Patent Application
20220044817
The present invention relates to a method and the like of creating a health level positioning map and a health function.
The preliminary stage before reaching onset of a disease from a healthy state is referred to as ahead sick. In view of concerns for national economic collapse due to increased medical expenses in a super aged society with a falling birthrate, development of a technology that enables visualization/quantification of a pre-disease state from a heathy state to prevent onset of a disease in advance is desired.
For example, Patent Literature 1 and Patent Literature 2 disclose a technology for evaluating the health status of a subject by acquiring data for a plurality of test items from a plurality of subjects and creating a function using a plurality of pieces of test data for the test items as variables.
Japanese Laid-Open Publication No. 2010-230428
Japanese Patent No. 6069598
Japanese Patent No. 5455071
Japanese Patent No. 5491749
However, the aforementioned conventional art evaluates risk by taking measurements that follow a path of heathy state, pre-disease state, and disease along a specific axis such as lifestyle diseases including diabetes and arteriosclerosis, cancer, dementia, sarcopenia, liver disease, or renal disease. For this reason, the conventional art has problems such as the inability to evaluate the degree of overall health of an individual.
The objective of one embodiment of the invention is to provide means of evaluating various health risks (e.g., health function creation method, health function creation apparatus, or the like that can create health function) to express the overall health level of an individual.
To solve the problem described above, a health function creation method according to one embodiment of the invention is a health function creation method for creating a health function of a group of subjects, comprising a first data acquisition step for acquiring first data related to health comprising future health diagnosis item data for the group of subjects, and a health function creation step for creating a heath function using the first data.
To solve the problem described above, a health function creation apparatus according to one embodiment of the invention is a health function creation apparatus for creating a health function of a group of subjects, comprising a data acquisition unit for acquiring first data related to health comprising future health diagnosis item data for the group of subjects, and a health function creation unit for creating a heath function using the first data.
The present invention provides, for example, the following items.
1. A method of creating a health level positioning map, comprising:
2. The method of item 1, wherein the first parameter set comprises an autonomic nerve parameter, a biological oxidation parameter, a less biological repair energy parameter, and an inflammation parameter.
3. The method of item 2, wherein the first parameter set further comprises a fundamental parameter, a cognitive function parameter, and a subjective parameter.
4. The method of any one of items 1 to 3, wherein the processing of the first data set comprises dimensionality reduction processing of the first data set.
5. The method of any one of items 1 to 4, wherein the processing of the first data set comprises standardization of the first data set and dimensionality reduction processing of the standardized data set.
6. The method of item 5, wherein the standardization of the first data set comprises:
7. The method of any one of items 4 to 6, wherein the dimensionality reduction processing is performed by multidimensional scaling.
8. The method of any one of items 1 to 7, further comprising:
9. The method of any one of items 1 to 8, wherein the health level positioning map is a two-dimensional or three-dimensional map.
10. A method of creating a health function for mapping a health level of a subject onto a health level positioning map, comprising:
11. The method of item 10, wherein the derivation is performed by machine learning.
12. The method of item 10 or 11, wherein the second parameter set does not comprise a result of an invasive test.
13. The method of item 12, wherein the second parameter set comprises:
14. The method of item 10 or 11, wherein the second parameter set comprises age, subjective evaluation on fatigue, fatigue duration, balance between a sympathetic nerve and a parasympathetic nerve, cognitive function, fat percentage, blood neutral fat, blood oxidative stress index (OSI, and blood CRP.
15. The method of any one of items 10 to 14, wherein the health function is a linear regression model or a nonlinear regression model using the second data set as an independent variable and a position on the health level positioning map of the at least some of the subjects as a dependent variable.
16. A method of estimating a health level of a user, comprising:
17. The method of item 16, comprising:
18. A method of evaluating an item for improving a health status, comprising:
19. A method of creating a health function for mapping a health level of a subject on a health level positioning map, comprising:
20. The method of item 19, wherein the first parameter set comprises an autonomic nerve parameter, a biological oxidation parameter, a less biological repair energy parameter, and an inflammation parameter.
21. The method of item 19 or 20, wherein the derivation is performed by machine learning.
22. The method of any one of items 19 to 21, wherein the second parameter set does not comprise a result of an invasive test.
23. The method of item 22, wherein the second parameter set comprises:
24. The method of any one of items 19 to 21, wherein the second parameter set comprises age, subjective evaluation on fatigue, fatigue duration, balance between a sympathetic nerve and a parasympathetic nerve, cognitive function, fat percentage, blood neutral fat, blood oxidative stress index (OSI, and blood CRP.
25. A method of estimating a health level of a user, comprising:
26. The method of item 25, wherein the second parameter set does not comprise a result of an invasive test.
27. The method of item 26, wherein the second parameter set comprises:
28. The method of item 25, wherein the second parameter set comprises age, subjective evaluation on fatigue, fatigue duration, balance between a sympathetic nerve and a parasympathetic nerve, cognitive function, fat percentage, blood neutral fat, blood oxidative stress index (OSI, and blood CRP.
29. The method of any one of items 25 to 28, wherein the health level positioning map is created using a first parameter set, and the first parameter set comprises an autonomic nerve parameter, a biological oxidation parameter, a less biological repair energy parameter, and an inflammation parameter.
30. The method of item 29, wherein the first parameter set further comprises a fundamental parameter, a cognitive function parameter, and a subjective parameter.
31. A system for creating a health level positioning map, comprising:
32. A system for creating a health function for mapping a health level of a subject on a health level positioning map, comprising:
33. A system for estimating a health level of a user, comprising:
34. A system for creating a health function for mapping a health level of a subject onto a health level positioning map, comprising:
35. A system for estimating a health level of a user, comprising:
36. A program for creating a health level positioning map, the program being executed in a computer system comprising a processing unit, the program causing the processing unit to perform processing comprising:
37. A program for creating a health function for mapping a health level of a subject on a health level positioning map, the program being executed in a computer system comprising a processing unit, the program causing the processing unit to perform processing comprising:
38. A program for estimating a health level of a user, the program being executed in a computer system comprising a processing unit, the program causing the processing unit to perform processing comprising:
39. A program for creating a health function for mapping a health level of a subject onto a health level positioning map, the program being executed in a computer system comprising a processing unit, the program causing the processing unit to perform processing comprising:
40. A program for estimating a health level of a user, the program being executed in a computer system comprising a processing unit, the program causing the processing unit to perform processing comprising:
One embodiment of the invention can provide means capable of evaluating various health risks (e.g., health level positioning map, health function, or the like).
The embodiments of the invention are described hereinafter with reference to the drawings. As used herein, “about” refers to a range of ±10% from the numerical value that is described subsequent to “about.”
The inventors of the present invention developed a new service for making a health status of a user viewable. This is a service, which acquires various pieces of data related to the health of a user, and provides the user with information regarding which region on a health level positioning map the health level of the user is positioned, based on the acquired data. The health level positioning map herein refers to a map with a plurality of regions characterized with information related to health.
Each region on a health level positioning map can be characterized by, for example, various states of health levels. For example, a region on a health level positioning map can be characterized as a young mental health disease risk group, and another region on the health level positioning map can be characterized as a middle age-senior life style disease risk group. A still another region on the health level positioning map can be characterized as a senior diabetes risk group. For example, a region on a health level positioning map can be characterized as a pre-disease state group for a certain health status, and another region can be characterized as a high risk group for the health status. Such characterizations are examples. Various characterizations can be applied. For example, as shown in
At step S1, user U undergoes testing at facility H such as a hospital or a specialized testing facility, and the result of the test is provided to a service provider P. The test at the facility H can include a noninvasive test such as medical consultation or cognitive function test in addition to an invasive test such as a blood test. Alternatively at step S1, the user U undergoes a simple test at facility C such as a company that is not a testing facility, and the result of the test is provided to the service provider P. The simple test at facility C can include, for example, only noninvasive tests such as medical consultation and a cognitive function test. At step S1, it is not necessary to administer both an invasive test and a noninvasive test administered at facility H and a noninvasive test administered at facility C. Only a test at facility H or only a simple test at facility C may be administered. As used herein, “invasive test” refers to a test that damages the body of a test subject (e.g., by blood collection using a syringe or tissue resection), and “noninvasive test” refers to a test involving no damage to the body of a test subject. A typical invasive test is a test for detecting the amount of components contained in blood or plasma. A typical noninvasive test is a test for detecting a component in a discharge (urine, breath, or saliva) of a subject, autonomic nervous function test, cognitive function test, questionnaire/VAS (Visual Analogue Scale), or the like. As used herein, “invasive parameter” refers to a parameter obtained through an invasive test, and “noninvasive parameter” refers to a parameter obtained through a noninvasive test.
The service provider P can generate information regarding which region on a health level positioning map the health level of the user is positioned, based on the provided test result of the user.
At step S2, information generated by the service provider P is provided to the user. The user can be aware of the user's own health status by referring to the information associated with a region on a health level positioning map in which the user's own health level is positioned. For example, if the user's health level belongs to a region characterized as health level B on a health level positioning map as shown in
The user can further be aware of a chronological change in the user's own health status by repeating the aforementioned step S1 to step S2 after a predetermined period has elapsed. For example, the user can be aware of the direction toward which the user's own health status is headed by referring to chronological changes in the position on a health level positioning map. If, for example, the user's health level belonged to a region characterized as health level B on a health level positioning map as shown in
The screen 1000 comprises a health level positioning map display section 1100 and a radar chart display section 1200.
The health level positioning map display section 1100 displays a health level positioning map. On the health level positioning map, the horizontal axis is associated with physical health, where a greater value on the horizontal axis indicates worse physical health, while the vertical axis is associated with mental health, where a greater value on the vertical axis indicates worse mental health.
A health level positioning map displayed on the health level positioning map display section 1100 includes 10 regions. Among the 10 regions, 1 region is characterized as health level A, 3 regions are characterized as health level B, and 6 regions are characterized as health level C. In this regard, health level A indicates a good health level, health level B indicates that the health level is normal, and health level C indicates a poor health level requiring caution.
The user can be aware of the user's own health status in accordance with which region on a health level positioning map the user's health level is positioned. In the example shown in
The radar chart display section 1200 displays a radar chart. The radar chart shows the health status in 6 levels from 0 to 5 from 6 viewpoints (musculoskeletal motor system, metabolic system, autonomic nervous system, sleep wake rhythm, mental health, and fatigue). The viewpoint of musculoskeletal motor system indicates the status of muscle or the like associated with the motor function. The viewpoint of metabolic system indicates the status of energy metabolism in the body, obesity, or the like. The viewpoint of autonomic nervous system indicates the status of the ability to regulate the nerves associated with concentration or relaxation. The viewpoint of sleep wake rhythm indicates the status of sleep, drowsiness, or the like. The viewpoint of mental health indicates the status such as depressed mood. The viewpoint of fatigue indicates the status of mental or physical exhaustion. The user can be aware of which viewpoint the user's health status is attributed to at a glance. Each axis of the radar chart displayed on the radar chart display section 1200 is characterized by information related to health, and the score of the user is mapped to each axis. Therefore, such a radar chart can also be considered as a type of a health level positioning map herein.
The aforementioned service can be materialized, for example, by the computer system 100 described below.
The computer system 100 is connected to a database unit 200. The computer system 100 is also connected to at least one user terminal apparatus 300 via a network 400.
The network 400 can be any type of network. The network 400 can be, for example, the Internet or LAN. The network 400 can be a wired network or a wireless network.
Examples of the computer system 100 include, but are not limited to, a computer (e.g., server apparatus) installed at a service provider providing a new service for making a health status of a user viewable. Examples of the user terminal apparatus 300 include, but are not limited to, a computer installed at a hospital (e.g., terminal apparatus), a computer installed in one room of an office that can administer a test (e.g., terminal apparatus), and a computer held by a user (e.g., terminal apparatus). In this regard, the computer (server apparatus or terminal apparatus) can be any type of computer. A terminal apparatus can be, for example, any type of terminal apparatus such as a smart phone, tablet, personal computer, or a smart glass.
The computer system 100 comprises an interface unit 110, a processing unit 120, and a memory unit 150.
The interface unit 110 exchanges information with the outside of the computer system 100. The processing unit 120 of the computer system 100 can receive information from the outside of the computer system 100 via the interface unit 110 and transmit information to the outside of the computer system 100. The interface unit 110 can exchange information in any manner.
The interface 110 comprises, for example, an input unit that enables input of information into the computer system 100. The mode through which an input unit enables input of information into the computer system 100 is not limited. If, for example, the input unit is a touch panel, a user can input information by touching the touch panel. Alternatively, if the input unit is a mouse, a user can input information by operating the mouse. Alternatively, if the input unit is a keyboard, a user can input information by pressing a key on the keyboard. Alternatively, if the input unit is a microphone, a user can input information by inputting an audio into the microphone. Alternatively, if the input unit is a camera, the input unit can input information captured by the camera. Alternatively, if the input unit is a data reader, information can be inputted by reading out information from a storage medium connected to the computer system 100. Alternatively, if the input unit is a receiver, information can be inputted by the receiver receiving the information from the outside of the computer system 100 via the network 400.
The interface unit 110 comprises, for example, an output unit that enables output of information from the computer system 100. The mode through which an output unit enables output of information from the computer system 100 is not limited. If, for example, the output unit is a display screen, information can be outputted onto the display screen. Alternatively, if the output unit is a speaker, information can be outputted by an audio from the speaker. Alternatively, if the output unit is a data writing apparatus, information can be outputted by writing in information into a storage medium connected to the computer system 100. Alternatively, if the output unit is a transmitter, information can be outputted by the transmitter transmitting the information to the outside of the computer system 100 via the network 400. In such a case, the type of network is not limited. For example, the transmitter can transmit information via the Internet or LAN.
The processing unit 120 executes processing of the computer system 100 and controls the overall operation of the computer system 100. The processing unit 120 reads out a program stored in the memory unit 150 and executes the program. This can cause the computer system 100 to function as a system that executes a desired step. The processing unit 120 can be implemented by a single processor or a plurality of processors.
The memory unit 150 stores a program required for executing the processing of the computer system 100, data required for executing the program, and the like. The memory unit 150 can store a program for causing the processing unit 120 to perform processing for creating a health level positioning map (e.g., program materializing the processing shown in
For example, data obtained from a plurality of subjects can be stored in the database unit 200. For example, data for a health level positioning map generated by the computer system 100 can be stored in the database unit 200. For example, a health function generated by the computer system 100 can be stored in the database unit 200.
The processing unit 120 comprises acquisition means 121, processing means 122, mapping means 123, clustering means 124, and characterization means 125.
The acquisition means 121 is configured to acquire a first data set with respect to a first parameter set described below for each of a plurality of subjects. For example, the acquisition means 121 acquire a dataset with respect to a plurality of items (e.g., 232 items in a certain embodiment) for each subject. The first parameter set can be a parameter set obtained by, for example, acquiring a data set with respect to an initial parameter set, finding a correlation between each data of the data set with respect to the initial parameter set, and extracting a parameter with a correlation coefficient that is equal to or greater than a predetermined threshold value. At this time, an extracted parameter set can be extracted to include the four basic parameters described below.
The acquisition means 121 can, for example, receive data for a plurality of subjects stored in the database unit 200 via the interface unit 110 to acquire the received data. The acquisition means 121 can, for example, receive data for a plurality of subjects stored in the database unit 200 from a computer system of a test facility (e.g., hospital, research laboratory, or the like) via the interface unit 110 to acquire the received data. The acquired first data set is passed along to the processing means 122 for subsequent processing.
A first data set with respect to a first parameter set can be stored in the database unit 200.
A first data set with respect to a first parameter set for each of a plurality of subjects is stored in the database unit 200. An ID is assigned to each of the plurality of subjects. For example, a set (first data set) of values of each parameter of the first parameter set such as age, muscle mass, BMI, fat percentage, speed of sound, osteoporosis index . . . is stored in the database unit 200.
Referring back to
The processing means 122 is configured to process a first data set acquired by the acquisition means 121. The processing means 122 can output first data by processing a first data set.
Processing by the processing means 122 can include any processing, as long as the outputted first data can be mapped. When creating a positioning map based on data for both males and females, it is preferable to apply a correction for the difference between sexes.
Processing by the processing means 122 can include, for example, dimensionality reduction processing. Dimensionality reduction processing is processing that converts an m-dimensional data into n-dimensional data, wherein m>n. Dimensionality reduction processing can be performed using, for example, multi-dimensional scaling (MDS), principal component analysis, multiple regression analysis, principle component analysis, machine learning, or the like, but the dimensionality reduction processing means is not limited thereto. Dimensionality reduction processing preferably reduces a first data set to two dimensional data or three dimensional data. This is because when two dimensional data or three dimensional data is mapped by the mapping means 123 described below, a map is created in a two dimensional space or a three dimensional space, so that a visually readily understandable map can be obtained. Dimensionality reduction processing can be performed using multidimensional scaling. This is because mapping first data obtained by multidimensional scaling by the mapping means 123 described below can result in a visually readily understandable map.
Processing by the processing means 122 can include, for example, standardization processing. Standardization processing is processing that aligns the scale of data for each parameter of a first data set. Standardization processing can be, for example, processing that computes a Z score (processing that corrects data so that the mean value is 0 and the standard deviation is 1), processing that computes a T score (processing that corrects data so that the mean value is 50 and the standard deviation is 10), or the like. The processing means 122 can be configured to perform standardization processing on data of a first data set with respect to all parameters in a first parameter set, or with respect to a specific parameter.
The processing means 122 can be configured to perform standardization processing on a first data set for all of the plurality of subjects, or for a specific population among the plurality of subjects. Examples of the specific population among the plurality of subjects include, but are not limited to, a male subject population, a female subject population, a young population (population of subjects under 40 years old), a middle age population (population of subjects who are 40 or older and younger than 60 years old), a senior population (population of subjects who are 60 or older), and the like. The processing means 122 can form any population from a plurality of subjects and perform standardization processing on the first data set of the population.
For example, the processing means 122 can classify a first data set from a plurality of subjects into a data set for a male subject and a data set for a female subject and standardize the data set for a male subject to perform standardization processing on a male subject population, or standardize the data set for a female subject to perform standardization processing on a female subject population, or both. Standardization processing on a male subject population and/or female subject population in this manner is preferable for a parameter in a first parameter set with a difference between males and females (e.g., blood neural fat concentration or the like), and is more preferable for a parameter in a first parameter set with a significant difference between males and females (e.g., blood red blood cell count or the like). This is because this eliminates the difference between males and females and enables the creation of a health level positioning map without a bias due to a difference between males and females.
Processing by the processing means 122 can include, for example, weighting processing. Weighting processing is processing for weighting at least some data in a first data set. For example, the processing can be configured to add weighting by adding a predetermine number to at least some data in a first data set, or by multiplying a predetermined number to at least some data in a first data set. The predetermined number that is added or multiplied can be constant or different for each data subjected to weighting processing. For example, a predetermined number can be varied so that larger or smaller weighting is added to data with a greater effect on a health function derived by the derivation means 133 described below, or alternatively, a predetermined number can be varied so that larger or smaller weighting is added to data with a lesser effect on a health function derived by the derivation means 133 described below.
The processing means 122 can be configured to perform weighting processing on a first data set for all of the plurality of subjects, or to perform weighting processing on a first data set for a specific population among the plurality of subjects. The processing means 122 can form any population from a plurality of subjects and perform weighting processing on a first data set of the population. A population subjected to weighting processing can be the same as or different from the aforementioned population subjected to standardization processing.
The mapping means 123 is configured to map an output of the processing means 122, i.e., first data, for each of the plurality of subjects. Mapping by the mapping means 123 is processing that associates n-dimensional first data with a position on an n-dimensional space. The mapping means 123 can output a map mapping first data of each of the plurality of subjects by mapping the first data. If, for example, the first data is two dimensional, the mapping means 123 can output a two dimensional map by mapping first data so that the first data is associated with a position on a two dimensional space, i.e., plane.
The clustering means 124 is configured to cluster first data mapped by the mapping means 123. Clustering by the clustering means 124 is processing for dividing mapped first data into a plurality of clusters and identifying each region to which the plurality of clusters belong. As used herein, “region” refers to a certain range within an n-dimensional space, having an n-dimensional range. The clustering means 124 can divide mapped first data into any number of clusters. For example, the clustering means 124 preferably divides mapped first data into at least three clusters. This is because a health level positioning map that is intuitively understandable by a user can be created by using three clusters (e.g., clusters such as good health level, normal health level, and poor health level as shown in
In one embodiment, the clustering means 124 can be configured to further cluster sub-first data acquired by the acquisition means 121. The clustering means 124 can divide sub-first data into a plurality of clusters and identify each region to which the plurality of clusters belong. The clustering means 124 can divide sub-first data into any number of clusters.
The characterization means 125 is configured to characterize at least some of a plurality of regions identified by the clustering means 124. The characterization means 125 can be configured, for example, to characterize at least some of a plurality of regions based on information inputted into the computer system 100 via the interface unit 110 by a health level positioning map creator. For example, a health level positioning map creator can analyze a characteristic of a subject corresponding to first data included in each of a plurality of regions and input information with which the regions should be characterized based on the result of analysis. Alternatively, the characterization means 125 can be configured to characterize at least some of a plurality of regions without depending on input by a health level positioning map creator. For example, the characterization means 125 can characterize at least some of a plurality of regions based on a relative position on a health level positioning map or based on machine learning.
In this manner, a health level positioning map with at least some of the plurality of regions characterized is created. In one embodiment, a health level positioning map from characterizing some of a plurality of regions identified by clustering sub-first data would be a health level positioning map for some of the plurality of subjects.
A health level positioning map created by the processing unit 120 is outputted, for example, to the outside of the computer system 100 via the interface unit 110. A health level positioning map can be transmitted, for example, to the database unit 200 via the interface 110 and stored in the database 200. Alternatively, the map can be transmitted to the processing unit 130 described below for creating a health function. As described below, the processing unit 130 can be a constituent element of the same computer system 100 as the processing unit 120, or a constituent element of another computer system.
The processing unit 130 comprises first acquisition means 131, second acquisition means 132, and derivation means 133.
The first acquisition means 131 is configured to acquire a health level positioning map. The acquired health level positioning map can be a health level positioning map created by the aforementioned processing unit 120 or a health level positioning map created in another manner, as long as the map is created using a first parameter set. The acquired health level positioning map is passed along to the derivation means 133 for subsequent processing.
The second acquisition means 132 is configured to acquire a second data set with respect to a second parameter set described below for at least some of the plurality of subjects. A second parameter set is a part of the first parameter set. For example, the second acquisition means 132 can acquire data for some of a plurality of subjects stored in the database unit 200 via the interface unit 110. The acquired second data set is passed along to the derivation means 133 for subsequent processing.
The derivation means 133 is configured to derive a health function that correlates a second data set acquired by the second acquisition means 132 with a position on a health level positioning map acquired by the first acquisition means 131. The derivation means 133 can derive a health function by, for example, machine learning, decision tree analysis, random forest regression, multiple regression analysis, principle component analysis, or the like. A health function can be derived, for example, for each axis of an n-dimensional health level positioning map. If, for example, a health level positioning map is two dimensional, a health function X for correlating a second data set with an X coordinate on the health level positioning map and a health function Y for correlating the second data set with a Y coordinate on the health level positioning map can be derived. The derivation means 133 can, for example, increase/decrease the number of variables of a health function to any number and create a plurality of health functions having the same degree of accuracy, i.e., health function group (hereinafter, also referred to as a multiple pattern heath function group). For example, the derivation means 133 can create (1) a health function using data for blood test items and data for other items as variables and (2) a health function using only data for blood test items as a variable, which has the same degree of accuracy to each other. The derivation means 133 can also create a health function group using data selected from a data group for items other than data for blood test items as a variable, as a multiple pattern health function group.
A health function can be, for example, a regression model. A regression model can be a linear regression model or a nonlinear regression model. The derivation means 133 can derive each coefficient of a regression model by machine learning using a second data set as an independent variable and a coordinate on a health level positioning map of a subject as a dependent variable for each of at least some of the plurality of subjects. When a second data set obtained from a subject is inputted into such a machine learned regression model dependent variable, a coordinate on the health level positioning map of the subject is outputted. The health level of the subject can be mapped onto the health level positioning map by using the outputted coordinate.
A health function can be, for example, a neural network model. A neural network model has an input layer, at least one hidden layer, and an output layer. The number of nodes of an input layer of a neural network model corresponds to the number of dimensions of inputted data. Specifically, the number of nodes of an input layer corresponds to the number of parameters in a second parameter set. A hidden layer of a neural network model can comprise any number of nodes. The number of nodes of an output layer of a neural network model corresponds to the number of dimensions of outputted data. Specifically, when an X coordinate on a health level positioning map is outputted from a neural network model, the number of nodes of an output layer would be 1. If, for example, n coordinates on an n-dimensional health level positioning map are outputted from a neural network model, the number of nodes of an output layer would be n. The derivation means 133 can derive a weighting coefficient of each node by machine learning using a second data set as an input supervisor data and a position on the health level positioning map of a subject as output supervisor data for each of at least some of the plurality of subjects.
For example, a set of (input supervisor data, output supervisor data) for machine learning can be (second data set with respect to a second parameter set for a first subject, coordinate on a health level positioning map of a first subject), (second data set with respect to a second parameter set for a second subject, coordinate on a health level positioning map of a second subject) . . . (second data set with respect to a second parameter set for an ith subject, coordinate on a health level positioning map of an ith subject) . . . or the like. If a second data set obtained from a subject is inputted into an input layer of such a machine learned neural network model, a coordinate on a health level positioning map of the subject is outputted to an output layer. A health level of the subject can be mapped onto a health level positioning map.
A health function created by the processing unit 130 is outputted, for example, to the outside of the computer system 100 via the interface unit 110. A health function can be transmitted, for example, to the database 200 via the interface 110 and stored in the database 200. Alternatively, the function can be transmitted to the processing unit 140 described below for processing to estimate a health level of a user. As described below, the processing unit 140 can be a constituent element of the same computer system 100 as the processing unit 130 or a constituent element of another computer system.
The processing unit 140 comprises third acquisition means 141, fourth acquisition means 142, output generation means 143, and output mapping means 144.
The third acquisition means 141 is configured to acquire a health function. A health function is a function that correlates a data set with respect to a second parameter set described above with a position on a health level positioning map. The acquired health function can be a health function created by the aforementioned processing unit 130 or a health function created in another manner, as long as the function can correlate a user data set with a position on a health level positioning map. The health level positioning map can be a health level positioning map created by the aforementioned processing unit 120 or a health level positioning map created in another manner, as long as the map is created using a first parameter set. The acquired health function is passed along to the output generation means 143 for subsequent processing.
The fourth acquisition means 142 is configured to acquire a user data set with respect to a second parameter set of a user. The fourth acquisition means 142 can acquire, for example, a user data set stored in the database unit 200 via the interface unit 110. Alternatively, the fourth acquisition means 142 can acquire, for example, a user data set via the interface unit 110 from a user terminal apparatus. The acquired user data set is passed along to the output generation means 143 for subsequent processing.
The output generation means 143 is configured to generate an output from a health function. The output generation means 143 generates an output from a health function by inputting a user data set acquired by the fourth acquisition means 142 into a health function acquired by the third acquisition means 141.
If, for example, a health function is a regression model as described above, a coordinate on a health level positioning map is outputted by inputting a user data set into an independent variable of the regression model.
If, for example, a health function is a neural network model as described above, a coordinate on a health level positioning map is outputted by inputting a user data set into an input layer of the neural network model.
The output mapping means 144 is configured to map an output generated by the output generation means 143 onto a health level positioning map. Since an output generated by the output generation means 143 is a coordinate, the output mapping means 144 can map the coordinate within an n-dimensional space of a health level positioning map.
An output mapped onto a health level positioning map by the processing unit 140 is outputted, for example, to the outside of the computer system 100 via the interface unit 110. An output can be transmitted, for example, to a terminal apparatus of a user via the interface unit 110.
Each of the aforementioned constituent elements of the computer system 100 can be comprised of a single hardware part or a plurality of hardware parts. If comprised of a plurality of hardware parts, the mode of connecting each hardware part is not limited. Each hardware part can be connected wirelessly or with a wire. The computer system 100 of the invention is not limited to a specific hardware configuration. The processing units 120, 130, and 140 comprised of analog circuits instead of digital circuits are also within the scope of the invention. The configuration of the computer system 100 of the invention is not limited to those described above, as long as the function thereof can be materialized.
A data set acquired at the data acquisition step 10 is send to the data processing step 20.
The data processing step 20 performs, for example, correction 21 and dimensionality reduction 22 as shown in
The correction 21 is correction of data acquired at the data acquisition step 10. Specifically, the correction 21 corrects each data acquired at the data acquisition step 10 so that a value of the data acquired at the data acquisition step 10 is, for example, a value within a predetermined range (e.g., mean value is 0 and standard deviation is 1). Corrected data is send to the dimensionality reduction 22.
The dimensionality reduction 22 reduces the dimension of a plurality of pieces of data passed along from the data acquisition step 10 or the correction step 21. Specifically, the dimensionality reduction 22 reduces the dimension of a plurality of pieces of data (multidimensional data) sent from the data acquisition step 10 or the correction step 21 to any dimension (two dimension in this embodiment) using multiple regression analysis, multidimensional scaling, principle component analysis, or machine learning. For example, data with reduced dimensionality can be in a form of a function. For example, this function is a function for computing an indicator associated with health. A function is, for example, a function using some or all data contained in first data as a variable, and is a function created by placing a greater weighting on data with a particularly significant effect among various disease factors. A function in this embodiment is created using some or all of data contained in first data using a linear or nonlinear model. In one embodiment, two dimensional function (in this embodiment, comprised of horizontal axis function X (hereinafter, referred to as a first function) and vertical axis function Y (hereinafter, referred to as a second function)) is created. The created function is passed along to the health evaluation map creation step 30 and the output step 70.
In this embodiment, the dimensionality reduction 22 creates a two dimensional function by reducing multidimensional data to two dimensions as described above. The first function and second function are functions using all or some of a plurality of pieces of data passed along from the data acquisition step 10 or the correction step 21 as a variable, and are functions computed by multiple regression analysis, multidimensional scaling, principle component analysis, or machine learning. In the present invention, “machine learning” refers to either machine learning with deep learning or machine learning without deep learning. The first function and the second function can have completely identical or completely different constituent variables, or some variables can overlap with each other.
The health evaluation map creation step 30 can be implemented by, for example, the mapping means 123 of the aforementioned processing unit 120. At the health evaluation map creation step 30, a map for evaluation health (hereinafter, referred to as a health evaluation map) is created using, for example, data processed by the data processing step 20, more specifically a function created by dimensionality reduction. In this embodiment, a function is two-dimensional, so that a health evaluation map is a two-dimensional map.
The clustering map creation step 40 can be implemented by, for example, the clustering means 124 and the characterization means 125 of the aforementioned processing unit 120. For example, a map from clustering a plurality of points plotted on a health evaluation map created at the health evaluation map creation step 30 into several clusters and characterizing a plurality of regions (hereinafter, referred to as a health level positioning map) is created at the clustering map creation step 40. At the clustering map creation step 40 in this embodiment, a plurality of plotted points are clustered into any number of clusters using a nonhierarchical clustering method (k-means in this embodiment). In the example shown in
A created health level positioning map can be outputted at the output step 70, passed along to the health function value computation step 50, or passed along to the health prediction positioning map creation step 60.
The health function value computation step 50 and the health prediction positioning map creation step 60 can be implemented by, for example, the aforementioned processing unit 130 and processing unit 140. At the health function value computation step 50, a health function is first created based on a health level positioning map. The number of variables of a health function can be optionally increased/decreased to create a plurality of health functions with the same degree of accuracy, i.e., a health function group. A value of a health function of a subject who is different from the subject from whom a first data set was acquired to create a health function is then computed. Specifically, a value of a health function of a subject is computed by applying the created health function on data of the subject acquired at the data acquisition step 10 (newly acquired data). Since a health function is a multidimensional function, a health function value is naturally a multidimensional value.
The health prediction positioning map creation step 60 creates a health prediction positioning map by plotting a value of a health function of a subject computed by the health function value computation step 50 on a health evaluation map created by the health evaluation map creation step 30 or a health level positioning map created by the clustering map creation step 40. This enables prediction of a health status of a subject from whom data was newly measured.
The health prediction positioning map creation step 60 can also create a health prediction positioning map by plotting a value of a health function computed at the health function value computation step 50 using data acquired with a time interval on the health evaluation map or the health level positioning map described above for a single subject. This enables evaluation of a degree in change of the health status of the subject.
Newly acquired data from a subject can also be used as data for updating a health function. At the health function value computation step 50, a health function can be updated by further using newly acquired data from a subject or data outputted from the correction step 21 using said data. This enables evaluation of a greater variety of health risks and increase in the accuracy of health risk evaluation.
Each step of data flow 1 (especially the data processing step 20, health evaluation map creation step 30, clustering map creation step 40, health function value computation step 50, and health prediction positioning map creation step 60) can be materialized with a logic circuit (hardware) formed into an integrated circuit (IC chip) or the like, or with a software. For the latter, the heath evaluation apparatus 1 comprises a computer for executing an instruction of a program, which is software materializing each function. Such a computer comprises, for example, one or more processors, and a computer readable recording medium for storing the program described above. The objective of the invention is accomplished by the processor reading out the program from the recording medium and executing the program in the computer. For example, a CPU (Central Processing Unit) can be used as the processor. As the recording medium, a “non-transient tangible medium” such as ROM (Read Only Memory) and the like, as well as tape, disk, card, semiconductor memory, programmable logic circuit, and the like can be used. This can also further comprise a RAM (Random Access Memory) or the like for deploying the program. The program can also be supplied to the computer via any transmission medium (communication network, broadcast wave, or the like) that is capable of transmitting the program. In one embodiment of the invention, the program can be materialized in a form of a data signal embedded into a carrier wave, actualized by electronic transmission.
According to the data flow described above, a health level positioning map and/or health function can be created using various data. Specifically, a health level positioning map and/or health function that can comprehensively compute the health level of a subject, instead of an existing health indicator associated with an individual disease, can be created. Therefore, a health level positioning map and/or health function that is capable of evaluating various health risks (i.e., capable of evaluating the overall health level of a subject) can be created.
A health level positioning map and/or health function can also be flexibly made by combining various data measured at the data acquisition step 10. Specifically, the versatility in the selection of items measured at the data acquisition step 10 is very high.
In one embodiment, a health function is created by machine learning using a second data set as input data. This enables the creation of a health function which is capable of making an indicator of a health status of a subject more accurately.
In one embodiment, data is corrected between sexes before creating a health level positioning map with a first data set. This enables the creation of a health level positioning map that is compatible for either sex.
In one embodiment of the invention, first data can be in a form comprising only data acquired by noninvasive measurement. Such a configuration can acquire data without injuring a subject through a blood test or the like.
In one embodiment, a map for evaluating health can be created by using a function created at the data processing step 20. Specifically, if a function is a multidimensional vector, a map for evaluating health is created by plotting points determined by a function into a multidimensional space for a plurality of subjects. This enables visual inspection of the health status of subjects.
In one embodiment, points plotted in a multidimensional space are clustered. By referring to clustered data on subjects belonging to each cluster, the type of subject population can be identified for each cluster to characterize the clusters. As a result, the health status of a measured subject can be predicted by plotting newly measured data of the subject on a health level positioning map.
In the example shown in
The inventors focused on physiological or biochemical mechanisms shared by deterioration in health and chronic fatigue in order to evaluate the overall health level of a subject. Although not wishing to be bound by any theory, there can be common mechanisms among fatigue to chronic fatigue, deterioration in health, senescence, and disease development based on medical research on fatigue and follow up study on deterioration in health. The common mechanisms are:
Comprehensive evaluation of the parameters (1) to (4) enables the determination of a state of deteriorated health (i.e., “ahead sick”) that has not developed into a disease.
In a representative embodiment, a first parameter set for acquiring a first data set in the present invention can comprise a “biological oxidation parameter,” “less repair energy parameter,” “inflammation parameter,” and “autonomic nerve function parameter,” based on the theory described above. As used herein, the four parameters, i.e., (1) “biological oxidation parameter,” (2) “less repair energy parameter,” (3) “inflammation parameter,” and (4) “autonomic nerve function parameter,” are also collectively referred to as the four basic parameters.
Reactive Oxygen Species (ROS) denature many biopolymers constituting a cell such as DNAs, lipids, proteins, and enzymes in the body by oxidation, thus damaging cellular functions. Denaturation by oxidation due to reactive oxygen species is understood to lead to various diseases and senescence.
Meanwhile, antioxidant enzymes such as superoxide dismutase (SOD) and catalase and antioxidants such as CoQ10, vitamin C, and vitamin E are present in the body in order to prevent cellular dysfunction due to reactive oxygen species.
Therefore, measurement of “biological oxidation parameter” in the present invention can comprise measurement of oxidative damage due to reactive oxygen species, measurement of antioxidant capability, or measurement of the balance between oxidative damage and antioxidant capability. Oxidative damage due to reactive oxygen species can be measured by directly measuring the amount of reactive oxygen species, or by measuring oxidative damage on proteins, lipids, or nucleic acids.
A method of measuring oxidative damage is well known in the art. Those skilled in the art can appropriately select and measure a subject of measurement. In the present invention, examples of specific markers used as an indicator for oxidative damage due to reactive oxygen species include, but are not limited, d-ROMs (Derivatives of Reaction Oxygen Metabolites) that directly measure the amount of reactive oxygen species in blood, carbonylated protein content (PCC; Protein Carbonyl Content), which is an indicator of oxidative damage on proteins, 4-hydroxynonenal and isoprostane, which are indicators of oxidative damage on lipids, 8-OHdG (8-hydroxy-dioxyguanosine), which is an indicator of oxidative damage on nucleic acids, and the like.
A method of measuring an antioxidant capability is well known in the art. Those skilled in the art can appropriately select and measure a subject of measurement. In the present invention, examples of specific markers used as an indicator for antioxidant capability include, but are not limited to, BAP (Biological Antioxidant Potential), which quantifies the ability to reduce to iron, serum thiol status, glutathione measurement, vitamin C content measurement, total CoQ10 content, ratio of reduced form of CoQ10, and the like. The total CoQ10 content and ratio of reduced form of CoQ10 can be measured by, for example, LC-MS/MS (specific example can include computing from the concentration of reduced or oxidized CoQ10 detected by multiple reaction monitoring).
In the present invention, examples of a marker used as an indicator for the balance between oxidative damage and antioxidant capability include, but are not limited to, OSI (Oxidation Stress Index). The OSI in the present invention is d-ROMs/BAP.
Preferred biological oxidation parameters in the invention include BAP, total CoQ10 content, ratio of reduced form of CoQ10, and OSI.
Typically, a biological oxidation parameter is a parameter that can be measured by a noninvasive test.
Even if a biological tissue is damaged by oxidation, the body is equipped with a mechanism for repairing the damaged tissue. ATP is required for tissue damage in the body. However, reduced ATP production would delay the repair of the damaged tissue. Reduced ATP production also leads to delayed recovery from fatigue. The “less repair energy parameter” in the present invention is any parameter indicating that ATP production is reduced.
ATP is produced via glycolysis, TCA cycle, and electron transport chain, but the largest amount of ATP is produced at the last electron transport chain. CoQ10 is a coenzyme with a critical role in the electron transport chain. Thus, examples of the “less repair energy parameter” in the invention include, but are not limited to, total CoQ10 content, ratio of reduced form of CoQ10, and metabolites of glycolysis or TCA cycle (e.g., pyruvic acid, lactic acid, citric acid, isocitric acid, succinic acid, fumaric acid, malic acid, and the like). It should be noted that the total CoQ10 content and ratio of reduced form of CoQ10 are a “biological oxidation parameter” due to having an antioxidant capability as well as a “less repair energy parameter” due to contributing to ATP production. The preferred “less repair energy parameter” in the invention can include the total CoQ10 content and ratio of reduced form of CoQ10.
The method of measuring the total CoQ10 content or ratio of reduced form of CoQ10 is described above. A metabolite of glycolysis or TCA cycle can be measured by extracting a compound reflecting glycolysis or TCA cycle from metabolone analysis.
Typically, a less repair energy parameter is a parameter that can be measured by a noninvasive test.
Many tissue damages due to oxidation in the body result in many instances of local inflammation due to immune responses. The type of inflammation parameter and measurement method thereof are known in the art. Those skilled in the art can suitably select and measure an inflammation parameter.
Examples of the inflammation parameter in the invention include, but are not limited to, CRP (C-Reactive Protein), WBC (white blood cell count), albumin, red blood cell count, interleukin-1β, interleukin-6, and the like.
Typically, an inflammation parameter is a parameter that can be measured by a noninvasive test.
If deterioration in health is closely inspected in chronological order, the autonomic nerve function (especially the parasympathetic nerve function) initially decreases, then the quality of sleep decreases, then fatigue accumulates, and loss of motivation, depressive tendency, immune system malfunction such as allergies, endocrine system abnormality such as irregular menstruation, digestive system abnormality, or the like is observed. Therefore, an abnormality in the autonomic nerve function is an important parameter for the understanding of the initial stage of deterioration in health.
In the present invention, an autonomic nerve function is evaluated from a heartbeat parameter. Specific examples of the heartbeat parameter used in the invention include but are not limited to the following.
*Mean HR
*Activity of the entire autonomic nervous system; TP (Total Power=ms2)
*Overall activity of the sympathetic nervous function; VLF (very low frequency ms2)
*Activity of sympathetic nerve; LF (lower frequency ms2)
*Activity of parasympathetic nerve; HF (high frequency ms2)
*Balance between a sympathetic nerve and a parasympathetic nerve; LF/HF ratio
The heartbeat parameter described above representing the autonomic nerve function can be measured by a method known in the art, but can also be measured by an accurate methodology for simultaneously measuring electrocardiographic waves and plethysmographic waves to analyze variation in heartbeats (see Japanese Patent No. 5455071 and Japanese Patent No. 5491749, which are incorporated herein by reference). For example, the autonomic nerve function of the invention can be measured by using a simplified autonomic nervous system measuring apparatus FMCC-VSM301 (Fatigue Science Laboratory Inc., Osaka, Japan) that can simultaneously measure electrocardiographic waves and plethysmographic waves.
The preferred “autonomic nerve function parameter” in the invention can be, but is not limited to, mean HR, TP, LF, HF, LF/HF, or the like. Since the values of HF and LF have large dispersion that is not distributed under a normal distribution, it is preferable to evaluate TP, LF, HF, and LF/HF associated with HF or LF among the parameters described above by applying logarithmic conversion. Therefore, the more preferred “autonomic nerve function parameter” in the invention includes means HR, ln(TP), ln(LF), ln(HF), and ln(LF/HF).
Typically, an autonomic nerve function parameter is a parameter that can be measured by a noninvasive test.
The first parameter set of the invention can comprise one or more of the following parameters in order to create a positioning map that more suitably evaluates the overall health level of a subject.
Fundamental parameters include known parameters representing the physical condition or health status of a subject. The fundamental parameters of the invention include, but are not limited to, age, height, boy weight, waist circumference, body composition, bone density, blood pressure, muscle strength, and the like.
Body composition includes, but is not limited to, muscle mass, BMI (body weight/height=Body Mass Index), fat percentage, and the like.
A method of measuring bone density is known, such as the MD method for capturing an image of a boned of a hand and measuring the bond density from the picture, ultrasound method for measuring the heel bone using ultrasound waves, QCT using CT scan, and DEXA (Dual energy X-ray absorptiometry) using X-rays and a computer. A parameter obtained from any of the measuring methods can be used in the present invention.
Examples of the bone density parameters in the invention include, but are not limited to, Speed of Sound (SOS), Osteoporosis Index (e.g., OSIRIS (Osteoporosis Index of Risk)), young adult comparative % (YAM=Young Adult Mean; % of BMD (Bone Mineral Density; bone density=bone mass/area (unit: g/cm2) value of subject in comparison to BMD value of young individuals when BMD value of young individuals (reference value) is assumed to be 100%), T score (value from defining an indicator with mean BMD value (reference value) of young individuals as 0 and standard deviation as 1 SD), and the like. Methods of measuring these fundamental parameters are well known in the art.
Blood pressure is conventionally measured in the art. Systolic blood pressure can be used as the fundamental parameter of the invention.
Muscle strength is conventionally measured in the art. Muscle mass and mean grip strength of left and right hands can be used as the fundamental parameter of the invention.
The fundamental parameters of the invention can preferably comprise age, muscle mass, BMI, fat percentage, SOS, OI, systolic blood pressure, and mean grip strength of left and right hands.
A fundamental parameter is a parameter that can be measured by a noninvasive test.
It is preferable that a commonly used hematological parameter is further included in a first parameter set in order to evaluate renal excretion/liver, gallbladder, and pancreas/detoxification function system of a subject in addition to the biological oxidation parameters described above.
Examples of such hematological parameters include, but are not limited to, HbA1c (hemoglobin A1c; glycated protein from glucose binding to hemoglobin), ALP (alkaline phosphatase), ALT (alanine aminotransferase), AST (aspartate aminotransferase), BS (blood sugar level), BUN (blood urea nitrogen), CK (creatine kinase; plasma muscle cell enzyme that can be used in the evaluation of motor/skeletal/muscle function system), G-GT (gamma-glutamyl transpeptidase), HDL-C (HDL-cholesterol), HGB (hemoglobin), LD (lactate dehydrogenase), LDL-C (LDL cholesterol), TG (triglyceride; neutral fat), T-P (total protein), UA (uric acid), amylase, albumin, potassium, creatinine, chlorine, cortisol, sodium, eGFR, vitamin (e.g., vitamin B1), mineral (iron, copper, calcium, etc.) content, and the like.
Typically, a hematological parameter is a parameter that can be measured by an invasive test.
According to the studies of the inventors, accumulation of fatigue leads to decreased cognitive function. Therefore, the first parameter of the invention can further comprise a cognitive function parameter.
The cognitive function parameter of the invention can be obtained by TMT (Trail Making test), which is a simple cognitive function test of tracing indicators such as numbers from 1 to 25 written on a piece of paper with a pencil in order, ATMT (Advanced Trail Making Test) for performing TMT on a touch panel, modified ATMT developed by the inventors (K. Mizuno et al. Brain & Development 33 (2011) 412-420), or the like. Although TMT, ATMT, and modified ATMT have differences in methodology, subjects of measurement are the same, which are all cognitive function parameters of the invention.
As modified ATMT, the inventors prepared, for example, five problems for evaluating various elements of cognitive function, which can be used alone or in combination. Each of the cognitive problems can be evaluated by total reaction time or total number of total number of correct answers.
A cognitive function parameter is a parameter that can be measured by a noninvasive test.
According to the studies conducted by the inventors, accumulation of fatigue can affect the blood vessel and skin. Therefore, the first parameter set of the invention preferably can comprise a blood vessel and skin parameter.
Examples of blood vessel parameters include, but are not limited to, blood vessel age, mean value of capillary length, cloudiness of blood vessel, number of blood vessels, and the like. The mean value of capillary length, cloudiness of blood vessel, and number of blood vessels can preferably be readily measured by image processing on the capillary course at the nail bed of the finger. Image processing on the capillary course and measurement of these parameters can be performed with, for example, a capillary scope manufactured by At Co., Ltd. (Osaka, Japan).
Examples of skin parameters include, but are not limited to, moisture content in the skin of an arm, amount of moisture evaporation, gloss, and the like. Methods of measuring moisture content in the skin of an arm, amount of moisture evaporation, and gloss are well known in the art.
A blood vessel and skin parameter is a parameter that can be measured by a noninvasive test.
The first parameter set of the invention can also comprise a subjective evaluation parameter of a subject in addition to an objective parameter obtained by the measurement described above. Statuses such as physical, fatigue, and mental states of a subject that cannot be sufficiently understood from measurement of various components can be reflected in a health level positioning map by adding subjective evaluation to a parameter set in addition to objective parameters from various measurement values.
In one embodiment, the subjective evaluation parameter in the invention can comprise subjective evaluation on fatigue, sleep, or mental state.
Subjective evaluation of fatigue can comprise one or more of subjective evaluation of fatigue duration, question related to a disorder due to fatigue, fatigue VAS (Visual Analogue Scale), Chalder Fatigue Scale (CFQ), fatigue symptom score computed using 11 items in CFQ (CFQ 11) (Tanaka M et al., Psychol Rep_2010, 106, 2, 567-575), questionnaire or VAS related to presenteeism, and questionnaire related to fatigue. “Question related to a disorder due to fatigue” refers to a question that checks subjective evaluation of a subject with respect to presence/absence of a cause and effect relationship between fatigue and some type of a disorder. “Question related to a disorder due to fatigue” can be, for example, a question on a disease thought to be the cause of fatigue, whether the subject feels that fatigue is impeding with work, household chores, or school work, or the like. “Questionnaire related to fatigue” asks whether the subject is aware of any symptom of fatigue, which can be, for example, whether the subject feels lethargy, whether the subject feels that fatigue remains even after a night of sleep, or the like. For example, WHO's Health and Work Performance Questionnaire (HPQ), Work Limitations Questionnaire (WLQ), or the like can be used as a questionnaire on presenteeism.
Subjective evaluation of sleep can comprise one or more of time of sleep (sleep and wake time), mean sleep time, VAS related to drowsiness, and questionnaire related to the quality of sleep.
Subjective evaluation of mental state can include one or more of VAS and questionnaire related to depression, and VAS and questionnaire related to enthusiasm. “Questionnaire related to depression” refers to questions related to any symptom of depression, which can include whether the subject feels melancholy, whether the subjects feels tiresome about interacting with others, or the like. One example thereof is a K6 total score (indicator commonly used as an indicator representing a mental problem developed by Kessler et al.)
The “questionnaire” described above can be evaluated by a response to a specific question or evaluated by converting responses to a large number of questions into a score.
A subjective parameter is a parameter that can be measured by a noninvasive test.
The first parameter set of the invention can comprise a living condition parameter in addition to an objective parameter and subjective evaluation parameter. The living condition parameter in the invention is a fact about the living conditions of a subject. Examples thereof can include years of education, period of marriage, presence/absence of cohabitant, smoking (yes/no, frequency, and/or amount), drinking (yes/no, frequency, and/or amount), working hours, exercise (yes/no, frequency, and/or amount), meals (whether the subject feels meals are frequent or early, frequency of meals after dinner, frequency of skipping breakfast, etc.), medical history, drug dosing, supplement dosing, and the like.
A living condition parameter is a parameter that can be measured by a noninvasive test.
The first parameter set of the invention can include a parameter based on data associated with cerebral function/neuropsychiatric system evaluation, circulatory/respiratory function system evaluation, or renal excretion/liver, gallbladder, and pancreas/detoxification function system evaluation.
For example, data associated with cerebral function/neuropsychiatric system evaluation can comprise data such as communication function, amount of activity (during the day or during sleep), brain anatomy measurement by MRI (Magnetic Resonance Imaging) (can measure decrease in function corresponding to a contracted site of brain tissue), fMRI at rest, and nerve fiber bundle course anisotropy (size and durability of nerve fiber bundle), in addition to those described above.
Data associated with circulatory/respiratory function system evaluation can include blood flow volume (can be measured, for example, with a Doppler blood flow meter), expirated gas component analysis (NO (asthma), acetone (diabetes), or the like), or the like, in addition to those described above. Data associated with expirated gas component analysis can be measured by mass spectrometry or ion mobility spectrometry.
Data associated with renal excretion/liver, gallbladder, and pancreas/detoxification function system evaluation can include data from skin gas component analysis or the like in addition to those described above. Data associated with skin gas component analysis can be measured by mass spectrometry or a high sensitivity variable laser detector.
In one embodiment, the first parameter set of the invention can comprise a biological oxidation parameter, less repair energy parameter, inflammation parameter, and autonomic nerve parameter (four basic parameters). By creating a health level positioning map based on data with respect to a first parameter set including the four basis parameters, the health level positioning map enables determination of a deteriorated health state (i.e., “ahead sick”) that has not resulted in a disease.
In another embodiment, the first parameter set of the invention can comprise four basic parameters, a fundamental parameter, a cognitive function parameter, and a subjective parameter. Although not wishing to be bound by any theory, it is understood that a wide ranging fatigue or mental states can be evaluated by adding a fundamental parameter, a cognitive function parameter, and a subjective parameter in addition to the four basic parameters. A subjective parameter preferably comprises evaluation on one or more of fatigue, sleep, and mental state, and more preferably comprises at least evaluation on fatigue.
In still another embodiment, the first parameter set of the invention can comprise four basic parameters, a fundamental parameter, a cognitive function parameter, a subjective parameter, and a hematological parameter. Although not wishing to be bound by any theory, a wide ranging health levels comprising additional different viewpoint can be evaluated by further adding a hematological parameter that can accurately evaluate a function of the endocrine system to the first parameter set, in addition to the four basic parameters, and a fundamental parameter, cognitive function parameter, and subjective parameter for evaluating fatigue or mental state.
In still another embodiment, the first parameter set of the invention can comprise four basic parameters, a fundamental parameter, a cognitive function parameter, a subjective parameter, a hematological parameter, a blood vessel and skin parameter, and a living condition parameter.
The first parameter set of the invention typically comprises both an invasive parameter and a noninvasive parameter. This would reflect information on an invasive parameter in a positioning map which is the base of evaluation, even when the second parameter set described below is comprised of only noninvasive parameters. As a result, evaluation including the effect of a parameter that can be obtained only through an invasive test can be performed by testing with only a noninvasive parameter. This is a significant effect of the inventions of the present application.
In one embodiment, a second parameter set can comprise the following.
(1) age;
(2) fat percentage;
(3) neutral fat (TG);
(4) CRP;
(5) OSI;
(6) subjective evaluation on fatigue;
(7) balance of autonomic nerve (e.g., LF/HF or logarithmic value thereof);
(8) a cognitive function.
In a preferred embodiment of the invention, a second parameter set can be comprised of only noninvasive parameters. However, a positioning map for evaluating the second parameter set can be a map created by including an invasive parameter, so that evaluation that essentially comprises also the effect of an invasive parameter can be performed by evaluating the overall health of a subject using the second parameter set comprised of only noninvasive parameters. This is a significant effect of the invention. Further, estimation of a health level of a user from such a parameter set that does not comprise an invasive test result enables a user to conveniently find the user's own health level, regardless of the testing site. A user can find the user's own health level from a simple test performed, for example, at a company, drug store, community center, café, home, and the like in addition to hospitals.
A second parameter set comprised of only noninvasive parameters can comprise the following:
(1) age;
(2) BMI;
(3) fat percentage;
(4) SOS;
(5) systolic blood pressure;
(6) subjective evaluation on fatigue;
(7) subjective evaluation on depression;
(8) activity of parasympathetic nerve (e.g., HF or a logarithmic value thereof);
(9) activity of an entire autonomic nervous system (e.g., TP or a logarithmic value thereof); and
(10) cognitive function.
At step S501, the acquisition means 121 of the processing unit 120 acquires a first data set with respect to a first parameter set for each of a plurality of subjects. The acquisition means 121 can acquire data, for example, by receiving data on the plurality of subjects stored in the database 200 via the interface unit 110. The acquisition means 121 can acquired data, for example, by receiving data on the plurality of subjects stored in the database unit 200 from the interface unit 110 from a computer system of a test facility (e.g., hospital, research laboratory, or the like).
At step S502, the processing means 122 of the processing unit 120 can obtain first data by processing a first data set acquired at step S501. Processing by the processing means 122 can comprise, for example, at least one of dimensionality reduction processing, standardization processing, and weighting processing on the first data set.
Preferably, processing by the processing means 122 comprises dimensionality reduction processing on the first data set. This is because such processing can reduce the dimension of a multidimensional and complex first data set to a more readily understandable data and consequently a health level positioning map. At this time, dimensionality reduction processing preferably reduces the first data set to two dimensional data or three dimensional data. This is because a health level positioning map created from two dimensional data or three dimensional data would be a map of a two-dimensional space or a three-dimensional space and visually readily understandable. By performing dimensionality reduction processing, (1) contribution of each measurement item to a value of each axis of a health level positioning map can be found, and (2) data on a health level positioning map can be more clearly understood by excluding data for a measurement item that is not significantly associated with a value of each axis of a health level positioning map.
More preferably, processing by the processing means 122 comprises standardization processing on a first data set and dimensionality reduction processing on the standardized data set. This is because this eliminates a scalar difference between parameters of a first data set, and the effect of each parameter of the first data set on the health level positioning map is considered equally, so that a health level positioning map that is highly accurate and readily understandable can be created. In particular, both data obtained from a male subject and data obtained from a female subject can be evaluated on the same health level positioning map by performing standardization processing on a parameter arising from a difference between males and females.
More preferably, processing by the processing 122 comprises weighting processing on a first data set, standardization processing on the weighted data set, and dimensionality reduction processing on the standardized data set. This is because this enables the magnitude of effect of each parameter of the first data set on a health level positioning map to be emphasized and creates a health level positioning map that is even more accurate and readily understandable.
Standardization processing by the processing means 122 can be configured to be performed, for example, on all parameters of a first parameter set, or on a specific parameter. Weighting processing by the processing means 122 can be configured to be performed, for example, on a first data set of all of the plurality of subjects, or on a first data set of a specific population of the plurality of subjects.
At step S503, the mapping means 123 of the processing unit 120 maps first data obtained at step S502 for each of the plurality of subjects. The mapping means 123 maps n- dimensional first data onto an n-dimensional space. The mapping means 123 can output a map mapping first data of each of the plurality of subjects by mapping the first data for each of the plurality of subjects. If, for example, the first data is two-dimensional, the mapping means 123 can output a two-dimensional map by mapping the first data onto a position on a two-dimensional space, i.e., a plane.
At step S504, the clustering means 124 of the processing unit 120 clusters the first data mapped at step S503 to identify a plurality of regions. The clustering means 124 can identify any number of regions by dividing the mapped first data into any number of clusters.
At step S505, the characterization means 125 of the processing unit 120 characterizes at least some of the plurality of regions identified at step S504. The characterization means 125 can be configured, for example, to characterize at least some of the plurality of regions based on information inputted into the computer system 100, or to automatically characterize at least some of the plurality of regions independent of any input. For example, the characterization means 125 can characterize at least some of the plurality of regions based on a relative position on a health level positioning map, or based on machine learning.
A health level positioning map with at least some of the plurality of regions characterized is created by the aforementioned processing 500. The created health level positioning map can be utilized in processing 510, processing 600, or processing 700 described below.
At step S511, the processing unit 120 receives an input selecting some of the plurality of regions on the health level positioning map created by the processing 500. An input selecting some of the plurality of regions is inputted, for example, via the interface unit 110 from the outside of the computer system 100.
At step S512, the acquisition means 121 of the processing unit 120 acquires first data mapped to the selected region. The acquired first data can be referred to as sub-first data.
At step S513, the clustering means 124 of the processing unit 120 clusters the sub-first data acquired at step S512 to identify a plurality of regions. The processing at step S513 can be the same as the processing at step S504.
At step S514, the characterization means 125 of the processing unit 120 characterizes at least some of the plurality of regions identified at step S513. The processing at step S514 can be the same as the processing at step S505.
A health level positioning map is created for some of the plurality of subjects by the processing 510 described above. A health level positioning map for some of the plurality of subjects can be a health level positioning map focused on a specific population in the plurality of subjects, such as a male subject population, a female subject population, a young population (population of subjects under 40 years old), a middle age population (population of subject who are 40 or older and younger than 60 years old), or a senior population (population of subjects who are 60 or older). A plurality of regions on a health level positioning map focused on a specific population can have characterization that is different from a plurality of regions on a health level positioning map of all of the plurality of subjects, and can be utilized for analyzing a health status from a different viewpoint.
At step S601, the processing unit 130 prepares a health level positioning map. For example, the processing unit 130 can prepare a health level positioning map by acquiring a health level positioning map with the first acquisition means of the processing unit 130. A health level positioning map is created based on a first data set with respect to a first parameter set described above for a plurality of subjects. A health level positioning map that is prepared can be a health level positioning map created by the processing 500 or processing 510, or a health level positioning map created in another manner, as long as the map is created using the first parameter set.
At step S602, the second acquisition means 132 of the processing unit 130 acquires a second data set with respect to a second parameter set for at least some of the plurality of subjects. The second parameter set is a part of the first parameter set. For example, the second acquisition means 132 can acquire data for some of the plurality of subjects stored in the database unit 200 via the interface unit 110.
At step S603, the derivation means 133 of the processing unit 130 derives a health function that correlates the second data set of at least some of the plurality of subjects acquired at step S602 with a position on a health level positioning map of at least some of the plurality of subjects. For example, the derivation means 133 can derive a health function by machine learning. A health function can be derived, for example, for each axis of an n-dimensional health level positioning map.
A health function can be, for example, a regression model or a neural network model. If a health function is a regression model, the derivation means 133 can derive each coefficient of the regression model by machine learning using a second data set as an independent variable and a coordinate on a health level positioning map of a subject as a dependent variable for each of at least some of the plurality of subjects. If a health function is a neural network model, the derivation means 133 can derive a weighting coefficient of each node by machine learning using a second data set as input supervisor data and a position on a health level positioning map of a subject as output supervisor data for each of at least some of the plurality of subjects.
A health function for mapping a health level of a subject on a health level positioning map is created by the aforementioned processing 600. The created health function can be utilized in the processing 700 described below. By not including a result of an invasive test in a second parameter set when creating a health function, the resulting health function would be able to identify a position on a health level positioning map from data for a parameter set that does not comprise a result of an invasive test and estimate a health level of a user. Such estimation of a health level of a user from a parameter set that does not comprise a result of an invasive test enables the user to find the user's own health level regardless of the testing site. A user would be able to find the user's own health level from a simple test performed, for example, at a company, drug store, community center, café, home, and the like in addition to hospitals.
At step S701, the processing unit 140 prepares a health function. For example, the processing unit 140 can prepare a health function by acquiring a health function with the third acquisition means 141 of the processing unit 140. A health function is a function that correlates a data set with respect to a second parameter set with a position on a health level positioning map. The acquired health function can be a health function created by the processing 600 or a health function created in another manner, as long as the function can correlate a user data set with a position on a health level positioning map. The health level positioning map can be a health level positioning map created by the processing 500 or the processing 510 or a health level positioning map created in another manner, as long as the map is created using a first parameter set.
At step S702, the fourth acquisition means 142 of the processing unit 140 acquires a first user data set with respect to a second parameter set of a user. The fourth acquisition means 142 can acquire the first user data set stored in the database unit 200, for example, via the interface unit 110. Alternatively, the fourth acquisition means 142 can acquire the first user data set, for example, via the interface unit 110 from a terminal apparatus of a user.
At step S703, the output generation means 143 of the processing unit 140 obtains a first output by inputting a first user data set into a health function. If, for example, a health function is a regression model, a coordinate on a health level positioning map is outputted as a first output by inputting a first user data set into an independent variable of the regression model. If, for example, a health function is a neural network model, a coordinate on a health level positioning map is outputted as a first output by inputting a first user data set into an input layer of the neural network model.
At step S704, the output mapping means 144 of the processing unit 140 maps a first output onto a health level positioning map. Since the output obtained at step S703 is a coordinate on a health level positioning map, the output mapping means 144 can map the coordinate onto an n-dimensional space of the health level positioning map.
The health status of a user can be estimated from characterization of a region to which a health level of the user is mapped by mapping the health level of the user onto a health level positioning map by the aforementioned processing 700. For example, by using a health function that does not comprise a result of an invasive test in a second parameter set, a position on a health level positioning map can be identified from data for a parameter set that does not comprise a result of an invasive test to estimate a health level of a user. Such estimation of a health level of a user from a parameter set that does not comprise a result of an invasive test enables the user to find the user's own health level regardless of the testing site. A user would be able to find the user's own health level from a simple test performed, for example, at a company, drug store, community center, café, home, and the like in addition to hospitals.
At step S705, the fourth acquisition means 142 of the processing unit 140 acquires a second user data set with respect to a second parameter set of a user. At step S705, this is performed after at least a predetermined period has elapsed from step S702. The fourth acquisition means 142 can acquire a second user data set stored in the database unit 200, for example, via the interface unit 110. Alternatively, the fourth acquisition means 142 can acquire a second user data set, for example, via the interface unit 110 from a terminal apparatus of a user.
At step S706, the output generation means 143 of the processing unit 140 obtains a second output by inputting a second user data set into a health function. If, for example, a health function is a regression model, a coordinate on a health level positioning map is outputted as a second output by inputting a second user data set into an independent variable of the regression model. If, for example, a health function is a neural network model, a coordinate on a health level positioning map is outputted as a second output by inputting a second user data set into an input layer of the neural network model.
At step S707, the output mapping means 144 of the processing unit 140 maps a second output onto a health level positioning map. Since the output obtained at step S706 is a coordinate on a health level positioning map, the output mapping means 144 can map a coordinate onto an n-dimensional space of the health level positioning map.
Through steps S705 to S707, a health status of a user after a predetermined period can be estimated. For example, a chronological change in the health status can be identified by comparing the health status estimated at steps S701 to S704 with the health status estimated at steps S705 to S707.
Furthermore, a future health status can be predicted based on the direction of a chronological change on a health level positioning map. For example, if a health level belongs to a region characterized as healthy on a health level positioning map as a result of mapping a first output and a health level belongs to a region characterized as healthy but has approached a region characterized as a life style disease risk group as a resulting of mapping a second output after a predetermined period has elapsed, the health status can be predicted as heading toward a direction of a life style disease risk.
In one embodiment, the processing 700 can be utilized to evaluate an item for improving a health status. For example, a chronological change on a health level positioning map can be identified by asking a user to use the item for improving a health status for a predetermined period and comparing a first output before the predetermined period with a second output after the predetermined period has elapsed. A chronological change on a health level positioning map due to use of an item for improving a health status reflects the effect of the item for improving a health status. The quality of the effect of the item for improving a health status can be evaluated by comparing with a chronological change on a health level positioning map when the item for improving a health status was not used.
Furthermore, an item for improving a health status can be recommended to a user based on the direction of a chronological change on a health level positioning map due to use of the item for improving a health status. For example, an item for improving a health status is generally effective for a user with a chronological change on a health level positioning map in the opposite direction from the direction of a chronological change on a health level positioning map due to use of the item for improving a health status. Therefore, an item having a chronological change in the opposite direction from the chronological change on a health level positioning map of a user identified by the processing of steps S701 to S707 can be recommended to the user.
Although the examples described above while referring to
Although the processing at each step shown in
The present invention is not limited to each of the embodiments described above. Various modifications can be made within the scope specified in the claims. Embodiments obtained from appropriately combining each technical means disclosed in different embodiments are also encompassed by the technical scope of the invention.
In this Example, initial data for 232 items related to health was first acquired for 720 subjects. The initial data was acquired at Riken Kobe Campus IIB. The 232 items included both an invasive test and a noninvasive test. The values of the acquired initial data were then corrected so that the mean value was 0 and the standard deviation was 1. A portion of the data was then extracted using a method of extracting, when a plurality of pieces of data with a correlation therebetween greater than a predetermined value (specifically, a correlation coefficient of 0.9) are available, only one piece of the plurality of pieces of data among the corrected data. This portion of data included data for four basic parameters. In this Example, when there are a plurality of pieces of data with a correlation coefficient greater than 0.9, one piece of the plurality of pieces of data was extracted, but the present invention is not limited thereto. The value of the correlation coefficient can be appropriately changed. A health function capable of computing a health level of a subject more comprehensively can be created by using a correlation coefficient of 0.9 or greater.
As a result of the extraction, 81 items were extracted. The 81 items fall under the “first parameter set” in the invention. The 81 items included four basic parameters, fundamental parameter, cognitive function parameter, subjective parameter, hematological parameter, blood vessel and skin parameter, and living condition parameter.
Next, multidimensional data (i.e., 81 dimensional data) was reduced to two dimensions by multidimensional scaling, and data for 692 subjects whose data were complete among the 720 members was plotted on a two-dimensional plane. A plot distribution pattern of 692 plots plotted on the two-dimensional plane was clustered by k-means into 10 clusters (
A function (health function) capable of suitable arrangement on the health level positioning map created in Example 1 with fewer number of test items (second parameter set) than the first parameter set was identified by machine learning.
As described above,
A subject group contained in group G2 was a subject group that was old with high blood γ-GTP/blood ALT/blood neutral fat/blood HbA1c/blood high sensitivity CRP values. Therefore, it can be understood that if a subject who has been newly measured for data belongs to group G2, the subject is likely to have a pre-disease state of a life style disease.
A subject group contained in group G3 was a subject group that was old with a high blood sugar level. Therefore, it can be understood that if a subject who has been newly measured for data belongs to group G3, the subject is likely to have a pre-disease state of diabetes.
The regions of first function X≤about 4 and second function Y≤about 2 generally indicate a healthy group without any problems in the health status.
The above groups G1 to G3 are several examples that can be evaluated with the health evaluation apparatus of the invention. Various other health risks can also be additionally evaluated.
A function capable of suitable arrangement on a health level positioning map with another second parameter set was attempted to be identified by machine learning.
A health level positioning map was created using a first parameter set of 76 items from data for 965 subjects by adding more subjects to Example 1. In the same manner as Example 2, data for half of the subjects were used for supervisor data, and data for the remaining half of subjects was used for evaluation of the obtained function.
As a result, a health function from a second parameter set of 50 items (R2 of X axis=0.9595; R2 of Y axis=0.9615), health function from a second parameter set of 21 items (R2 of X axis=0.9601; R2 of Y axis=0.8706), and health function from a second parameter set of 9 items (R2 of X axis=0.8889; R2 of Y axis=0.8207) were identified (
The 9 items were age, fatigue questionnaire score, fatigue duration, ln(LF+HF), cognitive function, fat percentage, blood neutral fat, OSI, and CRP.
If health can be evaluated with only data from a noninvasive test, users can conveniently find their own health level regardless of the testing site. In this regard, an attempt was made to identify a second parameter set comprised of only noninvasive parameters. As a result, parameters for the following 14 items were identified, and a health function using such a parameter set was identified (
*age
*BMI
*fat percentage
*SOS
*systolic blood pressure
*subjective evaluation on depression
*presenteeism
*subjective evaluation on fatigue
*activity of parasympathetic nerve (ln(HF))
*activity of an entire autonomic nervous system (ln(TP)), and
*cognitive problem.
Therefore, users can conveniently find their own health level by using the health level positioning map of the invention and the 14 noninvasive parameters described above.
This Example studied whether a change in the health status of a subject can be observed using the health level positioning map created in Example 1. Mapping positions on the health level positioning map were compared between before dosing of reduced CoQ10 and after dosing for 3 months. The results are shown in
As is apparent from
100 Computer system
110 Interface unit
120, 130, 140 Processing unit
150 Memory unit
200 Database unit
300 User terminal apparatus
400 Network
The various embodiments described above can be combined to provide further embodiments. All of the U.S. patents, U.S. patent application publications, U.S. patent applications, foreign patents, foreign patent applications and non-patent publications referred to in this specification and/or listed in the Application Data Sheet are incorporated herein by reference, in their entirety. Aspects of the embodiments can be modified, if necessary to employ concepts of the various patents, applications and publications to provide yet further embodiments.
These and other changes can be made to the embodiments in light of the above-detailed description. In general, in the following claims, the terms used should not be construed to limit the claims to the specific embodiments disclosed in the specification and the claims, but should be construed to include all possible embodiments along with the full scope of equivalents to which such claims are entitled. Accordingly, the claims are not limited by the disclosure.
| Number | Date | Country | Kind |
|---|---|---|---|
| 2018-207611 | Nov 2018 | JP | national |
| Filing Document | Filing Date | Country | Kind |
|---|---|---|---|
| PCT/JP2019/043062 | 11/1/2019 | WO | 00 |
