Prediction device, prediction method, and program

Abstract

A prediction device for predicting the health-oriented behavior tendency of a subject for prediction, the prediction device comprising: a health-oriented behavior estimation model storage unit for storing a health-oriented behavior estimation model for estimating a health-oriented behavior on the basis of the relationship between a health-oriented behavior and a personal characteristic; a personal characteristic criterion computing unit for computing a plurality of personal characteristic criteria on the basis of personal characteristic data indicating a personal characteristic of the subject for prediction; and a health-oriented behavior prediction unit that uses the health-oriented behavior estimation model to predict a health-oriented behavior tendency based on the plurality of personal characteristic criteria that have been computed.

Description

[Technical field]

[0001] The present invention relates to a prediction device, a learning device, a prediction method, a learning method, and a program. [Background technology]

[0002] It is an important issue to understand the series of goal-oriented actions that individuals take to achieve their own health. In recent years, with the widespread use of healthcare applications and fitness devices, it has become possible to observe various logs related to the behavioral process of people to become healthier on a large scale. For example, in many weight management applications, a user first sets a target weight and then records his / her weight for a certain period of time. In this case, data related to a series of goal-oriented actions in weight management can be measured from the recorded data, such as the user's initial weight (initial value), the target weight relative to the current value (goal difficulty), the actual weight loss relative to the current value (degree of execution), and the actual weight loss relative to the target weight (degree of achievement).

[0003] Based on such logs, technology has been developed to analyze the relationship between health-conscious behavior and human characteristics in order to understand the behavioral patterns that lead users to a healthier state. Here, human characteristics refer to properties that a user has inherently or statically at a certain point in time, and are mainly detected by self-report questionnaire surveys, etc. As conventional technology, methods have been devised to estimate health-conscious behavior, such as the height of a diet goal (Non-Patent Document 1) or actual weight loss (Non-Patent Document 2), from human characteristics such as gender, age, and BMI (Body Mass Index). [Prior art documents] [Non-patent literature]

[0004] [Non-Patent Document 1] Mitchell Gordon, Tim Althoff, and Jure Leskovec. 2019. Goal-setting And Achievement In Activity Tracking Apps: A Case Study Of MyFitnessPal. In The World Wide Web Conference (WWW '19). [Non-Patent Document 2] Zhiwei Wang, Tyler Derr, Dawei Yin, and Jiliang Tang. 2017. Understanding and Predicting Weight Loss with Mobile Social Networking Data. In Proceedings of the 2017 ACM on Conference on Information and Knowledge Management (CIKM '17). Summary of the Invention [Problem to be solved by the invention]

[0005] However, the above conventional technologies place emphasis on the process by which the user achieves the goal, and are unable to evaluate the degree to which the state reached by achieving the goal is retained by the user afterwards (the degree of retention).In addition, although basic information such as age, gender, and BMI are taken into account as human characteristics for predicting the health-conscious behavior of the user, other information such as demographic information, psychological characteristics, cognitive characteristics, health habits, and work productivity are not included, and sufficient prediction accuracy based on human characteristics has not been achieved.

[0006] The disclosed technology aims to improve the accuracy of predicting health-conscious behavioral trends. [Means for solving the problem]

[0007] The disclosed technology is a prediction device for predicting a health-conscious behavioral tendency of a prediction target, comprising: It was trained to take a personal characteristic scale based on the scale of answer values ​​to questions as input and output health-conscious behavior tendencies.a health-conscious behavior estimation model storage unit for storing a health-conscious behavior estimation model; a human characteristic data preprocessing unit that converts the human characteristic data, which indicates the human characteristics of a prediction target, into a predetermined format according to the nature of the human characteristic data as a scale of answer values ​​to questions included in the human characteristic data; a human characteristic scale calculation unit that calculates a plurality of human characteristic scales based on the human characteristic data; and By inputting the above into the health-conscious behavior estimation model, A prediction device comprising: a health-conscious behavior prediction unit that predicts a tendency of health-conscious behavior. Effect of the Invention

[0008] The accuracy of predicting health-conscious behavioral trends can be improved. [Brief description of the drawings]

[0009] [Figure 1] FIG. 2 is a functional configuration diagram of the learning device. [Diagram 2] FIG. 2 is a functional configuration diagram of a prediction device. [Diagram 3] 13 is a flowchart showing an example of the flow of a learning process. [Figure 4] 13 is a flowchart showing an example of the flow of a prediction process. [Diagram 5] FIG. 4 is a diagram showing an example of health goal data. [Figure 6] FIG. 4 is a diagram showing an example of health behavior data. [Figure 7] FIG. 13 is a diagram illustrating an example of preprocessed health behavior data. [Figure 8] FIG. 13 is a diagram showing an example of a calculation result of a health-conscious behavioral tendency. [Figure 9] FIG. 11 is a diagram showing an example of human characteristic data. [Figure 10] FIG. 13 is a diagram illustrating an example of preprocessed human characteristic data. [Figure 11] FIG. 1 is a diagram illustrating an example of a personal characteristic scale. [Figure 12] FIG. 2 illustrates an example of a hardware configuration of a computer. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0010] Hereinafter, an embodiment of the present invention (the present embodiment) will be described with reference to the drawings. The embodiment described below is merely an example, and the embodiment to which the present invention is applied is not limited to the following embodiment.

[0011] In the text of this specification, for convenience of description, the character "^" is placed before the character. "^r" is an example.

[0012] (Outline of this embodiment) The prediction device according to the present embodiment estimates the initial value, goal difficulty, execution level, achievement level, and fixation level of the scale of the health-conscious behavior of the prediction target (person) based on the human characteristic data. Also, the learning device according to the present embodiment learns the parameters of the health-conscious behavior estimation model used by the prediction device based on the data of the learning target (person). The human characteristic data is data indicating the human characteristics of the prediction target or learning target, and refers to the inherent or static properties that the prediction target or learning target has at a certain point in time, and is mainly data detected by a self-report questionnaire survey or the like.

[0013] (Example of functional configuration of learning device) 1 is a functional block diagram of a learning device. Learning device 10 includes health goal data storage unit 101, health behavior data storage unit 102, health behavior data preprocessing unit 103, health conscious behavior tendency calculation unit 104, personal characteristic data storage unit 105, personal characteristic data preprocessing unit 106, personal characteristic scale calculation unit 107, health conscious behavior estimation model construction unit 108, health conscious behavior estimation model learning unit 109, and health conscious behavior estimation model storage unit 110.

[0014] Health goal data storage unit 101 stores health goal data in which the target value in the health management of the learning subject, the time when the target value was set (declared time), and the period (goal period) self-declared by the learning subject as the time required to achieve the target value are linked with a user ID. The user ID is an identifier for identifying the learning subject.

[0015] The health behavior data storage unit 102 stores health behavior data in which a log of physical information (eg, weight, body fat percentage, BMI, etc.) that is the subject of health management to be learned is linked to a user ID along with the recording time.

[0016] The personal characteristic data storage unit 105 stores personal characteristic data in which answers to questions in a self-report questionnaire survey or the like for grasping the personal characteristics of the learning subject are linked to a user ID.

[0017] The health behavior data preprocessing unit 103 preprocesses the health behavior data. Specifically, the health behavior data preprocessing unit 103 determines a period to be analyzed based on a target period recorded in the health goal data, and extracts health behavior data corresponding to the determined period. The extracted health behavior data is called preprocessed health behavior data.

[0018] The health-conscious behavioral tendency calculation unit 104 calculates a value indicating a health-conscious behavioral tendency based on the health goal data and the preprocessed health behavior data.

[0019] The human characteristic data preprocessing unit 106 performs preprocessing on the human characteristic data. Specifically, the human characteristic data preprocessing unit 106 converts the human characteristic data into a predetermined format in accordance with the nature of the response values ​​to the questions in the questionnaire survey as a scale. The converted human characteristic data is called preprocessed human characteristic data.

[0020] The human characteristic measure calculation unit 107 calculates a plurality of human characteristic measures based on the preprocessed human characteristic data.

[0021] The health-conscious behavior estimation model construction unit 108 constructs a health-conscious behavior estimation model. The health-conscious behavior estimation model is an estimation model for estimating health-conscious behavior based on the relationship between the health-conscious behavior and human characteristics.

[0022] The health-conscious behavior inference model learning unit 109 learns about the health-conscious behavior inference model based on the value indicating the health-conscious behavior tendency calculated by the health-conscious behavior tendency calculation unit 104 and the specific human characteristic scale calculated by the human characteristic scale calculation unit 107, and outputs learned parameters of the health-conscious behavior inference model.

[0023] The health-conscious behavior inference model storage unit 110 stores the health-conscious behavior inference model constructed by the health-conscious behavior inference model construction unit 108 and parameters of the health-conscious behavior inference model learned by the health-conscious behavior inference model learning unit 109.

[0024] (Example of functional configuration of a prediction device) 2 is a functional configuration diagram of the prediction device 20. The prediction device 20 includes a personal characteristic data preprocessing unit 201, a personal characteristic scale calculation unit 202, a health-conscious behavior estimation model storage unit 203, and a health-conscious behavior prediction unit 204.

[0025] The human characteristic data pre-processing unit 201 performs pre-processing on the human characteristic data of the prediction target in the same manner as the human characteristic data pre-processing unit 106 of the learning device 10, and outputs the pre-processed human characteristic data.

[0026] The human characteristic measure calculation unit 202 calculates a plurality of human characteristic measures based on the preprocessed human characteristic data of the prediction target by calculations similar to those performed by the human characteristic measure calculation unit 107 of the learning device 10.

[0027] The health-conscious behavior estimation model storage unit 203 stores the health-conscious behavior estimation model constructed by the health-conscious behavior estimation model construction unit 108 and parameters of the health-conscious behavior estimation model learned by the health-conscious behavior estimation model learning unit 109, which are stored in the health-conscious behavior estimation model storage unit 110 of the learning device 10.

[0028] The health-conscious behavior prediction unit 204 uses the health-conscious behavior estimation model and parameters of the learned health-conscious behavior estimation model to predict a tendency of health-conscious behavior based on the multiple human characteristic scales calculated by the human characteristic scale calculation unit 202, and outputs data indicating the prediction result.

[0029] (Example of learning device operation) Next, a description will be given with reference to the drawings of an example of the operation of the learning device 10. The learning device 10 starts a learning process in response to a user operation or the like, or periodically.

[0030] 3 is a flowchart showing an example of the flow of the learning process. The health action data preprocessing unit 103 receives and processes the health action data from the health action data storage unit 102 and the health goal data from the health goal data storage unit 101 (step S100). Specifically, the health action data preprocessing unit 103 determines a period to be analyzed based on a goal period recorded in the health goal data, and extracts health action data corresponding to the determined period. The extracted health action data is regarded as preprocessed health action data.

[0031] Next, the health-conscious behavioral tendency calculation unit 104 receives and processes the health goal data from the health goal data storage unit 101 and the preprocessed health behavior data from the health behavior data preprocessing unit 103 (step S110). Specifically, the health-conscious behavioral tendency calculation unit 104 calculates a value indicating the health-conscious behavioral tendency based on the health goal data and the preprocessed health behavior data.

[0032] Next, the human characteristic data preprocessing unit 106 receives and processes the human characteristic data from the human characteristic data storage unit 105 (step S120). Specifically, the human characteristic data preprocessing unit 106 converts the human characteristic data into a predetermined format in accordance with the nature of the data as a scale of response values ​​to questions in the questionnaire survey. The converted human characteristic data is regarded as preprocessed human characteristic data.

[0033] Next, the human characteristic measure calculation unit 107 receives and processes the preprocessed human characteristic data from the human characteristic data preprocessing unit 106 (step S130). Specifically, the human characteristic measure calculation unit 107 calculates a specific human characteristic measure based on the preprocessed human characteristic data.

[0034] Next, the health-conscious behavior estimation model construction unit 108 constructs a model (step S140). The constructed model is a health-conscious behavior estimation model for estimating health-conscious behavior based on the relationship between the health-conscious behavior and human characteristics.

[0035] Next, the health-conscious behavior estimation model learning unit 109 receives the health-conscious behavior tendency from the health-conscious behavior tendency calculation unit 104, the personal characteristic scale from the personal characteristic scale calculation unit 107, and the model from the health-conscious behavior estimation model construction unit, learns the model, and stores the learned model in the health-conscious behavior estimation model storage unit (step S150). Specifically, the health-conscious behavior estimation model learning unit 109 learns the health-conscious behavior estimation model based on the value indicating the health-conscious behavior tendency calculated by the health-conscious behavior tendency calculation unit 104 and the specific personal characteristic scale calculated by the personal characteristic scale calculation unit 107, and stores the learned parameters of the health-conscious behavior estimation model in the health-conscious behavior estimation model storage unit 110.

[0036] (Example of the operation of the prediction device) Next, a description will be given with reference to the drawings of an example of the operation of the prediction device 20. The prediction device 20 starts a prediction process in response to a user operation or the like, or periodically.

[0037] 4 is a flowchart showing an example of the flow of the prediction process. The human characteristic data pre-processing unit 201 receives and processes human characteristic data as input (step S200). Specifically, similar to the processing of step S120 of the learning processing by the learning device 10, the human characteristic data pre-processing unit 201 converts the human characteristic data into a predetermined format in accordance with the nature of the response values ​​as a scale for the question items in the questionnaire survey. The converted human characteristic data is regarded as pre-processed human characteristic data.

[0038] Next, the human characteristic measure calculation unit 107 receives and processes the preprocessed human characteristic data from the human characteristic data preprocessing unit 106 (step S201). Specifically, the human characteristic measure calculation unit 107 calculates a specific human characteristic measure based on the preprocessed human characteristic data, similar to the processing in step S130 of the learning processing by the learning device 10.

[0039] Next, the health-conscious behavior prediction unit 204 receives the personal characteristic scale from the personal characteristic scale calculation unit 202 and the trained model from the health-conscious behavior estimation model storage unit 203, calculates the health-conscious behavior tendency according to the model, and outputs the calculation result (step S202). Specifically, the health-conscious behavior prediction unit 204 predicts the health-conscious behavior tendency based on the specific personal characteristic scale calculated by the personal characteristic scale calculation unit 202 using the health-conscious behavior estimation model and parameters of the trained health-conscious behavior estimation model, and outputs data indicating the prediction result.

[0040] (Examples of various data formats according to this embodiment) Next, examples of formats of various data handled by the learning device 10 and the prediction device 20 according to this embodiment will be described.

[0041] 5 is a diagram showing an example of health goal data 901. Health goal data 901 includes items such as a user ID, a goal value, a declared time, and a goal period.

[0042] The value of the item "User ID" is an identifier for identifying the learning subject. The value of the item "Target value" is the target value for the learning subject's health management. The value of the item "Declared time" is the time when the learning subject set the target value. The value of the item "Target period" is the period self-declared by the learning subject as the time required to achieve the target value.

[0043] 6 is a diagram showing an example of health behavior data 902. Health behavior data 902 includes items such as a user ID, a log ID, a recording time, and weight.

[0044] The value of the item "User ID" is an identifier for identifying the subject of learning. The value of the item "Log ID" is an identifier for identifying the log. The value of the item "Recorded Time" is the time when the log was recorded. The value of the item "Weight" is the weight of the subject of learning as an example of physical information that is the subject of health management.

[0045] 7 is a diagram showing an example of preprocessed health action data 903. Preprocessed health action data 903 includes items such as a user ID, a log ID, a recording time, and weight.

[0046] The value of the item "User ID" is an identifier for identifying the learning subject. The value of the item "Log ID" is an identifier for identifying the log. Note that the value of the item "Log ID" is reassigned in preprocessing by the health behavior data preprocessing unit 103, and therefore may not necessarily match the value of the item "Log ID" of the health behavior data 902 shown in FIG. 6. The value of the item "Recording time" is the time when the log was recorded. The value of the item "Weight" is the weight as an example of physical information that is the subject of health management for the learning subject.

[0047] 8 is a diagram showing an example of a calculation result of a health-conscious behavioral tendency. Each value of the calculation result 904 is a value indicating a health-conscious behavioral tendency calculated by the health-conscious behavioral tendency calculation unit 104. Specifically, the calculation result 904 is a value indicating a plurality of health-conscious behavioral tendencies including a current value for each user ID, a goal difficulty level, an execution frequency, an execution intensity, an achievement frequency, an achievement speed, a degree of establishment, and the like.

[0048] 9 is a diagram showing an example of human characteristic data. The human characteristic data 905 is data in which answers to questions in a self-report questionnaire survey or the like for grasping the human characteristics of a prediction target or a learning target are linked to a user ID.

[0049] For example, the example in Figure 9 is an answer to the following question:

[0050] Q1_1-Q1_5: Questions about demographic information (Q1_1: Gender, Q1_2: Age, Q1_3: BMI, Q1_4: Marital status, Q1_5: Number of people in household) Q2_1, Q2_2: Questions about psychological characteristics (Q2_1: "Do you like talking to people?", Q2_2: "Are you bad at being with a large group of people?" / Answer options: Not at all, Not at all, Neither, Yes, Very much) Q3_1, Q3_2: Questions about cognitive characteristics (Q3_1: "If there is a lottery ticket with a 50% chance of winning 100,000 yen, how much would you be willing to pay to buy the ticket?", Q3_2: "If there is a probability of precipitation, what percentage or more would make you bring an umbrella?") Q4_1, Q4_2: Questions about health habits (Q4_1: "Have you had trouble sleeping at night recently?", Q4_2: "Do you ever eat greasy foods?" / Answer options: never, rarely, sometimes, often) Q5_1, Q5_2: Questions about work productivity (Q5_1: "Please rate your work performance over the past year on a scale of 0 to 10" / Q5_2: "Please rate your work performance over the past 4 weeks on a scale of 0 to 10")

[0051] 10 is a diagram showing an example of preprocessed human characteristic data 906. Preprocessed human characteristic data 906 is obtained by converting human characteristic data 905 into a predetermined format.

[0052] For example, the example in FIG. 10 is obtained by preprocessing the human characteristic data 905 shown in FIG. 9 by the human characteristic data preprocessing unit 106 according to the following correspondence relationships.

[0053] Gender:(Q1_1 * :Q1_1)=(0:female) or (1:male) Age:Q1_2 * =Q1_2 BMI:Q1_3 * =Q1_3 Marital status:(Q1_4 *:Q1_4) = (0: unmarried) or (1: married) Number of people in household:Q1_5 * =Q1_5 Psychological characteristics: (Q2_1 * :Q2_1) = (1: Not at all), (2: Not at all), (3: Neither), (4: Yes) or (5: Very much so) (Q2_2 * :Q2_2) = (5: Not at all), (4: Not at all), (3: Neither), (2: Yes) or (1: Very much so) Cognitive characteristics: Q3 * =Q3 Both Q3_1 and Q3_2 follow the above transformation.

[0054] Health habits:(Q4 * :Q4) = (0: never), (1: rarely), (2: sometimes), or (3: often) Both Q4_1 and Q4_2 follow the above transformation.

[0055] Work Productivity: Q5 * =Q5 Both Q5_1 and Q5_2 follow the above transformation.

[0056] Fig. 11 is a diagram showing an example of a human characteristic scale. A calculation result 907 of the human characteristic scale includes various values ​​indicating the human characteristic scale calculated by the human characteristic scale calculation unit 107. For example, the calculation result 907 shown in Fig. 11 is calculated by the following calculation based on the preprocessed human characteristic data 906 shown in Fig. 10.

[0057] Gender:Q1_1 * Age:Q1_2 * BMI:Q1_3 * Marriage:Q1_4 * Housemate:Q1_5 * Extra version: (Q2_1 * +Q2_2 * ) / 2 Risk aversion:1-Q3_1 * / 50000 Note that the denominator in the second term of the above equation is the expected reward obtained from this lottery.

[0058] Risk taking:1-Q3_2 * / 100 The second term in the above equation is the value obtained by converting the reported probability of precipitation (%) into a probability value.

[0059] Sleep:Q4_1 * Oil:Q4_2 * Work performance:(Q5_1 * +Q5_2 * ) / 2

[0060] (Details of each process according to this embodiment) Next, the learning process and prediction process according to the present embodiment will be described in detail. i Then, the health behavior data vector is y = (y0, ,y N-1 ) T , y i The time when i In addition, the target value of the health goal data is g, and the reporting time is t g , and the goal period is s. Here, the health goal data is assumed to remain constant unless it is updated.

[0061] When health goal data is updated during a goal period, or when new health goal data is set after the end of a goal period, the learning device 10 updates the target value, declared time, and goal period, and then executes the processing described below.

[0062] The process by the health behavior data preprocessing unit 103 (step S100 of the learning process shown in FIG. 3) will be described. The health behavior data preprocessing unit 103 performs the process ofg From g In other words, the preprocessed health behavior data vector y obtained by the health behavior data preprocessing unit 103 is extracted. * is calculated by the following formula (1).

[0063]

number

[0064] Next, the process performed by the health-conscious behavioral tendency calculation unit 104 (step S110 of the learning process shown in FIG. 3) will be described. The health-conscious behavioral tendency calculation unit 104 calculates the preprocessed health behavior data vector y * and the target value g to the current value r (s) , target difficulty r (d) , execution degree r (p) , achievement level r (a) , Fixation rate r (f) Extract each of them.

[0065] Among these, the current state value refers to the first recorded data among the preprocessed behavioral data. The current state value of the learning object u is calculated by the following formula (2).

[0066]

number

[0067] In addition, the target difficulty is the current value r u (s) The health-conscious behavioral tendency calculation unit 104 calculates the goal difficulty level by the following formula (3).

[0068]

number

[0069]

number

[0070] Moreover, the degree of execution is a value expressing the degree to which an action was actually taken toward the target value during the target period until the target value was achieved. Here, as an example, the degree of execution is composed of the frequency of execution and the intensity of execution. As long as the definition of the degree of execution is followed, the scales that make up the degree of execution are not limited to these. The frequency of execution refers to the frequency at which an action was taken toward the target value, and the intensity of execution refers to the average amount of change per action toward the target value. For example, assuming that the Lth log is the log output when the target value was achieved for the first time, the health-conscious behavior tendency calculation unit 104 calculates the preprocessed health behavior data vector y * Let y be the vector obtained by extracting the first L elements from p Let y p is calculated by the following formula (4).

[0071]

number

[0072] In this case, the execution frequency r (p,f) is calculated by the following formula (5).

[0073]

number

[0074] Also, the execution intensity r (p,s) is calculated by the following formula (6).

[0075]

number

[0076]

number

[0077]

number

[0078] Moreover, the degree of achievement is a value that expresses the degree to which the set target value has been achieved. In this example, the degree of achievement is composed of the frequency of achievement and the speed of achievement. As long as the definition of the degree of achievement is followed, the scales that make up the degree of achievement are not limited to these. The frequency of achievement refers to the frequency with which the value that achieved the target value was recorded during the target period. The speed of achievement refers to the speed at which the target value is achieved for the first time. Frequency of achievement r (a,f) is calculated by the following formula (7).

[0079]

number

[0080] Also, if we assume that the Lth log is the log output when the target value is achieved for the first time, then the achieved speed r (a,v) is calculated by the following formula (8).

[0081]

number

[0082]

number

[0083] The degree of fixation is a value indicating the degree to which a state close to the target value is maintained during the target period. * The health-conscious behavior tendency calculation unit 104 calculates the degree of establishment by evaluating the degree of establishment r (f) y * It is calculated using a kernel density function based on the frequency distribution of

[0084] The health-conscious behavior tendency calculation unit 104 calculates y0 * ,···,y N-1 * Assuming that the samples are obtained from independent identical distributions with probability density function f, the kernel density estimate of the kernel function K and smoothing parameter h is calculated as shown in the following equation (9).

[0085]

number

[0086]

number

[0087] The kernel function is

[0088]

number

[0089] The health-conscious behavior tendency calculation unit 104 calculates the current state value r (s) , target difficulty r (d) , execution degree r(p) , achievement level r (a) and fixation rate r (f) to the health-conscious behavior inference model learning unit 109.

[0090] Next, a description will be given of step S120 of the learning process by the human characteristic data pre-processing unit 106 of the learning device 10. The process described below is also the same as step S200 of the prediction process by the human characteristic data pre-processing unit 201 of the prediction device 20.

[0091] The personal characteristic data preprocessing unit 106 converts the answer values ​​according to the nature of the scale of the answers to the questions in the questionnaire survey. The scale of the answers includes nominal scale, ordinal scale, interval scale, and ratio scale, and each is converted as follows.

[0092] The nominal scale is a scale used to distinguish attributes and categories. In the example of the human characteristic data 905 shown in Fig. 9, the answer values ​​for gender (Q1_1) and marital status (Q1_4) correspond to the nominal scale. The human characteristic data pre-processing unit 106 converts the answer values ​​corresponding to the nominal scale into values ​​preset in the learning device 10.

[0093] For example, the human characteristic data preprocessing unit 106 converts "female" to 0 and "male" to 1. Note that the conversion method is not limited to the above as long as the attributes and categories corresponding to the answer values ​​can be distinguished even after conversion.

[0094] The ordinal scale is a scale that has meaning in the magnitude relationship but not in the difference or ratio. In the example of the human characteristic data 905 shown in Fig. 9, the answer values ​​for the psychological characteristics (Q2_1, Q2_2) and health habits (Q4_1, Q4_2) correspond to this scale. The human characteristic data preprocessing unit 106 converts the answer values ​​corresponding to the ordinal scale so that the order relationship between the answer values ​​is maintained.

[0095] For example, the human characteristic data preprocessing unit 106 converts "not at all" to 1, "not" to 2, "neither" to 3, "true" to 4, and "very true" to 5. Note that the conversion method is not limited to the above as long as the order of the answer values ​​is maintained even after conversion.

[0096] Interval scales and ratio scales are called quantitative data and are scales with equal intervals. In the example of the personal characteristic data 905 shown in Fig. 9, the answer values ​​for age (Q1_2), number of people in the household (Q1_5), cognitive characteristics (Q3_1, Q3_2), and work productivity (Q5_1, Q5_2) correspond to interval scales or ratio scales. The personal characteristic data preprocessing unit 106 does not convert these answer values.

[0097] Next, a description will be given of the process of step S130 of the learning process by the human characteristic measure calculation unit 107. The process described below is also the same as the process of step S201 of the prediction process by the human characteristic measure calculation unit 202 of the prediction device 20.

[0098] The human characteristic scale calculation unit 107 calculates the human characteristic scale of the learning target using the converted answer value obtained by the human characteristic data pre-processing unit 106. The calculation method follows a method predetermined in the learning device 10.

[0099] For example, in the example shown in FIG. 11, the human characteristic scale calculation unit 107 calculates a human characteristic scale called "Extraversion" as Q2_1 in the preprocessed human characteristic data. * and Q2_2 * It is calculated by taking the average of the responses.

[0100] For example, assuming that the human characteristic scale calculation unit 107 has calculated the human characteristic scales of M items, the human characteristic scale of item i is calculated as x i Then, the personal characteristic scale vector x of each user is expressed as x i |i=1,...,M) T Output.

[0101] The process of step S140 of the learning process by the health-conscious behavior estimation model construction unit 108 will be described. The health-conscious behavior estimation model aims to predict the health-conscious behavior tendency from the personal characteristic scale, and can be constructed using any method that follows the supervised learning method. Linear regression is used here as an example. Linear regression generally uses a linear regression model in which the objective variable is y and the explanatory variable is x i , x i The coefficient of β i , where α is the constant term and ε is the error term, it is written as in the following equation (11).

[0102]

number

[0103]

number

[0104] Here, the health-conscious behavior tendency r of user u is h (h=(s),(d),(p,f),(p,s),(a,f),(a,v),(f)) as r u h Let x be the personal characteristic scale of user u. u,i A vector consisting of x u =(1,x u,1 ,···,x u,M ) T Health-conscious behavioral trends h The constant term in the regression of h , the coefficient of the personal characteristic scale of item i is β i h Let the parameter vector with these elements be β h =(α h ,β1 h , ,β Mh ) T Health-conscious behavioral trends h The error term for user u in the regression of u h These variables are written as in equation (13) based on equation (12).

[0105]

number

[0106]

number

[0107]

number

[0108] The health-conscious behavior inference model construction unit 108 constructs the health-conscious behavior inference model as the following formula (15).

[0109]

number

[0110] Next, the process of step S150 of the learning process by the health-conscious behavior estimation model learning unit 109 will be described. The health-conscious behavior estimation model learning unit 109 receives the health-conscious behavior tendency r h (h=(s),(d),(p,f),(p,s),(a,f),(a,v),(f)), a personal characteristic scale vector x is received from the personal characteristic scale calculation unit 107, and f(X) is received from the health-conscious behavior estimation model construction unit 108. Parameter β h The learning procedure is explained below.

[0111] The health-conscious behavior estimation model learning unit 109 calculates the personal characteristic scale x of the item i of the user u. u,iis standardized according to the following equation (16).

[0112]

number

[0113] Next, the health-conscious behavior estimation model learning unit 109 obtains parameters that minimize the error between the estimation result of the health-conscious behavior estimation model and the true value. An example of calculating the parameters by the least squares method is shown. The health-conscious behavior estimation model learning unit 109 calculates parameters by using the element x u,i Z u,i The standardized human characteristic matrix replaced by is denoted as Z, and is constructed as shown in the following equation (17).

[0114]

number

[0115] The health-conscious behavior tendency vector estimated from the standardized human characteristic matrix Z is expressed as ^r h Then, the estimation error vector ε h is written as the following equation (18).

[0116]

number

[0117] The health-conscious behavior estimation model learning unit 109 calculates the error vector ε h The parameter β that minimizes h is obtained by solving the optimization problem shown in Eq. (19) below.

[0118]

number

[0119] Therefore, the following equation (20) is obtained, and the parameter β for the health-conscious behavior tendency h is h is learned.

[0120]

number

[0121] Although an example of learning parameters using the least squares method has been given, the method of learning parameters is not limited to the above as long as the method is a method of obtaining parameters that minimize the error between the estimation result of the health-conscious behavior estimation model and the true value.

[0122] The health-conscious behavior estimation model learning unit 109 uses the obtained parameter β h into the health-conscious behavior inference model f, and the health-conscious behavior inference model into which the parameters have been substituted is stored in the health-conscious behavior inference model storage unit 110.

[0123] Next, the process of step S202 of the prediction process by the health-conscious behavior prediction unit 204 will be described. It is assumed that the linear regression model f(X) is stored in the health-conscious behavior estimation model storage unit 203 as a health-conscious behavior estimation model. The health-conscious behavior prediction unit 204 predicts the personal characteristic scale x′ u,i Once we obtain the health-seeking behavior estimation model f(X), we obtain the personal characteristic scale x′ u,i is transformed into the following equation (21).

[0124]

number

[0125]

number

[0126] Estimation results of health-conscious behavior^r h =(^r h |u=1,...,N) T is the health-seeking behavior estimation model f(X) and the trained parameter vector β h =(Z T Z) -1 Z T r h From this, the following equation (23) is obtained.

[0127]

number

[0128] (Example of Hardware Configuration According to the Present Embodiment) The learning device 10 and the prediction device 20 can be realized, for example, by making a computer execute a program describing the processing contents described in this embodiment. Note that this "computer" may be a physical machine or a virtual machine on the cloud. When a virtual machine is used, the "hardware" described here is virtual hardware.

[0129] The above program can be recorded on a computer-readable recording medium (such as a portable memory) and can be stored or distributed. The above program can also be provided via a network such as the Internet or e-mail.

[0130] Fig. 12 is a diagram showing an example of the hardware configuration of the computer. The computer in Fig. 12 includes a drive device 1000, an auxiliary storage device 1002, a memory device 1003, a CPU 1004, an interface device 1005, a display device 1006, an input device 1007, an output device 1008, and the like, all of which are interconnected via a bus B.

[0131] The program for realizing the processing on the computer is provided by a recording medium 1001 such as a CD-ROM or a memory card. When the recording medium 1001 storing the program is set in the drive device 1000, the program is installed from the recording medium 1001 to the auxiliary storage device 1002 via the drive device 1000. However, the program does not necessarily have to be installed from the recording medium 1001, but may be downloaded from another computer via a network. The auxiliary storage device 1002 stores the installed program as well as necessary files, data, etc.

[0132] When a program start instruction is received, the memory device 1003 reads out and stores the program from the auxiliary storage device 1002. The CPU 1004 realizes the functions related to the device according to the program stored in the memory device 1003. The interface device 1005 is used as an interface for connecting to a network. The display device 1006 displays a GUI (Graphical User Interface) or the like according to a program. The input device 1007 is composed of a keyboard, a mouse, a button, a touch panel, or the like, and is used to input various operation instructions. The output device 1008 outputs the calculation result. Note that the above computer may be equipped with a GPU (Graphics Processing Unit) or a TPU (Tensor processing unit) instead of the CPU 1004, or may be equipped with a GPU or a TPU in addition to the CPU 1004. In that case, the processes may be shared and executed, for example, the GPU or the TPU executes processes that require special calculations such as neural networks, and the CPU 1004 executes other processes.

[0133] (Effects of this embodiment) The learning device 10 according to the present embodiment performs learning taking into consideration not only basic items such as gender, age, and BMI, but also various human characteristics such as psychological characteristics, cognitive characteristics, health habits, and work productivity. This can improve the estimation accuracy of a model that estimates health-conscious behavioral tendencies. Furthermore, the prediction device 20 performs prediction taking into consideration not only basic items such as gender, age, and BMI, but also various human characteristics such as psychological characteristics, cognitive characteristics, health habits, and work productivity. This can improve the prediction accuracy of health-conscious behavioral tendencies.

[0134] Furthermore, the learning device 10 according to the present embodiment calculates various values ​​indicating the tendency of health-conscious behavior, including the degree of fixation, based on the health behavior data and the health goal data, and learns parameters of the health-conscious behavior estimation model by machine learning based on the calculated values. By evaluating various values ​​indicating the tendency of health-conscious behavior, including the degree of fixation, it is possible to evaluate health-conscious behavior from a new perspective that has not been evaluated in the past. This allows the tendency of health-conscious behavior to be evaluated from more diverse angles, and the prediction accuracy to be further improved.

[0135] (Summary of the embodiment) This specification describes at least the prediction device, learning device, prediction method, learning method, and program described in the following sections. (Section 1) A prediction device for predicting a health-conscious behavioral tendency of a prediction target, comprising: a health-conscious behavior estimation model storage unit that stores a health-conscious behavior estimation model for estimating health-conscious behavior based on a relationship between the health-conscious behavior and a personal characteristic; A human characteristic scale calculation unit that calculates a plurality of human characteristic scales based on human characteristic data indicating a human characteristic of a prediction target; a health-conscious behavior prediction unit that predicts a tendency of health-conscious behavior based on the calculated plurality of human characteristic measures by using the health-conscious behavior estimation model; Prediction device. (Section 2) a human characteristic data pre-processing unit that converts the human characteristic data into a predetermined format in accordance with a property of the human characteristic data as a scale of answer values ​​to question items included in the human characteristic data; The human characteristic scale calculation unit calculates the plurality of human characteristic scales based on the human characteristic data converted into the format. 2. The prediction device according to claim 1. (Section 3) the personal characteristic scale calculation unit converts the format of the personal characteristic data according to any one of a nominal scale, an ordinal scale, an interval scale, and a ratio scale as the scale of the response value; 3. The prediction device according to claim 2. (Section 4) A learning device for learning parameters of a health-conscious behavior estimation model for estimating health-conscious behavior, comprising: a health-conscious behavior estimation model storage unit that stores a health-conscious behavior estimation model for estimating health-conscious behavior based on a relationship between the health-conscious behavior and a personal characteristic; a health-conscious behavior tendency calculation unit that calculates a value indicating a health-conscious behavior tendency of the learning subject based on the health goal data and health behavior data of the learning subject; a human characteristic scale calculation unit that calculates a plurality of human characteristic scales based on human characteristic data indicating the human characteristics of the learning subject; a health-conscious behavior inference model learning unit that learns parameters of the health-conscious behavior inference model based on the plurality of personal characteristic measures and the value indicating the health-conscious behavior tendency. Learning device. (Section 5) the health-conscious behavioral tendency calculation unit calculates values ​​indicating a plurality of health-conscious behavioral tendencies including a degree of consolidation indicating a degree to which a state close to a target value is maintained during a target period; 5. A learning device as described in claim 4. (Section 6) A prediction method executed by a prediction device storing a health-conscious behavior estimation model for estimating health-conscious behavior based on a relationship between the health-conscious behavior and a human characteristic, comprising: Calculating a plurality of human characteristic measures based on human characteristic data indicative of a human characteristic to be predicted; and predicting a health-conscious behavior tendency based on the calculated plurality of personal characteristic measures using the health-conscious behavior prediction model. Forecasting methods. (Section 7) A learning method executed by a learning device storing a health-conscious behavior estimation model for estimating health-conscious behavior based on a relationship between the health-conscious behavior and a human characteristic, comprising: calculating a value indicating a health-conscious behavioral tendency of the learning subject based on the health goal data and health behavior data of the learning subject; Calculating a plurality of human characteristic measures based on human characteristic data indicative of the human characteristics of the learning subject; learning parameters of the health-conscious behavior estimation model based on the plurality of personal characteristic measures and the value indicating the health-conscious behavior tendency; How to learn. (Section 8) A program for causing a computer to function as each unit in the prediction device described in any one of paragraphs 1 to 3, or a program for causing a computer to function as each unit in the learning device described in paragraph 4 or 5.

[0136] Although the present embodiment has been described above, the present invention is not limited to such a specific embodiment, and various modifications and variations are possible within the scope of the gist of the present invention described in the claims. [Explanation of symbols]

[0137] 10 Learning Device 20 Prediction Device 101 Health goal data storage unit 102 Health behavior data storage unit 103 Health behavior data preprocessing unit 104 Health-oriented behavior trend calculation department 105 Human Characteristics Data Storage Unit 106 Human characteristics data preprocessing section 107 Human characteristics scale calculation part 108 Health-oriented behavior estimation model construction department 109 Health-oriented behavior estimation model learning part 110 Health-oriented behavior estimation model storage section 201 Human characteristics data preprocessing unit 202 Human characteristics scale calculation section 203 Health-oriented behavior estimation model storage section 204 Health-Oriented Behavior Prediction Department 1000 Drive Device 1001 Recording media 1002 Auxiliary storage device 1003 Memory device 1004 CPU 1005 Interface device 1006 Display device 1007 Input Device 1008 Output device

Claims

1. A prediction device for predicting a health-conscious behavioral tendency of a prediction target, comprising: a health-conscious behavior estimation model storage unit that stores a health-conscious behavior estimation model trained to output a health-conscious behavior tendency using a personal characteristic scale based on a scale of answer values ​​to question items as an input; a human characteristic data pre-processing unit that converts the human characteristic data, which indicates the human characteristics of a prediction target, into a predetermined format in accordance with the nature of the human characteristic data as a scale of answer values ​​to questions included in the human characteristic data; a human characteristic scale calculation unit that calculates a plurality of human characteristic scales based on the human characteristic data converted into the format; a health-conscious behavior prediction unit that predicts a tendency of the health-conscious behavior by inputting the calculated plurality of personal characteristic measures into the health-conscious behavior estimation model; Prediction device.

2. the personal characteristic scale calculation unit converts the format of the personal characteristic data according to any one of a nominal scale, an ordinal scale, an interval scale, and a ratio scale as the scale of the response value; The prediction device according to claim 1 .

3. A prediction method executed by a prediction device storing a health-conscious behavior estimation model trained to output a health-conscious behavior tendency using a personal characteristic scale based on a scale of answer values ​​to question items as an input, the prediction method comprising: A step of converting the human characteristic data indicating the human characteristics of the prediction target into a predetermined format according to the nature of the human characteristic data as a scale of answer values ​​to questions included in the human characteristic data; calculating a plurality of human characteristic measures based on the human characteristic data converted into the format; and predicting a health-conscious behavior tendency by inputting the calculated plurality of personal characteristic measures into the health-conscious behavior estimation model. Forecasting methods.

4. A program for causing a computer to function as each unit in the prediction device according to claim 1 or 2.

Images