Information processing device

Abstract

An information processing device 100 according to the present disclosure comprises: an acquisition unit 121 that acquires a model which is generated in each transitional period and is machine-trained to output a procedure with respect to a person in response to the input of a plurality of types of features representing a state of the person; a collection unit 122 that collects a first output when a predetermined number of types of the features are input to each model in each transitional period, and a second output when some of the predetermined number of types of the features are input thereto; and a setting unit 123 that, on the basis of the first output and the second output, sets a type that is associated with each model in each transitional period. Thus, the information processing device may be used for assisting a user's decision making, for example.

Description

Information processing device

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

[0002] There is an increasing demand for personalized medicine, which determines the treatment method according to the condition of each patient. Personalized medicine aims to provide treatment appropriate for each patient's condition using personal genetic information, medical information, etc. Here, Patent Document 1 describes the use of a computer to select a treatment method.

[0003] Special Publication No. 2016-514291

[0004] However, as described above, when selecting a treatment method, it is necessary to collect various features, such as biometric information that indicates the patient's condition. Because collecting patient information is costly and time-consuming, it is necessary to efficiently collect patient information in order to quickly and easily select an appropriate treatment method. This problem is not limited to treatment, but also arises when providing methods such as training, exercise, and dieting, making it necessary to efficiently collect information on subjects for policy proposals.

[0005] Therefore, an object of the present disclosure is to provide an information processing device that can improve the efficiency of collecting information on subjects for policy proposals.

[0006] An information processing device according to one embodiment of the present disclosure includes: an acquisition unit that acquires models that are generated for each elapsed period and that have been trained to output measures for a person when multiple types of feature quantities that represent the state of the person are input; a collection unit that collects, for each of the models for each elapsed period, a first output when a predetermined number of the types of feature quantities are input, and a second output when a portion of the predetermined number of the types of feature quantities are input; and a setting unit that sets the type associated with each of the models for each elapsed period based on the first output and the second output.

[0007] Moreover, an information processing method that is one form of the present disclosure has the following configuration: obtain models that are generated for each elapsed period and that are trained to output measures for a person when multiple types of feature quantities that represent the state of the person are input; collect, for each of the models for each elapsed period, a first output when a predetermined number of the types of feature quantities are input, and a second output when a part of the predetermined number of types of feature quantities are input; and set the type associated with each of the models for each elapsed period based on the first output and the second output.

[0008] A program according to one embodiment of the present disclosure has the following configuration: acquire models that are generated for each elapsed period and that have been trained to output measures for a person when multiple types of feature quantities that represent the state of the person are input; collect, for each of the models for each elapsed period, a first output when a predetermined number of the types of feature quantities are input, and a second output when a part of the predetermined number of types of feature quantities are input; and set the type associated with each of the models for each elapsed period based on the first output and the second output.

[0009] By configuring the present invention as described above, it is possible to efficiently collect information on subjects for proposing measures.

[0010] FIG. 1 is a block diagram showing the configuration of an information processing device according to a first embodiment of the present disclosure. FIG. 2 is a diagram showing the state of data processing by the information processing device disclosed in FIG. 1. FIG. 3 is a diagram showing the state of data processing by the information processing device disclosed in FIG. 1. FIG. 4 is a diagram showing the state of data processing by the information processing device disclosed in FIG. 1. FIG. 5 is a diagram showing the state of data processing by the information processing device disclosed in FIG. 1. FIG. 6 is a flowchart showing the operation of the information processing device disclosed in FIG. 1. FIG. 7 is a block diagram showing the hardware configuration of an information processing device according to a second embodiment of the present disclosure. FIG. 8 is a block diagram showing the configuration of an information processing device according to a second embodiment of the present disclosure.

[0011] <Embodiment 1> A first embodiment of the present invention will be described with reference to Figures 1 to 8. Figure 1 is a diagram for explaining the configuration of an information processing device, and Figures 2 to 8 are diagrams for explaining the processing operation of the information processing device.

[0012] [Configuration] The information processing device 10 of the present disclosure is used to propose a treatment method according to the condition of each individual patient during treatment. In particular, the information processing device 10 has a function of identifying the type of patient information representing the required patient condition for a treatment decision model that outputs a treatment method generated corresponding to each elapsed period during treatment. Note that the information processing device 10 of the present disclosure may also be used to provide measures (e.g., treatments, menus, actions, suggestions) that can be implemented according to the condition of a target person, not limited to treatment, in training, exercise, diet, etc.

[0013] The information processing device 10 is composed of one or more information processing devices each having a calculation device and a storage device. As shown in Fig. 1, the information processing device 10 includes a treatment model generation unit 11, an empirical data generation unit 12, and a feature type setting unit 13, which are constructed by the calculation device executing a program. The information processing device 10 also includes a data storage unit 16 and a model storage unit 17, which are formed in the storage device. Each component will be described in detail below.

[0014] The treatment model generation unit 11 (acquisition unit) generates a treatment decision model (model) that outputs a treatment method (measure) according to the condition of a patient (person). Specifically, the treatment model generation unit 11 first acquires patient information from the data management device 20 as learning data for machine learning, accepts input of the patient information, and stores it in the data storage unit 16. For example, the patient information as learning data includes the patient's condition history and treatment history for each elapsed period. As an example, FIG. 2 shows a conceptual diagram of patient information. In this example, the patient information for patient P1 first includes a "day" setting indicating the elapsed period since the start of treatment, and each stage, representing each elapsed period, is set, such as "Day 1," "Day 2," etc. The patient information includes the patient's condition history and treatment history for each stage, including the patient's condition history "A_1," "B_1," and "C_1" at a stage such as "Day 1," the patient's condition history "Treatment A," the patient's condition history "A_2," "B_2," and "C_2" at a stage such as "Day 2," and the patient's treatment history "Treatment B." Here, the patient's condition history is made up of multiple types of feature quantities that represent the patient's condition. For example, the types of feature quantities include information such as the patient's age, height, weight, medical history, and measured biological information such as the patient's body temperature, blood pressure, heart rate, and blood component values. In the example of Figure 2, different types are represented by different letters such as "A_", "B_", and "C_". In addition, in the example of Figure 2, different treatment methods are represented by different letters such as "Treatment a", "Treatment b", and so on. The patient information includes information on each of multiple patients P1, P2, P3, ...

[0015] 3, the treatment model generation unit 11 learns the patient information of multiple patients P1, P2, P3, ... for each stage, which is each elapsed period, and generates treatment decision models "D*_1," "D*_2," and "D*_3" for each stage. Specifically, when generating a treatment decision model for the "day 1" stage, for example, the treatment model generation unit 11 learns to input all types of feature quantities "A_1," "B_1," and "C_1," which are the condition history of each patient on "day 1," and to output the treatment history "treatment a" of each patient on "day 1," thereby generating a treatment decision model "D*_1" for "day 1." Furthermore, when generating a treatment decision model for the "day 2" stage, for example, the treatment model generation unit 11 inputs all types of feature quantities "A_1," "B_1," "C_1," "A_2," "B_2," and "C_2" that are the condition history of each patient up to "day 2," i.e., the condition history of each patient on "day 1" and "day 2," and learns to output the treatment history "treatment b" of each patient on "day 2," thereby generating a treatment decision model "D*_2" for "day 2." Note that in the above description, only "A_," "B_," and "C_" are shown as all types (a predetermined number of types) of feature quantities that are the condition history input during learning, but in reality, there may be many more types. Furthermore, the number of types of feature quantities input during learning is not limited to all types provided, but may be a preset number (a predetermined number).

[0016] The treatment decision model generated in this manner is configured to output a treatment method for a given stage by inputting the condition history of a patient for whom a treatment method has not been determined at that stage. For example, as shown in FIG. 4 , by inputting multiple types of feature quantities "A_1," "B_1," "C_1," "A_2," "B_2," and "C_2," which are the condition history up to "Day 2" for Patient P, for whom a treatment method for "Day 2" has not been determined, i.e., the condition history for "Day 1" and "Day 2," into a treatment decision model "D*_2" for the corresponding stage, "Day 2," the treatment method "Treatment b" for "Day 2" is output. Note that the elapsed period corresponding to the above-described stage is not limited to a period in days, but may be any period, such as a period in months.

[0017] Furthermore, the therapeutic model generation unit 11 prioritizes the types of influential features for each of the treatment decision models "D*_1," "D*_2," and "D*_3" for each stage generated as described above. For example, as the therapeutic model generation unit 11 performs machine learning of the treatment decision model described above, weights are assigned to each type of feature, and the priority is set according to the weight values. As an example, as shown in FIG. 5 , for the treatment decision model "D*_1" at the "Day 1" stage, priority feature types are assigned in the order of priority: feature types "A_1," "D_1," "B_1," "E_1," and "F_1." For the treatment decision model "D_2" at the "Day 2" stage, priority feature types are assigned in the order of priority: feature types "C_2," "D_2," "B_1," "A_1," and "A_2." In this case, the priority feature types at the "Day 2" stage also include the feature types "B_1" and "A_1" for "Day 1." The priority of the types of feature amounts set in each treatment decision model may be set by any method.

[0018] The treatment model generation unit 11 then stores the treatment decision models “D*_1,” “D*_2,” and “D*_3” for each stage generated as described above and the priority feature types corresponding to each of them in the model storage unit 17. Note that the treatment model generation unit 11 may also receive and acquire input of the treatment decision models for each stage prepared in advance and the priority feature types corresponding to each of them, and store them in the model storage unit 17.

[0019] The empirical data generator 12 (collection unit) collects, for each patient, a first output f', i.e., a treatment method, when all types of feature quantities are input, and a second output f', i.e., a treatment method, when a portion of all types of feature quantities are input, for each treatment decision model for each stage. The empirical data generator 12 then outputs a determination result y indicating whether the first output f' and the second output f match. At this time, the empirical data generator 12 changes the number and combination of some of the types of feature quantities and inputs the resulting combinations of feature quantities to each treatment decision model. For example, for the treatment decision model "D*_1" for the "Day 1" stage shown in FIG. 5 , combinations of the prioritized feature quantities are set in increasing order of priority, starting with the first feature quantity, and the following combinations of feature quantities are input: ("A_1"), ("A_1", "D_1"), ("A_1", "D_1", "B_1"), ("A_1", "D_1", "B_1", "E_1"), etc. In this case, the number of types to be combined is one or more and is less than the total number of types (less than a predetermined number). Note that feature quantities of types not included in the combination are input as "0." Then, it is determined whether the second output f for each combined type of feature quantity matches the first output f' for all types of feature quantities, and if f = f', the determination result is set to y = 1. In this way, the empirical data generation unit 12 generates empirical data (aggregated data) consisting of (X': set of feature quantities of the combined types, y': set of determination results). Here, an example of the empirical data (X', y') is shown below. "For the first day's stage" ("A_1", "0", "0", "0", ..., 0) ("A_1", "D_1", "0", "0", ..., 0) ("A_1", "D_1", "B_1", "0", ..., 1) ("A_1", "D_1", "B_1", "E_1", ..., 1) "For the second day's stage" ("C_2", "0", "0", ..., 0) ("C_2", "D_2", "0", "0", ..., 0) ("C_2", "D_2", "B_1", "0", ..., 1) ("C_2", "D_2", "B_1", "A_1", ..., 1)

[0020] In this way, the empirical data generator 12 generates empirical data (X', y') for each stage using the patient information of each patient. Note that, in the above example, the empirical data generator 12 generates the combined types of feature quantities to be input to each treatment decision model for each stage by increasing the number of types in order of priority and changing the number and combinations of types, but the combined types of feature quantities may be generated by randomly changing the number of types to be combined or randomly changing the combinations of types.

[0021] The feature type setting unit 13 learns the relationship between the empirical data (X', y') for each stage and generates a binary discriminant model H (second model) that discriminates the value of y (0 or 1) for the combined type of feature. As a result, by inputting the feature of each combination type into the binary discriminant model H, it outputs whether the treatment method (second output) output when the feature of that combination type is input into the treatment decision model matches the treatment method (first output) output when all types of feature are input into the treatment decision model. In other words, the feature of a combination type for which the output by the binary discriminant model H is y = 1 matches the treatment method determined from all types of feature, and therefore it can be determined that a treatment can be determined using the feature of that combination type.

[0022] Based on the above-described concept, the feature type setting unit 13 uses the binary discriminant model H generated for each stage to calculate the required number of priority feature types set in the treatment decision model for that stage. Specifically, for each patient, the feature type setting unit 13 inputs a different combination of feature types from the patient information for that stage into the binary discriminant model H for that stage, and checks the number of feature types for which the output is y = 1. Then, for each stage, the unit 13 calculates the average number of types for which the output is y = 1 for all patients, and sets this average as the required number of priority feature types. Through the above-described process, as an example, as shown in FIG. 6 , the required number of priority feature types set in the treatment decision model for the "first day" stage is calculated to be "2," and the required number of priority feature types set in the treatment decision model for the "second day" stage is calculated to be "4." Alternatively, the feature type setting unit 13 may calculate the minimum number of feature types for which the output is y = 1 for each patient, and set the required number as the average number of minimum feature types across all patients. However, the method for calculating the required number is not limited to the above-described method, and any method may be used, such as using the most frequent number of types for which the output y=1 as the required number.

[0023] The feature type setting unit 13 further resets each priority feature type based on the required number of priority feature types set in the treatment decision model for each stage as described above. In particular, the feature type setting unit 13 sets some of the priority feature types set in the treatment decision model for a later stage, among the stages located before and after in time, to the priority feature types set in the treatment decision model for the earlier stage.

[0024] Here, specific processing by the feature type setting unit 13 will be described using, as an example, the priority feature types set in the treatment decision model for the "Day 1" stage and the priority feature types set in the treatment decision model for the "Day 2" stage shown in FIG. 7 . Assume that the required number of priority feature types for the "Day 1" stage is "2," and the required number of priority feature types for the "Day 2" stage is "4." In this case, the feature type setting unit 13 first checks the priority feature types for the later stage, "Day 2," and extracts feature types representing the patient's condition history for the earlier stage, "Day 1," from the priority feature types ranked from first to fourth, which correspond to the required number "4." The feature type setting unit 13 then extracts "B_1" and "A_1," which are the feature types ranked third and fourth in priority for the later stage, "Day 2." Next, the feature type setting unit 13 checks the priority feature type of the previous stage, "Day 1," and checks whether the feature types "B_1" and "A_1" extracted from the stage of "Day 2" are present among the priority feature types ranked from first to second, which corresponds to the required number "2" of priorities. Since the feature type "B_1" is not present among the priority feature types ranked from second to third, which corresponds to the required number "2" of priorities of the previous stage, "Day 1," the feature type setting unit 13 inserts and sets the feature type "B_1" as the priority feature type of "Day 1." That is, the feature type setting unit 13 sets "B_1" as the feature type that should be acquired with priority in the stage of "Day 1." At this time, as indicated by the arrow in FIG. 7 , the feature type setting unit 13 inserts and sets the extracted feature type "B_1" in the intermediate position from first to second, which corresponds to the required number "2," of the priority feature types of "Day 1," that is, between the first and second priorities.

[0025] The method by which the feature type setting unit 13 extracts feature types to be inserted from a later stage into a previous stage is not limited to the above-described method. For example, when generating empirical data as described above, the feature type setting unit 13 may extract feature types from some of the feature types used in the input if the determination result in the empirical data is y = 1. In addition, the extracted feature type may be inserted into the priority feature type of another stage at any position, for example, at the beginning or end of the ranks from 1 to the required number, or at a position other than the required number. Furthermore, the priority feature type of each stage does not necessarily need to be prioritized, and the required number does not need to be calculated. A type extracted from the priority feature type of a given stage based on the empirical data as described above may be inserted into the priority feature type of another stage. Furthermore, the priority feature type of each stage does not need to be initially set. A feature type extracted based on the empirical data as described above may be newly set as the priority feature type.

[0026] [Operation] Next, the operation of the information processing device 10 described above will be described with reference to the flowchart in Fig. 8. The information processing device 10 first acquires patient information as learning data from the data management device 20 (step S1). As shown in Fig. 2, the patient information as learning data includes, for each elapsed period, multiple types of feature quantities (e.g., "A_1") representing the patient's condition history and a treatment history (e.g., "treatment a").

[0027] Next, the information processing device 10 uses the above-mentioned patient information to generate treatment decision models (such as "D*_1" and "D*_2") that output treatment methods in response to input of the patient's condition history for each stage, which is an elapsed period such as "day 1" and "day 2," as shown in Fig. 3 (step S2). At this time, the information processing device 10 further sets priority feature types by prioritizing influential feature types for each treatment decision model for each stage, as shown in Fig. 5 (step S3).

[0028] Next, using the patient information described above, the information processing device 10 collects, for each patient, a first output f', which is the output when all types of feature quantities are input, i.e., the treatment method, and a second output f', which is the output when some of all types of feature quantities are input, i.e., the treatment method, for each treatment decision model for each stage, along with a determination result y that determines whether the first output f' and the second output f match. At this time, the information processing device 10 changes the number and combination of some of the types of feature quantities and inputs the combined types of feature quantities into each treatment decision model. As a result, the information processing device 10 generates empirical data consisting of (X': a set of combined types of feature quantities, y': a set of determination results) (step S4).

[0029] Next, the information processing device 10 learns the relationship between the empirical data (X', y') for each stage and generates a binary discriminant model H that discriminates the value of the judgment result y (0 or 1) for the combined types of feature quantities (step S5).The information processing device 10 then uses the binary discriminant model H generated for each stage to calculate the required number of priority feature quantities set in the treatment decision model for that stage (step S6).For example, for each patient, the information processing device 10 inputs different combinations of feature quantities from the patient information for that stage into the binary discriminant model H for each stage, and calculates the required number of priority feature quantities by checking the number of feature quantities that result in an output of y = 1.

[0030] Next, the information processing device 10 resets each priority feature type based on the required number of priority feature types set in the treatment decision model for each stage (step S7). For example, as shown in FIG. 7, the information processing device 10 sets some of the priority feature types set in the treatment decision model for the later stage "Day 2" as the priority feature types set in the treatment decision model for the earlier stage "Day 1". The set priority feature types may be presented to a user, such as a medical professional, by being output to a user terminal (not shown). This allows the user to confirm the priority feature types and make a decision to modify them.

[0031] As described above, the information processing device 10 of the present disclosure first generates empirical data including a determination result indicating whether a treatment method output when all types of feature quantities representing a person's condition are input into a treatment decision model for each elapsed period matches a treatment method output when a subset of all types of feature quantities are input. Then, based on the empirical data, the types of feature quantities required for each treatment decision model for each elapsed period can be prioritized for that model, thereby streamlining information collection regarding the types of feature quantities required. In particular, the types of feature quantities required for a later elapsed period can also be prioritized for a previous elapsed period, ensuring that such types of feature quantities are reliably acquired in the previous elapsed period. This allows treatment method suggestions to be received with the same accuracy as when all types of feature quantities are input, even when fewer types of feature quantities are input. As a result, highly accurate treatment methods can be obtained even with a limited number of feature quantities, reducing the cost and time required to collect patient information and enabling appropriate treatment methods to be quickly and easily proposed. In this way, the information processing device 10 can support the decision-making of medical professionals, such as doctors, for example.

[0032] <Embodiment 2> Next, a second embodiment of the present disclosure will be described with reference to Fig. 9 and Fig. 10. Fig. 9 and Fig. 10 are block diagrams showing the configuration of an information processing device in embodiment 2. Note that this embodiment shows an outline of the configuration of the information processing device described in the above-mentioned embodiment.

[0033] First, the hardware configuration of the information processing device 100 in this embodiment will be described with reference to Fig. 9. The information processing device 100 is configured as a general information processing device, and is equipped with the following hardware configuration, for example: CPU (Central Processing Unit) 101 (arithmetic unit); ROM (Read Only Memory) 102 (storage device); RAM (Random Access Memory) 103 (storage device); programs 104 loaded into RAM 103; a storage device 105 that stores the programs 104; a drive device 106 that reads and writes data from and to a storage medium 110 external to the information processing device; a communication interface 107 that connects to a communication network 111 external to the information processing device; an input / output interface 108 that inputs and outputs data; and a bus 109 that connects the various components.

[0034] 9 shows an example of the hardware configuration of the information processing device 100, and the hardware configuration of the information processing device is not limited to the above-described case. For example, the information processing device may be configured with only a part of the above-described configuration, such as excluding the drive device 106. Furthermore, the information processing device may use a GPU (Graphics Processing Unit), a DSP (Digital Signal Processor), an MPU (Micro Processing Unit), an FPU (Floating Point Number Processing Unit), a PPU (Physics Processing Unit), a TPU (Tensor Processing Unit), a quantum processor, a microcontroller, or a combination thereof, instead of the above-described CPU.

[0035] The information processing device 100 can be equipped with an acquisition unit 121, a collection unit 122, and a setting unit 123 shown in FIG. 10 by having the CPU 101 acquire and execute the program group 104. The program group 104 is stored in advance in the storage device 105 or the ROM 102, for example, and is loaded into the RAM 103 and executed by the CPU 101 as needed. The program group 104 may be supplied to the CPU 101 via the communication network 111, or may be stored in advance in the storage medium 110, and the drive device 106 may read the program and supply it to the CPU 101. However, the acquisition unit 121, the collection unit 122, and the setting unit 123 described above may be constructed using dedicated electronic circuits for realizing such means.

[0036] The acquisition unit 121 acquires a model that is generated for each elapsed period and that has been trained to output measures for a person by inputting a plurality of types of feature quantities that represent the state of the person.

[0037] The collection unit 122 collects, for each model for each elapsed period, a first output when a predetermined number of types of feature quantities are input, and a second output when a portion of the predetermined number of types of feature quantities are input.

[0038] The setting unit 123 sets a type associated with each model for each elapsed period based on the first output and the second output.

[0039] With the above-described configuration, the present disclosure can set the types of features required for a model for each elapsed period. This makes it possible to efficiently collect information about a person for policy proposals, and even when a small number of features are input, it is possible to propose highly accurate policies that appropriately correspond to the person's condition. Furthermore, it is possible to reduce the cost and time required to collect a small number of types of features, making it possible to quickly and easily propose appropriate policies.

[0040] The above-described program can be stored and supplied to a computer using various types of non-transitory computer-readable media. Non-transitory computer-readable media include various types of tangible storage media. Examples of non-transitory computer-readable media include magnetic recording media (e.g., flexible disks, magnetic tapes, hard disk drives), magneto-optical recording media (e.g., magneto-optical disks), CD-ROMs (Read Only Memory), CD-Rs, CD-R / Ws, and semiconductor memories (e.g., mask ROMs, PROMs (Programmable ROMs), EPROMs (Erasable PROMs), flash ROMs, and RAMs (Random Access Memory)). The program may also be supplied to a computer by various types of transitory computer-readable media. Examples of transitory computer-readable media include electrical signals, optical signals, and electromagnetic waves. The transitory computer-readable media can be supplied to a computer via wired communication paths such as electric wires and optical fibers, or via wireless communication paths.

[0041] Although the present disclosure has been described above with reference to the above-described embodiments, the present disclosure is not limited to the above-described embodiments. Various modifications that are understandable to those skilled in the art can be made to the configuration and details of the present disclosure within the scope of the present disclosure. Furthermore, at least one or more of the functions of the acquisition unit 121, collection unit 122, and setting unit 123 described above may be executed by an information processing device installed and connected anywhere on a network, i.e., may be executed by so-called cloud computing.

[0042] <Supplementary Notes> Some or all of the above embodiments may be described as in the following supplementary notes. The following provides an outline of the configurations of an information processing device, an information processing method, and a program according to the present disclosure. However, the present disclosure is not limited to the following configurations. (Supplementary Note 1) An information processing device comprising: an acquisition unit that acquires models that are generated for each elapsed period and that are machine-learned to output measures for a person when multiple types of feature quantities representing the state of the person are input; a collection unit that collects, for each of the models for the elapsed period, a first output when a predetermined number of the types of feature quantities are input, and a second output when a portion of the predetermined number of types of feature quantities are input; and a setting unit that sets the type associated with each of the models for the elapsed period based on the first output and the second output. (Supplementary Note 2) The information processing device according to Supplementary Note 1, wherein the collection unit collects, for each of the models for the elapsed period, the second outputs when feature quantities are input in each case where some of the predetermined number of types are changed. (Supplementary Note 3) The information processing device according to Supplementary Note 2, wherein the collection unit collects the second outputs when the feature quantities are input for each of the models for each of the elapsed periods, with the number and / or combination of some of the types among the predetermined number being changed. (Supplementary Note 4) The information processing device according to Supplementary Note 2 or 3, wherein the collection unit collects aggregated data including whether the second outputs match the first outputs when the feature quantities are input for each of the models corresponding to the same elapsed period, with the number and / or combination of some of the types among the predetermined number being changed, and the setting unit sets the types to be associated with each of the models for each of the elapsed periods based on the aggregated data. (Supplementary Note 5) The information processing device according to Supplementary Note 4, wherein the setting unit sets a required number of the types to be associated with each of the models for each of the elapsed periods based on the aggregated data, and sets the types based on the required number.(Supplementary Note 6) The information processing device according to Supplementary Note 5, wherein the types are set in association with each of the models for each of the elapsed periods and are prioritized in advance, and the setting unit resets the types based on the priority and the required number of the types associated with each of the models for each of the elapsed periods. (Supplementary Note 7) The information processing device according to Supplementary Note 6, wherein the setting unit resets the types associated with the models of a previous elapsed period based on the priority and the required number set for the types associated with the models of a previous elapsed period and the priority and the required number set for the types associated with the models of a later elapsed period, of the elapsed periods located before and after in time. (Supplementary Note 8) The information processing device according to any one of Supplementary Notes 5 to 7, wherein the setting unit generates a second model for each elapsed period based on the aggregated data, taking as input the feature quantities for each case in which some of the types among the predetermined number are changed, and outputting whether the first output and the second output match, and sets the required number based on an output when the feature quantities for each case in which some of the types among the predetermined number are changed are input to the second model. (Supplementary Note 9) An information processing method comprising: acquiring a model generated for each elapsed period and machine-learned to output a measure for a person by inputting multiple types of feature quantities representing a person's state; collecting, for each of the models for each elapsed period, a first output when a predetermined number of the types of feature quantities are input and a second output when some of the types of feature quantities among the predetermined number are input; and setting the type associated with each of the models for each elapsed period based on the first output and the second output. (Supplementary Note 10) The information processing method according to Supplementary Note 9, further comprising: collecting the second outputs when the feature quantities are input for each of the models for each of the elapsed periods, in each case where some of the predetermined number of types are changed.(Supplementary Note 11) The information processing method according to Supplementary Note 10, comprising collecting aggregated data including whether the second output and the first output match when the feature amounts in each case of changing some of the predetermined number of types are input to the models corresponding to the same elapsed period, and setting the types to be associated with each of the models for each of the elapsed periods based on the aggregated data. (Supplementary Note 12) The information processing method according to Supplementary Note 11, comprising setting a required number of the types to be associated with each of the models for each elapsed period based on the aggregated data, and setting the types based on the required number. (Supplementary Note 13) The information processing method according to Supplementary Note 12, wherein the types are set to be associated with each of the models for each elapsed period and prioritized in advance, and the types are reset based on the priority and the required number of the types associated with each of the models for each of the elapsed periods. (Supplementary Note 14) The information processing method according to Supplementary Note 13, wherein the types associated with the models of a previous elapsed period are reset based on the priority order and the required number set for the types associated with the models of a previous elapsed period and the priority order and the required number set for the types associated with the models of a later elapsed period, among the elapsed periods located before and after in time. (Supplementary Note 15) The information processing method according to any of Supplements 12 to 14, wherein for each elapsed period, a second model is generated based on the aggregated data, and the feature amounts for each case in which some of the types among the predetermined number are changed are input, and the second model outputs whether or not the first output matches the second output, and the required number is set based on the output when the feature amounts for each case in which some of the types among the predetermined number are changed are input into the second model.(Supplementary Note 16) A computer-readable storage medium storing a program that causes a computer to execute the following processes: obtaining a machine-learned model that is generated for each elapsed period and that outputs measures for a person when multiple types of feature quantities that represent the state of the person are input; collecting, for each of the models for each elapsed period, a first output when a predetermined number of the types of feature quantities are input, and a second output when a part of the predetermined number of types of feature quantities are input; and setting the type associated with each of the models for each elapsed period based on the first output and the second output.

[0043] REFERENCE SIGNS LIST 10 Information processing device 11 Treatment model generation unit 12 Empirical data generation unit 13 Feature type setting unit 16 Data storage unit 17 Model storage unit 100 Information processing device 101 CPU 102 ROM 103 RAM 104 Program group 105 Storage device 106 Drive device 107 Communication interface 108 Input / output interface 109 Bus 110 Storage medium 111 Communication network 121 Acquisition unit 122 Collection unit 123 Setting unit

Claims

1. an acquisition unit that acquires a machine-learned model that outputs a measure for a person by inputting multiple types of feature amounts that are generated for each elapsed period and represent the state of the person; a collection unit that collects, for each of the models for each elapsed period, a first output when a predetermined number of the types of feature quantities are input, and a second output when a part of the predetermined number of the types of feature quantities are input; a setting unit that sets the type associated with each of the models for each elapsed period based on the first output and the second output; An information processing device comprising:

2. 2. The information processing device according to claim 1, the collecting unit collects the second outputs when the feature amounts in each case where some of the predetermined number of types are changed are input to each of the models for each elapsed period. Information processing device.

3. 3. The information processing device according to claim 2, the collection unit collects, for each of the models for each elapsed period, each of the second outputs when the feature amounts in each case where the number and / or combination of some of the types among the predetermined number are changed are input. Information processing device.

4. 3. The information processing device according to claim 2, the collecting unit collects aggregated data including whether or not each of the second outputs matches the first output when the feature amounts in each case where some of the types among the predetermined number are changed are input to the model corresponding to the same elapsed period; the setting unit sets the type associated with each of the models for each elapsed period based on the aggregated data. Information processing device.

5. 5. The information processing device according to claim 4, the setting unit sets a required number of the types associated with each of the models for each elapsed period based on the aggregated data, and sets the types based on the required number. Information processing device.

6. 6. The information processing device according to claim 5, the types are set in association with the respective models for each elapsed period, and are prioritized in advance; the setting unit resets the types based on the priority order and the required number of the types associated with each of the models for each elapsed period. Information processing device.

7. 7. The information processing device according to claim 6, the setting unit resets the type associated with the model in the previous elapsed period based on the priority order and the required number set for the type associated with the model in the previous elapsed period and the priority order and the required number set for the type associated with the model in the later elapsed period, among the elapsed periods located before and after in time. Information processing device.

8. 6. The information processing device according to claim 5, the setting unit generates, for each elapsed period, a second model based on the aggregated data, which receives as input the feature amounts for each case in which some of the types among the predetermined number are changed, and outputs whether or not the first output and the second output match, and sets the required number based on an output when the feature amounts for each case in which some of the types among the predetermined number are changed are input to the second model. Information processing device.

9. A machine-learned model is obtained that outputs measures for a person by inputting multiple types of feature quantities that are generated for each elapsed period and represent the person's state, and collecting, for each of the models for each elapsed period, a first output when a predetermined number of the types of feature quantities are input, and a second output when a part of the predetermined number of the types of feature quantities are input; setting the type associated with each of the models for each elapsed period based on the first output and the second output; Information processing methods.

10. A machine-learned model is obtained that outputs measures for a person by inputting multiple types of feature quantities that are generated for each elapsed period and represent the person's state, and collecting, for each of the models for each elapsed period, a first output when a predetermined number of the types of feature quantities are input, and a second output when a part of the predetermined number of the types of feature quantities are input; setting the type associated with each of the models for each elapsed period based on the first output and the second output; A program that causes a computer to perform a process.