prediction result regarding the customer's health condition or a proposal based on the result. It can be sent to smartphone M.

[0282] 3 Learning system for predictive model 110Z9 3.1 Overview Next, we will use machine learning to develop the predictive model 110Z included in the integrated health prediction program 100Z. A method for generating the model will be described with reference to FIG. 23. The following description will be given assuming that 900Z generates a prediction model 110Z. RAM 900Z may be implemented as a single program or may be linked by data input and output. It may be implemented as multiple programs that can operate in conjunction with one another.

[0283] The integrated health prediction program 100Z shown in Figure 15 uses six types of prediction models. These prediction models include 10Z of the training data of 950Z. The only difference is the data 954Z. The following mainly describes the generation of such a menopausal prediction model 110a.

[0284] 3.2 Training data 950Z The model learning program 900Z is a teaching program containing various data collected from a large number of subjects. We use teacher data 950Z to perform machine learning and build a predictive model 110Z. The data 950Z includes purchase data 952Z relating to the purchasing behavior of each subject and the results of those subjects. This includes health data 954Z, which is data related to the health of individuals.

[0285] The purchase data 952Z relates to information similar to that shown in the specific example with reference to FIG. That is, in this embodiment, for each subject ID, Item 150a relating to the category and figures such as purchase amount, purchase quantity, and number of purchases during the period The purchase data 952Z can include the actual result data 150b, which is the data of the food. The food and / or beverage is suitable as a product related to the health status of the subject or customer. Therefore, it is preferable that the purchase data 952Z includes a purchase history of food and / or beverages.

[0286] Health data 954Z describes the subject's health status. This health status includes: The condition may be related to subjective symptoms, or may be a condition recognized as a result of measurement. That is, at least one of the following: health awareness, mental health, cognitive function, and health checkup results Any of these can be used as health data 954Z. In 54Z, the judgement value for each condition segment of the subject is described as the health condition of the subject. The condition segment is defined for each combination of the type and level (in other words, magnitude) of health condition. The type and level of health condition in this embodiment refers to the physical and mental disorders of the corresponding individual. The judgment value is an index that indicates the health condition, and is divided into 2 classes, 3 classes, etc. The "type and level of health condition" mentioned here can be expressed in terms of medical rigor. It can also be understood as the type and level of health concerns that an individual has, rather than as a result of the be.

[0287] Here, as an example, the model learning program 900Z generates the menopause prediction model 110a. In cases where the symptoms of the Simplified Menopause Index (SMI) are used to determine the level of menopause, In this case, the health data 954Z includes the simplified menopausal index to which the subject belongs. Labels indicating symptoms of several SMIs can be included in health data 954Z. It is possible to express the menopausal index (SMI) as two classes. There are two types: "H" which indicates a relatively large value, and "other than H". In comparison, the symptoms are subdivided into "M" which indicates moderate symptoms and "L" which indicates relatively small symptoms ( In other words, it is possible to express the symptoms as three classes in total. The subject group that participates in this study is called the "H group."

[0288] In addition, in the labeling of the symptom severity in the health data 954Z, two classes and Instead of three classes, it may be further subdivided into four or more classes, or Instead of the class, it is also possible to use a numerical value indicating the possibility of corresponding to menopausal symptoms.

[0289] That is, the training data 950Z includes information about each product category of each subject and a predetermined Purchase data 952Z, including information on purchase amounts, purchase quantities, and number of purchases during the period; , and health data 954Z including at least information on the label indicating the health status. Here, the model learning program 900Z, for example, Symptoms), sleep status (disturbed sleep), lifestyle-related diseases (lack of physical activity), hydration status (degree of hydration) ), immune status (immunocompromised level), and nutritional status (nutrition deficiency level) - six predictive models 110Z If we want to generate a prediction model 110Z, we need to use the information corresponding to each of these prediction models in the health data. In supervised learning, the teacher The teacher data 950Z is roughly classified into explanatory variables and objective variables. , purchase data 952Z corresponds to the explanatory variable, and health data 954Z corresponds to the target variable. In the embodiment, only the purchase data 952Z is ​​used as the explanatory variable, but this is not limited to this. For example, it is possible to add the subject's attributes, such as gender and age, to the explanatory variables. can be.

[0290] Below, the model learning program 900Z can improve sleep quality by providing appropriate health data 954Z. Sleep state, lifestyle-related diseases, immune status, hydration status, nutritional status, dementia, oxygen utilization, vascular health, Hair health, intestinal environment, promoting exercise effects, appetite stimulation, eye health, skin health, oral hygiene, and mental health Health, depression, stress, headaches, stiff shoulders, ear health, joint health, temperature homeostasis, fatigue, bone Prediction models for various symptoms, such as menopausal syndrome, high blood pressure, and weather-related pain. It is possible to generate Z. Below, we will focus on women's health (menopausal symptoms), sleep, etc. Condition (disturbed sleep), lifestyle-related diseases (lack of physical activity), hydration status (degree of deficiency), immune status Examples of indices that can be used as indicators of the level of immune depression and nutritional status are shown below. do.

[0291] 3.2.1 Women's health status (menopausal symptoms) As an index showing the degree of menopausal symptoms, for example, the Simplified Menopausal Index SM The Simplified Menopausal Index (SMI) is a measure of the severity of various symptoms listed. Each subject was asked to complete the relevant items in advance, and the results were scored. The Simplified Menopausal Index (SMI) is said to be an index that reflects the symptoms specific to Japanese menopausal women. Therefore, depending on the country or region, other indicators related to menopausal symptoms, such as The Menopause Rating Scale, Bowser The Kupperman index, Menopausal symptom rating scale ( Green Climacteric Scale), PSST (The premen strual symptoms screening tool), MRS (Menop ause rating scale), WHQ (The Women's Healt) h Questionnaire), VAS(Visual analogue sca le), HFRDI(Hot Flash Related Daily Interf erence Scale), HFCS (Hot Flash Composite S core), and MENQOL (Menopause-Specific Quality of Life) y of life), or the Japanese Women's Menopausal Symptom Assessment Table, as appropriate. It is also possible that this is the case.

[0292] Regarding women's health, in addition to menopausal symptoms, we also provide information on premenstrual syndrome (PMS, PMDD) Symptoms can also be considered. In other words, to predict PMS, it is necessary to By using machine learning to measure the degree of purchase as the objective variable and the purchasing data 952 as the explanatory variable, All we need to do is build the health prediction model 110Z.

[0293] The health data 954Z may also include numerical data of these indicators. It is also possible to include only labels that indicate the severity of symptoms based on the numerical values ​​of these indicators. The same applies to the subsequent indicators of sleep state, lifestyle-related diseases, etc.

[0294] 3.2.2 Sleep state (sleep disturbance) Indicators of sleep include poor sleep quality and / or sleep rhythm, inadequate sleep duration, It is thought that these are indicators that show either of the following. It may be subjective symptoms or measured values ​​obtained as a result of measuring each control. Sleepiness was assessed using the Athens Sleep Scale (AIS), the Pittsburgh Sleep Questionnaire (Pittsburgh Sleep Quality Index), and the rgh Sleep Quality Index: PSQI), 3-dimensional sleep scale (3 Dimensional Sleep Scale (3DSS), Insomnia Severity Questionnaire (Insomnia Severity Index: ISI) The severity of symptoms is analyzed individually and in combination for each item, and the impact on daily life is assessed. The health status of the individual is assigned to a category and classified. Questionnaire (Munich ChronoType Questionnaire: μMCT) Q) to calculate the median sleep time for weekdays and weekends, and the difference between them is used to calculate the social jet. This can be understood by calculating the Social Jet Lag (SJL). do.

[0295] 3.2.3 Lifestyle-related diseases (lack of physical activity) The indicators of physical activity were selected from the following: lack of exercise, obesity, and weight change over a certain period of time. More specifically, the weight, BMI (Body Mass Index), and Mass Index), number of steps, and / or metabolic syndrome diagnostic reference value The diagnostic criteria for metabolic syndrome are based on the tibial circumference, hypertriglyceridemia, and / or hypo-HDL cholesterol. The thresholds established for the diagnosis of fasting hyperglycemia, the maximum and / or minimum blood pressure values, and The threshold is equal to or greater than one of the thresholds.

[0296] In addition, when machine learning is used to generate predictive models for lifestyle-related diseases and exercise status, purchase data is used. In addition to the data set 952Z, information on the subjects' smoking status and sedentary behavior was included as explanatory variables. Machine learning may also be performed.

[0297] 3.2.4 Hydration status (degree of deficiency) Indicators of fluid status include serum Na, BUN / creatinine ratio, and / or urine Osmolality, urine specific gravity, urine color, dehydration assessment scale, and subjective symptoms associated with lack of fluids are considered. This information can be obtained in the same manner as in the previous example of menopausal symptoms. For example, multiple subjects can be examined for a certain period of time, for example, one year, to record the changes in their bodies that they have experienced through health checkups or physical examinations. It can be obtained from clinical test results obtained in the questionnaire or a separate health questionnaire. Subjective symptoms associated with a decrease can be obtained through questionnaires, etc. The severity of symptoms can be assessed by the serum Na level. , urine osmolality, urine specific gravity, BUN / creatinine ratio, water intake, physical activity, urine color Both can be understood from one or more of the following:

[0298] Based on these indicators, you may be experiencing or likely to experience a worsening health condition. It is possible to find out from epidemiological information in the literature about conditions that may be associated with urinary incontinence. For example, if the serum sodium level is 1 People with serum Na levels above 42mEq / L had higher cognitive impairment than people with serum Na levels below 142mEq / L. It has been found that there is a high risk of developing disorders and high blood pressure, and it is evaluated as a high health risk. It is possible to do this.

[0299] In addition, when machine learning is performed to generate a prediction model for moisture status, purchasing data 952Z is ​​added. Subjects' water intake, alcohol intake, physical activity, urine color, urine osmolality, urine specific gravity, serum Na level, BU Machine learning may be performed by including one or more pieces of information, such as the N / creatinine ratio, as explanatory variables.

[0300] 3.2.5 Immune status (immunocompromised level) Indicators of immune status include, for example, susceptibility to catching a cold and / or its frequency, cold symptoms, and Symptoms can be used to measure the susceptibility and frequency of catching a cold. The subjective symptoms of each subject may be obtained from the medical history of the subject. The severity of symptoms is calculated in three or more categories, and the severity of symptoms is related to the likelihood of catching a cold. , SIgA concentration, allergic symptoms, oral environment, immune status, exercise status, stress status The symptoms may be severe sleep symptoms, moderate sleep symptoms, mild sleep symptoms, or cold-like symptoms.

[0301] 3.2.6 Nutritional status (degree of malnutrition) The indicators of nutritional status may be subjective symptoms of each subject obtained by questionnaire, etc. The degree of symptoms may be measured by the following method, or may be measured by measuring each control. The presence or absence of subjective symptoms associated with the condition, the Dietary Variety Score (DVS), and the Short Form of the Appetite Scale (SNAQ) :Simplified nutritional appetite questio At least one of the following: Average Heart Rate (HR) and BMI (Body Mass Index) It is possible.

[0302] 3.3 Machine Learning Methods The model learning program 900Z according to this embodiment uses a machine learning method such as Logis However, logistic regression is just one example, and other methods, e.g. For example, random forests, decision trees, gradient boosting, and support vector regression. , linear regression, partial least squares (PLS) It is also possible to adopt regression, Gaussian process regression, neural networks, etc. When using logistic regression, accuracy can be further improved by using the L1 regularization method. Alternatively, logistic regression (ridge regression) using L2 regularization instead of L1 regularization may be used. Regression) may be used. Elastic regularization employs both L1 and L2 regularization. You can also use the internet.

[0303] Logistic regression takes an input and calculates the probability (or probability of occurrence) of a certain event. This is a method to determine whether an input corresponds to a certain event by calculating the probability that the input does not correspond to a certain event. The logistic regression model is, for example, the logit function, which is expressed by the following equation (1 In equation (1), the left side is the natural logarithm. Also, i is the i-th The subjects, p is the probability that the target variable event occurs, and b1, b2...b n is the partial regression coefficient, b 0 is the constant term, x1, x2…x n indicates explanatory variables, respectively.

[0304]

number

[0305] If the right-hand side of equation (1) is z, equation (1) can be transformed into equation (2) which expresses probability p. Cut.

[0306]

number

[0307] Here, the sigmoid function obtained by equation (2) is set as "φ(z)", and depending on its output value, The output y is classified as one of two values. Equation (3) is 1 if φ(z) is 0.5 or more. If it is less than 0.5, it is classified into class 0. 0.5 is an example, and a value other than 0.5 may be used as the threshold. In other words, Equation (3) can be expressed as Equation (2 If z in equation (2) is 0 or greater, it is classified into class 1. If z in equation (2) is less than 0, it is classified into class 0. This means classifying

number

[0308] Next, we introduce the "likelihood" used for learning in logistic regression. The likelihood function L, which represents the likelihood, is given by the following equation (4 ) where P represents a probability value.

number

[0309] The likelihood function L indicates the probability of making a correct decision for all events. By calculating the weights, the probability of the event you want to predict can be output more accurately. The likelihood function L is multiplied by (-1) to invert the likelihood function. The degree function is the error function in logistic regression. The weights at which the error function is at its minimum value are That is, partial regression coefficients b1, b2...b n To find the error function, we use b1, b2...b n That Apply partial differentiation and gradient descent to each of them. This allows us to learn logistic regression. will be carried out.

[0310] As the likelihood function, a logarithmic likelihood function using the natural logarithm of equation (4) may be used. The optimal function is obtained by multiplying the log likelihood function by (-1) and minimizing the function with the inverted sign. weight can be found.

[0311] The above-mentioned calculation algorithm for logistic regression is well known. Software applications for rhythmic machine learning are readily available. The principle is as described above, but with available software, it is possible to calculate the Even if a person does not know the details of the specific analysis method, the logistic loop according to this embodiment can be easily understood. It is possible to achieve machine learning using regression.

[0312] 3.4 How to determine product categories for health prediction Each product and service category in the purchasing data 952Z included in the training data 950Z The numerical data such as the purchase amount, purchase quantity, and purchase frequency of the product are calculated by the above-mentioned formula (1). corresponds to the explanatory variable x in

[0313] As described above, the health data 954Z included in the teacher data 950Z in this embodiment For each symptom, such as menopausal symptoms or sleep disturbances, the patient is classified into the H stratum or assigned to the L stratum. Each subject was given a label of "H" or "L" (or "H", "M", or "L") indicating whether the subject was In the above formula (3), if it is the H layer, it outputs "1" and if it is the H layer or lower, it outputs "2". By classifying the cases so that "0" is output, the likelihood function can be calculated. The elements that make up the explanatory variable x are not limited to the above-mentioned purchase data 952Z. For example, For example, it may be possible to include data indicating age groups or gender as explanatory variables.

[0314] Machine learning was performed using the training data 950Z for a large number of subjects, and a predictive model was created. By constructing this, we can identify the product groups (categories) that H-class customers often purchase. It is possible.

[0315] Figure 24 shows that many of the subjects in the H and non-H groups chose the following regarding menopausal symptoms (purchasing The top of Figure 24 shows the product categories that most of the subjects in the H tier purchase. The product categories selected by the subjects other than those in the H layer are shown at the bottom. It is being done.

[0316] Figure 25 shows the results of menopausal symptoms found from the items purchased by subjects in the H group. The figures in the rightmost column of Figure 25 show the relationship between menopausal symptoms and health issues. This shows the degree (standard deviation) of how likely it is that a woman will choose it. Symptom-specific deviation scores for issues you want to address even if it means spending time on them - Understanding women's health, beauty and lifestyle This is a quote from the "Research Project Report - Behavioral Observation Survey Results."

[0317] These results suggest that subjects in the H group who are concerned about menopausal symptoms actively choose specific products ( The H-class subjects were interested in products that deal with dry skin and internal dryness, and vague symptoms. I am purchasing goods.

[0318] Figure 26 shows an example of a product group (product category) for women with menopausal symptoms who belong to the H group. As shown in the figure, subjects in the H tier purchased more frequently than subjects in other tiers. You can decide the product category to purchase, and purchase such products (quantity, price) It is estimated that subjects with a high frequency of menopausal symptoms are likely to belong to the H stratum. Similarly, subjects who do not belong to the H stratum tend to purchase more frequently than subjects in the H stratum. You can also determine the product categories that you want to purchase, and then you can see the frequency of purchases in those product categories. It can be assumed that subjects with low levels of menopausal symptoms are unlikely to belong to the H stratum.

[0319] Therefore, the model learning program 900Z in this embodiment uses the purchasing data 952Z Among these, data on product categories that can contribute to the estimation of whether or not a product belongs to the H tier By using this data to build the prediction model 110Z, we can accurately estimate customers belonging to the H stratum. We are making it possible for people to do so.

[0320] 3.5 Structure of Model Learning Program 900Z The functional configuration of the model learning program 900Z will be described below with reference to FIG. FIG. 23 is a diagram showing the functional configuration of the model learning program 900Z. The RAM 900Z includes an input unit 910Z, a product category determination unit 912Z, a learning unit 914Z, and and an output unit 916Z. The model learning program 900Z is a single program. It may be realized as a single program or as multiple programs working together.

[0321] The input unit 910Z receives the training data 950Z. The HDD and S that are built into the information processing device (computer) on which the learning program 900Z is executed The teacher data 950Z can be read from a storage medium such as an SD card, or it can be read from a LAN or internet. Teacher data is transmitted from external information processing devices connected via a network such as the Internet. It may also receive input from the 950Z.

[0322] The product category determination unit 912Z is ​​used for machine learning of the prediction model 110Z to be learned. Select the product category for purchase data 952Z. To select a product category, follow the steps above. Since it has been described in .4, detailed explanation will be omitted here. When generating a plurality of prediction models 110Z, the product category determination unit 912Z For each prediction model 110Z, different product categories can be selected.

[0323] As mentioned above, when machine learning the predictive model 110Z, we did not narrow down the product category. It is also possible to create a prediction model 110Z using the entire purchase data 952Z. In this case, the product category determination unit 912Z is ​​not necessary.

[0324] The learning unit 914Z determines the purchasing data of the product category determined by the product category determination unit 912Z. Machine learning is performed using data 952Z to generate prediction model 110Z. The method for generating 0Z has been described in 3.3 above, so a detailed explanation will be omitted here. .

[0325] The output unit 916Z outputs the prediction model 110Z generated by the learning unit 914Z to a built-in storage medium. Other information processing devices connected via a body or a network (for example, The output is made to the information processing device 800Z described above. It is possible to output the function that constitutes 110Z, or to output the prediction model. It is also possible to output it as a parameter of the function that makes up Rule 110Z.

[0326] 3.6 Processing flow The processing flow of the model learning program 900Z will be explained below with reference to FIG. 27. FIG. 27 shows the prediction model 110Z created by the model learning program 900Z. 10 is a flowchart showing the procedure for creating the image forming apparatus.

[0327] The input section 910Z of the model learning program 900Z is The teacher data 950Z is acquired (S1301). Purchase data 952Z showing purchase history and health data 954Z showing health status (degree of symptoms) As mentioned above, the purchasing data 952Z includes, for example, JICFS classification (classification code ) and product category, as well as the corresponding number of purchases, purchase amount, purchase quantity, etc. It may contain value data.

[0328] Next, the product category determination unit 912Z uses JICFS The classification and product category are determined (S1303). The determination method is explained in 3.4 above. As explained above, the health data obtained as health data 954Z has a large impact on the estimation of health status. The product category determination section 912Z can select the appropriate JICFS classification and product category. In addition, we used the entire 952Z purchase data without narrowing down the product categories to create a prediction model. When creating 10Z, the processing of S1303 is not required.

[0329] The learning unit 914Z uses the purchasing data 952Z as an explanatory variable and the health data 954Z as an objective variable. Machine learning is performed as a number (S1305). The machine learning method is detailed in 3.3 above. Since the explanation has been given using a specific example, the explanation will be omitted here.

[0330] The output unit 916Z outputs the prediction model 110Z generated by the learning unit 914Z in S1305. and output it to a built-in storage medium or to other information processing devices connected via a network. The prediction model 110Z is output as the function that constitutes the prediction model 110Z. As mentioned above, the information may be input or output in the form of parameters, etc.

[0331] 3.7 Hardware Configuration Referring to FIG. 28, the information processing device 14 that executes the model learning program 900Z A specific example of the hardware configuration of the model learning program 900 will be described. 1 is a hardware configuration diagram of an information processing device 1400Z in which the information processing device 14 Z is executed. 00Z uses the model learning program 900Z to provide advice on menopausal symptoms and sleep status of customers. A prediction model 110Z is generated to make predictions regarding the

[0332] The information processing device 1400Z is a computer system that can be generally available. In addition, even if it is installed at the business premises of the operator of the Health Prediction System 1Z, it will not be treated as a cloud service. The points that may be provided as the above-described information processing device 800Z are the same as those of the above-described information processing device 800Z. In this embodiment, an information processing device 800Z that performs health prediction and a prediction model 110Z that generates the prediction model are provided. The following description will be focused on the case where the information processing device 1400Z is a separate device, but the two may be the same device. It may also be realized as a device.

[0333] The information processing device 1400Z includes a control unit 1410Z, a communication I / F unit 1420Z, and a storage device. It is equipped with a 1430Z.

[0334] The control unit 1410Z may include a CPU, a ROM, a RAM, etc. It is possible to execute the model learning program 900Z stored in the memory device 1430Z. Therefore, the information processing device 1400Z has the above-mentioned functions in addition to the functions of a general computer. The control unit 1410Z can execute various processes related to the construction of the prediction model 110Z. The included arithmetic circuit does not have to be a CPU, but can be realized by various processors such as MPU or GPU. Also, it may be realized by multiple processors instead of one processor. stomach.

[0335] The communication I / F unit 1420Z is connected to an input I / F unit 820Z via, for example, Ethernet (registered trademark ) communication terminal, USB (registered trademark) terminal, IEEE802.11, 4G, or 5G standards, etc. This can be realized by a communication circuit that performs communication in accordance with the standard. 0Z can be connected to communication networks such as intranets and the Internet, and The training data 950Z prepared by the operator of the learning system 9 is received. The management device 1400Z may communicate directly with other devices via the communication I / F unit 1420Z. Communication may be via an access point or the like.

[0336] The storage device 1430Z is a computer required to operate the information processing device 1400Z. The storage device 1430Z is a storage medium for storing programs and data. The storage device 1430Z may be a RAM such as a DRAM or an SRAM. It may be provided with a temporary memory element configured as a memory area for the control unit 1410Z. The storage device 1430Z may store the model learning program 900Z and The forecast model 110Z generated by the program 900Z (the table that defines the forecast model 110Z) Contains the rules and / or parameters.

[0337] 4. Effects of this embodiment In the health prediction system 1Z according to this embodiment, the integrated health prediction program 100Z It is possible to predict multiple health risks of a customer from their purchasing data 150Z. Based on the results of each individual forecast, we can provide detailed forecasts for each customer, including current and future forecasts. Suggestions for improving and / or maintaining health can be made.

[0338] 5. Supplementary Notes The configurations of the above-described embodiments may be combined or some of the components may be replaced. The configuration of the present invention is not limited to the above-described embodiment. Various modifications may be made within the scope of the present invention without departing from the spirit and scope of the invention.

[0339] For example, in the above embodiment, the integrated health prediction program 100Z includes six types of health prediction models. Although we have decided to include Dell, including six types is just an example. It may be a single health prediction, or it may include seven or more health predictions. The health prediction program 100Z provides the results for each health symptom from Type 1 to Type N (N is an integer of 2 or more). The kth (1≦k≦N) health symptom may include a prediction model 110Z. The health prediction model 110Z for the product category to which at least one product belongs is The input is the purchasing data 150Z, and the output is the degree of the kth health symptom. A health prediction model for this kth type (1≦k≦N) health symptom11 0Z uses the purchasing data 952Z relating to the purchasing history of each of the multiple subjects as explanatory variables, The health data 954Z collected from each of the subjects, which indicates the degree of health status, is used as the dependent variable. It can be built by machine learning.

[0340] In addition, when the degree of health symptoms is expressed numerically, the upper limit and The upper and lower limit values ​​are merely examples and may be changed as appropriate. The number of classes to be defined can also be determined arbitrarily by those skilled in the art. It can be determined.

[0341] As will be readily appreciated by those skilled in the art having the benefit of this disclosure, the various examples described above may be inconsistent. As long as they do not create a shield, they can be used in various combinations with each other. do.

[0342] Considering the many possible embodiments to which the principles of the invention disclosed herein may be applied, The various illustrated embodiments are merely preferred examples and are not intended to limit the scope of the invention as claimed. It is understood that the technical scope should not be considered limited to these preferred examples. In fact, the technical scope of the claimed invention is determined by the attached claims. Therefore, all that falls within the technical scope of the invention described in the claims The inventors hereby claim a patent for the above invention. [Explanation of symbols]

[0343] 1. Communication Systems 2. Communication network (communication line) 10, 10A, 10B Server equipment 20, 20A, 20B terminal equipment 11, 21 Central Processing Unit (CPU)

Claims

1. When executed by at least one processor, obtaining target behavior data identifying a history of behaviors performed by a target user; By inputting the target behavior data into an estimation model generated by performing supervised learning, target condition data identifying a hydration state of the target user is output from the estimation model, the target condition data including at least one of data identifying the serum Na value of the target user, data identifying the urine osmolality of the target user, data identifying the urine specific gravity of the target user, data identifying the urine color of the target user, data identifying the BUN / creatinine ratio of the target user, data identifying a dehydration assessment scale of the target user, and data identifying the presence or absence of subjective symptoms associated with dehydration of the target user.

20. A computer program product for causing the at least one processor to function in accordance with claim 19.

2. The estimation model is generated by inputting a plurality of sets of training data into a learning model, the training data including sample behavior data that identifies a history of behaviors performed by one sample user corresponding to the set, and sample status data that identifies a hydration status of one sample user corresponding to the set.

2. The computer program product of claim 1, further comprising:

3. A plurality of sets of training data are input to a learning model, each set of training data including sample behavior data identifying a history of behaviors performed by one sample user corresponding to the set, and sample status data identifying a hydration status of one sample user corresponding to the set, and the training data is connected to the estimation model via a communication line.

2. The computer program product of claim 1, further comprising:

4. Acquire a plurality of sets of training data, each set of training data including sample behavior data identifying a history of behaviors performed by one sample user corresponding to the set, and sample status data identifying a hydration status of one sample user corresponding to the set; The plurality of sets of training data are input to a learning model for learning, thereby generating the estimation model.

2. The computer program product of claim 1, further comprising:

5. determining at least one target product and / or at least one target service corresponding to the target status data from among a plurality of products and / or a plurality of services; 5. A computer program product according to claim 1, which causes the at least one processor to function in such a way that:

6. inputting the target state data into another estimation model generated by performing supervised learning, and outputting proposal data identifying the at least one target product and / or the at least one target service from the other estimation model; 6. The computer program product of claim 5, further comprising: a processor configured to:

7. inputting the target condition data into a search table that associates the target condition data with the plurality of products and / or the plurality of services, thereby acquiring proposal data that identifies the at least one target product and / or the at least one target service from the search table; 6. The computer program product of claim 5, further comprising: a processor configured to:

8. The computer program product of claim 1 , wherein the at least one processor comprises a central processing unit (CPU), a microprocessor, and / or a graphics processing unit (GPU).

9. at least one processor; The at least one processor: obtaining target behavior data identifying a history of behaviors performed by a target user; By inputting the target behavior data into an estimation model generated by performing supervised learning, target condition data identifying a hydration state of the target user is output from the estimation model, the target condition data including at least one of data identifying the serum Na value of the target user, data identifying the urine osmolality of the target user, data identifying the urine specific gravity of the target user, data identifying the urine color of the target user, data identifying the BUN / creatinine ratio of the target user, data identifying a dehydration assessment scale of the target user, and data identifying the presence or absence of subjective symptoms associated with dehydration of the target user. An information processing device characterized by being configured as follows.

10. The information processing device according to claim 9, which is a terminal device or a server device.

11. 11. The information processing device according to claim 9 or 10, wherein the at least one processor comprises a central processing unit (CPU), a microprocessor, and / or a graphics processing unit (GPU).

12. 1. A method performed by at least one processor executing computer readable instructions, comprising: The at least one processor executes the instructions to: acquiring target behavior data identifying a history of behaviors performed by a target user; inputting the target behavior data into an estimation model generated by performing supervised learning, and outputting from the estimation model target condition data identifying a hydration state of the target user, the target condition data including at least one of data identifying the serum Na value of the target user, data identifying the urine osmolality of the target user, data identifying the urine specific gravity of the target user, data identifying the urine color of the target user, data identifying the BUN / creatinine ratio of the target user, data identifying a dehydration assessment scale of the target user, and data identifying the presence or absence of subjective symptoms associated with dehydration in the target user; A method comprising:

13. The method of claim 12 , wherein the at least one processor comprises a central processing unit (CPU), a microprocessor, and / or a graphics processing unit (GPU).

14. 1. A method performed by at least one processor executing computer readable instructions, comprising: The at least one processor executes the instructions to: an acquisition step of acquiring a plurality of sets of training data, each set of training data including sample behavior data identifying a history of behaviors performed by one sample user corresponding to the set, and sample status data identifying a hydration status of one sample user corresponding to the set; The plurality of sets of training data are input into a learning model for learning, an estimation model configured to input target behavior data identifying a history of behaviors performed by a target user, and output target condition data identifying a hydration state of the target user, the target condition data including at least one of data identifying a serum Na value of the target user, data identifying a urine osmolality of the target user, data identifying a urine specific gravity of the target user, data identifying a urine color of the target user, data identifying a BUN / creatinine ratio of the target user, data identifying a dehydration assessment scale of the target user, and data identifying the presence or absence of subjective symptoms associated with dehydration in the target user; a generation stage for generating A method comprising:

15. an estimation step of inputting target behavior data identifying a history of behaviors performed by a target user into the estimation model, and outputting target state data identifying a hydration state of the target user from the estimation model; 15. The method of claim 14, further comprising:

16. The method of claim 14 or claim 15, wherein the at least one processor comprises a central processing unit (CPU), a microprocessor, and / or a graphics processing unit (GPU).

Images