Path estimation device, path estimation method, and path estimation program

Abstract

This invention provides a path estimation device that can improve the accuracy of estimating the trends in health-related data. [Solution] The path estimation device 10 comprises an acquisition unit 11, a generation unit 13, a path estimation unit 14, and an output unit 15. The acquisition unit 11 acquires health data for multiple individuals. The generation unit 13 generates a probability distribution of occurrence of health data based on the health data for each of the multiple individuals at multiple points in time. The path estimation unit 14 estimates the paths between the health data of the target individuals at different points in time on the generated probability distribution. The output unit 15 outputs information about the estimated paths. With this configuration, the path estimation device 10 can support decision-making based on the estimation results of the transition of health data.

Description

Technical Field

[0005]

[0001] The present disclosure relates to a route estimation device and the like.

Background Art

[0002] For improving health conditions or preparing for future life, an estimation of future health conditions may be performed. The estimation of future health conditions is, for example, performed based on data indicating the current health condition. Also, in order to grasp the health condition in more detail, a transition of the health condition from the present to the future may be estimated.

[0003] <00?0012>The health improvement route exploration device of Patent Document 1 estimates a health index based on measurement values measured by a health check or the like. Then, the health improvement route exploration device of Patent Document 1 identifies a route from the current health index to the improved health index based on the probability distribution of the estimated health index.

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0005] In the technology described in Patent Document 1, it may be difficult to improve the estimation accuracy of the transition of health-related data. <000003? An object of the present disclosure is to provide a route estimation device or the like that can improve the estimation accuracy of the transition of health-related data in order to solve the above problems.

Means for Solving the Problems

[0007] To solve the above problems, the path estimation device of this disclosure comprises: acquisition means for acquiring health data of multiple persons; generation means for generating a probability distribution of occurrence of health data based on health data of multiple persons at multiple time points; path estimation means for estimating the path between the health data of target persons at different time points on the generated probability distribution; and output means for outputting information about the estimated path.

[0008] The path estimation method of this disclosure acquires health data of multiple individuals, generates a probability distribution of occurrence of health data based on the health data of each individual at multiple time points, estimates the path between the health data of the target individuals at different time points on the generated probability distribution, and outputs information about the estimated path.

[0009] The path estimation program of this disclosure causes a computer to perform the following processes: acquiring health data of multiple individuals; generating a probability distribution of occurrence of health data based on the health data of each individual at multiple points in time; estimating the paths between the health data of the target individuals at different points in time on the generated probability distribution; and outputting information about the estimated paths. [Effects of the Invention]

[0010] According to this disclosure, it is possible to improve the accuracy of estimating trends in health-related data. [Brief explanation of the drawing]

[0011] [Figure 1] This figure shows an example of the configuration of the path estimation system described in this disclosure. [Figure 2] This figure shows an example of the configuration of the path estimation device in this disclosure. [Figure 3] This figure schematically illustrates an example of the process for generating the probability distribution of occurrence in this disclosure. [Figure 4]A diagram schematically showing an example of a process for estimating a path on a probability of occurrence distribution in the present disclosure. [Figure 5] A diagram schematically showing an example of a process for estimating a path on a probability of occurrence distribution in the present disclosure. [Figure 6] A diagram showing an example of an estimation result of a path in the present disclosure. [Figure 7] A diagram showing an example of an estimation result of a path in the present disclosure. [Figure 8] A diagram showing an example of an estimation result of a path in the present disclosure. [Figure 9] A diagram showing an example of an estimation result of a path in the present disclosure. [Figure 10] A diagram showing an example of an estimation result of a path in the present disclosure. [Figure 11] A diagram showing an example of an estimation result of a path in the present disclosure. [Figure 12] A diagram showing an example of an estimation result of a path in the present disclosure. [Figure 13] A diagram showing an example of an estimation result of a path in the present disclosure. [Figure 14] A diagram showing an example of a display screen of an estimation result of a path in the present disclosure. [Figure 15] A diagram showing an example of a display screen of an estimation result of a path in the present disclosure. [Figure 16] A diagram showing an example of an operation flow of a path estimation device in the present disclosure. [Figure 17] A diagram showing an example of a hardware configuration of a path estimation device in the present disclosure.

Mode for Carrying Out the Invention

[0012] Embodiments of the present disclosure will be described in detail with reference to the drawings. FIG. 1 is a diagram showing an example of the configuration of a route estimation system. The route estimation system includes a route estimation device 10, a terminal device 20, and a data management device 30. The route estimation device 10 is connected to the terminal device 20 via a network, for example. The route estimation device 10 is connected to the data management device 30 via a network, for example. Also, there may be a plurality of terminal devices 20 and data management devices 30, respectively. The number of the terminal devices 20 and the data management devices 30 can be set as appropriate.

[0013] The route estimation system estimates, for example, the route between data related to the health of a target person at different times. For example, the route estimation system estimates the route between data related to the health of a target person at the current time and a future time. The route is, for example, the transition of the values of health-related data between the data.

[0014] The route estimation system estimates, for example, the route between data related to the health of a target person on the occurrence probability distribution of health-related data. The occurrence probability distribution of health-related data is generated based on, for example, two or more items of health-related data. The occurrence probability distribution of health-related data is generated based on, for example, time-series data related to the health of a plurality of persons. Health-related data is, for example, data indicating the health state of a target person. For example, health-related data is a measured value of the physical state. Also, health-related data may be an index calculated from the measured value of the physical state.

[0015] The route between data related to the health of a target person is, for example, the transition of data between the data related to the health of the target person plotted on the occurrence probability distribution of health-related data. That is, the route estimation system estimates how the data related to the health of the target person transitions on the occurrence probability distribution of health-related data. Also, the route estimation system may estimate the route of health-related data between adjacent time points for data at a plurality of time points in time series, for example.

[0016] The path between health data for a subject is, for example, the progression of data over time intervals finer than the time intervals at which the data exists. For example, if the data are measured or estimated values ​​at one-year intervals, the path between health data for a subject is the progression of data over one month. For example, if the probability distribution of occurrence is divided into grids, and the health data at time t1 is in grid P1, and the health data at time t2 is in grid P2, the path between health data for a subject is the order of grids that the data passes through between P1 and P2 over time. The subject is, for example, the person for whom the path between health data at different time points is estimated. The grid represents, for example, one interval obtained by dividing each variable used in the probability distribution of occurrence into predetermined values ​​on the probability distribution. The predetermined values ​​are, for example, set for each variable used in the probability distribution of occurrence. The predetermined values ​​are set so that, for example, there is a meaningful difference when the section containing the data changes for each variable used in the probability distribution of occurrence. In this way, by estimating the paths between data related to the health of a subject, the path estimation system can improve, for example, the accuracy of estimating data trends. In this case, accuracy refers to the degree of precision in representing changes in health data as trends. For example, the more continuously the health data changes, the higher the accuracy, and the less discontinuous the changes, the lower the accuracy. For example, by estimating the paths between health data, the continuously changing nature of health data can be represented with precision, and therefore, the path estimation system can improve, for example, the accuracy of estimating data trends.

[0017] Here, a specific example of the configuration of the path estimation device 10 will be described. Figure 2 shows an example of the configuration of the path estimation device 10. The path estimation device 10 basically comprises an acquisition unit 11, a generation unit 13, a path estimation unit 14, and an output unit 15. The path estimation device 10 may further comprise, for example, a data estimation unit 12 and a storage unit 16.

[0018] The acquisition unit 11 acquires health data for multiple individuals. The acquisition unit 11 acquires health data, for example, with information indicating the attributes of the individual corresponding to each piece of health data associated with it. Attributes are, for example, information indicating groups where differences in attributes may result in differences in the trends of health data. For example, the trends in health data may differ between individuals living in cold regions and individuals living in warm regions. In such cases, information indicating place of residence may be used as an attribute. Attributes are, for example, information on one or more items from age, gender, place of residence, nationality, occupation, medical history, and family medical history. Attributes are not limited to those listed above. The acquisition unit 11 acquires health data for each of the multiple individuals from, for example, the data management device 30.

[0019] Health data includes, for example, data indicating physical condition. Health data may also include indicators calculated from measured values. Health data includes, for example, data on one or more items from the following: health checkup results, interview results, hospital test values, vital data, presence or absence of disease, probability of disease onset, physician's findings, motor function, progression of dementia, progression of frailty, need for care, and degree of care needed. Health checkup results include, for example, data on one or more items from the following: height, weight, vision, blood pressure, waist circumference, hearing, blood test measurements, imaging results, and physician's interview results measured during the health checkup. The progression of dementia is, for example, an indicator showing the degree of dementia progression. For example, the higher the degree of dementia progression, the higher the risk of daily life being impaired due to dementia. The progression of frailty is, for example, an indicator showing the degree of frailty progression. For example, the higher the degree of frailty progression, the higher the risk of daily life being impaired due to frailty.

[0020] Health-related data may include expenses necessary for maintaining a healthy state or expenses necessary depending on the health state. Expenses necessary for maintaining a healthy state or expenses necessary depending on the health state include, for example, medical expenses or living expenses. For example, medical expenses or living expenses may increase as the progression of dementia increases. Similarly, medical expenses or living expenses may increase as the progression of frailty increases. Expenses necessary for maintaining a healthy state or expenses necessary depending on the health state are not limited to those mentioned above. Furthermore, health-related data is not limited to those mentioned above.

[0021] The acquisition unit 11 may acquire time-series data of health-related data for each of multiple individuals. For example, if the health-related data consists of the results of a health checkup, and each individual undergoes a health checkup once a year, the acquisition unit 11 may acquire the annual health checkup results for each of the multiple individuals. For example, the acquisition unit 11 may acquire data showing the trend of health checkup results as each of the multiple individuals increases in age, as time-series data of health-related data.

[0022] The acquisition unit 11 acquires, for example, data related to the health of a target person. The acquisition unit 11 acquires, for example, information specifying the target person from the terminal device 20. The acquisition unit 11 then acquires, for example, information related to the health of the specified target person from the data management device 30. The information specifying the target person may include the attributes of the target person. The acquisition unit 11 may also acquire data related to the health of the target person from the terminal device 20.

[0023] The acquisition unit 11 may acquire information specifying a target point. The target point is, for example, a grid that is the endpoint of a path when healthily estimating a path between functional data on an occurrence probability distribution. The acquisition unit 11 acquires information specifying a target point from, for example, a terminal device 20. The acquisition unit 11 may acquire information specifying a grid that is the starting point of a path when healthily estimating a path between functional data on an occurrence probability distribution. The acquisition unit 11 acquires information specifying a grid that is the starting point of a path from, for example, a terminal device 20.

[0024] The acquisition unit 11 may acquire information specifying the grids to be masked. Masking is, for example, a process of excluding grids set as targets for masking from the estimation of paths between health-related data. That is, masking is, for example, a process of setting grids that block health-related paths from passing through. The acquisition unit 11 acquires information specifying the grids to be masked from, for example, the terminal device 20. The acquisition unit 11 may also acquire information specifying grids to be excluded from masking among the grids to be masked. The acquisition unit 11 may acquire information specifying the variables to be masked as information specifying the masking targets. The acquisition unit 11 acquires information specifying grids to be excluded from masking from, for example, the terminal device 20.

[0025] The data estimation unit 12 estimates time-series data of health information for each of several individuals, based on their respective health data. The time-series data estimated from the health data for each of several individuals is used, for example, to generate the probability density distribution of health information in the generation unit 13. The time-series data estimated from the health data for each of several individuals is, for example, data from a later point in the time series than the health data used as the basis for the estimation. In other words, the data estimation unit 12 estimates time-series data of health information for each of several individuals at a point in the future than when the health data was measured.

[0026] The data estimation unit 12 estimates time-series data for each of several individuals' health-related data, for example, using a data estimation model. The data estimation model is a machine learning model that estimates time-series data of health-related data from input health-related data.

[0027] The data estimation unit 12 estimates time-series data of a target person's health based on, for example, data related to the target person's health. The data estimation unit 12 estimates time-series data of a target person's health using, for example, a data estimation model. The data estimation model for estimating time-series data of a target person's health is, for example, the same as the data estimation model for estimating time-series data of health for each of multiple individuals. The data estimation model for estimating time-series data of a target person's health may be, for example, a different machine learning model from the data estimation model for estimating time-series data of health for each of multiple individuals. A different machine learning model is, for example, a machine learning model in which at least one of the training data and the training algorithm is different.

[0028] The data estimation model is a machine learning model that uses health-related data as input data to estimate time-series data of health-related data at a later point in the time series than the input data. The data estimation model may also be a machine learning model that uses time-series data of health-related data as input data to estimate time-series data of health-related data at a later point in the time series than the input data. For example, if the health-related data is the result of a health checkup, the data estimation unit 12 estimates the results of health checkups for each year for the next 20 years, based on the results of health checkups for the past 5 years up to the current year. The data estimation model may be generated for each attribute of the person to be estimated. The data estimation model may be generated for example by deep learning using a neural network. The machine learning algorithm for generating the data estimation model is not limited to the above. Furthermore, the data estimation model may be generated for example by a device outside the path estimation device 10. The data estimation model may also be generated by a learning means (not shown) within the path estimation device 10.

[0029] The data estimation unit 12 may estimate time-series data of health-related data using a function that calculates health-related data. For example, the data estimation unit 12 calculates health-related data at each time point in the time series that is the target of estimation, using a function that uses the elapsed time from the present and health-related data at the present as explanatory variables. Then, the data estimation unit 12 estimates time-series data of health-related data by generating time-series data based on the calculated health-related data at each time point in the time series that is the target of estimation. The function that calculates health-related data is generated, for example, by regression analysis using time-series data of health-related data. The algorithm used to generate the function that calculates health-related data is not limited to the above.

[0030] The generation unit 13 generates a probability distribution of health data based on health data for multiple individuals at multiple points in time. The generation unit 13 generates a probability distribution of health data based, for example, on time-series data of health for multiple individuals estimated by the data estimation unit 12. The generation unit 13 generates a probability distribution of health data for multiple individuals based, for example, on time-series data of health for multiple individuals estimated by a data estimation model. The generation unit 13 may also generate a probability distribution of health data based on time-series data of health for multiple individuals acquired as actual measurement data. Furthermore, the generation unit 13 may generate a probability distribution of health data based on data estimated by a data estimation model and actual measurement data.

[0031] For example, when using two data items from health-related data as variables in the probability distribution, the generation unit 13 generates the probability distribution on a grid with the two data items as axes. For example, if the health-related data are the progression of dementia and the progression of frailty, the generation unit 13 classifies the health-related data into grids separated by each indicator. Also, for example, if the health-related data are mean blood pressure and HbA1c, the generation unit 13 classifies the health-related data into grids separated by 5 mmHg intervals for mean blood pressure data and 0.1 percent intervals for HbA1c data. Then, the generation unit 13 generates the probability distribution by calculating the probability of occurrence in each grid. The generation unit 13 calculates the probability of occurrence in each grid, for example, by dividing the number of data items classified in each grid by the total number of data items. The health-related data used to generate the probability distribution may include three or more items. Furthermore, the items from the health-related data used as variables in the probability distribution are not limited to those described above.

[0032] The generation unit 13 generates a probability distribution of the occurrence of health data, for example, by using time-series data of health data for each of several individuals as independent data regardless of the individual or the time point in the time series. That is, the generation unit 13 generates a probability distribution of the occurrence of health data, for example by using health data for each of several individuals at multiple time points as independent data. For example, if there is health data for n time points for each of M individuals, the generation unit 13 generates a probability distribution of the occurrence of health data using M × n data points.

[0033] For example, if the health data for multiple individuals is for each year, the generation unit 13 generates a probability distribution of the health data by using the data for the same individual but for different years as independent data. For example, if the health data for person A over n years is A1, A2, ..., An, and the health data for person B over n years is B1, B2, ..., Bn, the generation unit 13 generates a probability distribution of the health data using A1, A2, ..., An, B1, B2, ..., Bn. By generating a probability distribution in this way, the time-series progression of the health data can be reflected in the probability distribution.

[0034] Figure 3 schematically illustrates an example of the process for generating an occurrence probability distribution. In the example in Figure 3, the occurrence probability distribution is generated based on the progression of frailty and dementia for each of several individuals. In the example in Figure 3, the generation unit 13 generates an occurrence probability distribution with the progression of frailty and dementia as axes, for example. In the example in Figure 3, the generation unit 13 generates an occurrence probability distribution using, for example, five years' worth of data for each of several individuals. In the example occurrence probability distribution in Figure 3, the darker the grid color, the higher the probability of occurrence. The generation unit 13 generates an occurrence probability distribution using, for example, one year's worth of data for each of several individuals as one data point.

[0035] The generation unit 13 generates a probability distribution for each combination of health data items. The combination of health data items used to generate the probability distribution is set, for example, based on the magnitude of the impact on daily life if a symptom related to that item occurs. Alternatively, the combination of health data items used to generate the probability distribution may be set, for example, based on the magnitude of the medical expenses incurred if a symptom related to that item occurs. Alternatively, the combination of health data items used to generate the probability distribution may be set based on the incidence rate of diseases related to that item.

[0036] The generation unit 13 may generate probability distributions for each attribute. For example, the generation unit 13 may generate probability distributions for each gender. The generation unit 13 may also generate probability distributions for each age group that serves as the starting point for estimation. For example, when estimating the data path of health for a person in their 50s until they reach 70 years old, the generation unit 13 generates a probability distribution using the health data of the person in their 50s and the predicted values ​​of health data for each increasing age predicted from that health data. The generation unit 13 may also generate a probability distribution for each region of residence. The attributes used to generate the probability distributions are not limited to those mentioned above. The generation unit 13 stores the generated probability distributions in the storage unit 16, for example, associating them with the attributes.

[0037] The path estimation unit 14 estimates the paths between health data of a subject at different points in time, using a generated probability distribution. For example, if the health data is yearly, the path estimation unit 14 estimates the paths between the yearly data.

[0038] The path estimation unit 14 extracts, for example, path patterns between two points for which a path between data points is to be estimated. The path estimation unit 14 extracts path patterns such as those that do not pass through the same grid more than once. The path estimation unit 14 may also extract path patterns that do not progress in a direction in which health-related data improves. For example, if the health-related data is mean blood pressure, the path estimation unit 14 estimates a path pattern that progresses either in a direction in which the value increases or in a direction in which the value remains the same. This is because mean blood pressure tends to increase with age, for example. The path estimation unit 14 may also extract paths as path patterns that pass through fewer than a predetermined number of grids. The predetermined number is set, for example, based on the number of grids present between two points for which a path is to be estimated. For example, if the two points for which a path is to be estimated are at diagonal corners of a 3x3 grid, the probability of passing through a path outside the 3x3 grid is low, and the probability of passing through the same point multiple times within the grid is also low. In such cases, the predetermined number is set to, for example, "4," which is the number of grids required to reach one point from the other while satisfying the above conditions.

[0039] The path estimation unit 14 calculates, for example, the probability of passing through each of the extracted path patterns as the pass probability. The path estimation unit 14 calculates the pass probability for each path by multiplying it by the probability of occurrence of each grid on the path. Then, the path estimation unit 14 estimates, for example, the path with the highest pass probability as the path between the two points of the target. The path estimation unit 14 may estimate multiple paths to the target point of the health data of the target person on the occurrence probability distribution. For example, the path estimation unit 14 estimates paths whose pass probability satisfies a criterion as candidate paths between data. The criterion for pass probability is set, for example, to broadly extract possible paths while ensuring that the number of extracted candidates is large enough to consider each of the extracted candidates.

[0040] Furthermore, for example, if data exists at multiple points in a time series, the path estimation unit 14 estimates the path between two consecutive points in the time series. For example, suppose there is data for five years, and the data for the first year is d1, the data for the second year is d2, the data for the third year is d3, the data for the fourth year is d4, and the data for the fifth year is d5. In this case, the path estimation unit 14 estimates the path between d1 and d2, between d2 and d3, between d3 and d4, and between d4 and d5. Then, the path estimation unit 14 estimates the overall path between d1 and d5 by connecting the paths between two consecutive points in the time series. Alternatively, the path estimation unit 14 may estimate a new path by branching off from a partial path between two consecutive points in the time series and connecting it to another path in the vicinity of that partial path.

[0041] The path estimation unit 14 may estimate paths to multiple target points for health data of the subject person on the probability distribution of occurrence. A target point is, for example, a grid corresponding to the highest estimated age when health data for each age of the subject person is estimated. A target point may also be, for example, a grid corresponding to the estimated health data for an age specified by the user when estimating health data as the subject person ages. A target point may also be, for example, a grid corresponding to a health data value that is preferable to reach as the subject person ages. Furthermore, the target point may be specified by the user. Information specifying the target point is acquired, for example, by the acquisition unit 11 from the terminal device 20.

[0042] Figure 4 plots the yearly data of the subject individuals on the probability distribution of occurrence shown in the example in Figure 3. In the example in Figure 4, probability distributions are generated with the progression of frailty and the progression of dementia as axes, respectively, based on the progression of frailty and the progression of dementia for each of the multiple individuals. In addition, in the example in Figure 4, five years' worth of data for the subject individuals is plotted on the probability distribution with the progression of frailty and the progression of dementia as axes. In the example in Figure 4, each of the subject individuals' data is plotted as a circle on the probability distribution.

[0043] Figure 5 shows an example of the estimated paths between data points on the probability distribution of occurrence shown in Figure 4. In the example in Figure 5, probability distributions are generated with the degree of frailty progression and the degree of dementia progression as axes, respectively, based on the degree of frailty progression and the degree of dementia progression for each of several individuals. In the example in Figure 5, five years' worth of data for the subject is plotted year by year on the probability distribution of occurrence with the degree of frailty progression and the degree of dementia progression as axes. In the example in Figure 5, the estimated paths between the yearly data points for the subject plotted on the probability distribution of occurrence are shown by solid lines. In the example in Figure 5, the estimated paths between the yearly data points for the subject are shown on the probability distribution of occurrence with the degree of frailty progression and the degree of dementia progression as axes, respectively.

[0044] Figure 6 shows an example of estimation results when estimating multiple pathways between health data of the subjects. In the example in Figure 6, probability distributions are generated with the degree of frailty progression and the degree of dementia progression as axes, respectively, based on the degree of frailty progression and the degree of dementia progression for each of the multiple subjects. In the example in Figure 6, the estimation results for two pathways from the starting grid S to the ending grid G ​​are shown. In the example estimation results in Figure 6, the estimation results for the two pathways are shown as a solid line and a dashed line.

[0045] The path estimation unit 14 may estimate multiple paths by masking at least one grid point on the probability distribution through which the estimated paths pass. For example, the path estimation unit 14 estimates a new path after masking at least one grid point on the probability distribution through which the estimated paths pass. Masking means, for example, excluding the masked grid from the path candidates. The path estimation unit 14 sets the grid on the estimated path to a masked state. The masked grid is, for example, a grid that is excluded from the extraction of path candidates. The path estimation unit 14 terminates the process of estimating paths between data points if, for example, as a result of masking, it is no longer possible to extract new path patterns.

[0046] Figures 7, 8, and 9 show examples of processes for masking grids on estimated paths between data when multiple paths are estimated. In the examples in Figures 7, 8, and 9, probability distributions are generated with the degree of frailty progression and the degree of dementia progression as axes, respectively, based on the degree of frailty progression and the degree of dementia progression for each of the multiple individuals. In the examples in Figures 7, 8, and 9, the path estimation unit 14 estimates, for example, the path from grid S to grid G.

[0047] Figure 7 shows an example of estimation results where the path with the highest probability of being traversed is estimated as the first path. In the example in Figure 7, the grid indicated by the diagonal lines is the grid on the estimated path. Figure 8 shows an example of estimation results where the path with the highest probability of being traversed is estimated as the second path when the grids traversed by the first path are masked. In the example in Figure 8, the grid indicated in black is the masked grid. In the example in Figure 8, the grid indicated by the diagonal lines is the grid on the path estimated as the second path. Figure 9 also shows an example where the grids traversed by the first and second paths are masked. In the example in Figure 9, the grid indicated in black is the masked grid. In the example in Figure 9, the grids in both directions from the starting grid S are the grids indicated in black. Therefore, in the state shown in the example in Figure 9, there are no grids to proceed to from the starting grid S, so the path estimation unit 14 cannot estimate the third and subsequent paths. For this reason, when a state like the one in Figure 9 occurs, the path estimation unit 14 does not, for example, estimate a new path.

[0048] The path estimation unit 14 may exclude at least one of the grids on the occurrence probability distribution, specifically the grids around the starting point and the grids around the ending point, from the masking target. Figure 10 shows an example where the grids around the starting point and the grids around the ending point are excluded from the masking target. In the example in Figure 10, occurrence probability distributions are generated with the progression of frailty and the progression of dementia as axes, respectively, based on the progression of frailty and the progression of dementia for each of several individuals. Also in the example in Figure 10, the path estimation unit 14 estimates, for example, a path from grid S to grid G. In the example in Figure 10, the grids around the starting point and the grids around the ending point are excluded from the masking target, and an example is shown where the grids traversed by the first and second paths are masked. For example, in the state of the example in Figure 9, since there are only paths that go around grids that are also masked around the grid of the ending point, the accuracy of path estimation decreases when estimating the third and subsequent paths. On the other hand, in the example in Figure 10, the grids around the starting point and the grids around the ending point are excluded from masking, allowing for the estimation of more paths than when masking is performed.

[0049] The path estimation unit 14 may estimate the path between data points by masking at least one variable-side point on the probability distribution of occurrence. For example, the path estimation unit 14 estimates the path between data points by masking at least one variable-side point from a path branching point on the probability distribution of occurrence. For example, if the probability distribution of occurrence has two variables, the grid from the starting grid toward the direction of one variable is masked on the side of one variable. The direction to be masked is set to, for example, the side on which the progression of health-related data is undesirable. For example, when estimating a path that avoids the progression of dementia, the path estimation unit 14 masks the variable side that indicates the progression of dementia. The variable to be masked is specified, for example, by the user.

[0050] Figure 11 shows an example of estimating a path by masking points on one of two variables. In the example in Figure 11, probability distributions are generated with the degree of frailty progression and the degree of dementia progression as axes, respectively, based on the degree of frailty progression and the degree of dementia progression for each of several individuals. In the example in Figure 11, the path estimation unit 14 estimates, for example, the path from grid S to grid G. In the example in Figure 11, the grids that move towards the side where the degree of dementia progression increases are masked. Therefore, in the example in Figure 11, the path estimation unit 14 estimates the path on the side where the degree of frailty progression increases. In this way, by masking one direction, the path estimation unit 14 can estimate, for example, the path in cases where the subject person has a favorable progression. Alternatively, by masking one direction, the path estimation unit 14 may estimate the path in cases where the subject person has a progression that should be avoided.

[0051] Figure 12 also shows an example of estimating a path between data by masking the grid on the side of one of the two variables from a branching point. In the example in Figure 12, probability distributions are generated with the degree of frailty progression and the degree of dementia progression as axes, respectively, based on the degree of frailty progression and the degree of dementia progression for each of several individuals. In the example in Figure 12, the path estimation unit 14 estimates, for example, the path from grid S to grid G. In the example in Figure 12, near branching point D, the grid that leads to the side where the degree of dementia progression increases is masked. Therefore, in the example in Figure 12, the path estimation unit 14 estimates the path on the side where the degree of frailty progression increases. In this way, by masking one direction near the branching point, the path estimation unit 14 can estimate, for example, the path in cases where the subject person makes a favorable transition after the branching point. Furthermore, by masking one direction near the branching point, the path estimation unit 14 may estimate the path that the target person should avoid after the branching point.

[0052] The path estimation unit 14 may estimate the path to the target point of the health data of the subject on the probability distribution of occurrence, based on the probability distribution of occurrence and the estimated medical expenses at each point on the probability distribution of occurrence. For example, the path estimation unit 14 estimates a path in which the estimated medical expenses are lower than those of other paths. For example, the path estimation unit 14 estimates a path in which the estimated medical expenses are lower than those of other paths among paths in which the probability of passing the path satisfies the criteria. The relationship between the health data used as a variable in the probability distribution of occurrence and medical expenses is set, for example, as table-formatted data. For example, the path estimation unit 14 estimates the path in which the cumulative value of medical expenses along the path is the lowest.

[0053] Figure 13 shows an example of a path estimated based on the probability distribution of occurrence and the estimated medical costs at each point on the probability distribution. In the example in Figure 13, the estimated path results are shown on a map composed of the probability distribution of occurrence, with the progression of dementia and the progression of frailty as variables, and medical costs. In Figure 13, two paths are estimated after the branching point. In the example in Figure 13, the rate of increase in medical costs differs between the left path and the right path. Therefore, in the example in Figure 13, the user can determine the preferred path by considering the rate of increase in medical costs.

[0054] The output unit 15 outputs information about the estimated path. For example, the output unit 15 outputs the estimated path superimposed on the probability distribution. The output unit 15 may also output candidate estimated paths based on the probability of passing through each path as information about the estimated path. For example, the output unit 15 outputs candidate estimated paths in a different display format for each probability of passing through each path between the data. For example, the output unit 15 outputs candidate estimated paths by changing at least one of the color and line shape for each probability of passing through each path between the data. The output unit 15 may also output candidate estimated paths by changing at least one of the color and line shape for each probability stage of passing through each path between the data. How each path between the data is output is not limited to the above.

[0055] The output unit 15 may output the target path of the subject person superimposed on the probability distribution as information about the estimated path. For example, the output unit 15 superimposes the estimated path between data points and the target path of the subject person onto the probability distribution in a manner that makes them distinguishable from each other. For example, the output unit 15 outputs the estimated path between data points and the target path of the subject person in a display format in which at least one of the color and line shape is different. The display format of the estimated path between data points and the target path of the subject person is not limited to the above.

[0056] The output unit 15 outputs information about the estimated route to, for example, the terminal device 20. The output unit 15 may also output information about the estimated route to a display device (not shown) connected to the route estimation device 10.

[0057] The output unit 15 may output a display screen for setting a target point. On the display screen for setting a target point, for example, if the user performs a click operation on the probability distribution, the grid where the click was performed will be set as the target point. The output unit 15 may also output a display screen for specifying grids to be masked. On the display screen for specifying grids to be masked, for example, if the user performs a click operation on the probability distribution, the grid where the click was performed will be set as the target for masking. Alternatively, for example, if the user performs a click operation on a grid that is subject to masking on the probability distribution, the grid where the click was performed may be excluded from the target for masking.

[0058] Figure 14 shows an example of a display screen showing the estimated paths between data. In the example screen in Figure 14, the subject's name, current age, and the results of the most recent health checkup are displayed. In the example screen in Figure 14, the estimated path from the age of a 60-year-old subject to the age of 80 is displayed on a probability distribution where the progression of dementia and the progression of frailty are variables. In addition, in the example screen in Figure 14, advice regarding the subject's health is displayed at the bottom of the screen. For example, the output unit 15 outputs a display screen showing the estimated paths between data, as shown in the example in Figure 14, to the terminal device 20. The terminal device 20 then outputs the estimated paths between data to a display device (not shown), for example. By outputting a display screen showing the estimated paths between data, as shown in the example in Figure 14, the path between data related to the subject's health can be presented to the subject.

[0059] Figure 15 shows an example of a display screen that shows the estimated path between data points and the target path. In the example display screen in Figure 15, the subject's name, current age, and the results of the latest health checkup are displayed. The example display screen in Figure 15 shows the estimated path for a 60-year-old subject until they reach the age of 80. The example display screen in Figure 15 also shows the path to the grid containing the target value when the subject reaches the age of 80. For example, the output unit 15 outputs a display screen of the estimated path results, as shown in the example in Figure 15, to the terminal device 20. The terminal device 20 then outputs the estimated path between data points to a display device (not shown), for example. By outputting a display screen of the estimated path results, as shown in the example in Figure 15, it is possible to present the subject with information showing the difference between the path between the subject's health data and the path to reach the target health state.

[0060] The storage unit 16 stores, for example, data relating to the process of estimating paths between data at different time points in time on a probability distribution. The storage unit 16 stores, for example, health data of multiple individuals used to generate the probability distribution. The storage unit 16 stores, for example, time-series data of each of the health data of multiple individuals. The storage unit 16 stores, for example, the probability distribution generated by the generation unit 13. The storage unit 16 stores, for example, health data of a target person. The storage unit 16 stores, for example, time-series data of the health data of a target person. The storage unit 16 stores, for example, the estimation results of paths between data at different time points. The storage unit 16 stores, for example, a data estimation model. The data estimation model may be stored in storage means outside the path estimation device 10.

[0061] The terminal device 20 is used, for example, by a person who utilizes the route estimation results between data at different points in time. The terminal device 20 acquires information regarding the route estimation results from the route estimation device 10, for example. The terminal device 20 then outputs information regarding the route estimation results to a display device (not shown), for example. The terminal device 20 may also acquire information specifying a target person, which is input by the user's operation. The information specifying a target person is, for example, information that identifies each target person or the attributes of the target person. The information specifying a target person is not limited to the above. When acquiring information specifying a target person, the terminal device 20 outputs the information specifying a target person to the acquisition unit 11 of the route estimation device 10.

[0062] The terminal device 20 may acquire information specifying a target grid, which is input by the user. When acquiring information specifying a target grid, the terminal device 20 outputs the information specifying the target grid to the acquisition unit 11 of the path estimation device 10. The terminal device 20 may also acquire information specifying a starting grid for the path, which is input by the user. When acquiring information specifying a starting grid for the path, the terminal device 20 outputs the information specifying a starting grid for the path to the acquisition unit 11 of the path estimation device 10.

[0063] The terminal device 20 may acquire information specifying the grid to be masked, which is input by the user. When acquiring information specifying the grid to be masked, the terminal device 20 outputs the information specifying the grid to be masked to the acquisition unit 11 of the path estimation device 10. The terminal device 20 may also acquire information specifying the grid to be excluded from masking, which is input by the user. When acquiring information specifying the grid to be excluded from masking, the terminal device 20 outputs the information specifying the grid to be excluded from masking to the acquisition unit 11 of the path estimation device 10.

[0064] The results of the estimation of pathways between data are used by, for example, the subject, a person who advises the subject on health matters, or a person who advises the subject on financial matters. Those who advise the subject are, for example, healthcare professionals, insurance agents, human resources personnel, financial planners, or financial institution representatives. Healthcare professionals include doctors, nurses, physical therapists, pharmacists, laboratory technicians, or counselors. Healthcare professionals are not limited to those listed above. Similarly, those who advise the subject are not limited to those listed above. A person using the terminal device 20 can, for example, easily make health decisions regarding the subject by referring to the estimation results of pathways between data at different points in time.

[0065] The terminal device 20 can be, for example, a personal computer, a tablet computer, a smartphone, or a smartwatch. The information processing device used in the terminal device 20 is not limited to those mentioned above.

[0066] The data management device 30 stores, for example, health-related data. The data management device 30 stores, for example, health-related data, associating it with the date the health-related data was measured and the attributes of the person corresponding to the health-related data. The data management device 30 outputs the health-related data to, for example, the acquisition unit 11 of the path estimation device 10.

[0067] When health-related data consists of health checkup results, the data management device 30 stores, for example, the date the health checkup was conducted, the attributes of the person who underwent the health checkup, and the health checkup results in association with each other. Attributes include, for example, information on one or more items from age, gender, place of residence, nationality, occupation, medical history, and family medical history. Attributes are not limited to those listed above. The date the health checkup was conducted may be indicated by the month, year, or fiscal year in which the health checkup was conducted. The data management device 30 may also store the health checkup results as a database classified based on at least one of the date the health checkup was conducted and the attributes of the person who underwent the health checkup.

[0068] The data management device 30 may, for example, store the results of health examinations as anonymized information or pseudonymized information. Anonymized information is, for example, information that has been processed so that an individual cannot be identified even when cross-referenced with other information. Pseudonymized information is, for example, information that cannot identify an individual on its own, but can identify an individual when cross-referenced with other information.

[0069] The data management device 30 stores, for example, the results of health examinations conducted in a predetermined group. The predetermined group is, for example, a group that undergoes health examinations. The predetermined group is, for example, a local government, a company, an organization, a cooperative, a school, or a health insurance association. The predetermined group is not limited to the above. The data management device 30 may also store the results of health examinations for multiple groups.

[0070] The process by which the path estimation device 10 estimates the path between data points will be described. Figure 16 shows an example of the operation flow in the process by which the path estimation device 10 estimates the path between data points.

[0071] The acquisition unit 11 acquires health data of multiple individuals (step S11).

[0072] When health data for multiple individuals is obtained, the generation unit 13 generates a probability distribution of the occurrence of health data based on the health data for each of the multiple individuals at multiple points in time (step S12).

[0073] Once a probability distribution of health-related data is generated, the path estimation unit 14 estimates the paths between health-related data of the subject at different time points on the generated probability distribution (step S13).

[0074] Once the pathway between health data of the subject is estimated, the output unit 15 outputs information about the estimated pathway (step S14). The output unit 15 outputs information about the estimated pathway to, for example, the terminal device 20.

[0075] Each process in the path estimation device 10 may be distributed and executed across multiple information processing devices connected via a network. For example, the processes in the data estimation unit 12 and the generation unit 13, and the processes in the path estimation unit 14 may be performed in separate information processing devices. Alternatively, for example, the processes in the data estimation unit 12, and the processes in the generation unit 13 and the path estimation unit 14 may be performed in separate information processing devices. The choice of which information processing device performs each process in the path estimation device 10 can be set as appropriate.

[0076] The path estimation device 10 generates a probability distribution of occurrence of health data based on health data of multiple individuals at multiple points in time. The path estimation device 10 then estimates the paths between the health data of the target individuals at different points in time on the generated probability distribution. In this way, the path estimation device 10 can improve the accuracy of estimating the progression of health data by estimating the paths between the health data of the target individuals at different points in time on the probability distribution generated based on health data of multiple individuals at multiple points in time. In other words, the path estimation device 10 can accurately represent how health data changes by, for example, estimating the paths between health data.

[0077] By estimating multiple paths to estimated future health data for the subject, the path estimation device 10 can, for example, present a path for the subject's health data. Furthermore, by estimating a path to a target point for the subject's health data, the path estimation device 10 can, for example, output a path to bring the subject's health status closer to the target point.

[0078] By excluding pre-estimated grids near the starting and ending points on the probability distribution from masking and estimating the paths of health-related data, the path estimation device 10 can estimate more paths than, for example, when masking pre-estimated grids near the starting and ending points. It can estimate the paths of health-related data from a wide range of candidates. Therefore, the path estimation device 10 can estimate the paths of health-related data from a wide range of candidates. For this reason, the path estimation device 10 can improve the accuracy of path estimation, for example, when estimating paths for multiple health-related data. That is, the path estimation device 10 can accurately represent how health-related data changes along the paths of multiple health-related data.

[0079] Furthermore, by masking undesirable sides in the probability distribution of occurrence and estimating the path of health-related data, the path estimation device 10 can output a path that moves health-related data toward a desirable side, for example. Therefore, the path estimation device 10 can, for example, support actions for improving the health of users who refer to the estimated path of health-related data. In addition, by having each of the above configurations, the path estimation device 10 can support decision-making based on the estimated results of the progression of health-related data.

[0080] By outputting medical expenses along with the path of health-related data, the path estimation device 10 can output, for example, estimated results of future health status and information on the medical expenses required for each estimated health status. Therefore, the path estimation device 10 can easily grasp the future progression of health status and the necessary medical expenses. Furthermore, by estimating the path that minimizes medical expenses, the path estimation device 10 can output, for example, a path that advances health status in the direction that minimizes medical expenses. Therefore, the path estimation device 10 can, for example, support the decision-making of users who refer to the path estimation results.

[0081] Each process in the path estimation device 10 can be implemented by executing a computer program on a computer. Figure 17 shows an example of the configuration of a computer 100 that executes the computer program that performs each process in the path estimation device 10. The computer 100 includes a CPU (Central Processing Unit) 101, memory 102, storage device 103, input / output interface 104, and communication interface 105.

[0082] The CPU 101 reads and executes computer programs that perform various processing tasks from the storage device 103. The CPU 101 may be composed of a combination of multiple CPUs. Alternatively, the CPU 101 may be composed of a combination of a CPU and another type of processor. For example, the CPU 101 may be composed of a combination of a CPU and a GPU (Graphics Processing Unit). The memory 102 is composed of DRAM (Dynamic Random Access Memory) or the like, and temporarily stores computer programs executed by the CPU 101 and data being processed. The storage device 103 stores computer programs executed by the CPU 101. The storage device 103 is composed of, for example, a non-volatile semiconductor storage device. Other storage devices such as hard disk drives may be used for the storage device 103. The input / output interface 104 is an interface that receives input from the operator and outputs display screens, etc. The communication interface 105 is an interface that sends and receives data between the terminal device 20, the data management device 30, and other information processing devices. The terminal device 20 and the data management device 30 can also be configured similarly to the computer 100.

[0083] The computer programs used to execute each process can also be stored and distributed on a computer-readable recording medium that non-temporarily stores data. Examples of recording media include magnetic tapes for data recording and magnetic disks such as hard disks. Optical discs such as CD-ROMs (Compact Disc Read Only Memory) can also be used as recording media. Non-volatile semiconductor memory devices may also be used as recording media.

[0084] Some or all of the above embodiments may also be described as follows, but are not limited to the following:

[0085] [Note 1] A means of acquiring health data for multiple individuals, A generation means for generating a probability distribution of the occurrence of health data based on health data of each of the aforementioned multiple individuals at multiple points in time, A path estimation means for estimating the paths between data on the health of a subject at different points in time, on the generated probability distribution of occurrence, Output means for outputting information about the estimated route and A path estimation device equipped with the following features.

[0086] [Note 2] The path estimation means estimates a new path by masking at least one grid of points on the occurrence probability distribution through which the estimated path passes. The path estimation device described in Appendix 1.

[0087] [Note 3] The path estimation means estimates the path by masking at least one variable-side grid among the points on the occurrence probability distribution. The path estimation device described in Appendix 2.

[0088] [Note 4] The path estimation means estimates the path to the target point of the health data of the subject person on the probability distribution, based on the probability distribution and the medical expenses estimated at each grid on the probability distribution. A path estimation device as described in any of the appendices 1 to 3.

[0089] [Note 5] The path estimation means estimates multiple paths to the target point of the health data of the subject person on the probability distribution of occurrence. A path estimation device as described in any of the appendices 1 to 4.

[0090] [Note 6] The path estimation means estimates the path to each of the multiple target points of the health data of the subject person on the probability distribution of occurrence. A path estimation device as described in any of the appendices 1 to 5.

[0091] [Note 7] The output means outputs the estimated path superimposed on the probability distribution of occurrence. A path estimation device as described in any of the appendices 1 to 6.

[0092] [Note 8] The system further includes a data estimation means for estimating the time-series data of the health of each of the multiple individuals by using a machine learning model that estimates time-series data of health data at a later point in time than the input data, based on the input health data. The generation means generates a probability distribution of occurrence of health data for the multiple persons based on the time-series data for each of the multiple persons' health data. A path estimation device as described in any of the appendices 1 to 7.

[0093] [Note 9] The route estimation means estimates a route in which the estimated medical cost is lower than that of other routes. The path estimation device described in Appendix 4.

[0094] [Note 10] The output means outputs estimated candidate paths based on the probability of passing through each path. A path estimation device as described in any of the appendices 1 to 9.

[0095] [Note 11] The path estimation means estimates the path by masking at least one variable-side point from the path branching point among the points on the probability distribution of occurrence. The path estimation device described in Appendix 3.

[0096] [Note 12] The output means further superimposes the target path of the target person onto the probability distribution and outputs it. The path estimation device described in Appendix 7.

[0097] [Note 13] We obtained health data from multiple individuals, Based on the health data of each of the aforementioned individuals at multiple points in time, a probability distribution of the occurrence of the health data is generated. The pathways between health data of the subject at different points in time are estimated on the generated probability distribution. Outputs information about the estimated route. Path estimation method.

[0098] [Note 14] The process of obtaining health data from multiple individuals, A process to generate a probability distribution of the occurrence of health data based on health data of each of the aforementioned multiple individuals at multiple points in time, A process for estimating the pathways between data on the health of a subject at different points in time, using the generated probability distribution of occurrence, A process to output information about the estimated route and A path estimation program that causes a computer to perform the following steps.

[0099] Furthermore, some or all of the configurations described in Appendices 2 to 12, which are dependent on Appendice 1 above, may also be dependent on Appendices 13 and 14 in the same way as those described in Appendices 2 to 12. Moreover, not limited to Appendices 1, 13, and 14, some or all of the configurations described as appendices may also be dependent on various hardware, software, various recording means for recording software, or systems, without departing from the embodiments described above.

[0100] Although the present disclosure has been described above with reference to embodiments, the present disclosure is not limited to the embodiments described above. Various modifications to the structure and details of the present disclosure can be made as can be understood by those skilled in the art within the scope of the present disclosure. Furthermore, each embodiment can be combined with other embodiments as appropriate. [Explanation of symbols]

[0101] 10. Path estimation device 11 Acquisition Department 12 Data Estimation Unit 13 Generation part 14. Path estimation unit 15 Output section 16 Memory section 20 Terminal devices 30 Data Management Devices 100 Computers 101 CPU 102 memory 103 Storage device 104 Input / Output Interfaces 105 Communication I / F

Claims

1. A means of acquiring health data for multiple individuals, A generation means for generating a probability distribution of the occurrence of health data based on health data of each of the aforementioned multiple individuals at multiple points in time, A path estimation means for estimating the paths between data on the health of a subject at different points in time, on the generated probability distribution of occurrence, Output means for outputting information about the estimated route and A path estimation device equipped with the following features.

2. The path estimation means estimates a new path by masking at least one grid on the probability distribution through which the estimated path passes. The path estimation device according to claim 1.

3. The path estimation means estimates the path by masking at least one variable-side grid among the points on the occurrence probability distribution. The path estimation device according to claim 2.

4. The path estimation means estimates the path to the target point of the health data of the subject person on the probability distribution, based on the probability distribution and the medical expenses estimated at each grid on the probability distribution. A path estimation device according to any one of claims 1 to 3.

5. The path estimation means estimates multiple paths to the target point of the health data of the subject person on the probability distribution of occurrence. A path estimation device according to any one of claims 1 to 3.

6. The path estimation means estimates the path to each of the multiple target points of the health data of the subject person on the probability distribution of occurrence. A path estimation device according to any one of claims 1 to 3.

7. The output means outputs the estimated path superimposed on the probability distribution of occurrence. A path estimation device according to any one of claims 1 to 3.

8. The system further includes a data estimation means for estimating the time-series data of the health of each of the multiple individuals by using a machine learning model that estimates time-series data of health data at a later point in time than the input data, based on the input health data. The generation means generates a probability distribution of occurrence of health data for the multiple persons based on the time-series data for each of the multiple persons' health data. A path estimation device according to any one of claims 1 to 3.

9. We obtained health data from multiple individuals, Based on the health data of each of the aforementioned individuals at multiple points in time, a probability distribution of the occurrence of the health data is generated. The pathways between health data of the subject at different points in time are estimated on the generated probability distribution. Outputs information about the estimated route. Path estimation method.

10. The process of obtaining health data from multiple individuals, A process to generate a probability distribution of the occurrence of health data based on health data of each of the aforementioned multiple individuals at multiple points in time, A process for estimating the pathways between data on the health of a subject at different points in time, using the generated probability distribution of occurrence, A process to output information about the estimated route and A path estimation program that causes a computer to perform the following steps.

Images