Prediction device, prediction method, and prediction program

Abstract

To accurately predict the effects of interventions on factors for which there is no prior intervention history through intervention programs. [Solution] The prediction device includes a storage device that stores intervention effect performance information representing the actual intervention effect of an intervention program that provides interventions for a predetermined health condition to each participant in the past, for at least one of a plurality of factors that may cause the said health condition, and a processing device that performs intervention effect prediction processing that predicts the intervention effect for factors for which there is no intervention history by the intervention program, based on the intervention effect performance information and the relationships between each factor, and output processing that outputs the prediction results.

Description

Technical Field

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

Background Art

[0002] In an intervention project for preventing the aggravation of diseases or preventing care in local governments or the like, there is a need to accurately select people with high effects of the intervention on the condition of factors such as medical receipts or the results of health checkups.

[0003] Patent Document 1 discloses an intervention processing system that estimates the intervention effect obtained as a result of an intervention for a user who receives a content distribution service using an intervention model using the feature amount of the user and the feature amount of the intervention, and generates an intervention material used for a new intervention based on the estimated intervention effect.

[0004] Patent Document 2 discloses a system for selecting the most cost-effective intervention or treatment for a specific patient by using data such as medical records of other patients related to the specific patient.

[0005] Patent Document 3 discloses an analysis system that refers to health information, medical information, and a disease transition model, predicts the change in the state when an intervention is implemented and when it is not implemented, and calculates medical expenses.

Prior Art Documents

Patent Documents

[0006]

Patent Document 1

Patent Document 2

Patent Document 3

Summary of the Invention

Problems to be Solved by the Invention

[0007] However, since the content of services differs for each intervention program, and the intervention results for the same program are limited, it is difficult to select factors that are likely to benefit from intervention for each intervention program. Furthermore, the aforementioned technology has the problem of being unable to predict the intervention effect on factors for which there is no intervention history.

[0008] This invention has been made in view of these circumstances, and its purpose is to provide a prediction device, prediction method, and prediction program that can accurately predict the intervention effect on factors for which there is no track record of intervention through intervention projects. [Means for solving the problem]

[0009] One of the present inventions for solving the above problems is a prediction device comprising: a storage device that stores intervention effect performance information representing the actual intervention effect of an intervention project that provides interventions for a predetermined health condition to each participant, for at least one of a plurality of factors that may cause the said health condition in the past; and an intervention effect prediction process that predicts the intervention effect for factors for which there is no prior intervention record by the intervention project, based on the intervention effect performance information and the relationships between each factor; and an output process that outputs the result of the prediction. [Effects of the Invention]

[0010] According to the present invention, it is possible to accurately predict the intervention effect on factors for which there is no prior intervention history through intervention programs. Other configurations and effects will be clarified by the following description of the embodiments. [Brief explanation of the drawing]

[0011] [Figure 1] This is a diagram showing an example of a prediction system configuration. [Figure 2] This figure shows an example of the functional configuration of an intervention effect prediction server. [Figure 3] This figure shows an example of information on preventable risk factors. [Figure 4] It is a diagram showing an example of personal factor information. [Figure 5] It is a diagram showing an example of past business performance information. [Figure 6] It is a processing flowchart for explaining the preventive target factor onset risk calculation process. [Figure 7] It is a diagram showing an example of preventive target factor onset risk information. [Figure 8] It is a processing flowchart for explaining the factor - to - factor causal relationship calculation process. [Figure 9] It is a processing flowchart for explaining the details of the cause factor extraction process. [Figure 10] It is a diagram showing an example of factor - to - factor causal relationship information. [Figure 11] It is a diagram showing the factor - to - factor causal relationship information in a graph display. [Figure 12] It is a processing flowchart for explaining the intervention effect performance calculation process. [Figure 13] It is a diagram showing an example of intervention effect performance information. [Figure 14] It is a processing flowchart for explaining the intervention effect prediction process. [Figure 15] It is a processing flowchart for explaining the details of the calculation process of the relevance degree (A) of the causal relationship. [Figure 16] It is a processing flowchart for explaining the details of the calculation process of the relevance degree (B) of the preventive target factor onset risk. [Figure 17] It is a diagram showing an example of relevance degree information. [Figure 18] It is a processing flowchart for explaining the details of the intervention effect prediction process. [Figure 19] It is a diagram showing an example of intervention effect prediction information. [Figure 20] It is a diagram showing an example of the factor intervention effect presentation screen.

Embodiments for Carrying Out the Invention

[0012] Embodiments of the present invention will be described in detail below with reference to the drawings. The following description and drawings are illustrative for illustrating the present invention, and have been omitted and simplified as appropriate for clarity of explanation. The present invention can also be carried out in various other forms. Unless otherwise specified, each component may be singular or plural. The positions, sizes, shapes, and ranges of the components shown in the drawings may not represent their actual positions, sizes, shapes, and ranges in order to facilitate understanding of the invention. Therefore, the present invention is not necessarily limited to the positions, sizes, shapes, and ranges disclosed in the drawings. In the following explanation, various types of information may be described using terms such as "table," "list," and "queue," but these types of information may also be represented using other data structures. To indicate independence from data structure, "XX table," "XX list," etc., may be referred to as "XX information." When describing identification information, terms such as "identification information," "identifier," "name," "ID," and "number" will be used, but these terms are interchangeable. When there are multiple components with the same or similar function, they may be described using the same symbol but with different subscripts. However, if it is not necessary to distinguish between these multiple components, the subscripts may be omitted in the description. Furthermore, while the following explanation may describe the processes performed by executing a program, the processor (e.g., CPU, GPU) executes the program, performing defined processes using memory resources (e.g., memory) and / or interface devices (e.g., communication ports) as appropriate. Therefore, the processor may be the primary entity performing the processes. Similarly, the primary entity performing the processes by executing a program may be a controller, device, system, computer, or node having a processor. The primary entity performing the processes by executing a program may be an arithmetic unit, and may include dedicated circuits (e.g., FPGAs or ASICs) that perform specific processes. A program may be installed from its program source into a device such as a computer. The program source may be, for example, a program distribution server or a computer-readable storage medium. If the program source is a program distribution server, the program distribution server includes a processor and storage resources for storing the program to be distributed, and the processor of the program distribution server may distribute the program to other computers. Furthermore, in the following description, two or more programs may be implemented as a single program, or one program may be implemented as two or more programs.

[0013] <Prediction System> Figure 1 shows an example of the configuration of Prediction System 1. Prediction System 1 is an information processing system that predicts the effectiveness of interventions for individuals who are eligible for intervention (such as those receiving care or treatment, who are eligible for intervention through intervention programs) that are implemented by local governments or other organizations to address specific health conditions (conditions requiring care, conditions of severe illness, etc.). The intervention programs aim to prevent the worsening of conditions such as requiring care or illness. In this embodiment, the factors are, for example, factors that can cause specific health conditions (conditions requiring care, conditions of severe illness), such as the health condition, disease, pathology, symptoms, or habits of individuals eligible for intervention. For example, the factors may be items from medical claims, health checkups, health questionnaires, etc.

[0014] As shown in Figure 1, the prediction system 1 comprises an intervention effect prediction server 10 (prediction device) and a user terminal 20. The intervention effect prediction server 10 and the user terminal 20 are connected via a wired or wireless communication network N, such as the Internet, LAN (Local Area Network), WAN (Wide Area Network), or a dedicated line.

[0015] The user terminal 20 is, for example, a personal computer, tablet terminal, or smartphone, an information processing device related to an intervention condition selector U that selects the conditions (factors in this embodiment) of people who are the target of intervention in an intervention project implemented by a local government or the like.

[0016] The intervention effect prediction server 10 is an information processing device (server device) that predicts the intervention effect for each factor in a predetermined intervention project. The intervention effect prediction server 10, as an example, includes a processing device 11 such as a CPU (Central Processing Unit), a storage device 12 such as an HDD (Hard Disk Drive) or SSD (Solid State Drive), a memory 13 such as RAM (Random Access Memory) or ROM (Read Only Memory), an input device 14 such as a keyboard, mouse, or touch panel, an output device 15 such as a display or printer, and a communication device 16 consisting of a NIC (Network Interface Card), wireless communication module, USB (Universal Serial Interface) module, or serial communication module. The processing device 11, storage device 12, memory 13, input device 14, output device 15, and communication device 16 are interconnected via a bus.

[0017] Figure 2 shows an example of the functional configuration of the intervention effect prediction server 10. This figure shows the input / output relationships of each data stored in the storage device 12 and each functional module implemented by the processing device 11.

[0018] The memory device 12 stores information on preventable factors 121, personal factor information 122, past project performance information 123, information on the risk of developing preventable factors 124 (described later), information on causal relationships between factors 125 (described later), information on the effectiveness of interventions 126 (described later), information on the degree of relevance 127 (described later), and information on the predicted effectiveness of interventions 128 (described later).

[0019] (Information on preventable factors) Figure 3 shows an example of information on preventable factors. Preventable factor information 121 represents the factors targeted for prevention in the intervention program (hereinafter referred to as "preventable factors"). The following explanation uses the example of an intervention program being an "exercise class for dementia prevention" and the preventable factor being "dementia." The preventable factors are selected in advance by the intervention condition selector U.

[0020] (Personal factor information) Figure 4 shows an example of personal factor information. Personal factor information 122 is a database that manages the history of factors possessed by each person eligible for intervention in each fiscal year. Personal factor information 122 has data items such as ID 1221, which sets the identification information of the person eligible for intervention; fiscal year 1222, which sets the fiscal year; and factor presence / absence 1223, which sets the presence or absence of each factor (in the example shown, "low BMI", "diabetes", and "dementia").

[0021] (Past business performance information) Figure 5 shows an example of past project performance information. Past project performance information 123 is a database that manages the results of intervention projects implemented in the past. Past project performance information 123 has the following data items: project 1231 which sets the intervention projects implemented in the past, implementation period 1232 which sets the year in which the intervention projects were implemented, ID 1233 which sets the identification information of the person who could be intervened, and performance 1234 which sets whether or not the person who could be intervened participated in the intervention project ("participated" or "did not participate").

[0022] The intervention effect prediction server 10 implements the functions of the prevention target factor onset risk calculation unit 111, the factor causal relationship calculation unit 112, the intervention effect performance calculation unit 113, the intervention effect prediction unit 114, and the intervention effect prediction presentation unit 115 by having the CPU 11 read the program stored in the storage device 12 into the RAM of the memory 13 and execute it.

[0023] The preventable factor onset risk calculation unit 111 calculates the probability (hereinafter referred to as "preventable factor onset risk") that a person who is eligible for intervention and possesses each factor will have (develop) a preventable factor at a predetermined time in the future, based on the preventable factor information 121 and personal factor information 122, and sets the calculated preventable factor onset risk in the preventable factor onset risk information 124, which will be described later. The preventable factor onset risk is the risk of developing a preventable factor from each factor.

[0024] The factor-causal relationship calculation unit 112 derives causal relationships between factors based on the personal factor information 122, and sets the derived causal relationships in the factor-causal relationship information 125, which will be described later. The causal relationships between factors are, for example, the relationships between factors related to disease onset or health risk.

[0025] The intervention effect calculation unit 113 calculates the actual intervention effect on the factors possessed by participants in the intervention program based on the preventable target factor information 121, the individual factor information 122, and the past program performance information 123, and sets the calculated actual intervention effect in the intervention effect performance information 126, which will be described later.

[0026] The intervention effect prediction unit 114 derives the relationship between factors that have not been intervened in the intervention program and factors that have been intervened in the intervention program, based on the risk information 124 for the preventable factor onset, the causal relationship information 125 between factors, and the actual intervention effect information 126, and sets the derived relationship in the correlation information 127, which will be described later. In addition, the intervention effect prediction unit 114 predicts the intervention effect on factors that have not been intervened in the intervention program, based on the risk information 124 for the preventable factor onset, the causal relationship information 125 between factors, and the actual intervention effect information 126, and sets the predicted intervention effect in the intervention effect prediction information 128, which will be described later.

[0027] The intervention effect prediction display unit 115 performs output processing to output a prediction of the intervention effect for factors for which there is no intervention record from the intervention project, based on the prevention target factor onset risk information 124, inter-factor causal relationship information 125, intervention effect performance information 126, correlation information 127, and intervention effect prediction information 128.

[0028] The programs that implement the functions of the preventable factor onset risk calculation unit 111, the factor causal relationship calculation unit 112, the intervention effect performance calculation unit 113, the intervention effect prediction unit 114, and the intervention effect prediction presentation unit 115 may be pre-recorded on an external storage medium or may be introduced as needed via a predetermined communication network. Furthermore, these programs can be recorded and distributed, for example, on a portable or fixed recording medium. Next, we will explain the processing performed in each functional unit of the intervention effect prediction server 10.

[0029] <Calculation process for the risk of developing preventable factors> Figure 6 is a processing flow diagram illustrating the process for calculating the risk of developing preventable factors. The process shown in this figure is executed, for example, when a predetermined input is made to the intervention effect prediction server 10 by a user (e.g., the person who selects the intervention conditions U), when personal factor information 122 is updated, or at predetermined timings (e.g., a predetermined time, a predetermined time interval).

[0030] First, the risk calculation unit 111 for preventing the onset of the preventable factor extracts a list of factors from the personal factor information 122 (S101).

[0031] Next, the preventable factor onset risk calculation unit 111 performs the S102 process for each of the extracted factors.

[0032] Specifically, the preventable factor onset risk calculation unit 111 generates a model (e.g., a function or a trained machine learning model, etc.) from the personal factor information 122, with the presence or absence of the factor in a predetermined year as the explanatory variable and the presence or absence of the preventable factor in a predetermined future period (e.g., within 3 years) from the predetermined year as the dependent variable. Based on the generated model, it calculates the probability that a person with the factor will develop the preventable factor "dementia" within a predetermined period (e.g., within 3 years) (hereinafter referred to as the "preventable factor onset risk") (S102).

[0033] The preventable factor onset risk calculation unit 111 performs the process in S102 for all extracted factors, and then sets the preventable factor risk for each factor in the preventable factor onset risk information 124 (S103). After that, the preventable factor onset risk calculation process is terminated.

[0034] (Information on the risk of developing preventable factors) Figure 7 shows an example of information on the risk of developing a preventable factor. The information on the risk of developing a preventable factor 124 represents the risk of developing a preventable factor for each factor. The information on the risk of developing a preventable factor 124 includes data items for Factor 1241, on which the extracted factor is set; Preventable Factor 1242, on which the preventable factor is set; and Preventable Factor Risk 1243, on which the calculated risk of developing a preventable factor is set.

[0035] <Calculation process for inter-factor causal relationships> Figure 8 is a processing flow chart illustrating the process for calculating causal relationships between factors. The process shown in this figure is executed at the same time as, for example, the process for calculating the risk of developing a preventable factor.

[0036] First, the factor causal relationship calculation unit 112 performs the following S201 to S204 processes for each of the factors in the personal factor information 122.

[0037] First, the factor-causal relationship calculation unit 112 generates a model (e.g., a function or a trained machine learning model, etc.) for the relevant factor (hereinafter referred to as "Factor 1") from the personal factor information 122, with the presence or absence of Factor 1 in a predetermined year as the explanatory variable and the presence or absence of the preventable target factor "dementia" at a predetermined time in the future from the predetermined year (e.g., within 3 years) as the dependent variable. Based on the generated model, it calculates the risk (probability) that a person with Factor 1 will have the preventable target factor "dementia" within a predetermined period (e.g., within 3 years) (S201).

[0038] Next, the factor-causal relationship calculation unit 112 determines whether the calculated risk is above a certain level (a predetermined threshold) (S202). If the risk is below a certain level (S202: No), it is determined that the first factor is not a causative factor of the preventable factor "dementia," and the processing S201 to S204 is performed on the next factor in the personal factor information 122.

[0039] On the other hand, the factor-causal relationship calculation unit 112 extracts the first explanatory variable as the causal factor of the preventable factor (outcome factor) if the risk is above a certain level (S202: Yes) (S203).

[0040] Next, the factor-inter-factor causal relationship calculation unit 112 uses the first factor extracted as a causal factor as an outcome factor and performs a causal factor extraction process to extract the causal factors of the first factor (S204).

[0041] <Causal Factor Extraction Process> Figure 9 is a process flow diagram illustrating the details of the causal factor extraction process. The process shown in this figure is a detailed description of the process described in S204 above.

[0042] The factor causal relationship calculation unit 112 performs the following S2041 to S2044 processes for each of the factors in the personal factor information 122 other than the outcome factors.

[0043] First, the factor-inter-causal relationship calculation unit 112 generates a model (e.g., a function or a trained machine learning model, etc.) for the relevant factor (hereinafter referred to as "Factor 2") from the personal factor information 122, with the presence or absence of Factor 2 in a specified year as the explanatory variable and the presence or absence of the outcome factor within a specified number of years (e.g., 3 years) from the specified year as the dependent variable. Based on the generated model, it calculates the risk (probability) that a person with Factor 2 will have the outcome factor within a specified period (e.g., within 3 years) (S2041).

[0044] Next, the factor-inter-causal relationship calculation unit 112 determines whether the calculated risk is above a certain level (a predetermined threshold) (S2042). If the risk is below a certain level (S2042: No), it is determined that the second factor is not a causal factor of the outcome factor, and the processes S2041 to S2044 are executed for the next factor.

[0045] On the other hand, the factor-inter-factor causal relationship calculation unit 112 extracts the second explanatory variable as the causal factor of the outcome factor (S2043) if the risk is above a certain level (S2042: Yes).

[0046] Next, the factor-inter-causal relationship calculation unit 112 uses the second factor extracted as a causal factor as an outcome factor and performs a causal factor extraction process to extract the causal factors of the second factor (S2044). That is, the factor-inter-causal relationship calculation unit 112 recursively uses the causal factors as outcome factors and extracts the causal factors of the outcome factor whose risk is above a certain level.

[0047] The factor-causal relationship calculation unit 112 performs the processes S201 to S204 for all factors, and then sets (configures) the extracted causal and consequent factor combinations in the factor-causal relationship information 125 (S205). After that, the factor-causal relationship calculation process is terminated.

[0048] (Information on interfactor causal relationships) Figure 10 shows an example of interfactor causal relationship information. Interfactor causal relationship information 125 is data representing the causal relationship between each factor. Interfactor causal relationship information 125 has data items for causal factors 1251, where the extracted causal factors are set, and for outcome factors 1252, where the extracted outcome factors are set.

[0049] Figure 11 is a graph illustrating the causal relationships between factors. This figure shows an example of the data example shown in Figure 10, with the causal factor as the starting point of the arrow and the outcome factor as the ending point of the arrow. In the example shown, the causal factors for the preventable factor "dementia" are "stroke," "fracture," "malnutrition," and "mental illness." Furthermore, the causal factor for "stroke," which is a causal factor for the preventable factor "dementia," is "diabetes." Furthermore, the causal factors for "fracture," which is a causal factor for the preventable factor "dementia," are "low BMI," "no exercise habits," and "slow walking speed." Furthermore, the causal factor for "malnutrition," which is a causal factor for the preventable factor "dementia," is "low BMI." Furthermore, the causal factor for "mental illness," which is a causal factor for the preventable factor "dementia," is "slow walking speed."

[0050] <Processing for calculating intervention effect results> Figure 12 is a processing flow chart illustrating the process for calculating the intervention effect. The process shown in this figure is executed at the same time as, for example, the process for calculating the risk of developing the preventable factor.

[0051] First, the intervention effect calculation unit 113 extracts the IDs of past participants in the relevant project, "Exercise Class for Dementia Prevention," from the past project performance information 123, and then performs the following S301 process for each of the extracted participants.

[0052] Specifically, the intervention effect calculation unit 113 uses machine learning to calculate the change in the risk of each individual possessing the target preventive factors before and after the implementation of the relevant project, "Exercise Class for Dementia Prevention," based on the individual factor information 122 of the participant in question (S301).

[0053] For example, the intervention effect calculation unit 113 uses personal factor information 122 as training data and pre-generates a trained model using machine learning, inputting the factors possessed by each person eligible for intervention and outputting the risk of possessing the preventable factors. The trained model is constructed based on algorithms such as neural networks, decision trees, random forests, and support vector machines (SVMs). A neural network is a neural network having an input layer that receives input data, one or more hidden layers that extract and output features from the input data, and an output layer that outputs output data, such as a Convolutional Neural Network (CNN).

[0054] The intervention effect calculation unit 113 then inputs the factors the participant possessed before the intervention (for example, in the year prior to the intervention project) into a trained model to obtain the risk of possessing the preventable factors before the intervention, and inputs the factors the participant possessed after the intervention (for example, in the year following the intervention project) into the trained model to obtain the risk of possessing the preventable factors after the intervention. The intervention effect calculation unit 113 then calculates the change in the participant's risk of possessing the preventable factors by subtracting the risk of possessing the preventable factors after the intervention from the risk of possessing the preventable factors before the intervention.

[0055] The intervention effect calculation unit 113 performs the processing in S301 for all participants, and then extracts "factors" in which a certain number of participants (a predetermined threshold) or more possess the factor in past relevant projects, "exercise classes for dementia prevention," as "factors with intervention results" (S302).

[0056] Next, the intervention effect calculation unit 113 performs the following S303-S304 processes for all extracted "factors with intervention results".

[0057] First, the intervention effect calculation unit 113 calculates, for each implementation period (year) of the project "Exercise Class for Dementia Prevention," the number of participants who possess the "factors with intervention results," and the average value of the change in the risk of possessing the preventable factors (S303).

[0058] Next, the intervention effect calculation unit 113 calculates the intervention effect in "factors with a track record of intervention" by taking the difference between the amount of future cost reductions for future care or medical treatment due to changes in the risk of possessing the preventable factors (i.e., the total cost reductions incurred when the person eligible for intervention uses not only the care service in question but also all providers of care or medical treatment; hereinafter sometimes referred to as "medical care reduction costs") and the intervention costs required by the intervention service, for each implementation period (year) of the project "Exercise Class for Dementia Prevention" (S304).

[0059] Here, the intervention effect calculation unit 113 calculates the intervention cost as A*N*k / 100 yen and the medical / nursing care reduction cost as B*N*k / 100*m / 100 yen, assuming that the participation rate is k%, the risk improvement rate (average value of the change in risk) is m%, the number of people with the factor is N, the intervention cost per person is A yen, and the future medical / nursing care cost per person (costs related to medical or nursing care) is B yen. The intervention effect calculation unit 113 then calculates the intervention effect as B*N*k / 100*m / 100 - A*N*k / 100.

[0060] The intervention effect calculation unit 113 performs the S303-S304 processes on all extracted "factors with intervention experience," and then sets the calculated number of participants, the change in the risk of possessing the preventable target factor, and the estimated intervention effect in the intervention effect information 126 (S305). After that, it terminates the process.

[0061] (Information on the effectiveness of interventions) Figure 13 shows an example of intervention effectiveness performance information. Intervention effectiveness performance information 126 is data that represents the actual effectiveness of the intervention. Intervention effectiveness performance information 126 includes the following data items: project 1261 where the intervention project is set, implementation period 1262 where the implementation period (year) is set, factors with intervention performance 1263 where the extracted "factors with intervention performance" are set, number of people 1264 where the number of people with "factors with intervention performance" is set, participants 1265 where the number of participants in the intervention project out of the number of people 1264 is set, risk improvement rate 1266 where the change in the risk of possessing the preventable target factor is set, and effect estimation 1267 where the estimated intervention effect is set.

[0062] <Intervention effect prediction processing> Figure 14 is a processing flow chart illustrating the intervention effect prediction process. The processes shown in this figure are executed, for example, after the processes of calculating the risk of developing the preventable factor, calculating the causal relationship between factors, and calculating the actual intervention effect.

[0063] First, the intervention effect prediction unit 114 extracts a list of "factors without intervention experience" from the difference between the list of "factors" obtained from the prevention target factor onset risk information 124 and the list of "factors with intervention experience" from the intervention effect performance information 126 (S401).

[0064] Next, the intervention effect prediction unit 114 performs the following processing S402 to S407 for all extracted "factors without prior intervention experience".

[0065] First, the intervention effect prediction unit 114 performs the following S402 to S404 processes for all "factors with a track record of intervention".

[0066] Specifically, first, the intervention effect prediction unit 114 calculates the degree of causal relationship (A) between "factors without intervention history" and "factors with intervention history" (S402).

[0067] <Calculation process for the degree of causal relationship (A)> Figure 15 is a process flow diagram illustrating the details of the calculation process for the degree of causal relationship (A). The process shown in this diagram is a detailed representation of the process described in S402 above.

[0068] First, the intervention effect prediction unit 114 calculates the degree of association (similarity) of the causal factors (S4021). For example, the intervention effect prediction unit 114 takes the number of causal factors j common to "factors without intervention experience" and "factors with intervention experience" as the degree of association of the causal factors, divides the number of common causal factors j by the sum of the number of causal factors n that are not common to "factors with intervention experience" among the causal factors of "factors without intervention experience" and the number m that are not common to "factors without intervention experience" among the causal factors of "factors with intervention experience," and multiplies the result by 100. Specifically, the intervention effect prediction unit 114 calculates the correlation between causal factors as j / (j + n + m)*100 when the causal factors of "factors without intervention experience" are {c1, c2, …, cj, f1, f2, …, fn} and the causal factors of "factors with intervention experience" are {c1, c2, …, cj, g1, g2, …, gm}, and j factors c1, c2, …, cj are common.

[0069] Next, the intervention effect prediction unit 114 calculates the degree of association (similarity) of the outcome factors (S4022). For example, the intervention effect prediction unit 114 takes the number of outcome factors k common to "factors without intervention experience" and "factors with intervention experience" as the degree of association of the causal factors, divides the number of common outcome factors k by the sum of the number of outcome factors q of "factors without intervention experience" that are not common to "factors with intervention experience" and the number p of outcome factors of "factors with intervention experience" that are not common to "factors without intervention experience", and multiplies the result by 100. Specifically, the intervention effect prediction unit 114 calculates the correlation between the outcome factors as k / (k + q + p)*100 when the outcome factors for "factors without intervention experience" are {b1, b2, …, bk, s1, s2, …, sq} and the outcome factors for "factors with intervention experience" are {b1, b2, …, bk, t1, t2, …, tp}, and k factors b1, b2, …, bk are common.

[0070] Next, the intervention effect prediction unit 114 calculates the degree of causal relationship (A) (S4023). For example, the intervention effect prediction unit 114 uses the average value of the degree of causal relationship and the degree of causal relationship of the outcome factor as the degree of causal relationship (A). Specifically, the intervention effect prediction unit 114 calculates the degree of causal relationship (A) as (j / (j + n + m) + k / (k + q + p)) / 2 * 100. Note that if there are no causal factors (or outcome factors) that are either "factors without intervention experience" or "factors with intervention experience," the intervention effect prediction unit 114 uses the degree of causal relationship (A) as the degree of causal relationship of the outcome factor (or causal factor). That is, the intervention effect prediction unit 114 calculates the similarity of causal factors and outcome factors between each factor (the proportion of causal factors and outcome factors that are common between each factor) as the similarity of causal relationships between each factor (degree of causal relationship (A)).

[0071] For example, if the outcome factor for the factor "no exercise habits" (which has no prior intervention history) is {fracture}, and the outcome factors for the factor "low BMI" (which has prior intervention history) are {fracture, malnutrition}, then only "fracture" is common to both factors. There are no other outcome factors common to either "no exercise habits" or "low BMI," and there is one outcome factor common to both factors (malnutrition). Therefore, the intervention effect prediction unit 114 calculates the correlation between the outcome factors as 1 / (1+0+1)*100 = 50%. Also, since neither "no exercise habits" nor "low BMI" has a causal factor, the intervention effect prediction unit 114 sets the correlation (A) of the causal relationship to the same 50% as the correlation of the outcome factors.

[0072] After calculating the degree of causal relationship (A), the intervention effect prediction unit 114 calculates the degree of association (B) of the risk of developing the preventable factor between "factors without intervention history" and "factors with intervention history" (S403).

[0073] <Calculation process for the degree of association (B) of the risk of developing the disease due to preventable factors> Figure 16 is a process flow diagram illustrating the details of the calculation process for the association degree (B) of the risk of developing the preventable factor. The process shown in this figure is a detailed representation of the process described in S403 above.

[0074] The intervention effect prediction unit 114 refers to the risk information 124 for the preventable factor and calculates the association degree (B) of the risk of developing the preventable factor by subtracting the difference between the risk of developing the preventable factor for "factors without intervention history" and the risk of developing the preventable factor for "factors with intervention history" from 100 (S4031). Specifically, if the risk of developing the preventable factor for "factors without intervention history" is k%, and the risk of developing the preventable factor for "factors with intervention history" is s%, the intervention effect prediction unit 114 sets the association degree (B) of the risk of developing the preventable factor to (100 - |k - s|). For example, if the risk of developing the preventable factor for the factor "no exercise habits" (without intervention history) is 21%, and the risk of developing the preventable factor for the factor "low BMI" (with intervention history) is 20%, the intervention effect prediction unit 114 sets the association degree (B) of the risk of developing the preventable factor to (100 - |21-20|) = 99%. In other words, the intervention effect prediction unit 114 defines the similarity of the risk of developing the preventable factor as the association degree (B) of the risk of developing the preventable factor.

[0075] After calculating the correlation (B) between the risk of developing the preventable factor, the intervention effect prediction unit 114 calculates the correlation (association) between "factors without intervention history" and "factors with intervention history" by averaging the correlation (A) for causal relationship and the correlation (B) for the risk of developing the preventable factor (S404). For example, the correlation (A) between the factor without intervention history, "no exercise habits," and the factor with intervention history, "low BMI," is 50%, and the correlation (B) for the risk of developing the preventable factor is 99%. Therefore, the intervention effect prediction unit 114 calculates the correlation as (50% + 99%) / 2 = 75% (rounded to the nearest whole number).

[0076] The intervention effect prediction unit 114 sets the calculated causal relationship correlation (A), the calculated correlation of the risk of developing the preventable target factor (B), and the calculated correlation into the correlation information 127.

[0077] (Relevance information) Figure 17 shows an example of relevance information. Relevance information 127 is data representing the relationship between "factors without intervention history" and "factors with intervention history." Relevance information 127 includes the following data items: Factors without intervention history 1271, for which "factors without intervention history" is set; Factors with intervention history 1272, for which "factors with intervention history" is set; Preventable target factors 1273, for which preventable target factors are set; Relevance 1274, for which the calculated relevance is set; Risk of onset 1275, for which the calculated relevance of preventable target factor onset risk (B) is set; and Causal relationship relevance 1276, for which the calculated relevance of causal relationship (A) is set.

[0078] After performing the S402-S404 processes on all "factors with a track record of intervention," the intervention effect prediction unit 114 extracts the "factors with a track record of intervention" that are most highly correlated with the "factors without a track record of intervention" (S405).

[0079] Next, the intervention effect prediction unit 114 determines whether the correlation of the extracted "factors with a track record of intervention" is above a certain level (a predetermined threshold) (S406). If the correlation is below a certain level (S406: No), the process S402 to S407 is executed for the next "factor without a track record of intervention". In other words, the intervention effect prediction unit 114 determines that there are no "factors with a track record of intervention" related to the "factor without a track record of intervention", and does not predict the intervention effect of the "factor without a track record of intervention".

[0080] On the other hand, the intervention effect prediction unit 114 predicts the intervention effect of the "factor with no prior intervention history" if the degree of relevance is above a certain level (S406: Yes) (S407).

[0081] <Prediction process for intervention effect> Figure 18 is a process flow diagram illustrating the details of the intervention effect prediction process. The process shown in this figure is a detailed description of the process in S407 described above.

[0082] First, the intervention effect prediction unit 114 predicts that the participation rate and risk improvement rate of "factors without prior intervention experience" will be the same as the participation rate and risk improvement rate of "factors with prior intervention experience" which have the highest correlation (S4071).

[0083] Next, the intervention effect prediction unit 114 predicts the intervention effect by subtracting the total intervention cost from the medical and nursing care cost reduction cost (S4072). Here, if the predicted participation rate (predicted participation rate) is k%, the predicted risk improvement rate (predicted risk improvement rate) is m%, the number of people with the intervention factor is N, the intervention cost required per person is A yen, and the future medical and nursing care cost required per person is B yen, the intervention effect prediction unit 114 calculates the total intervention cost as A*N*k / 100 yen, the medical and nursing care cost reduction cost as B*N*k / 100*m / 100 yen, and the effect prediction as B*N*k / 100*m / 100 - A*N*k / 100. For example, if the predicted participation rate for the factor "no exercise habits" (which has no prior intervention history) is 10%, the predicted risk improvement rate is 5%, there are 12,000 people with the factor "no exercise habits", the intervention cost per person is 10,000 yen, and the future medical and nursing care cost per person is 500,000 yen, then the intervention effect prediction unit 114 calculates the total intervention cost as 10,000 yen * (12,000 people * 10 / 100) = 12 million yen, the medical and nursing care reduction cost as 500,000 yen * (12,000 people * 10 / 100 * 5 / 100) = 30 million yen, and the predicted effect as 30 million yen - 12 million yen = 18 million yen.

[0084] After performing the processes S402 to S407 for all "factors with no prior intervention history," the intervention effect prediction unit 114 sets the calculated information in the intervention effect prediction information 128 (S408). Then, the intervention effect prediction process is terminated.

[0085] (Information predicting intervention effects) Figure 19 shows an example of intervention effect prediction information. Intervention effect prediction information 128 is data representing the prediction of the intervention effect for "factors with no prior intervention history". Intervention effect prediction information 128 includes the following data items: project 1281 where the intervention project is set, factor 1282 where "factors with no prior intervention history" are set, number 1283 where the number of people who can be intervened and have "factors with no prior intervention history" are set, participant prediction 1284 where the predicted number of participants in the intervention project is set, risk improvement rate prediction 1285 where the predicted risk improvement rate is set, and effect estimation prediction 1286 where the calculated effect prediction is set.

[0086] (Screen showing predicted intervention effects of factors) Figure 20 shows an example of a screen displaying the intervention effect of a factor. After the intervention effect prediction processing, for example, when a predetermined input is made to the user terminal 20 by the intervention condition selector U, the intervention effect prediction display unit 115 displays the factor intervention effect performance display screen 510 shown in this figure on the user terminal 20 based on the factor causal relationship information 125, the risk information of the onset of the preventable target factor 124, the actual intervention effect information 126, the degree of relevance information 127, and the intervention effect prediction information 128.

[0087] The intervention effectiveness results display screen 510 includes a display area 511 for the intervention project "FY23 Exercise Class for Dementia Prevention," a display area 512 for the preventable factor "Dementia," and a display area 513 for a table showing the risk of onset (risk of onset of preventable factor) and intervention results for each "factor with intervention results." The intervention results include participants (number of participants), risk improvement (change in the risk of onset of preventable factor), and effect estimation (intervention effect).

[0088] Furthermore, the intervention effect prediction display unit 115 may prioritize the display of the "factor with a proven track record" with the highest estimated effect on the intervention effect results display screen 510 over other "factors with a proven track record." "Prioritizing display" includes displaying the "factors with a proven track record" in order from those with the highest estimated effect, or highlighting the "factor with a proven track record" with the highest estimated effect. "Highlighting" includes changing the display color, changing the font size, changing the background color, or adding a mark, etc. In the illustrated example, the "low BMI" factor with the highest estimated effect is highlighted. This allows the intervention condition selector U to easily identify the factor with the highest estimated effect. The intervention effect results display screen 510 also has selection buttons that accept selection input for each "factor with a proven track record."

[0089] When the intervention effect prediction display unit 115 receives input from the selection button for a specific "factor with a track record of intervention" on the intervention effect performance display screen 510, it refers to the correlation information 127 and displays the intervention effect prediction display screen 520 for "factors without a track record of intervention" that are related to the selected "factor with a track record of intervention". In the illustrated example, the intervention effect prediction display screen 520 for "factors without a track record of intervention" whose correlation with "low BMI", the factor with the highest estimated effect of intervention, is above a predetermined threshold is displayed on the user terminal 20.

[0090] The intervention effect prediction display screen 520 displays a display area 521 for the preventable target factor "dementia," a display area 522 for the selected factor with a history of intervention "low BMI," and a display area 523 for a table showing the estimated effect for each "factor without a history of intervention," as well as the basis used for the prediction (relationship (degree of relevance) with the factor with a history of intervention "low BMI," the risk of onset (risk of onset for the preventable target factor), causal factors, and outcome factors). For reference, the table may also display the estimated effect (actual), risk of onset, causal factors, and outcome factors for the selected factor with a history of intervention "low BMI."

[0091] By referring to the intervention effect prediction display screen 520, the intervention condition selector U can easily identify "factors without prior intervention experience" that are predicted to have a high intervention effect in the intervention project. In the illustrated example, the factors "no exercise habit" and "slow walking speed," which are related to the factor "low BMI" (a factor with a high intervention experience) and have a high effect estimation, can be identified as factors with a high intervention effect.

[0092] As described above, the intervention effect prediction server 10 of this embodiment performs an intervention effect prediction process that predicts the intervention effect on factors for which there is no prior intervention record by an intervention project, based on intervention effect performance information 126 representing the past intervention effect of an intervention project that provides intervention for a predetermined health condition (a state requiring care, a state of worsening condition) for each participant, and the relationship between each factor (relevance information 127), and output processing that outputs the prediction results.

[0093] In other words, the intervention effect prediction server 10 of this embodiment predicts the intervention effect of factors for which there is no prior intervention record, based on the actual intervention effect of factors for which there is a track record of intervention by the intervention project, and the relationships between each factor. This makes it possible to select factors with high intervention effect from among a large number of factors for each intervention project, even if there is little track record of intervention by the intervention project.

[0094] Furthermore, in the intervention effect prediction processing, the intervention effect prediction server 10 calculates the relationship between each factor based on either the risk of developing a preventable target factor (a pre-selected factor) or the causal relationship between each factor.

[0095] This allows for more accurate prediction of intervention effects, for example, by using the intervention history of factors with similar risk factors for developing preventable factors or similar causal relationships between factors to predict the intervention effect of factors that have not yet been intervened in.

[0096] Furthermore, the intervention effect prediction server 10 of this embodiment performs an interfactor causal relationship calculation process based on personal factor information 122 representing the history of factors possessed by each person who can be intervened in by the intervention project. If the probability that a person who can be intervened in who possesses the first factor will possess the second factor at a predetermined time in the future is above a predetermined threshold, the server generates interfactor causal relationship information 125 in which the first factor is the causal factor of the second factor and the second factor is the consequence factor of the first factor. In the intervention effect prediction process, the server refers to the interfactor causal relationship information 125 and calculates the relationship between each factor based on the proportion of causal and consequence factors that are common among the factors.

[0097] In this way, by identifying the causal relationships between each factor using its causal and consequent factors, the relationships between each factor can be calculated with greater accuracy.

[0098] Furthermore, the intervention effect prediction server 10 of this embodiment performs a preventive factor onset risk calculation process based on personal factor information 122, which calculates the probability that a person who is eligible for intervention and has a predetermined factor will have the preventive factor at a predetermined time in the future, as the preventive factor onset risk.

[0099] This allows for more accurate prediction of intervention effects, for example, by using the intervention track record for factors with similar risk of developing the target factors of an intervention program to predict the intervention effect for factors that have not yet been intervened in.

[0100] Furthermore, the intervention effect prediction server 10 of this embodiment refers to the personal factor information 122 and performs an intervention effect calculation process that calculates the change in the probability that a participant with a predetermined factor has a preventable target factor as a result of the intervention of the intervention project, as a change in the risk of developing the preventable target factor for that factor. In the intervention effect prediction process, it predicts the intervention effect on the factor that has not been intervened by calculating the change in the risk of developing the preventable target factor for the factor that has not been intervened by, as a change in the risk of developing the preventable target factor for that factor that has not been intervened by.

[0101] This allows us to predict the effectiveness of interventions for factors that have not been previously intervened in, based on changes in the risk of developing preventable factors for factors that have been previously intervened in.

[0102] Furthermore, the intervention effect prediction server 10 of this embodiment determines the intervention effect for factors without prior intervention history by calculating the difference between the amount of future cost reduction for care or medical treatment of participants and the intervention cost required by the intervention project, based on changes in the risk of developing the preventable target factor for factors without prior intervention history.

[0103] This allows us to calculate the cost-effectiveness of the intervention program as the intervention effect.

[0104] Furthermore, the intervention effect prediction server 10 of this embodiment refers to personal factor information 122 and uses the change in the probability that a participant with a predetermined factor has the target preventive factor due to interventions in past intervention projects as a change in the risk of developing the target preventive factor. It calculates the difference between the amount of cost reduction required for future care or medical treatment of the participant based on the change in the risk of developing the target preventive factor and the intervention cost required by the intervention project as the actual intervention effect for the predetermined factor, and executes an intervention effect calculation process that stores intervention effect result information 126 representing the change in the risk of developing the target preventive factor and the actual intervention effect.

[0105] In other words, according to the intervention effect prediction server 10 of this embodiment, the actual intervention effect of the intervention project can be calculated based on the personal factor information 122 and the intervention effect performance information 126 can be automatically generated. This eliminates the need to manually create the intervention effect performance information 126.

[0106] Furthermore, in the output processing, the intervention effect prediction server 10 of this embodiment outputs the predicted intervention effect for factors without prior intervention experience, the relationship with factors with prior intervention experience in the intervention project, the risk of developing the preventable target factor, the causal factor, and the outcome factor, respectively.

[0107] This allows the selection officer U for the intervention conditions to easily confirm not only the prediction of the intervention effect for factors with no prior intervention history, but also the basis used for that prediction (the relationship with factors with a track record of intervention in the intervention program, the risk of developing the preventable factor, causal factors, and outcome factors).

[0108] The present invention is not limited to the embodiments described above, and can be implemented using any components without departing from its spirit. The embodiments and modifications described above are merely examples, and the present invention is not limited to these as long as the features of the invention are not impaired. Furthermore, although various embodiments and modifications have been described above, the present invention is not limited to these. Other embodiments conceivable within the scope of the technical idea of ​​the present invention are also included within the scope of the present invention.

[0109] For example, some of the hardware components of each device in this embodiment may be provided in other devices.

[0110] Furthermore, each program of the intervention effect prediction server 10 or the user terminal 20 may be installed on other devices, a program may consist of multiple programs, or multiple programs may be integrated into a single program. [Explanation of Symbols]

[0111] 1. Prediction System 10. Intervention Effect Prediction Server 20 User Terminals 111 Calculation Unit for Risk of Onset of Preventable Factors 112. Inter-factor causal relationship calculation unit 113 Intervention Effect Calculation Department 114 Intervention Effect Prediction Unit 115 Intervention Effect Prediction Presentation Section 121 Information on preventable factors 122 Personal factor information 123 Past Business Performance Information 124 Preventable Factors: Risk Information for Onset 125 Inter-factor causal relationship information 126 Information on the effectiveness of interventions 127 Relevance Information 128 Intervention Effect Prediction Information

Claims

1. A storage device that stores intervention effect performance information representing the actual effectiveness of interventions on at least one of multiple factors that may cause a given health condition for past participants, in an intervention program that provides interventions for a predetermined health condition to each participant, and An intervention effect prediction process that predicts the intervention effect on factors for which there is no intervention record under the intervention project, based on the aforementioned intervention effect performance information and the relationships between each factor, Output processing to output the results of the prediction, A prediction device equipped with a processing unit that performs the following.

2. The aforementioned processing apparatus is In the intervention effect prediction process, the risk of developing a preventable target factor, which is a pre-selected factor, or the relationship between each factor is calculated based on the causal relationship between each factor. The prediction device according to claim 1.

3. The memory device stores personal factor information representing the history of factors possessed by each person who can be intervened in by the intervention project, The aforementioned processing apparatus is Based on the aforementioned personal factor information, if the probability that a person who has the first factor and is eligible for intervention will have the second factor at a predetermined time in the future is greater than or equal to a predetermined threshold, a factor-causal relationship calculation process is performed to generate factor-causal relationship information in which the first factor is the causal factor of the second factor and the second factor is the consequence factor of the first factor. In the intervention effect prediction process, the relationship between each factor is calculated based on the proportion of causal and consequent factors common to each factor, by referring to the causal relationship information between factors. The prediction device according to claim 2.

4. The memory device stores personal factor information representing the history of factors possessed by each person who can be intervened in by the intervention project, The aforementioned processing apparatus is Based on the aforementioned personal factor information, a process is performed to calculate the probability that a person who is eligible for intervention and possesses a predetermined factor will have the aforementioned preventable factor at a predetermined time in the future, as the preventable factor onset risk. The prediction device according to claim 2 or 3.

5. The memory device stores personal factor information representing the history of factors possessed by each person who can be intervened in by the intervention project, The aforementioned processing apparatus is Referencing the aforementioned personal factor information, an intervention effect calculation process is performed to calculate the change in the probability that a participant with a predetermined factor has a preventable target factor as a result of the intervention of the intervention project, as a change in the risk of developing the preventable target factor for the predetermined factor. In the aforementioned intervention effect prediction process, the change in the risk of developing the target factor for the factor most highly associated with the factor that has not been intervened is used as the change in the risk of developing the target factor for the factor that has not been intervened, and the intervention effect on that factor is predicted. The prediction device according to claim 1.

6. The aforementioned processing apparatus is In the intervention effect prediction process described above, the difference between the reduction in future costs required for care or medical treatment of participants, based on changes in the risk of developing the preventable target factor for factors that have not been previously intervened, and the intervention cost required by the intervention program, is defined as the intervention effect for the factors that have not been previously intervened. The prediction device according to claim 5.

7. The memory device stores personal factor information representing the history of factors possessed by each person who can be intervened in by the intervention project, The aforementioned processing apparatus is Referencing the aforementioned personal factor information, the change in the probability that a participant with a predetermined factor has the target factor due to past interventions is defined as the change in the risk of developing the target factor. The difference between the reduction in future costs required for the participant's care or medical treatment based on the change in the risk of developing the target factor and the intervention cost required by the intervention is calculated as the actual intervention effect for the predetermined factor. An intervention effect calculation process is then performed to store the intervention effect information, which represents the change in the risk of developing the target factor and the actual intervention effect, in the storage device. The prediction device according to claim 1.

8. The aforementioned processing apparatus is In the output processing described above, the predicted intervention effect for factors without prior intervention history, the relationship with factors that have prior intervention history in the intervention program, the risk of developing the preventable target factor, the causal factor, and the outcome factor are output, respectively. The prediction device according to claim 3.

9. Information processing device, Intervention effect performance information representing the past intervention effect of an intervention program that provides interventions for a predetermined health condition to each participant, for at least one of several factors that may cause the said health condition in past participants, and intervention effect prediction processing that predicts the intervention effect for factors for which there is no prior intervention performance by the intervention program, based on the relationships between each factor. Output processing to output the results of the prediction, A prediction method that performs this task.

10. In an information processing device, Intervention effect performance information representing the past intervention effect of an intervention program that provides interventions for a predetermined health condition to each participant, for at least one of several factors that may cause the said health condition in past participants, and intervention effect prediction processing that predicts the intervention effect for factors for which there is no prior intervention performance by the intervention program, based on the relationships between each factor. Output processing to output the results of the prediction, A prediction program that executes the prediction.

Images