Information processing device, method, and program

Abstract

The present disclosure relates to estimating a missing value of body information. An information processing device according to one embodiment of the present disclosure comprises at least one memory and at least one processor. The at least one processor executes inputting of body information in which the value of a first attribute is missing to at least one trained model to thereby acquire an estimated value of the first attribute, and inputting of body information in which the value of a second attribute different from the first attribute is missing to the at least one trained model to thereby acquire an estimated value of the second attribute, the first and second attributes representing body information other than basic information.

Description

Information processing device, method, and program

[0001] The present disclosure relates to an information processing device, a method, and a program.

[0002] Conventionally, physical information such as test results from health checkups has been utilized. Specifically, services have been provided that use values ​​of multiple attributes of physical information (e.g., results of multiple tests). However, not all individuals measure the values ​​of all attributes of physical information, and values ​​of some attributes of physical information may be missing. For example, optional tests may not be taken during a health checkup, or health checkup items may differ between facilities.

[0003] JP 2012-064087 A

[0004] Handling Incomplete Heterogeneous Data using VAEs, [online], Internet <URL: https: / / arxiv.org / pdf / 1807.03653.pdf>

[0005] The problem of this disclosure is to estimate missing values ​​of physical information.

[0006] An information processing device that is one embodiment of the present disclosure includes at least one memory and at least one processor, and the at least one processor executes the following: inputting physical information in which a value of a first attribute is missing into at least one trained model to obtain an estimated value of the first attribute; and inputting physical information in which a value of a second attribute different from the first attribute is missing into the at least one trained model to obtain an estimated value of the second attribute, wherein the first attribute and the second attribute are physical information other than basic information.

[0007] FIG. 1 is an example of an overall configuration according to an embodiment of the present disclosure; FIG. 2 is a functional block diagram of an information processing device according to an embodiment of the present disclosure; FIG. 3 is a diagram for explaining learning data according to an embodiment of the present disclosure; FIG. 4 is a diagram for explaining learning data according to an embodiment of the present disclosure; FIG. 5 is a diagram for explaining physical information including missing values ​​and physical information including estimated values ​​according to an embodiment of the present disclosure; FIG. 6 is a diagram showing an estimation example according to an embodiment of the present disclosure; FIG. 7 is a flowchart of a learning process according to an embodiment of the present disclosure; FIG. 8 is a flowchart of an estimation process according to an embodiment of the present disclosure; and FIG. 9 is a hardware configuration diagram of an information processing device according to an embodiment of the present disclosure.

[0008] Hereinafter, embodiments of the present disclosure will be described with reference to the drawings.

[0009] <Explanation of Terms> In this specification, "physical information" refers to information related to the human mind and body, and includes information that can serve as an index of physical information. For example, physical information is information collected from the results of a health checkup, medical prescription information (e.g., information regarding medical procedures performed by a medical institution on a patient, names of injuries and illnesses, medications, etc.), responses to a physical questionnaire, etc. For example, "physical information" can include "basic information," "information regarding physical condition," and "information regarding psychological condition." Note that the present disclosure can also be applied to the physical information of animals other than humans, at least with regard to "basic information" and "information regarding physical condition."

[0010] In this specification, "basic information" refers to basic information about the body that can be recognized by the person. In this embodiment, the "basic information" preferably includes gender, age, height, and weight.

[0011] In this specification, "information relating to physical condition" refers to information that indicates the physical condition of a person. For example, "information relating to physical condition" refers to information on general blood, liver function, lipids, metabolic system, blood pressure, etc., which are test items in a health checkup, and includes scores, numerical values, and classification information obtained by measuring all or part of the body, and scores, numerical values, and classification information obtained from a medical interview or questions, and does not include basic information.

[0012] For example, "information about physical condition" includes physique / body composition values ​​(body fat, muscle mass, visceral fat, etc.), blood test values ​​(triglycerides, total cholesterol, HDL cholesterol, LDL cholesterol, blood glucose, HbA1c, ALT, red blood cell count, aspartate aminotransferase, alanine aminotransferase, gamma-glutamyl transpeptidase, alkaline phosphatase), various hormones (cortisol, triiodothyronine, thyroxine, dehydroepiandrosterone, etc.), and blood glucose levels. Drososterone, testosterone, estradiol, progesterone, follicle-stimulating hormone, luteinizing hormone, prolactin, etc.), vascular function (arteriosclerosis index, lower limb arterial stenosis / occlusion index, etc.), glucose metabolism function (blood glucose, insulin, glucagon, GLP-1, GIP), cognitive function (functional test (Cognitrax (registered trademark)), blood D or L-amino acids), motor function (grip strength, gross motor ability, walking function (speed, stride length, pitch), etc.), immune-related index (lymphoid subset analysis, NK cell activity, cytokine analysis, etc.), disease information (presence or absence of disease, medical treatment, medication, etc.), activity level (calories burned per day, number of steps per day), sleep-related indicators (score from medical interview questionnaire), productivity (score from questionnaire), diet / nutritional status (calories burned, PFC balance, etc.), menstrual-related indicators (stage and disability level assessment based on medical interview and questions), menopausal disorder-related indicators (stage and disability level assessment based on questions), hair (diameter, degree of waviness), skin condition (skin blood flow rate, skin glycation level, amount of ceramide), body odor (quantification of odor components by gas chromatography), Oriental medicine constitution classification (clustering based on questionnaire), biomicrobiota information (bacterial quantity and composition ratio in the oral cavity, intestines, scalp, and skin), progression of hair thinning and alopecia (score from photographic assessment), DNA and RNA information, overactive bladder / urinary incontinence (score from questionnaire), etc.

[0013] In this specification, "information related to mental state" refers to information that indicates a person's mental state, and is information that can be assessed by interviews, questionnaires, tests, or markers, and can be scored, given numerically, or classified. For example, "information related to mental state" is information related to personality traits (Big-5 theory - openness, conscientiousness, extraversion, agreeableness, neuroticism), stress (stress response score (stress level) of the simplified version of the Occupational Stress Questionnaire), mental fatigue, mental health (bipolar disorder), happiness (score on a QOL questionnaire), etc.

[0014] In this specification, each attribute included in "information on physical condition," "information on mental condition," and "basic information" is expressed by a score, a numerical value, or a classification, and it is known that there is a correlation between each attribute. For example, "activity amount" is presumed to be correlated with "basic information" such as age, weight, and gender, or "information on physical information" related to physical conditions such as vascular function, glucose metabolism, and cognitive function, and furthermore, it is suggested to be correlated with "information on mental condition" such as stress and mental fatigue.

[0015] <Overall Configuration> Fig. 1 shows an example of an overall configuration according to an embodiment of the present disclosure. The information processing system 1 includes an information processing device 10, a business operator server 20, and a user terminal 30. Note that some or all of the information processing device 10, the business operator server 20, and the user terminal 30 may be implemented by a single device. Furthermore, there may be multiple information processing devices 10 and multiple business operator servers 20. Each of these will be described below.

[0016] <<Information Processing Device>> The information processing device 10 is a device for estimating missing values ​​of physical information. The information processing device 10 is made up of one or more computers. The information processing device 10 can transmit and receive data to and from the business operator server 20 via any network.

[0017] Specifically, the information processing device 10 inputs physical information in which a value of a first attribute is missing into at least one trained model to obtain an estimated value of the first attribute, and inputs physical information in which a value of a second attribute different from the first attribute is missing into the at least one trained model to obtain an estimated value of the second attribute. Preferably, the first attribute and the second attribute are physical information other than basic information. Here, "a missing value of an attribute" includes at least "a case in which the value of the attribute does not exist," "a case in which the value of the attribute is missing," "a case in which the value of the attribute has not been measured," "a case in which the value of the attribute has not been acquired," "a case in which the value of the attribute has not been input and a missing flag has been set for the value of the attribute," "a case in which the value of the attribute cannot be accessed," or "a case in which the value of the attribute has been lost."

[0018] <<Provider Server>> The provider server 20 is a server managed by a provider that provides estimated values ​​of physical information to the user 31. The provider server 20 is made up of one or more computers. The provider server 20 can send and receive data to and from the information processing device 10 and the user terminal 30 via any network.

[0019] Specifically, in response to a request from the user terminal 30, the business operator server 20 calls an API (Application Programming Interface) provided by the information processing device 10 and receives the estimated value of the physical information from the information processing device 10. Note that the business operator server 20 may receive the estimated value of the physical information from the information processing device 10 without using the API.

[0020] 1 illustrates a case in which the information processing system 1 includes the business entity server 20, but the user terminal 30 may receive the estimated value of physical information from the information processing device 10 without going through the business entity server 20. In this case, the information processing system 1 may be composed of the information processing device 10 and the user terminal 30, and the user terminal 30 can directly transmit and receive data to and from the information processing device 10.

[0021] <<User Terminal>> The user terminal 30 is a terminal operated by a user 31. For example, the user terminal 30 is a smartphone, a tablet, a personal computer, etc. The user terminal 30 can send and receive data to and from the business entity server 20 via any network.

[0022] Specifically, the user terminal 30 requests the business operator server 20 to estimate missing values ​​of the physical information. The user terminal 30 also receives estimated values ​​of the physical information from the business operator server 20.

[0023] The user terminal 30 transmits physical information in which a value of a first attribute is missing to the information processing device 10 and receives an estimated value of the first attribute from the information processing device 10, and transmits physical information in which a value of a second attribute different from the first attribute is missing to the information processing device 10 and receives an estimated value of the second attribute from the information processing device. It is preferable that the first attribute and the second attribute are physical information other than basic information. Here, the information processing device 10 may obtain the estimated value of the first attribute and the estimated value of the second attribute using one trained model, or may obtain the estimated value of the first attribute and the estimated value of the second attribute using multiple trained models. Furthermore, the user terminal 30 receives information of an additional attribute from the information processing device 10, transmits the value of the additional attribute to the information processing device 10, and receives a second estimated value different from the estimated value of the first attribute from the information processing device 10. The additional attribute is an attribute determined by the information processing device 10 based on the physical information in which a value of the first attribute is missing. However, this does not exclude the possibility of including basic information as the subject of estimation.

[0024] The number of times the user terminal 30 requests estimation of the missing value is not limited to multiple times as described above, and may be only once. In this case, the user terminal 30 can transmit physical information in which the value of a first attribute is missing to the information processing device 10 and receive an estimated value of the first attribute from the information processing device 10. The estimated value of the first attribute is calculated using at least one trained model provided in the information processing device 10, and the at least one trained model is a model trained using at least two datasets, each of which is missing a different attribute and has a value of at least one common attribute.

[0025] For example, the user terminal 30 can transmit physical information with some attribute values ​​missing to the information processing device 10 (may be via the business operator server 20). Alternatively, physical information with some attribute values ​​missing is stored in another device, and the other device can transmit the physical information from the other device to the information processing device 10 (may be via the business operator server 20) based on an instruction from the user terminal 30. Alternatively, physical information with some attribute values ​​missing is stored in the business operator server 20, and the business operator server 20 can transmit the physical information from the business operator server 20 to the information processing device 10 based on an instruction from the user terminal 30.

[0026] 2 is a functional block diagram of the information processing device 10 according to an embodiment of the present disclosure. The information processing device 10 includes a learning unit 110 and an estimation unit 120.

[0027] <<Learning Unit>> The learning unit 110 includes a learning data acquisition unit 111 and a machine learning unit 112 .

[0028] The learning data acquisition unit 111 acquires learning data for machine learning (specifically, a plurality of pieces of physical information).

[0029] The machine learning unit 112 generates the trained model 100 through machine learning using the learning data acquired by the learning data acquisition unit 111 (trains the trained model 100).

[0030] The trained model 100 is a model that, when inputted with physical information in which some attributes (which may be one attribute or multiple attributes) are missing, outputs estimated values ​​of the missing attributes. The trained model 100 is, for example, a model capable of estimating a joint probability distribution (e.g., HI-VAE (https: / / arxiv.org / pdf / 1807.03653.pdf) or TabTransformer (https: / / arxiv.org / pdf / 2012.06678.pdf)). The trained model 100 can also employ various methods for outputting estimated values. For example, various other machine learning models can be employed, such as neural networks such as multiple regression analysis, logistic regression models, multilayer perceptrons, convolutional neural networks (CNNs), and recurrent neural networks (RNNs), support vector machines using arbitrary kernel functions such as Gaussian kernels, random forests modeled as regression trees, models using hidden Markov models, statistical models, and probabilistic models. Also, models that combine various models to perform comprehensive judgments can be employed. Among these, models that can estimate joint probability distributions (e.g., HI-VAE and TabTransformer) are particularly preferred.

[0031] Here, the training data will be described. The training data is preferably a plurality of pieces of physical information. At least one trained model 100 is a model trained using at least two datasets, each of which is missing different attributes and has at least one common attribute value. The training data will be described in detail below with reference to FIG. 3. Note that "missing different attributes" typically means that, for two or more datasets, there are differences in the lists of attribute items of multiple attribute data included in each dataset. Specifically, this means that at least some attribute items do not overlap between the list of attribute items included in one dataset and the list of attribute items included in another dataset.

[0032] FIG. 3 is a diagram illustrating training data according to an embodiment of the present disclosure. In the embodiment of the present disclosure, the training data includes a plurality of datasets (each dataset includes one or more pieces of physical information. In the example of FIG. 3 , dataset 1 includes IDs: 001 to 005 (data relating to weight and body fat percentage, gender, and height), dataset 2 includes IDs: 006 to 010 (data relating to weight and blood glucose level, urinary sugar, smoking, and age), and dataset 3 includes IDs: 011 to 015 (data relating to age and stress level, stress level and personality traits, and blood glucose level)). The physical information indicated by each ID may be physical information of multiple different individuals, or may be physical information of the same individual acquired at different times or in different locations (e.g., different facilities).

[0033] The common attribute between Dataset 1 (IDs: 001-005) and Dataset 2 (IDs: 006-010) is weight (basic information). Dataset 1 is missing information on blood glucose level, urinary sugar, smoking, age, stress level, and personality traits, while Dataset 2 is missing information on gender, height, body fat percentage, stress level, and personality traits. The common attribute between Dataset 2 and Dataset 3 (IDs: 011-015) is age (basic information) + blood glucose level (body information). Dataset 3 is missing information on gender, height, body fat percentage, weight, urinary sugar, and smoking. Based on these Datasets 1-3, for example, a trained model based on IDs: 001-015 can be generated by executing the training method using HI-VAE described below in <Training Method Using HI-VAE>. Note that in the above example, Datasets 1-3 were all combined to create a single training model, but this example is not limited thereto. A desired training model may also be obtained by combining Datasets 1-3 in stages.

[0034] The multiple pieces of physical information used as training data include pieces of physical information with missing values ​​for different attributes. Referring to Figure 3, data set 1 is missing values ​​for blood glucose level, urinary sugar, smoking, age, stress level, and personality traits, data set 2 is missing values ​​for gender, height, body fat percentage, stress level, and personality traits, and data set 3 is missing values ​​for gender, height, body fat percentage, weight, urinary sugar, and smoking.

[0035] The plurality of pieces of physical information serving as learning data include pieces of physical information having at least one identical attribute value. Referring to FIG. 3 , Dataset 1 and Dataset 2 have the same attribute value (body weight, which is the overlapping portion in the example of FIG. 3 ), and Dataset 2 and Dataset 3 have the same attribute values ​​(blood glucose level and age, which are the overlapping portions in the example of FIG. 3 ). An attribute having a value in both one dataset (e.g., Dataset 2 in FIG. 3 ) and another dataset (e.g., Dataset 1 in FIG. 3 ) (i.e., body weight in FIG. 3 ) may be different from an attribute having a value in both the one dataset (e.g., Dataset 2 in FIG. 3 ) and another dataset (e.g., Dataset 3 in FIG. 3 ) (i.e., blood glucose level and age in FIG. 3 ).

[0036] For example, the datasets may be datasets acquired in different situations (e.g., a medical checkup and a questionnaire). Alternatively, the datasets may be datasets in which different attribute values ​​are acquired in the same situation (e.g., basic test items and specific test items in a medical checkup). Furthermore, each dataset may contain physical information acquired in a different location or at a different time. Furthermore, the datasets may contain physical information of the same person.

[0037] As described above, in this embodiment, by generating the trained model 100 using training data that integrates multiple datasets in which values ​​of different attributes are missing, it is possible to increase the amount of physical information that can be estimated by the trained model 100. Furthermore, since this embodiment can integrate datasets in which values ​​of different attributes are missing, it is possible to effectively use the results of multiple health checkups consisting of different test items held by each facility as a single training data set.

[0038] When training a model by combining different datasets, there must be at least one attribute (hereinafter referred to as a common attribute) that is common to the multiple datasets. For example, the weight, blood glucose level, and age in Figure 3 correspond to the common attributes. The existence of a common attribute can improve prediction accuracy even when predicting attributes specific to one dataset from attributes specific to another dataset.

[0039] Here, the "common attributes" that have values ​​common to the data sets preferably include one or more attributes selected from the "basic information." Furthermore, the "common attributes" preferably further include one or more attributes selected from the "information related to physical condition."

[0040] Alternatively, the training data may be randomly missing data and then the error function for restoring the data may be minimized. Artificially missing data allows for more robust estimation.

[0041] The learning data in FIG. 3 is merely an example, and any learning data of physical information, such as the learning data shown in FIG. 4, can be used.

[0042] When using a neural network as a model, optimization can be performed using methods commonly used for neural networks (e.g., Adam, Momentum SGD). For example, when using HI-VAE (https: / / arxiv.org / pdf / 1807.03653.pdf) as a model, there are hyperparameters such as the dimensions of the hidden variables, and these can be combined with hyperparameter search to determine the optimal ones for prediction.

[0043] <Learning Method Using HI-VAE> Below is an example of a learning method when using HI-VAE as a model. 1. Prepare training data containing multiple datasets. (A) As an example, the data for each ID included in the training data is a four-dimensional vector (attributes A to D). (1) Example: Attribute A: 10, Attribute B: 8, Attribute C: Missing, Attribute D: 5 (B) In the four-dimensional vector, the missing value (attribute C in the above example) is filled in with one of the following values. Note that the missing value may be different for each ID. (1) NULL (may be a specified symbol) (2) 0 (3) A numerical value determined based on the value of the attribute included in the training data (e.g., the average value) (C) If 0 as in (2) is used, an example of the above vector would be as follows: (1) Example: Attribute A: 10, Attribute B: 8, Attribute C: 0, Attribute D: 5 2. Some attributes in the training data are left missing. (A) As an example, the value of attribute D is made missing (the value is replaced with 0). (1) Example) Attribute A: 10, Attribute B: 8, Attribute C: 0, Attribute D: 0 (B) The missing attributes may be set randomly for each ID vector. 3. Multiple vectors corresponding to each ID generated in "2" above are input into HI-VAE to obtain multiple output vectors. (A) The output vector corresponding to the above example outputs not only the estimated value of attribute D, but also estimated values ​​for the other attributes A, B, and C. In other words, in the output vector, the values ​​of attributes A, B, and C are also replaced with estimated values. Examples of output vectors are as follows: (1) Example) Attribute A: 11, Attribute B: 7, Attribute C: 3, Attribute D: 4 4. The model parameters are updated to reduce the difference (error) between the vector generated in "2" above and the corresponding vector output by the model in "3" above. (A) Here, in the output vector in the above example, attribute C is a missing part that does not originally have a value (a part that does not have a correct answer), and therefore is not taken into consideration in calculating the error.

[0044] <Estimation Method Using HI-VAE> An example of an estimation method using HI-VAE as a model is described below. 1. An input vector is generated using physical information (with some attributes missing) received from the user terminal 30. (A) One of the following values ​​is entered for the missing attribute values: (1) NULL (which may be a predetermined symbol) (2) 0 (3) A numerical value (e.g., the average value) determined based on the attribute values ​​included in the training data (B) There may be multiple attributes with missing values. 2. A vector of "1" is input to the trained model (HI-VAE) to obtain an output vector. (A) In the output vector, estimated values ​​are entered for the missing attribute parts, as well as for the other attributes that were not missing. 3. Estimated values ​​are presented to the user terminal 30. (A) Here, the information processing device 10 may present to the user terminal 30 only estimated values ​​for the attributes that were missing in "1". (i) Furthermore, the information processing device 10 may combine the estimated value of the missing attribute in “1” with the physical information received from the user terminal 30 in “1” and present it to the user terminal 30.

[0045] Returning to the description of FIG.

[0046] <<Estimation Unit>> The estimation unit 120 includes a physical information acquisition unit 121 , a missing value estimation unit 122 , and an estimated value presentation unit 123 .

[0047] The physical information acquisition unit 121 acquires physical information in which some of the attributes of the physical information are missing values ​​(which may be one attribute or multiple attributes). Specifically, the physical information acquisition unit 121 acquires the name of each attribute and its value (which may be null if any value is missing).

[0048] The missing value estimation unit 122 inputs the physical information acquired by the physical information acquisition unit 121 (i.e., physical information in which values ​​of some attributes of the physical information are missing) into the trained model 100 to acquire estimated values ​​(e.g., expected values, variances) of the some attributes. The missing value estimation unit 122 may acquire at least one estimated value for the missing attribute, or may acquire values ​​having a predetermined range as estimated values. The missing value estimation unit 122 can also acquire the reliability of the estimated value. The missing value estimation unit 122 may acquire the estimated value using a trained model selected from multiple trained models. The missing value estimation unit 122 may also perform an estimation process using multiple trained models and use the estimated value with the highest reliability as the estimated value to be presented to the user. Furthermore, the missing value estimation unit 122 may select a trained model to be used for estimation from multiple trained models based on an instruction from the user terminal 30.

[0049] In addition, the missing value estimation unit 122 may estimate other information (for example, information on whether or not the active ingredient Q is effective) by inputting the estimated value into a predetermined calculation formula or into a separately trained machine learning model.

[0050] For example, HI-VAE can estimate not only points but also distributions. When training HI-VAE, the distribution for each attribute is determined in advance. Examples of distributions include Gaussian distributions (continuous values ​​such as weight), categorical distributions (discrete values ​​such as whether a person has a disease), Poisson distributions (natural numbers such as the number of times a person has had a disease), and log-normal distributions (positive real numbers and asymmetric distributions such as BMI). Based on this distribution, the system returns its parameters during estimation. For example, in the case of a Gaussian distribution, it returns the mean and variance parameters. The variance parameters can be used to estimate the deviation of the estimated value. For example, if the mean of the estimated weight is 60 and the variance is 1, this estimate indicates that there is a 95% or higher probability (confidence) that the actual weight falls within the range of 60 kg ± 2 kg. Similar confidence intervals can be calculated for other distributions (i.e., the estimated value has a specified range).

[0051] For example, it is possible to notify users that "By providing data on attribute X in addition to data on already measured attributes, attribute A can be estimated with higher accuracy." There are several possible methods for recommending attribute X. Here, we will explain a method using the information gain of HI-VAE as an example. Information gain is defined by the following formula.

[0052] α ES (x;D n ):=H(x * |D n )-E y|Dn,x H(x * |D n ∪{(x,y)})...Formula (1)

[0053] Here, x* corresponds to attribute A, x to the candidate attribute, y to the estimated value of the candidate attribute, D_n to the training data, and H(x | D_n) = -∫p(x | D_n) log p(x | D_n) dx to the entropy of the posterior distribution p(x | D_n). E represents the expected value, and αES represents the expected value of the information gain.

[0054] The first term in this equation represents the uncertainty of x* given the observation of D_n; the higher this value, the more difficult the estimation. The second term in this equation represents the uncertainty of x* when, in addition to D_n, a value of y is hypothetically observed for the unobserved attribute x. In other words, subtracting the second term from the first term can be interpreted as evaluating how much the uncertainty of x* decreases when x is added. The entropy and expected value in equation (1) generally do not have analytical solutions, but approximate solutions can be obtained using methods such as the Monte Carlo method. By calculating this information gain for candidate attributes and recommending the one with the highest value, the attribute that best reduces the uncertainty in estimating x* can be recommended. The estimated value presentation unit 123 presents the user with information about the attribute that best reduces uncertainty, thereby prompting the user to provide the value of that attribute (additional attribute). The missing value estimation unit 122 can further improve the estimation accuracy of attribute A by performing the estimation process again using the value of the additional attribute received from the user. Note that attribute A may be determined based on an instruction from the user, or a predetermined item may be set as attribute A.

[0055] The estimated value presenting unit 123 presents the estimated values ​​of the part of the attributes estimated by the missing value estimating unit 122. The estimated value presenting unit 123 may present not only the estimated values ​​of the part of the attributes but also the entire physical information including the estimated values. The estimated value presenting unit 123 may also provide information related to the physical information (e.g., services, products, etc.) based on the estimated values ​​of the attributes. Here, the estimated value presenting unit 123 may acquire the information related to the physical information by inputting the estimated values ​​into a predetermined calculation formula or into a separately trained machine learning model. Furthermore, the estimated value presenting unit 123 may acquire the information related to the physical information using physical information other than the estimated values.

[0056] After obtaining the estimated values, the information processing device 10 can delete the physical information with missing attribute values ​​(i.e., the physical information obtained from the user) and the estimated physical information from the storage device (memory) of the information processing device 10. This prevents the user's physical information from being leaked, allowing the user to use the information processing device 10 with peace of mind. The information processing system 1 may also delete the physical information with missing attribute values ​​and the estimated physical information not only from the information processing device 10 but also from other devices (such as the business server 20). The information processing system 1 may also delete the physical information with missing attribute values ​​and the estimated physical information from all devices other than the user terminal 30.

[0057] FIG. 5 is a diagram illustrating physical information including missing values ​​and physical information including estimated values ​​according to an embodiment of the present disclosure (note that XXX in the figure indicates judgment, score, numerical value, and classification information).

[0058] The "physical information including missing values" at the top of Figure 5 is physical information in which the values ​​of some attributes (which may be one attribute or multiple attributes) are missing. Explaining this with reference to Figure 5, in the physical information of ID: 100, the values ​​of attributes 5 to 10 are missing. In addition, in the physical information of ID: 200, the values ​​of attributes 1 to 3 are missing. When the "physical information including missing values" is input to the trained model 100, the values ​​of the missing attributes are estimated.

[0059] The "Physical information including estimated values" at the bottom of Fig. 5 is physical information including values ​​estimated by the trained model 100. Referring to Fig. 5, in the physical information of ID: 100, the values ​​of the missing attributes (attributes 5 to 10 in the example of Fig. 5) have been estimated. Also, in the physical information of ID: 200, the values ​​of the missing attributes (attributes 1 to 3 in the example of Fig. 5) have been estimated.

[0060] In this manner, in one embodiment of the present disclosure, physical information in which some attributes have missing values ​​can be received, and physical information in which the missing values ​​have been replaced with estimated values ​​can be returned.

[0061] The estimated value presenting unit 123 can present the estimated physical information in the form of table data (e.g., a table structure as shown in FIG. 5). For example, the estimated value presenting unit 123 can present the estimated attribute values ​​of the physical information in different display formats (e.g., by changing the display format on the screen, such as by changing the color, font, highlighting, or size of only the estimated values). For example, the estimated value presenting unit 123 can present the estimated attribute values ​​in a display format according to the reliability of the estimated attribute values ​​(e.g., by using different display formats for high and low reliability). Furthermore, the estimated value presenting unit 123 may present an index indicating the reliability in addition to the physical information.

[0062] Fig. 6 is a diagram illustrating an example of estimation according to an embodiment of the present disclosure. As shown in <Physical information including missing values> in Fig. 6 , when physical information in which values ​​of some attributes of the physical information (in the example of Fig. 6 , "visceral fat area," "triglyceride," "blood glucose level," and "stress level") are missing is input to trained model 100, the values ​​of the missing attributes are estimated as shown in <Physical information including estimated values> in Fig. 6 .

[0063] Examples of estimation of physical information will be described below. Note that a combination of two or more of the following estimation examples may be used.

[0064] <<Estimation Example 1>> The missing value estimation unit 122 can estimate the value of a missing attribute (e.g., estimate the current value). For example, the missing value estimation unit 122 can estimate the probability of a certain disease or estimate the value of a blood test without drawing blood.

[0065] <<Estimation Example 2>> The missing value estimation unit 122 can convert attribute values ​​(which may include not only estimated values ​​but also measured values) into indicators that are easy for the user to understand and display them. For example, the missing value estimation unit 122 can calculate a health age, a vascular age, a physical age, etc. based on the attribute values. The missing value estimation unit 122 may calculate the health age, the vascular age, the physical age, etc. by inputting the estimated values ​​into a predetermined calculation formula or into a separately trained machine learning model.

[0066] <<Estimation Example 3>> When a user selects an attribute to be estimated, the missing value estimation unit 122 can indicate attributes that should be measured to increase the reliability of the estimated value of that attribute. For example, the missing value estimation unit 122 can indicate attributes that should be measured to narrow the confidence interval in estimating the value of a certain attribute (e.g., if the estimated probability of diabetes is 40 to 80 percent, what should be measured to narrow the confidence interval). This can be achieved, for example, by a method that uses the information gain of the HI-VAE described above.

[0067] <<Estimation Example 4>> The missing value estimation unit 122 can extract attributes that change in conjunction with a change in the value of a certain attribute. For example, by extracting attributes that change in conjunction with a decrease in blood glucose level, it is possible to present to the user attributes that are effective for lowering blood glucose levels. Here, the target attribute, such as blood glucose level, may be determined based on an instruction from the user terminal 30.

[0068] <<Estimation Example 5>> The missing value estimation unit 122 can input a hypothetical value for a certain attribute and estimate the value of another attribute. In other words, it can show how the value of another attribute changes when a hypothetical value is input for a certain attribute (for example, whether the probability of having diabetes would be slightly lower if the weight were 5 kg lighter). In this way, by hypothesizing a causal relationship between a certain attribute and another attribute that changes as a certain attribute changes and verifying this through clinical trials or the like, it is possible to discover new relationships between attributes that appear unrelated at first glance. The missing value estimation unit 122 may input a hypothetical value for a certain attribute based on an instruction from the user terminal 30 and transmit the result to the user terminal 30.

[0069] <<Estimation Example 6>> The missing value estimation unit 122 can perform estimation using trained models (e.g., a model for smokers and a model for non-smokers) generated based on specific attributes. The missing value estimation unit 122 stores multiple trained models trained using attribute-based learning data in a storage device, and can switch between trained models used for estimation based on the attributes of the input physical information. Note that the attributes may be based on basic information. The missing value estimation unit 122 may determine the attributes to use for switching between trained models based on an instruction from the user terminal 30.

[0070] <<Estimation Example 7>> The missing value estimation unit 122 can obtain a representative value for a certain attribute as a value of a probability distribution and indicate the position of the user's value on the probability distribution. For example, the missing value estimation unit 122 can obtain a standard value for a certain attribute and indicate a relative index (e.g., the value for people of the same gender and age as the user, or the position of the user's result on the probability distribution).

[0071] <Processing Method> The learning process will be described below with reference to FIG. 7, and the inference process will be described with reference to FIG.

[0072] <<Learning Process>> FIG. 7 is a flowchart of a learning process according to an embodiment of the present disclosure.

[0073] In step 11 (S11), the learning data acquisition unit 111 acquires learning data for machine learning (specifically, a plurality of pieces of physical information).

[0074] In step 12 (S12), the machine learning unit 112 generates the trained model 100 by machine learning using the training data acquired in S11.

[0075] <<Inference Processing>> FIG. 8 is a flowchart of an estimation processing according to an embodiment of the present disclosure.

[0076] In step 21 (S21), the physical information acquisition unit 121 acquires physical information in which values ​​of some attributes (which may be one attribute or multiple attributes) of the physical information are missing.

[0077] In step 22 (S22), the missing value estimation unit 122 inputs the physical information acquired in S21 (i.e., physical information in which values ​​of some of the attributes of the physical information are missing) into the trained model 100, thereby obtaining estimated values ​​of the some of the attributes.

[0078] In step 23 (S23), the estimated value presenting unit 123 presents the estimated values ​​of the part of the attributes estimated in S22. Here, the estimated value presenting unit 123 may present a combination of the physical information acquired in S21 and the estimated values ​​acquired in S22, or may present only the estimated values ​​acquired in S22.

[0079] <Effect> In this way, in an embodiment of the present disclosure, by inputting physical information in which values ​​of some attributes are missing into a trained model, it is possible to estimate the values ​​of those attributes, so that even if the values ​​of all attributes have not been measured, it is possible to obtain the values ​​of all attributes.

[0080] Some or all of the devices (information processing device 10 or user terminal 30) in the above-described embodiments may be configured as hardware, or may be configured as software (program) information processing executed by a CPU (Central Processing Unit), GPU (Graphics Processing Unit), or the like. When configured as software information processing, software that realizes at least some of the functions of each device in the above-described embodiments may be stored on a non-transitory storage medium (non-transitory computer-readable medium) such as a CD-ROM (Compact Disc-Read Only Memory) or USB (Universal Serial Bus) memory, and the software information processing may be executed by loading the software into a computer. The software may also be downloaded via a communications network. Furthermore, all or part of the software processing may be implemented in a circuit such as an ASIC (Application Specific Integrated Circuit) or FPGA (Field Programmable Gate Array), thereby allowing the software information processing to be executed by hardware.

[0081] The storage medium that stores the software may be a removable medium such as an optical disk, or a fixed medium such as a hard disk, memory, etc. The storage medium may be provided inside the computer (such as a main storage device or auxiliary storage device) or outside the computer.

[0082] 9 is a block diagram showing an example of the hardware configuration of each device (information processing device 10 or user terminal 30) in the above-described embodiment. Each device may be realized as a computer 1000 including, for example, a processor 1001, a main storage device 1002 (memory), an auxiliary storage device 1003 (memory), a network interface 1004, and a device interface 1005, which are connected via a bus B.

[0083] Although the computer 1000 in FIG. 9 includes one of each component, it may also include multiple of the same component. Also, while FIG. 9 shows a single computer 1000, the software may be installed on multiple computers, with each of the multiple computers executing the same or different parts of the software. In this case, a distributed computing configuration may be used in which each computer communicates via a network interface 1004 or the like to execute processing. In other words, each device (information processing device 10 or user terminal 30) in the above-described embodiment may be configured as a system in which one or more computers execute instructions stored in one or more storage devices to realize functions. Furthermore, the system may be configured such that information transmitted from a terminal is processed by one or more computers provided on a cloud, and the processing results are transmitted to the terminal.

[0084] The various calculations of each device (information processing device 10 or user terminal 30) in the above-described embodiments may be executed in parallel using one or more processors, or using multiple computers via a network. Furthermore, the various calculations may be distributed to multiple processing cores within a processor and executed in parallel. Furthermore, some or all of the processes, means, etc. disclosed herein may be implemented by at least one of a processor and a storage device provided on a cloud that can communicate with the computer 1000 via a network. Thus, each device in the above-described embodiments may be implemented in the form of parallel computing using one or more computers.

[0085] The processor 1001 may be an electronic circuit (processing circuit, processing circuitry, CPU, GPU, FPGA, ASIC, etc.) that performs at least one of computer control and calculation. The processor 1001 may also be a general-purpose processor, a dedicated processing circuit designed to perform a specific calculation, or a semiconductor device that includes both a general-purpose processor and a dedicated processing circuit. The processor 1001 may also include an optical circuit or a calculation function based on quantum computing.

[0086] The processor 1001 may perform arithmetic processing based on data or software input from each device or the like configured internally by the computer 1000, and may output arithmetic results or control signals to each device or the like. The processor 1001 may control each component constituting the computer 1000 by executing the OS (Operating System) of the computer 1000, applications, etc.

[0087] Each device (information processing device 10 or user terminal 30) in the above-described embodiment may be realized by one or more processors 1001. Here, the processor 1001 may refer to one or more electronic circuits arranged on one chip, or may refer to one or more electronic circuits arranged on two or more chips or two or more devices. When multiple electronic circuits are used, the electronic circuits may communicate with each other via wire or wirelessly.

[0088] The main memory device 1002 may store instructions executed by the processor 1001, various data, etc., and information stored in the main memory device 1002 may be read by the processor 1001. The auxiliary memory device 1003 is a memory device other than the main memory device 1002. Note that these memory devices refer to any electronic component capable of storing electronic information and may be semiconductor memory. The semiconductor memory may be either volatile memory or non-volatile memory. The memory device for saving various data, etc. in each device (information processing device 10 or user terminal 30) in the above-described embodiments may be realized by the main memory device 1002 or the auxiliary memory device 1003, or may be realized by an internal memory built into the processor 1001. For example, the memory unit in the above-described embodiments may be realized by the main memory device 1002 or the auxiliary memory device 1003.

[0089] When each device (information processing device 10 or user terminal 30) in the above-described embodiment is configured with at least one storage device (memory) and at least one processor connected (coupled) to this at least one storage device, at least one processor may be connected to one storage device. Also, at least one storage device may be connected to one processor. Also, a configuration in which at least one processor among multiple processors is connected to at least one storage device among multiple storage devices may be included. Furthermore, this configuration may be realized by storage devices and processors included in multiple computers. Furthermore, a configuration in which a storage device is integrated with a processor (for example, a cache memory including an L1 cache and an L2 cache) may be included.

[0090] The network interface 1004 is an interface for connecting to a communication network N wirelessly or via a wire. The network interface 1004 may be an appropriate interface, such as one that conforms to an existing communication standard. Information may be exchanged with an external device 1010A connected via the communication network N through the network interface 1004. The communication network N may be any one of a wide area network (WAN), a local area network (LAN), a personal area network (PAN), etc., or a combination thereof, as long as information is exchanged between the computer 1000 and the external device 1010A. An example of a WAN is the Internet, an example of a LAN is IEEE 802.11 or Ethernet (registered trademark), and an example of a PAN is Bluetooth (registered trademark) or NFC (Near Field Communication), etc.

[0091] The device interface 1005 is an interface such as a USB that directly connects to the external device 1010B.

[0092] The external device 1010A is connected to the computer 1000 via a network. The external device 1010B is connected directly to the computer 1000.

[0093] For example, the external device 1010A or the external device 1010B may be an input device. The input device is, for example, a device such as a camera, a microphone, a motion capture device, various sensors, a keyboard, a mouse, or a touch panel, and provides acquired information to the computer 1000. Alternatively, the external device 1010A or the external device 1010B may be a device including an input unit, a memory, and a processor, such as a personal computer, a tablet terminal, or a smartphone.

[0094] Furthermore, the external device 1010A or the external device 1010B may be, for example, an output device. The output device may be, for example, a display device such as an LCD (Liquid Crystal Display) or an organic EL (Electro Luminescence) panel, or may be a speaker that outputs sound or the like. Alternatively, the output device may be a device including an output unit, a memory, and a processor, such as a personal computer, a tablet terminal, or a smartphone.

[0095] The external device 1010A or 1010B may be a storage device (memory). For example, the external device 1010A may be a network storage device, and the external device 1010B may be a storage device such as an HDD.

[0096] Furthermore, the external device 1010A or the external device 1010B may be a device having some of the functions of the components of each device (the information processing device 10 or the user terminal 30) in the above-described embodiments. That is, the computer 1000 may transmit some or all of the processing results to the external device 1010A or the external device 1010B, or may receive some or all of the processing results from the external device 1010A or the external device 1010B.

[0097] In this specification (including the claims), when the expression "at least one of a, b, and c" or "at least one of a, b, or c" (including similar expressions) is used, it includes any of a, b, c, ab, ac, bc, or abc. It may also include multiple instances of any element, such as aa, abb, aabbcc, etc. Furthermore, it also includes the addition of elements other than the enumerated elements (a, b, and c), such as having d, as in abcd.

[0098] In this specification (including claims), when expressions such as "using / using data as input / based on / according to / in response to data" (including similar expressions) are used, unless otherwise specified, this includes cases where the data itself is used, or where data that has been processed in some way (e.g., data with noise added, normalized data, features extracted from data, intermediate representations of data, etc.) is used. Furthermore, when a statement is made that a result is obtained "using data as input / based on / according to / in response to data" (including similar expressions), this includes cases where the result is obtained based solely on the data, or where the result is influenced by other data, factors, conditions, and / or states other than the data. Furthermore, when a statement is made that "data is output" (including similar expressions), this includes cases where the data itself is used as output, or where data that has been processed in some way (e.g., data with noise added, normalized data, features extracted from data, intermediate representations of various data, etc.) is used as output, unless otherwise specified.

[0099] When the terms "connected" and "coupled" are used in this specification (including the claims), they are intended as open-ended terms that include any of direct connection / coupling, indirect connection / coupling, electrically connection / coupling, communicatively connection / coupling, functionally connection / coupling, and physically connection / coupling. These terms should be interpreted appropriately according to the context in which they are used, but any connection / coupling form that is not intentionally or naturally excluded should be interpreted as being included in these terms without limitation.

[0100] In this specification (including the claims), the expression "A configured to B" may include the physical structure of element A having a configuration capable of performing operation B, and the permanent or temporary setting / configuration of element A being configured / set to actually perform operation B. For example, if element A is a general-purpose processor, it is sufficient that the processor has a hardware configuration capable of performing operation B, and is configured to actually perform operation B by setting a permanent or temporary program (instruction). Also, if element A is a dedicated processor, dedicated arithmetic circuit, etc., it is sufficient that the circuit structure of the processor is implemented to actually perform operation B, regardless of whether control instructions and data are actually attached.

[0101] Whenever words implying containing or possessing (e.g., "comprising / including," "having," etc.) are used in this specification (including the claims), they are intended to be open-ended terms that include containing or possessing things other than the object designated by the object of the term. When the object of such words implying containing or possessing does not specify a quantity or suggests a singular number (e.g., expressions using the articles "a" or "an"), the expression should be construed as not being limited to a specific number.

[0102] In this specification (including the claims), although expressions such as "one or more" and "at least one" are used in some places and expressions that do not specify a quantity or that imply a singular number (expressions using the articles "a" or "an") are used in other places, the latter expressions are not intended to mean "one." In general, expressions that do not specify a quantity or that imply a singular number (expressions using the articles "a" or "an") should be interpreted as not necessarily being limited to a specific number.

[0103] In this specification, when a particular advantage / result is described as being obtained with respect to a particular configuration of an embodiment, it should be understood that the same advantage / result can also be obtained with one or more other embodiments having the same configuration, unless otherwise stated. However, it should be understood that the presence or absence of the effect generally depends on various factors, conditions, and / or situations, and that the effect is not necessarily obtained with the configuration. The effect is merely obtained by the configuration described in the embodiment when various factors, conditions, and / or situations are satisfied, and the effect does not necessarily occur in a claimed invention that defines the same configuration or a similar configuration.

[0104] When terms such as "maximize" / "maximization" are used in this specification (including the claims), they include finding a global maximum, finding an approximation of a global maximum, finding a local maximum, and finding an approximation of a local maximum, and should be interpreted appropriately according to the context in which the terms are used. They also include finding approximations of these maxima probabilistically or heuristically. Similarly, when terms such as "minimize" / "minimization" are used, they include finding a global minimum, finding an approximation of a global minimum, finding a local minimum, and finding an approximation of a local minimum, and should be interpreted appropriately according to the context in which the terms are used. They also include finding approximations of these minima probabilistically or heuristically. Similarly, when terms such as "optimize" / "optimization" are used, they include finding a global optimum, finding an approximation of a global optimum, finding a local optimum, and finding an approximation of a local optimum, and should be interpreted appropriately according to the context in which the terms are used. It also includes finding approximations of these optimum values ​​probabilistically or heuristically.

[0105] In this specification (including claims), when multiple pieces of hardware perform a predetermined process, the pieces of hardware may cooperate to perform the predetermined process, or some of the hardware may perform all of the predetermined process. Furthermore, some of the hardware may perform part of the predetermined process, and other hardware may perform the rest of the predetermined process. In this specification (including claims), when an expression such as "one or more pieces of hardware perform a first process, and the one or more pieces of hardware perform a second process" (including similar expressions) is used, the hardware performing the first process and the hardware performing the second process may be the same or different. In other words, it is sufficient that the hardware performing the first process and the hardware performing the second process are included in the one or more pieces of hardware. Note that hardware may include an electronic circuit, a device including an electronic circuit, etc.

[0106] In this specification (including the claims), when multiple storage devices (memories) store data, each of the multiple storage devices may store only a portion of the data, or may store the entire data. Also, a configuration in which only some of the multiple storage devices store data may be included.

[0107] In this specification (including the claims), terms such as "first," "second," etc. are used merely as a way of distinguishing between two or more elements, and are not necessarily intended to impose technical meanings such as temporal aspect, spatial aspect, sequence, quantity, etc. on the subject. Thus, for example, a reference to a first element and a second element does not necessarily mean that only two elements may be employed therein, that the first element must precede the second element, that the first element must be present in order for the second element to be present, etc.

[0108] Although the embodiments of the present disclosure have been described in detail above, the present disclosure is not limited to the individual embodiments described above. Various additions, modifications, substitutions, partial deletions, etc. are possible within the scope of the conceptual idea and spirit of the present invention, which is derived from the content defined in the claims and their equivalents. For example, when numerical values ​​or formulas are used in the above-described embodiments, they are shown for illustrative purposes and do not limit the scope of the present disclosure. Furthermore, the order of each operation shown in the embodiments is also illustrative and does not limit the scope of the present disclosure.

[0109] This international application claims priority based on Japanese Patent Application No. 2022-028697, filed on February 25, 2022, the entire contents of which are incorporated herein by reference.

[0110] REFERENCE SIGNS LIST 1 Information processing system 10 Information processing device 20 Business operator server 30 User terminal 31 User 100 Trained model 110 Learning unit 111 Learning data acquisition unit 112 Machine learning unit 120 Estimation unit 121 Physical information acquisition unit 122 Missing value estimation unit 123 Estimated value presentation unit 1001 Processor 1002 Main storage device (memory) 1003 Auxiliary storage device (memory) 1004 Network interface 1005 Device interface 1010A External device 1010B External device

Claims

1. at least one memory; at least one processor; The at least one processor obtaining an estimate of a first attribute by inputting physical information missing a value of the first attribute into at least one trained model; acquiring an estimate of a second attribute different from the first attribute by inputting physical information missing a value of the second attribute into the at least one trained model; Run the at least one trained model is a model trained using at least two datasets; The two data sets each have different missing attributes and at least one common attribute value. Information processing device.

2. The at least two data sets are data sets measured at different locations or times. The information processing device according to claim 1 .

3. The at least two data sets are data sets measured in different situations. The information processing device according to claim 1 .

4. The at least two data sets are data sets in which values ​​of different attributes are obtained under the same circumstances. The information processing device according to claim 1 .

5. The information processing device according to claim 1 , wherein the first attribute and the second attribute are physical information other than basic information.

6. The basic information includes gender, age, height, and weight. The information processing device according to claim 5 .

7. The information processing device according to claim 1 , wherein the physical information includes information relating to a physical condition.

8. The information processing device according to claim 1 , wherein the physical information includes information about a psychological state.

9. The information processing device according to claim 1 , wherein the estimated value is a value having a predetermined range.

10. The at least one processor The information processing apparatus according to claim 1 , further comprising: a reliability level of the estimated value being acquired.

11. The at least one processor The information processing device according to claim 1 , wherein information relating to the physical information is provided based on the estimated value.

12. The at least one processor The information processing device according to claim 1 , wherein attributes to be measured in order to increase the reliability of an estimated value are presented.

13. The at least one processor generates the at least one trained model by performing machine learning using the at least two datasets. The information processing device according to claim 1 .

14. The at least one processor: determining additional attributes to be acquired based on the physical information; using the obtained value of the additional attribute to obtain a second estimate of the first attribute, the second estimate being different from the first estimate. The information processing device according to claim 1 .

15. The at least one processor deletes the physical information in which the value of the first attribute is missing, the estimated value of the first attribute, the physical information in which the value of the second attribute is missing, and the estimated value of the second attribute from the information processing device. The information processing device according to claim 1 .

16. The at least one processor presents the physical information with missing values ​​of the first attribute and the estimated values ​​of the first attribute in the form of table data. The information processing device according to claim 1 .

17. The at least one processor presents the estimated value of the first attribute in a display format different from other physical information. The information processing device according to claim 1 .

18. The at least one processor changes a display format of the estimated value of the first attribute based on a reliability of the estimated value of the first attribute. The information processing device according to claim 1 .

19. at least one memory; An information processing device comprising at least one processor, The at least one processor obtaining an estimate of a first attribute by inputting physical information missing a value of the first attribute into at least one trained model; acquiring an estimate of a second attribute different from the first attribute by inputting physical information missing a value of the second attribute into the at least one trained model; deleting the physical information in which the value of the first attribute is missing, the estimated value of the first attribute, the physical information in which the value of the second attribute is missing, and the estimated value of the second attribute from the information processing device; An information processing device that executes the above.

20. The at least one processor The information processing device according to claim 13 , wherein the physical information with missing values ​​of the first attribute and the estimated values ​​of the first attribute are presented in the form of table data.

21. The at least one processor The information processing device according to claim 13 or 14, wherein the estimated value of the first attribute is presented in a display form different from that of other physical information.

22. The at least one processor The information processing apparatus according to claim 13 , further comprising: a display mode of the estimated value of the first attribute being changed based on the reliability of the estimated value of the first attribute.

23. at least one memory; at least one processor; The at least one processor transmitting physical information having a missing value of a first attribute to at least one information processing device; receiving an estimate of the first attribute from the at least one information processing device; transmitting physical information in which a value of a second attribute different from the first attribute is missing to the at least one information processing device; receiving an estimate of the second attribute from the at least one information processing device; Run the terminal.

24. The at least one processor further comprises: receiving information of additional attributes from the at least one information processing device; transmitting a value of the additional attribute to the at least one information processing device; receiving a second estimate of the attribute from the at least one information processing device, the second estimate being different from the first estimate; Run the additional attribute is an attribute determined by the at least one information processing device based on physical information in which the value of the first attribute is missing; 18. The terminal of claim 17.

25. the estimated value of the first attribute and the estimated value of the second attribute are calculated using at least one trained model included in the at least one information processing device; the at least one trained model is a model trained using at least two datasets; The two data sets each have different missing attributes and at least one common attribute value.

19. A terminal according to claim 17 or 18.

26. at least one memory; at least one processor; The at least one processor transmitting physical information having a missing value of a first attribute to at least one information processing device; receiving an estimate of the first attribute from the at least one information processing device; Run the estimated value of the first attribute is calculated using at least one trained model included in the at least one information processing device; the at least one trained model is a model trained using at least two datasets; The two data sets are each missing different attributes and have at least one common attribute value.

27. A method executed by at least one processor, comprising: obtaining an estimate of a first attribute by inputting physical information missing a value of the first attribute into at least one trained model; acquiring an estimate of a second attribute different from the first attribute by inputting physical information missing a value of the second attribute into the at least one trained model; Including, the at least one trained model is a model trained using at least two datasets; The two data sets each have different missing attributes and at least one common attribute value. method.

28. At least one processor obtaining an estimate of a first attribute by inputting physical information missing a value of the first attribute into at least one trained model; acquiring an estimate of a second attribute different from the first attribute by inputting physical information missing a value of the second attribute into the at least one trained model; Execute the at least one trained model is a model trained using at least two datasets; The two data sets each have different missing attributes and at least one common attribute value. program.

29. A method executed by at least one processor, comprising: transmitting physical information having a missing value of a first attribute to at least one information processing device; receiving an estimate of the first attribute from the at least one information processing device; transmitting physical information in which a value of a second attribute different from the first attribute is missing to the at least one information processing device; receiving an estimate of the second attribute from the at least one information processing device; Including, method.

30. at least one memory; at least one processor; The at least one processor obtaining an estimate of a first attribute by inputting physical information missing a value of the first attribute into at least one trained model; acquiring an estimate of a second attribute different from the first attribute by inputting physical information missing a value of the second attribute into the at least one trained model; Run the at least one trained model is a model trained using at least two datasets; The two datasets are datasets each missing different attributes, have at least one common attribute value, and were measured at different locations or times; the first attribute and the second attribute are physical information other than basic information, The basic information includes gender, age, height, and weight.