Learning model generation device, test value prediction device, learning model generation method, test value prediction method, and program

Abstract

This learning model generation device 10 is equipped with a learning model generation unit 11 which, when a function expressing a change in an inspection value obtained by inspecting a person is set, generates a learning model in which the inspection value is the explanatory variable and the parameter is the objective variable, by performing machine learning using inspection values of sample people and parameters of the function for the sample people as training data.

Description

Learning model generation device, test value prediction device, learning model generation method, test value prediction method, and computer-readable recording medium

[0001] The present disclosure relates to a test value prediction device and a test value prediction method for predicting a user's future test values, and also to a learning model generation device and a learning model generation method for generating a learning model used to predict test values. Furthermore, the present disclosure relates to a computer-readable recording medium having recorded thereon a program for realizing these devices and methods.

[0002] In recent years, the use of big data in the medical and healthcare fields has become more common in order to address lifestyle habits. Big data analysis is expected to enable prediction of the future health status of individual employees and to recommend specific improvement measures.

[0003] For example, Non-Patent Document 1 discloses a system that predicts the values ​​of a user's test items using a learning model that has learned the relationship between test values ​​and lifestyle habits through machine learning. In the system disclosed in Non-Patent Document 1, when data indicating the user's lifestyle habits is input into the learning model, the learning model outputs the user's future test values. The user can know how their test values ​​will change in the future if they continue their current lifestyle habits.

[0004] Atsushi Taniguchi and Hiroshi Tamano, "Construction of a predictive model that takes into account the interpretability of age-related predictions in health checkup predictions," FIT2018 (17th Forum on Information Science and Technology), 2018, Vol. 2, pp. 45-50

[0005] When predicting test values ​​for a three-year period, the system disclosed in the above-mentioned Non-Patent Document 1 first predicts test values ​​from the present up to one year from now, and then uses the predicted test values ​​for one year to predict values ​​for one to two years from now.Furthermore, the system disclosed in the above-mentioned Non-Patent Document 1 uses the predicted test values ​​for the second year to predict values ​​for two to three years from now.

[0006] For this reason, in the system disclosed in the aforementioned Non-Patent Document 1, the initially predicted test values ​​one year from now, the test values ​​one year to two years from now, and the test values ​​two years to three years from now may each show different trends of change. As a result, the interpretability of the learning model is reduced. For example, suppose the test values ​​increase from the present to one year from now, decrease from one year to two years from now, and then increase again from two years to three years from now. In this case, the interpretability of the learning model is reduced, making it difficult to interpret the behavior of the learning model.

[0007] An example of an objective of the present disclosure is to improve interpretability by showing a certain trend in predicting a user's test values.

[0008] In order to achieve the above object, a learning model generation device according to one aspect of the present disclosure is characterized by comprising: a learning model generation unit that, when a function representing the progress of test values ​​obtained from testing a person is set, performs machine learning using the test values ​​of a sample person and parameters of the function for the sample person as training data to generate a learning model in which the test values ​​are used as explanatory variables and the parameters are used as target variables.

[0009] In order to achieve the above object, a test value prediction device according to one aspect of the present disclosure comprises: a test value acquisition unit that acquires test values ​​of a person to be predicted; a parameter prediction unit that inputs the acquired test values ​​into a learning model that performs machine learning to learn the relationship between test values ​​and parameters of a function that represents their trends, and predicts the parameters of the person to be predicted; and a test value prediction unit that inputs the predicted parameters into the function, and predicts future test values ​​of the person to be predicted.

[0010] In order to achieve the above object, a learning model generation method according to one aspect of the present disclosure is characterized by comprising: a learning model generation step of, when a function representing the progress of test values ​​obtained from testing a person is set, performing machine learning using the test values ​​of a sample person and parameters of the function for the sample person as training data to generate a learning model in which the test values ​​are used as explanatory variables and the parameters are used as target variables.

[0011] In order to achieve the above object, a test value prediction method according to one aspect of the present disclosure comprises: a test value acquisition step of acquiring test values ​​of a person to be predicted; a parameter prediction step of predicting the parameters of the person to be predicted by inputting the acquired test values ​​into a learning model that performs machine learning to learn the relationship between test values ​​and parameters of a function that represents their trends; and a test value prediction step of inputting the predicted parameters into the function to predict future test values ​​of the person to be predicted.

[0012] Furthermore, in order to achieve the above object, a first computer-readable recording medium according to one aspect of the present disclosure is characterized in that it records a program including instructions for causing a computer to execute a learning model generation step in which, when a function representing the progress of test values ​​obtained from testing a person is set, machine learning is performed using the test values ​​of a sample person and parameters of the function for the sample person as training data to generate a learning model in which the test values ​​are used as explanatory variables and the parameters are used as target variables.

[0013] Furthermore, in order to achieve the above object, a second computer-readable recording medium according to one aspect of the present disclosure is characterized in that it records a program including instructions for causing a computer to execute: a test value acquisition step of acquiring test values ​​of a person to be predicted; a parameter prediction step of inputting the acquired test values ​​into a learning model that performs machine learning to learn the relationship between test values ​​and parameters of a function that represents their trends, thereby predicting the parameters of the person to be predicted; and a test value prediction step of inputting the predicted parameters into the function, thereby predicting future test values ​​of the person to be predicted.

[0014] As described above, according to the present disclosure, a certain trend can be indicated in the prediction of a user's test values, thereby improving interpretability.

[0015] FIG. 1 is a configuration diagram showing a schematic configuration of an example of a learning model generation device. FIG. 2 is a configuration diagram specifically showing the configuration of an example of a learning model generation device. FIG. 3 is a diagram showing an example of a function. FIG. 4 is a diagram showing an example of training data. FIG. 5 is a flow diagram showing the operation of an example of a learning model generation device. FIG. 6 is a configuration diagram showing the configuration of a modified example of a learning model generation device. FIG. 7 is a configuration diagram showing a schematic configuration of an example of a test value prediction device. FIG. 8 is a configuration diagram specifically showing the configuration of an example of a test value prediction device. FIG. 9 is a flow diagram showing the operation of an example of a test value prediction device. FIG. 10 is a block diagram showing an example of a computer that realizes the learning model generation device and the test value prediction device.

[0016] First Embodiment In the first embodiment, a learning model generating device, a learning model generating method, and a program will be described below with reference to FIGS.

[0017] [Device Configuration] First, the schematic configuration of a learning model generation device will be described with reference to Fig. 1. Fig. 1 is a diagram showing the schematic configuration of an example of a learning model generation device.

[0018] 1, a learning model generation device 10 according to the first embodiment generates a learning model for predicting a user's future test score. As shown in FIG. 1, the learning model generation device 10 includes a learning model generation unit 11.

[0019] First, a function is set that represents the transition of test values ​​obtained from tests on people. In this case, the learning model generation unit 11 performs machine learning using the test values ​​of the sample people and the parameters of the function for the sample people (hereinafter referred to as "function parameters") as training data to generate a learning model. The learning model uses the test values ​​as explanatory variables and the parameters of the function as objective variables.

[0020] As described above, in the first embodiment, a function representing the transition of test values ​​is set, and a learning model is generated by machine learning the relationship between test values ​​and function parameters. Therefore, the function parameters are first predicted by the learning model, and future test values ​​are predicted from the predicted function parameters. Therefore, according to the first embodiment, a certain trend is indicated in the prediction of a user's test value, which makes it possible to improve interpretability.

[0021] Next, the configuration and functions of the learning model generation device 10 will be specifically described with reference to Figures 2 to 4. Figure 2 is a block diagram specifically showing the configuration of an example of the learning model generation device.

[0022] 2, in the first embodiment, the learning model generation device 10 is connected to a database 30 via a network or the like so as to be able to communicate data with the database 30. The database 30 stores training data for machine learning.

[0023] 2, the learning model generation device 10 includes a memory unit 12 in addition to the learning model generation unit 11. The memory unit 12 stores parameters 13 (hereinafter referred to as "model parameters") for constructing the learning model.

[0024] In the present invention, "generating a learning model" refers to generating or updating model parameters of the learning model. That is, the learning model generation device 10 generates or updates model parameters 13 of the learning model by executing machine learning, and stores the generated or updated model parameters 13 in the storage unit 12. As a result, a learning model is generated. Furthermore, the generation of a learning model is performed for each test item, i.e., for each test value.

[0025] In the first embodiment, a function that represents the transition of test values ​​after a set period has elapsed is used. A specific example of the function is the function expressed by the following equation 1.

[0026] (Math. 1) y = alog t + b

[0027] The functions used in the first embodiment will be described with reference to Fig. 3. Fig. 3 is a diagram showing an example of the functions.

[0028] The graph shown in Figure 3 has the test value on the vertical axis and the number of years elapsed on the horizontal axis. In the example of Figure 3, the set period mentioned above is one year. The reference time "0" is set arbitrarily. The function shown in Equation 1 above is set based on this set period of one year, and the vertical axis passing through one year serves as the baseline. In Figure 3, the black circles represent actual values. The function shown in Equation 1 above is set as an approximation curve of actual values ​​from two years onwards.

[0029] In the above equation 1, b corresponds to the actual value on the baseline, and a is a function parameter determined by test values ​​from two years onward. In other words, the function parameter a is a parameter determined by the changes in a person's past test values ​​over time, and is a value that differs for each person.

[0030] Therefore, in the first embodiment, machine learning is performed using the function parameter a in the above-mentioned Equation 1 as the objective variable. Furthermore, in the first embodiment, in addition to the above-mentioned test values, the contents of the medical questionnaire are also used as explanatory variables. Specific examples of the contents of the medical questionnaire include a person's lifestyle habits, such as whether they currently or previously smoke, the amount of alcohol they consume per day, and the frequency of exercise (the frequency of exercise per unit time).

[0031] The database 30 stores, as training data for each sample person, set data of the function parameter a, past test value data one year or later from the reference time, and the contents of the medical questionnaire at the time the test value was obtained, as shown in Fig. 4. Fig. 4 is a diagram showing an example of training data.

[0032] In the first embodiment, the learning model generation unit 11 performs machine learning using, as training data, the contents of the medical questionnaire of the sample person in addition to the test value and the function parameter a. Therefore, the learning model generation unit 11 reads the above-described group data as training data from the database 30, and performs machine learning using the read group data.

[0033] Specifically, the learning model generation unit 11 uses a machine learning engine to construct a learning model such as a neural network in memory, and inputs the test values ​​and the contents of the medical questionnaire into the constructed learning model.

[0034] Then, the learning model generation unit 11 calculates the difference between the output value of the learning model and the function parameter a serving as teacher data, and updates the model parameters 13 in the learning model so that the calculated difference approaches 0. After that, when the learning model generation unit 11 has finished updating the model parameters 13 using all the training data, it stores the final model parameters 13 in the storage unit 12.

[0035] [Device Operation] Next, the operation of the learning model generation device 10 in embodiment 1 will be described with reference to FIG. 5. FIG. 5 is a flow diagram showing an example of the operation of the learning model generation device. In the following description, reference will be made to FIGS. 1 to 4 as appropriate. In addition, in embodiment 1, a learning model generation method is implemented by operating the learning model generation device 10. Therefore, the description of the learning model generation method in embodiment 1 will be replaced by the following description of the operation of the learning model generation device 10.

[0036] First, the function shown in the above equation 1 is set as a premise. Furthermore, the database 30 stores, as training data, the function parameter a, past test value data from one year after the reference time, and the contents of the medical questionnaire at the time the test value was obtained for each sample person.

[0037] As shown in FIG. 5, first, the learning model generation unit 11 acquires a set of training data from the database 30 (step A1).

[0038] Next, the learning model generation unit 11 performs machine learning using the training data acquired in step A1, and updates the parameters of the learning model (step A2).

[0039] Specifically, in step A2, the learning model generation unit 11 uses a machine learning engine to construct a learning model such as a neural network in memory, and then inputs the test values ​​and the contents of the medical questionnaire contained in the training data into the constructed learning model.

[0040] Furthermore, the learning model generation unit 11 calculates the difference between the output value of the learning model and the function parameter a serving as training data. Subsequently, the learning model generation unit 11 updates the model parameters 13 of the learning model so that the calculated difference approaches zero.

[0041] Next, the learning model generation unit 11 determines whether machine learning has been completed for all training data (step A3).

[0042] If the result of the determination in step A3 is that machine learning has not been completed for all sets of training data, the learning model generation unit 11 executes step A1 again to obtain a new set of training data.

[0043] On the other hand, if the result of the judgment in step A3 is that machine learning has been completed for all sets of training data, the learning model generation unit 11 stores the model parameters 13 of the final learning model that has been updated using all training data in the memory unit 12 (step A4).

[0044] As described above, in the first embodiment, a function representing the transition of test values ​​one year from now and thereafter, which would otherwise be prone to poor interpretability in the past, is set, and a learning model is generated by machine learning the relationship between test values ​​and function parameters. When this learning model is used to predict future test values, as will be described in the second embodiment below, the learning model predicts function parameters, and future test values ​​are predicted from the predicted function parameters, thereby obtaining prediction results with high interpretability. In other words, according to the first embodiment, a certain trend can be indicated in the prediction of a user's test values, thereby improving interpretability.

[0045] [Program] The program in the first embodiment may be any program that causes a computer to execute steps A1 to A4 shown in Fig. 5. By installing and executing this program in a computer, the learning model generation device 10 and the learning model generation method in the first embodiment can be realized. In this case, the processor of the computer functions as the learning model generation unit 11 and performs processing.

[0046] In addition, in embodiment 1, the memory unit 12 may be realized by storing the data files that make up these in a memory device such as a hard disk provided in the computer, or may be realized by a memory device of another computer.

[0047] Examples of the computer include a general-purpose PC, a smartphone, and a tablet terminal device. The program in the first embodiment may be executed by a computer system constructed by a plurality of computers.

[0048] [Modification] Next, a modification of the first embodiment will be described with reference to Fig. 6. Fig. 6 is a configuration diagram showing the configuration of a modification of the learning model generation device.

[0049] Similar to the learning model generation device 10, the learning model generation device 20 in the modified example shown in FIG. 6 is also a device that generates a learning model for predicting a user's future test value.

[0050] 6, like the learning model generation device 10, the learning model generation device 20 also includes a learning model generation unit 11 and a storage unit 12. However, in a modified example, the learning model generation device 20 further includes a function setting unit 21 in addition to these.

[0051] The function setting unit 21 sets a function based on the changes in test values ​​for multiple people over a specific period. Specifically, even in this modified example, the function represents the changes in test values ​​after a set period has elapsed, and the set period is assumed to be one year. In this case, the test values ​​for the specific period are test values ​​for several years starting one year after the set period.

[0052] The function setting unit 21 first acquires test values ​​for a specific period of time for multiple people from the database 30. Next, the function setting unit 21 plots the changes in the acquired test values ​​for each person on a graph with the test values ​​on the vertical axis and the number of years elapsed on the horizontal axis. The function setting unit 21 then identifies an approximating function from the plotted test values ​​using, for example, the least squares method, and sets the identified function.

[0053] 1 and 2, the function is set in advance, but in this modification, the function setting unit 21 can set the function based on the test values ​​of each sample person. Therefore, according to this modification, a learning model with higher prediction accuracy can be generated.

[0054] Second Embodiment Next, in a second embodiment, a test value prediction device, a test value prediction method, and a program will be described with reference to FIGS.

[0055] [Device Configuration] First, the schematic configuration of a test value prediction device will be described with reference to Fig. 7. Fig. 7 is a diagram showing the schematic configuration of an example of a test value prediction device.

[0056] 7, a test value prediction device 40 according to the second embodiment is a device for predicting a user's future test values. As shown in FIG. 6, the test value prediction device 40 includes a test value acquisition unit 41, a parameter prediction unit 42, and a test value prediction unit 43.

[0057] The test value acquisition unit 41 acquires the test values ​​of the person to be predicted. The parameter prediction unit 42 inputs the acquired test values ​​into a learning model to predict parameters (function parameters) of the person to be predicted. The learning model performs machine learning to determine the relationship between the test values ​​and the function parameters of the function that represents their trends. The test value prediction unit 43 inputs the predicted function parameters into the function to predict the future test values ​​of the person to be predicted.

[0058] As described above, in the second embodiment, first, the parameters of the function representing the transition of the test value are predicted by the learning model, and then future test values ​​are predicted using the function into which the parameters have been input. Therefore, according to the second embodiment, a certain trend is indicated in the prediction of the user's test value, which makes it possible to improve interpretability.

[0059] Next, the configuration and functions of an example of the test value prediction device 40 will be specifically described with reference to Fig. 8. Fig. 8 is a configuration diagram specifically showing the configuration of an example of the test value prediction device.

[0060] As shown in FIG. 8 , in the second embodiment, a test value prediction device 40 includes a learning model 44 in addition to the test value acquisition unit 41, parameter prediction unit 42, and test value prediction unit 43 described above.

[0061] Also in the second embodiment, the function used is a function that represents the transition of the test value after the set period has elapsed, specifically the function expressed by the above-mentioned equation 1, and the function parameter used is the function parameter a shown in the equation 1. Note that if a function other than the function expressed by the above-mentioned equation 1 is used as the function in the first embodiment, then the function other than the function expressed by the above-mentioned equation 1 will also be used in the second embodiment.

[0062] The learning model 44 is the learning model generated in the first embodiment. The learning model 44 uses the function parameters of the function as the objective variables and the test values ​​and the contents of the medical questionnaire as the explanatory variables. Specifically, the learning model 44 is a neural network or the like constructed in a memory by the machine learning engine using the model parameters 13 updated in the first embodiment.

[0063] In addition, in Embodiment 2, the test value acquiring unit 41 acquires the medical questionnaire of the person to be predicted in addition to the test values ​​of the person to be predicted described above. Then, in Embodiment 2, the parameter predicting unit 42 inputs the acquired test values ​​and the contents of the medical questionnaire into the learning model 44 to predict the function parameter a of the person to be predicted.

[0064] In the second embodiment, the test value prediction unit 43 inputs the predicted function parameter a into the function expressed by the above-mentioned equation 1. Then, the test value prediction unit 43 inputs the time period value t into the function to which the function parameter a has been input, and predicts the future test value of the person to be predicted.

[0065] Thereafter, the test value prediction unit 43 presents the predicted test values ​​on the screen of a display device directly connected to the test value prediction device 40, the screen of a terminal device connected to the test value prediction device 40 via a network, or the like.

[0066] [Device Operation] Next, the operation of the test value prediction device 40 will be described with reference to FIG. 9. FIG. 9 is a flow diagram showing an example of the operation of the test value prediction device. In the following description, reference will be made to FIGS. 7 and 8 as appropriate. In addition, in the second embodiment, a test value prediction method is implemented by operating the test value prediction device 40. Therefore, the description of the test value prediction method in the second embodiment will be replaced by the following description of the operation of the test value prediction device 40.

[0067] As shown in FIG. 9, first, the test value acquisition unit 41 acquires the test values ​​and medical questionnaire of the person who is the prediction target (step B1).

[0068] Next, the parameter prediction unit 42 inputs the test values ​​and the contents of the medical questionnaire acquired in step B1 into the learning model 44, acquires the output of the learning model 44, and predicts the function parameter a of the person to be predicted (step B2). Specifically, when the test values ​​and the contents of the medical questionnaire acquired in step B1 are input, the learning model 44 outputs a predicted value of the function parameter a. The parameter prediction unit 42 makes a prediction by acquiring the output predicted value.

[0069] Next, the test value prediction unit 43 inputs the predicted function parameter a into the function expressed by the above equation 1 to predict the future test value of the person to be predicted (step B3). Specifically, the test value prediction unit 43 inputs the time period value t into the function to which the function parameter a has been input, to predict the future test value of the person to be predicted.

[0070] Thereafter, the test value prediction unit 43 presents the predicted test value on the screen of the display device, the screen of the terminal device, etc. (step B4).

[0071] As described above, in the second embodiment, the function parameter a is predicted by the learning model 44, and future test values ​​are predicted from the predicted function parameter a, so that highly interpretable prediction results can be obtained for test values ​​one year and beyond, which would have tended to have poor interpretability in the past. Therefore, according to the second embodiment, a certain trend can be shown in the prediction of a user's test values, making it possible to improve interpretability.

[0072] [Program] The program in embodiment 2 may be any program that causes a computer to execute steps B1 to B4 shown in Fig. 9. By installing and executing this program in a computer, it is possible to realize the test value prediction device 40 and the test value prediction method in embodiment 2. In this case, the processor of the computer functions as test value acquisition unit 41, parameter prediction unit 42, and test value prediction unit 43 and performs processing.

[0073] Examples of the computer include a general-purpose PC, a smartphone, and a tablet terminal device. The program in the second embodiment may be executed by a computer system constructed by a plurality of computers. In this case, for example, each computer may function as one of the test value acquiring unit 41, the parameter predicting unit 42, and the test value predicting unit 43.

[0074] [Physical Configuration] Here, a computer that realizes the learning model generating device 10 or the test value prediction device 40 by executing the programs in the first and second embodiments will be described with reference to Fig. 10. Fig. 10 is a block diagram showing an example of a computer that realizes the learning model generating device and the test value prediction device.

[0075] 10, the computer 110 includes a CPU (Central Processing Unit) 111, a main memory 112, a storage device 113, an input interface 114, a display controller 115, a data reader / writer 116, and a communication interface 117. These components are connected to each other via a bus 121 so as to be able to communicate data with each other.

[0076] Furthermore, the computer 110 may include a GPU (Graphics Processing Unit) or an FPGA (Field-Programmable Gate Array) in addition to or instead of the CPU 111. In this aspect, the GPU or FPGA can execute the programs in the embodiments.

[0077] The CPU 111 loads a program in the embodiment, which is composed of a group of codes and stored in the storage device 113, into the main memory 112 and executes each code in a predetermined order to perform various calculations. The main memory 112 is typically a volatile storage device such as a DRAM (Dynamic Random Access Memory).

[0078] The program in the embodiment is provided in a state stored in a computer-readable recording medium 120. The program in the embodiment may be distributed over the Internet connected via the communication interface 117.

[0079] Specific examples of the storage device 113 include a hard disk drive and a semiconductor storage device such as a flash memory. The input interface 114 mediates data transmission between the CPU 111 and input devices 118 such as a keyboard and a mouse. The display controller 115 is connected to a display device 119 and controls the display on the display device 119.

[0080] The data reader / writer 116 mediates data transmission between the CPU 111 and the recording medium 120, reads programs from the recording medium 120, and writes processing results from the computer 110 to the recording medium 120. The communication interface 117 mediates data transmission between the CPU 111 and other computers.

[0081] Specific examples of the recording medium 120 include general-purpose semiconductor storage devices such as CF (Compact Flash (registered trademark)) and SD (Secure Digital), magnetic recording media such as flexible disks, or optical recording media such as CD-ROMs (Compact Disk Read Only Memory).

[0082] The learning model generation device and the test value prediction device can be realized not by a computer on which a program is installed, but by hardware corresponding to each unit, for example, an electronic circuit. Furthermore, the learning model generation device and the test value prediction device may be realized in part by a program and in part by hardware. In the embodiments, the computer is not limited to the computer shown in FIG. 10 .

[0083] Some or all of the above-described embodiments can be expressed by (Supplementary Note 1) to (Supplementary Note 30) described below, but are not limited to the following descriptions.

[0084] (Supplementary Note 1) A learning model generation device comprising: a learning model generation unit that, when a function representing the progress of test values ​​obtained from tests on a person is set, performs machine learning using the test values ​​of a sample person and parameters of the function for the sample person as training data to generate a learning model in which the test values ​​are used as explanatory variables and the parameters are used as objective variables.

[0085] (Supplementary Note 2) The learning model generation device according to Supplementary Note 1, wherein the learning model generation unit performs machine learning using the contents of a medical questionnaire of the sample person as the training data, and generates the learning model in which the test values ​​and the contents of the medical questionnaire are used as explanatory variables and the parameters are used as objective variables.

[0086] (Supplementary Note 3) The learning model generation device according to Supplementary Note 2, wherein the medical questionnaire includes lifestyle habits of the sample person.

[0087] (Supplementary Note 4) The learning model generation device according to Supplementary Note 1, wherein the function is a function that represents a transition of the test value after a set period has elapsed.

[0088] (Supplementary Note 5) The learning model generation device according to Supplementary Note 1, further comprising a function setting unit that sets the function based on changes in test values ​​of a plurality of people over a specific period of time.

[0089] (Supplementary Note 6) The learning model generation device according to Supplementary Note 1, wherein the parameters of the function for the sample person are determined from changes in past test values ​​of the sample person over time.

[0090] (Supplementary Note 7) A test value prediction device comprising: a test value acquisition unit that acquires test values ​​of a person to be predicted; a parameter prediction unit that predicts the parameters of the person to be predicted by inputting the acquired test values ​​into a learning model that performs machine learning to learn the relationship between test values ​​and parameters of a function that represents their trends; and a test value prediction unit that inputs the predicted parameters into the function to predict future test values ​​of the person to be predicted.

[0091] (Supplementary Note 8) The test value prediction device described in Supplementary Note 7, wherein the learning model machine-learns the relationship between the test value and the parameters of a function representing the test value over time, as well as the relationship between the test value and the contents of a person's medical questionnaire, the test value acquisition unit further acquires the medical questionnaire of the person to be predicted, and the parameter prediction unit inputs the acquired test value and the contents of the medical questionnaire into the learning model to predict the parameters of the person to be predicted.

[0092] (Supplementary Note 9) The test value prediction device according to Supplementary Note 8, wherein the medical questionnaire includes information about the person's lifestyle habits.

[0093] (Supplementary Note 10) The test value prediction device according to Supplementary Note 7, wherein the function is a function that represents a transition of the test value after a set period has elapsed.

[0094] (Supplementary Note 11) A learning model generation method comprising: when a function representing the progress of test values ​​obtained from tests on a person is set, performing machine learning using the test values ​​of a sample person and parameters of the function for the sample person as training data to generate a learning model in which the test values ​​are used as explanatory variables and the parameters are used as objective variables.

[0095] (Supplementary Note 12) The learning model generation method according to Supplementary Note 11, wherein in the learning model generation step, machine learning is performed using the contents of a medical questionnaire of the sample person as the training data, and the learning model is generated using the test values ​​and the contents of the medical questionnaire as explanatory variables and the parameters as objective variables.

[0096] (Supplementary Note 13) The learning model generation method according to Supplementary Note 12, wherein the medical questionnaire includes lifestyle habits of the sample person.

[0097] (Supplementary Note 14) The learning model generation method according to Supplementary Note 11, wherein the function is a function that represents a transition of the test value after a set period has elapsed.

[0098] (Supplementary Note 15) The learning model generation method according to Supplementary Note 11, further comprising a function setting step of setting the function based on changes in test values ​​of a plurality of people over a specific period of time.

[0099] (Supplementary Note 16) The learning model generation method according to Supplementary Note 11, wherein the parameters of the function for the sample person are determined from changes in past test values ​​of the sample person over time.

[0100] (Supplementary Note 17) A test value prediction method comprising: a test value acquisition step of acquiring test values ​​of a person to be predicted; a parameter prediction step of predicting the parameters of the person to be predicted by inputting the acquired test values ​​into a learning model that performs machine learning to learn the relationship between test values ​​and parameters of a function that represents their trends; and a test value prediction step of predicting the future test values ​​of the person to be predicted by inputting the predicted parameters into the function.

[0101] (Supplementary Note 18) The test value prediction method according to Supplementary Note 17, wherein the learning model machine-learns the relationship between the test value and the parameters of a function representing the test value over time, as well as the relationship between the test value and the contents of a person's medical questionnaire; the test value acquisition step further acquires the medical questionnaire of the person to be predicted; and the parameter prediction step inputs the acquired test value and the contents of the medical questionnaire into the learning model to predict the parameters of the person to be predicted.

[0102] (Supplementary Note 19) The test value prediction method according to Supplementary Note 18, wherein the questionnaire includes lifestyle habits of the sample person.

[0103] (Supplementary Note 20) The test value prediction method according to Supplementary Note 17, wherein the function represents a transition of the test value after a set period has elapsed.

[0104] (Supplementary Note 21) A computer-readable recording medium having a program recorded thereon, the computer-readable recording medium including instructions to execute a learning model generation step in which, when a function representing the progress of test values ​​obtained from tests on a person is set, machine learning is performed using the test values ​​of a sample person and parameters of the function for the sample person as training data to generate a learning model in which the test values ​​are used as explanatory variables and the parameters are used as objective variables.

[0105] (Supplementary Note 22) The computer-readable recording medium according to Supplementary Note 21, wherein in the learning model generation step, machine learning is performed using the contents of a medical questionnaire of the sample person as the training data, and the learning model is generated in which the test values ​​and the contents of the medical questionnaire are used as explanatory variables and the parameters are used as target variables.

[0106] (Supplementary Note 23) The computer-readable recording medium according to Supplementary Note 22, wherein the questionnaire includes lifestyle habits of the sample person.

[0107] (Supplementary Note 24) The computer-readable recording medium according to Supplementary Note 21, wherein the function represents a transition of the test value after a set period has elapsed.

[0108] (Supplementary Note 25) The computer-readable recording medium according to Supplementary Note 21, wherein the program further causes the computer to execute a function setting step of setting the function based on changes in test values ​​of a plurality of people over a specific period of time.

[0109] (Supplementary Note 26) The computer-readable recording medium according to Supplementary Note 21, wherein the parameters of the function for the sample person are determined from changes in past test values ​​of the sample person over time.

[0110] (Supplementary Note 27) A computer-readable recording medium having recorded thereon a program including instructions for causing a computer to execute: a test value acquisition step of acquiring test values ​​of a person to be predicted; a parameter prediction step of predicting the parameters of the person to be predicted by inputting the acquired test values ​​into a learning model that performs machine learning to learn the relationship between test values ​​and parameters of a function that represents their trends; and a test value prediction step of inputting the predicted parameters into the function to predict future test values ​​of the person to be predicted.

[0111] (Appendix 28) A computer-readable recording medium according to Appendix 27, wherein the learning model machine-learns the relationship between the test values ​​and the parameters of a function representing the changes in the test values ​​and the parameters of the function representing the changes in the test values ​​and the parameters of the person's medical questionnaire, and further acquires the medical questionnaire of the person to be predicted in the test value acquisition step, and predicts the parameters of the person to be predicted in the parameter prediction step by inputting the acquired test values ​​and the contents of the medical questionnaire into the learning model.

[0112] (Supplementary Note 29) The computer-readable recording medium according to Supplementary Note 28, wherein the questionnaire includes lifestyle habits of the sample person.

[0113] (Supplementary Note 30) The computer-readable recording medium according to Supplementary Note 27, wherein the function represents a transition of the test value after a set period has elapsed.

[0114] Although the present invention has been described above with reference to the embodiments, the present invention is not limited to the above-described embodiments. Various modifications that can be understood by those skilled in the art can be made to the configuration and details of the present invention within the scope of the present invention.

[0115] This application claims priority based on Japanese Patent Application No. 2023-050360, filed March 27, 2023, the disclosure of which is incorporated herein in its entirety.

[0116] As described above, according to the present disclosure, it is possible to improve interpretability by showing a certain trend in predicting a user's test values. The present disclosure is useful in the medical field.

[0117] REFERENCE SIGNS LIST 10 Learning model generation device (first embodiment) 11 Learning model generation unit 12 Storage unit 13 Model parameters 20 Learning model generation device (variant of first embodiment) 21 Function setting unit 30 Database 40 Test value prediction device 41 Test value acquisition unit 42 Parameter prediction unit 43 Test value prediction unit 44 Learning model 110 Computer 111 CPU 112 Main memory 113 Storage device 114 Input interface 115 Display controller 116 Data reader / writer 117 Communication interface 118 Input device 119 Display device 120 Recording medium 121 Bus

Claims

1. a learning model generation unit that, when a function representing the transition of test values ​​obtained from tests on a person is set, executes machine learning using the test values ​​of a sample person and parameters of the function for the sample person as training data to generate a learning model in which the test values ​​are used as explanatory variables and the parameters are used as objective variables; A learning model generation device comprising:

2. the learning model generation unit performs machine learning using the content of the medical questionnaire of the sample person as the training data, and generates the learning model using the test value and the content of the medical questionnaire as explanatory variables and the parameter as a target variable. The learning model generation device according to claim 1 .

3. The questionnaire includes the lifestyle habits of the sample person. The learning model generating device according to claim 2 .

4. The function represents a transition of the test value after a set period has elapsed. The learning model generation device according to claim 1 .

5. The device further includes a function setting unit that sets the function based on changes in test values ​​of a plurality of people over a specific period. The learning model generation device according to claim 1 .

6. The parameters of the function for the sample person are determined from the time-series changes in the past test values ​​of the sample person. The learning model generation device according to claim 1 .

7. a test value acquisition unit that acquires test values ​​of a person who is a prediction target; a parameter prediction unit that inputs the obtained test values ​​into a learning model that performs machine learning to learn the relationship between the test values ​​and parameters of a function that represents a transition between the test values, and predicts the parameters of the person to be predicted; a test value prediction unit that inputs the predicted parameters into the function to predict a future test value of the person to be predicted; A test value prediction device comprising:

8. The learning model performs machine learning on the relationship between the test value and the content of a person's medical questionnaire in addition to the parameters of the function representing the test value and its transition, the test value acquisition unit further acquires a medical questionnaire of the person to be predicted, the parameter prediction unit inputs the acquired test values ​​and the contents of the medical questionnaire into the learning model to predict the parameters of the person to be predicted. The test value prediction device according to claim 7 .

9. The questionnaire includes a person's lifestyle habits, The test value prediction device according to claim 8 .

10. The function represents a transition of the test value after a set period has elapsed. The test value prediction device according to claim 7 .

11. When a function representing the transition of test values ​​obtained from tests on a person is set, machine learning is performed using the test values ​​of a sample person and parameters of the function for the sample person as training data to generate a learning model in which the test values ​​are used as explanatory variables and the parameters are used as objective variables. A learning model generation method characterized by:

12. Obtain the test results of the person to be predicted, inputting the acquired test values ​​into a learning model that performs machine learning on the relationship between the test values ​​and parameters of a function that represents their transition, and predicting the parameters of the person to be predicted; inputting the predicted parameters into the function to predict future test values ​​of the person to be predicted; A method for predicting a test value, comprising:

13. On the computer, When a function representing the transition of test values ​​obtained from tests on a person is set, machine learning is performed using the test values ​​of a sample person and parameters of the function for the sample person as training data to generate a learning model in which the test values ​​are used as explanatory variables and the parameters are used as objective variables. program.

14. On the computer, Obtain the test results of the person to be predicted, inputting the acquired test values ​​into a learning model that performs machine learning to learn the relationship between test values ​​and parameters of a function that represents a change in the test values, and having the model predict the parameters of the person to be predicted; The predicted parameters are input into the function to predict the future test value of the person to be predicted. program.

Images