JP7348691B2 - Intervention evaluation device - Google Patents

Description

The present invention relates to an intervention evaluation device.
More specifically, biological information is collected using brain image data output from a device that non-invasively takes three-dimensional images of the human brain, such as MRI (magnetic resonance imaging). The present invention relates to an intervention evaluation device that evaluates the usefulness of predetermined actions considered to be useful for brain health as a function of a brain information analysis device that performs analysis.

Originally, MRI was a non-invasive tool for discovering and diagnosing diseases of internal organs, including those of the brain. However, in recent years, as it has become a permanent practice to photograph people's brains using MRI and obtain three-dimensional brain image data, brain image data of various people has gradually been accumulated. It has become. Therefore, in recent years, neuroscientists have begun to explore the correlation between information obtained from brain imaging data and human health, not just brain diseases.
Hereinafter, in this specification, information obtained from brain image data will be collectively referred to as brain information. Brain information includes, for example, the amount of gray matter in a specific region of the brain, the anisotropy of nerve fibers in a specific region of the brain, etc., obtained as a result of image analysis of brain image data.

Patent Document 1 discloses a brain information analysis device and a brain health index calculation device by the inventors, which convert brain analysis information into easy-to-handle values and can be used as a health index.

Japanese Patent Application Publication No. 2018-33516

There are various methods for acquiring information for each part of the brain from brain image data. Regardless of the brain information acquisition method, the information obtained includes a large number of variables and is extremely detailed. However, this detailed information, that is, the large number of variables, is not compatible with so-called big data analysis methods that statistically infer overall trends in data.

MRI machines are expensive, so although Japan has relatively many MRI machines installed in international comparisons, it cannot be said that the number of MRI machines installed is large compared to other medical devices. Furthermore, since the operating cost is not cheap, it is not easy to accumulate a large amount of brain image data obtained by MRI. Therefore, at present, there are not so many data parameters on which to base statistical analysis. On the other hand, if information obtained from brain image data is replaced with variables, the number of variables becomes extremely large. This causes a phenomenon called overfitting, in which statistical analysis methods produce correct estimation results for known values, but fail to obtain correct estimation results for unknown values.
Furthermore, a large number of variables indicates that the indicators for judgment are complex. For this reason, the current state of detailed brain analysis information is only understandable to neuroscience experts and is not suitable for the general public to use as an indicator of brain health.

The inventors have solved these problems and invented a brain information analysis device and a brain health index calculation device that convert brain analysis information into easy-to-handle values that can be used as health indicators. This invention is the invention disclosed in Patent Document 1.

In Patent Document 1, the inventors analyzed brain image data of a large number of subjects and questionnaires regarding the subjects' health, lifestyle, etc., and found that brain health and physical health are closely related. Ta. Furthermore, the inventors have also found that when a plurality of subjects continuously perform predetermined actions to improve their physical health, the brain health index values of the subjects also tend to improve.

Based on the above, businesses such as companies, organizations, and individuals are promoting products and services with catchphrases such as "promote brain health" on the basis that the brain health index values of test subjects have improved. It becomes possible to sell services. For example, it is conceivable to attach a catchphrase such as "promote brain health" to various products and services such as food, exercise equipment, sports gyms, sanitary products, office desks, housing, etc. If the product or service in question truly contributes to brain health, this could be a major selling point for the product or service to consumers.
On the other hand, compared to physical health, brain health does not have clear characteristics, so it cannot be clearly determined whether a person's brain is healthy or not just by observing a person's appearance. For this reason, there are concerns that businesses will market their own products and services without any basis and labeling them as ``good for brain health.''
Therefore, there is a need for a system to objectively determine whether a business's products and services truly have an effect that contributes to human brain health.

In addition, in this specification, a corporation, group, or individual continuously performs prescribed acts or uses of equipment, etc., on multiple subjects for a certain period of time or more that is believed to be useful for improving brain health. is called "intervention." From now on, the legal entity, organization, or individual implementing this intervention will be referred to as the intervention implementer.

The present invention solves this problem and makes it possible to objectively judge whether or not a business's products and services contribute to a person's brain health at a low cost, without the need for subjects to undergo MRI imaging. The purpose is to provide an intervention evaluation device.

In order to solve the above problems, the intervention evaluation device of the present invention evaluates the effectiveness of intervention using objective numerical values. The intervention evaluation device uses the subject ID that uniquely identifies the subject receiving the intervention, the intervention ID that uniquely identifies the intervention that the subject receives, and the first daily life test in the pre-intervention schedule, which is the stage immediately before implementing the intervention. At the pre-intervention date, when you answered the environmental questionnaire and created the pre-intervention living environment questionnaire results, and at the post-intervention date, which is the stage immediately after implementing the intervention, you answered the second living environment questionnaire and created the pre-intervention living environment questionnaire results. It includes an intervention table that describes the relationship with the post-intervention schedule when creating the environmental questionnaire results, and a subject master that describes the relationship between the subject ID, the age of the subject, and the gender of the subject. Furthermore, based on the subject's estimated BHQ value or the subject's estimated BHQ difference derived from the subject's pre-intervention living environment questionnaire results and post-intervention living environment questionnaire result, the average value and estimated BHQ value are calculated for each subject's age and gender. An average/deviation calculation processing unit that calculates the standard deviation or the average value and standard deviation of the estimated BHQ difference, and the age of the subject and the average value and standard deviation of the estimated BHQ value, The apparatus includes a t-test calculation processing unit that calculates a t-test for each gender and outputs a p-value.

According to the present invention, it becomes possible to objectively judge whether or not a business's products and services contribute to a person's brain health at low cost, without requiring a subject to undergo MRI imaging.
Problems, configurations, and effects other than those described above will be made clear by the following description of the embodiments.

It is a schematic diagram showing a learning phase and an estimation phase of an intervention evaluation device concerning a first embodiment of the present invention. FIG. 1 is a block diagram showing the hardware configuration of an intervention evaluation device according to a first embodiment of the present invention. It is a block diagram showing the whole software function of the intervention evaluation device concerning a first embodiment of the present invention. FIG. 3 is a diagram showing field configurations of various tables. FIG. 2 is a block diagram showing the inside of a BHQ arithmetic processing section. It is a figure showing an example of a living environment questionnaire. It is a flowchart which shows the flow of the whole operation|movement in the learning phase of an intervention evaluation apparatus. It is a flowchart which shows the flow of the whole operation|movement in the estimation phase of an intervention evaluation apparatus. FIG. 2 is a block diagram showing details of data processing by a learning calculation processing section. 3 is a block diagram showing details of data processing by an estimation calculation processing unit 305. FIG. FIG. 2 is a block diagram showing details of data processing by an average/deviation calculation processing section. It is a schematic diagram showing a learning phase of an intervention evaluation device concerning a second embodiment of the present invention. It is a schematic diagram showing an estimation phase of an intervention evaluation device concerning a second embodiment of the present invention. It is a block diagram showing the whole software function of the intervention evaluation device concerning a second embodiment of the present invention. FIG. 3 is a diagram showing field configurations of an estimated BHQ difference table and a measured BHQ difference table. It is a flowchart which shows the flow of the whole operation|movement in the learning phase of an intervention evaluation apparatus. It is a flowchart which shows the flow of the whole operation|movement in the estimation phase of an intervention evaluation apparatus. FIG. 3 is a block diagram showing details of data processing by a differential learning calculation processing section. FIG. 2 is a block diagram showing details of data processing of a difference estimation calculation processing section. It is a block diagram showing details of data processing of an average/deviation calculation processing part and a t-test calculation processing part.

The intervention evaluation device according to the present invention calculates and outputs numerical basis for objectively determining whether an "intervention" really contributes to improving brain health.
Specifically, a first questionnaire is conducted for multiple subjects before the intervention, and then the intervention is carried out over a certain period of time. Then, after a certain period of time has passed since the intervention, a second questionnaire will be conducted. By inputting the responses from the first and second questionnaires into the intervention evaluation device, statistical analyzes such as t-tests for two paired samples or t-tests for two unpaired samples can be performed. It is possible to show whether there is a statistically significant difference based on the results. If a statistically significant difference is found, it is objectively recognized that the intervention is effective in improving brain health.

[First embodiment: intervention evaluation device 101: overall configuration]
FIG. 1A is a schematic diagram showing the learning phase of the intervention evaluation device 101 according to the first embodiment of the present invention.
First, the MRI apparatus 103 photographs the brain of the first subject 102 and outputs an MRI image file group 104.
The intervention evaluation device 101 takes in the MRI image file group 104 of the first subject 102 and the living environment questionnaire results 105, and performs a learning process. As a result, approximate function parameters 106 are generated or updated.
The intervention evaluation device 101 is a personal computer or server equipped with a large-capacity nonvolatile storage 206 (described later in FIG. 2) and a predetermined computing power.

An MRI image file group 104 obtained by photographing the brain of the first subject 102 using the MRI apparatus 103 is input to the intervention evaluation apparatus 101 and stored as a database. The MRI image file group 104 is linked by a subject ID that uniquely identifies all the first subjects 102. In particular, in order to store the MRI image file group 104, the intervention evaluation device 101 requires a large-capacity non-volatile storage 206.

The MRI image file group 104 is a plurality of image files obtained by photographing the brain of the first subject 102 by the MRI apparatus 103. A plurality of image files are obtained by photographing the brain of the first subject 102 in a virtual cross-section state, from the top of the head to the neck.
Then, the intervention evaluation apparatus 101 acquires brain information from the MRI image file group 104 obtained by imaging the first subject 102 using a predetermined image analysis process. Next, the intervention evaluation device 101 calculates the average value of this brain information, and stores it in the nonvolatile storage 206 in association with the subject ID of the first subject 102 and the shooting date of the MRI image file group 104.

The intervention evaluation device 101 according to the embodiment of the present invention calculates the average value of two types of brain information disclosed in Patent Document 1.
The average value of the first brain information is the average value of 116 gray matter volume data groups included in the brain. Hereinafter, the average value of the gray matter volume data group will be referred to as GM-BHQ (Grey-Matter Brain Healthcare Quotient). GM-BHQ is a single scalar value.
The average value of the second brain information is the average value of 48 nerve fiber anisotropy data groups. Hereinafter, the average value of the nerve fiber anisotropy data group 402 will be referred to as FA-BHQ (fractional anisotropy Brain Healthcare Quotient). FA-BHQ is also a single scalar value like GM-BHQ.

Then, on the same day as the MRI scan (in most cases, during the waiting time for the MRI scan), a living environment questionnaire 601 (see FIG. 6) is administered to the first subject 102. Then, the living environment questionnaire results 105, which are the response contents of the questionnaire, are input into the intervention evaluation device 101 as data for each item.

The above-described MRI imaging and living environment questionnaire 601 are performed on a plurality of first subjects 102, and the obtained data group is input to the intervention evaluation device 101 and accumulated. Then, when the learning process is executed, the approximate function parameters 106 are generated or updated. Then, the intervention evaluation device 101 can estimate GM-BHQ and FA-BHQ with respect to the response contents of the living environment questionnaire 601 using the approximation function parameters 106.

FIG. 1B is a schematic diagram showing the estimation phase of the intervention evaluation device 101 according to the first embodiment of the present invention.
A second subject 107, who is different from the time point in FIG. 1A, answers the first living environment questionnaire 601 on a pre-intervention schedule 108, which is the stage immediately before implementing the intervention, and creates a pre-intervention living environment questionnaire result 109. .
Next, the intervention implementer implements the intervention for the second subject 107 for a predetermined period. The predetermined period from the pre-intervention schedule to the post-intervention schedule is at least one month, preferably about three months to half a year.
Then, after a predetermined period of time has elapsed after the intervention, the second subject 107 answers the second living environment questionnaire 601 on a post-intervention schedule 110, and creates a post-intervention living environment questionnaire result 111.

The intervention evaluation device 101 takes in the pre-intervention living environment questionnaire results 109 on the pre-intervention schedule and the post-intervention living environment questionnaire results 111 on the post-intervention schedule of the second subject 107, and performs estimation processing using the approximation function parameters 106. I do. As a result, a first estimated BHQ corresponding to the pre-intervention living environment questionnaire result 109 of the second subject 107 and a second estimated BHQ corresponding to the post-intervention living environment questionnaire result 111 are obtained.
The first estimated BHQ is a set of the first estimated GM-BHQ based on the pre-intervention living environment questionnaire results 109 of the second subject 107 and the first estimated FA-BHQ.
The second estimated BHQ is a set of the second estimated GM-BHQ based on the post-intervention living environment questionnaire results 111 of the second subject 107 and the second estimated FA-BHQ.

The intervention evaluation device 101 executes this estimation calculation process of the first estimated BHQ and the second estimated BHQ for the plurality of second subjects 107, and calculates the first estimated BHQ for each age and gender of the second subjects 107. An estimated BHQ average value, a first estimated BHQ standard deviation, a second estimated BHQ average value, and a second estimated BHQ standard deviation are calculated.
The first estimated BHQ average value is a set of the first estimated GM-BHQ average value and the first estimated FA-BHQ average value for each age and gender of the second subject 107.
The first estimated BHQ standard deviation is a set of the standard deviation of the first estimated GM-BHQ and the standard deviation of the first estimated FA-BHQ for each age and gender of the second subject 107.
The second estimated BHQ average value is a set of the second estimated GM-BHQ average value and the second estimated FA-BHQ average value for each age and gender of the second subject 107.
The second estimated BHQ standard deviation is a set of the standard deviation of the second estimated GM-BHQ and the standard deviation of the second estimated FA-BHQ for each age and gender of the second subject 107.

Furthermore, from the estimated GM-BHQ average value and estimated FA-BHQ average value, estimated GM-BHQ standard deviation, and estimated FA-BHQ standard deviation for each age and gender, the p value of estimated GM-BHQ for each age and gender is calculated. and a statistical hypothesis test of the estimated FA-BHQ, and its probability (p value) is calculated.
For each age and gender, the first estimated BHQ average value, the first estimated BHQ standard deviation, the second estimated BHQ average value, the second estimated BHQ standard deviation, the estimated GM-BHQ p value, and the estimated The p-value of FA-BHQ is output to the non-volatile storage 206 or the like as the estimated BHQ average value and standard deviation for each age and gender and the p-value 112.

The first subject 102 is a subject for providing learning data in the learning phase of the intervention evaluation device 101. For this reason, the MRI apparatus 103 performs MRI imaging, and the living environment questionnaire results 105 are provided to the intervention evaluation apparatus 101.
The second subject 107 is a subject for estimating the BHQ value from the contents of the questionnaire and further determining the effectiveness of the intervention in the estimation phase of the intervention evaluation device 101. Therefore, MRI imaging using the MRI apparatus 103 is not performed. Instead, the second subject 107 receives the predetermined intervention between the pre-intervention date 108 and the post-intervention date 110. Then, the pre-intervention living environment questionnaire results 109 on the pre-intervention schedule 108 and the post-intervention living environment questionnaire results 111 on the post-intervention schedule 110 are provided to the intervention evaluation device 101.

[First embodiment: Intervention evaluation device 101: Hardware configuration]
FIG. 2 is a block diagram showing the hardware configuration of the intervention evaluation device 101 according to the first embodiment of the present invention. Note that the same applies to the intervention evaluation device 101 according to the second embodiment described later.
The intervention evaluation device 101, which is a general personal computer or server, includes a CPU 201, a ROM 202, a RAM 203, a display section 204 such as a liquid crystal display, an operation section 205 such as a keyboard and a mouse, and a non-volatile memory such as a hard disk device, which are connected to a bus 208. The real time clock (hereinafter referred to as "RTC") 207 outputs current date and time information.

In addition, a serial port 209 and a NIC (Network Interface Card) 210 are connected to the bus 208 for receiving the MRI image file group 104 and registering it in a database formed in the nonvolatile storage 206. The nonvolatile storage 206 stores an OS, a program for operating a personal computer or server as the intervention evaluation device 101, and various databases described later with reference to FIG.
Note that when the intervention evaluation device 101 is a server or a cloud computing environment, the NIC 210 is essential, but the display section 204 and the operation section 205 are not necessarily necessary, and the serial port 209 is not necessary. In that case, a separate terminal for operating the server or the like through the network may be provided.

[First embodiment: intervention evaluation device 101: entire software functions]
FIG. 3 is a block diagram showing the overall software functions of the intervention evaluation device 101 according to the first embodiment of the present invention.
An MRI image file group 104 and a questionnaire response group 302 are input to the input/output control unit 301 . Here, the questionnaire response group 302 includes the living environment questionnaire results 105, the pre-intervention living environment questionnaire results 109, and the post-intervention living environment questionnaire results 111 shown in FIG. There are various means for inputting the questionnaire response group 302 to the input/output control unit 301, such as the operation unit 205, the serial port 209, and the NIC 210.

The input/output control unit 301 has a data processing and calculation function, and performs data processing with a BHQ calculation processing unit 303, a learning calculation processing unit 304, an estimation calculation processing unit 305, an average/deviation calculation processing unit 306, and a t-test calculation processing unit 307. Give and receive.
The input/output control unit 301 also reads and writes to the subject master 308, brain information table 309, measurement BHQ table 310, approximate function parameters 106, questionnaire table 311, estimated BHQ table 312, intervention master 313, and intervention table 314. .

[First embodiment: Intervention evaluation device 101: Field configuration of various tables]
FIG. 4 is a diagram showing field configurations of various tables.
The subject master 308 has a subject ID field, a gender field, and a date of birth field.
The subject ID field stores a subject ID that uniquely identifies the subject.
A gender flag indicating the gender of the subject is stored in the gender field.
The date of birth of the subject is stored in the date of birth field.
The subject's age is calculated from the value of the date of birth field.
The first subject 102 and the second subject 107 are stored in the subject master 308 without being distinguished.

The brain information table 309 includes a subject ID field, a gray matter mass data group field, a nerve fiber anisotropy data group field, and an MRI imaging date field.
The subject ID field is the same as the same name field in the subject master 308.
The gray matter mass data group field stores the gray matter mass data group of the first subject 102 calculated by the gray matter mass calculation unit 501 (see FIG. 5) from the MRI image file group 104.
The nerve fiber anisotropy data group field stores the nerve fiber anisotropy data group of the first subject 102 calculated from the MRI image file group 104 by the nerve fiber anisotropy calculation unit (see FIG. 5).
The MRI imaging date field stores the date on which the first subject 102 was imaged by MRI (MRI imaging date).

The measurement BHQ table 310 has a subject ID field, an MRI imaging date field, a measurement GM-BHQ field, and a measurement FA-BHQ field.
The subject ID field is the same as the same name field in the subject master 308.
The MRI imaging date field is the same as the field with the same name in the brain information table 309.
The measurement GM-BHQ field stores the measurement GM-BHQ value of the first subject 102 on the MRI imaging date stored in the MRI imaging date field, which is calculated by the BHQ calculation processing unit 303.
The measurement FA-BHQ field stores the measurement FA-BHQ value of the first subject 102 on the MRI imaging date stored in the MRI imaging date field, which is calculated by the BHQ calculation processing unit 303.

In this specification, the GM-BHQ value and FA-BHQ value calculated from the MRI image file group 104 and the GM-BHQ value estimated from the pre-intervention living environment questionnaire results 109 and/or the post-intervention living environment questionnaire results 111 are used. and FA-BHQ value.
Hereinafter, the GM-BHQ value and FA-BHQ value calculated from the MRI image file group 104 will be referred to as the measured GM-BHQ value and the measured FA-BHQ value. Furthermore, the measured GM-BHQ value and the measured FA-BHQ value are collectively referred to as the measured BHQ value.
Similarly, the GM-BHQ value and FA-BHQ value estimated from the pre-intervention living environment questionnaire results 109 and/or the post-intervention living environment questionnaire results 111 are referred to as estimated GM-BHQ value and estimated FA-BHQ value. Also, the estimated GM-BHQ value and the estimated FA-BHQ value are collectively referred to as the estimated BHQ value.

The approximation function parameters 106 are parameters (matrix data) for forming an approximation function based on the learning algorithm, are generated or updated by the learning calculation processing unit 304, and are referenced by the estimation calculation processing unit 305.

The questionnaire table 311 has a subject ID field, a questionnaire reception date field, a questionnaire item field, and a questionnaire value field.
The subject ID field is the same as the same name field in the subject master 308.
The questionnaire reception date field stores the date on which the subject's questionnaire was received.
The questionnaire item field stores an item ID that uniquely identifies the questionnaire item.
The questionnaire value field stores a value indicating the response content of the subject in the questionnaire item corresponding to the item ID.
Questionnaire items include 2, 3, 4, 5, or 9 choices.
In the questionnaire table 311, the living environment questionnaire results 105 answered by the first subject 102, the pre-intervention living environment questionnaire results 109 and the post-intervention living environment questionnaire results 111 answered by the second subject 107 are stored without distinction.

The estimated BHQ table 312 has a subject ID field, a questionnaire reception date field, an estimated GM-BHQ field, and an estimated FA-BHQ field.
The subject ID field is the same as the same name field in the subject master 308.
The questionnaire reception date field is the same as the field with the same name in the questionnaire table 311.
The estimated GM-BHQ field stores the estimated GM-BHQ value of the second subject 107 on the questionnaire reception date stored in the questionnaire reception date field, which is estimated by the estimation calculation processing unit 305.
The estimated FA-BHQ field stores the estimated FA-BHQ value of the second subject 107 on the questionnaire reception date stored in the questionnaire reception date field, which is calculated by the estimation calculation processing unit 305.

In the intervention evaluation device 101, the first subject 102 in the learning phase and the second subject 107 in the estimation phase are treated as different subjects.
The first subject 102 in the learning phase has a record in the measurement BHQ table 310. The MRI imaging date, which is the value in the MRI imaging date field of the measurement BHQ table 310, is equal to the questionnaire reception date, which is the value in the questionnaire reception date field of the questionnaire table 311.
In contrast, the second subject 107 in the estimation phase does not have a record in the measurement BHQ table 310.

The intervention master 313 has an intervention ID field, an intervention implementer field, an intervention purpose field, an intervention content field, and other fields.
The intervention ID field stores an intervention ID that uniquely identifies an intervention.
The intervention implementer field stores the name of the intervention implementer who implements the intervention.
The intervention purpose field stores an action that is intended to improve BHQ through intervention. Specifically, there are four actions: exercise, learning, rest, and dialogue.
The intervention content field stores sentences that describe the specific content of the intervention.
The other field stores various information that cannot be written in the above fields, such as the period during which the intervention was performed, the content of the intervention, the person performing the intervention, and the like.

The intervention table 314 has a subject ID field, a pre-intervention questionnaire reception date field, a post-intervention questionnaire reception date field, and an intervention ID field.
The subject ID field is the same as the same name field in the subject master 308.
The pre-intervention questionnaire reception date field stores the pre-intervention schedule 108, which is the date on which the subject submitted the first questionnaire.
The post-intervention questionnaire reception date field stores the post-intervention date 110, which is the date on which the subject submitted the second questionnaire.
The intervention ID field is the same as the field with the same name in the intervention master 313.

There are a variety of possible interventions that interventionists believe can help improve brain health.
They can be food, luxury items, exercise equipment, facilities that encourage exercise, puzzles, etc.
The intervention ID is provided to uniquely identify the various interventions that such an intervention implementer performs on a subject.
That is, the intervention ID is identification information for identifying multiple subjects who have undergone a certain intervention.

A single subject may receive multiple interventions. Therefore, the subject ID and intervention ID have a one-to-many relationship.
A single intervention for a given subject always has a pre-intervention questionnaire reception date, which is before the intervention, and a post-intervention questionnaire reception date, which is after the intervention.

[First embodiment: Intervention evaluation device 101: BHQ calculation processing unit 303]
FIG. 5 is a functional block diagram showing the functions of the BHQ calculation processing section 303.
The BHQ calculation processing section 303 includes a gray matter amount calculation section 501, a nerve fiber anisotropy calculation section 502, and an average value calculation section 503.
The MRI image file group 104 is read into a gray matter amount calculation unit 501 and a nerve fiber anisotropy calculation unit 502.
The gray matter volume calculation unit 501 reads the MRI image file group 104 and outputs the gray matter volume data group 504.
The nerve fiber anisotropy calculation unit 502 reads the MRI image file group 104 and outputs the nerve fiber anisotropy data group 505.

In addition to the gray matter mass data group 504 and the nerve fiber anisotropy data group 505, the subject ID 506 and MRI imaging date 507 input from the operation unit 205 or the like are recorded in the brain information table 309.
Then, the average value calculation unit 503 calculates the average values of the gray matter mass data group 504 and the nerve fiber anisotropy data group 505, respectively, and outputs the measured GM-BHQ value and the measured FA-BHQ value of the first subject 102. .
These measured GM-BHQ values and measured FA-BHQ values of the first subject 102 are recorded in the measured BHQ table 310 together with the subject ID 506 and the MRI imaging date 507.

[First embodiment: Intervention evaluation device 101: Living environment questionnaire 601]
FIG. 6 is a diagram showing an example of a living environment questionnaire 601.
The living environment questionnaire 601 includes basic information about the subject, questions about the subject's health, questions about the subject's exercise, questions about the subject's dialogue, questions about the subject's diet, and questions about the subject's learning. It is being

The subject's basic information includes items such as date of birth, gender, height, weight, and body fat percentage.
After the date of birth is input into the intervention evaluation device 101, it is converted into the subject's age using the current date obtained from the RTC 207.
Gender is male or female, and is input as a value of 0 or 1 in the intervention evaluation device 101.
The height, weight, and body fat percentage are input to the intervention evaluation device 101 as they are.

Other items in the questionnaire regarding the subject's health, exercise, dialogue, diet, and learning are configured with 2 choices, 3 choices, 4 choices, 5 choices, or n choices, respectively.
For example, if there are two choices, a value of 0 or 1 is input to the intervention evaluation device 101.
In the case of three choices, the intervention evaluation device 101 inputs a value of 0, 1, or 2.
Similarly, in the case of n choices, values of 0, 1, . . . n-1 are input to the intervention evaluation device 101.

The learning calculation processing unit 304 uses a supervised learning algorithm, but the supervised learning algorithm only accepts numerical data as feature vectors and teacher data. Therefore, although the questions for each item in the living environment questionnaire 601 are written in natural language, all the answers are stored in the questionnaire table 311 as numerical data.

[First embodiment: intervention evaluation device 101: learning phase]
FIG. 7 is a flowchart showing the overall flow of operations in the learning phase of the intervention evaluation device 101.
When the process is started (S701), first, an MRI operator (not shown) performs MRI imaging of the first subject 102 to obtain an MRI image file group 104 of the first subject 102 (S702).
Next, the input/output control unit 301 provides the MRI image file group 104 of the first subject 102 to the BHQ calculation processing unit 303. The BHQ calculation processing unit 303 calculates the GM-BHQ and FA-BHQ of the first subject 102 from the MRI image file group 104 of the first subject 102, and stores them in the measurement BHQ table 310 (S703).

Next, the input/output control unit 301 converts the questionnaire of the first subject 102 into data and stores it in the questionnaire table 311 (S704).
Then, the input/output control unit 301 reads out the questionnaire contents of the first subject 102 from the questionnaire table 311 and provides it to the learning calculation processing unit 304 . Similarly, the input/output control unit 301 reads the GM-BHQ and FA-BHQ of the first subject 102 from the measurement BHQ table 310 and provides them to the learning calculation processing unit 304. The learning calculation processing unit 304 executes a learning process using the questionnaire contents of the first subject 102 as a feature vector and the GM-BHQ and FA-BHQ of the first subject 102 as teacher data to generate the approximate function parameters 106, or Update (S705).
Then, the series of processing ends (S706).

[First embodiment: intervention evaluation device 101: estimation phase]
FIG. 8 is a flowchart showing the overall flow of operations in the estimation phase of the intervention evaluation device 101.
When the process starts (S801), the input/output control unit 301 first searches the intervention table 314 using the intervention ID that specifies the intervention to be evaluated, and finds the subject ID of the second subject 107 to be evaluated. Extract and list them (S802).
Next, the input/output control unit 301 receives the pre-intervention living environment questionnaire results 109 of the second subject 107 on the pre-intervention date 108 before the intervention and the post-intervention living environment questionnaire results 111 on the post-intervention date 110 after the intervention. Each is converted into data and stored in the questionnaire table 311 (S803).

Next, the input/output control unit 301 extracts the pre-intervention living environment questionnaire results 109 and the post-intervention living environment questionnaire results 111 of the second subject 107 from the questionnaire table 311 and provides them to the estimation calculation processing unit 305 . The estimation calculation processing unit 305 executes estimation processing using the approximation function parameters 106 using the questionnaire contents of the second subject 107 as a feature vector, and calculates the first estimated GM-BHQ value and estimated FA-BHQ of the second subject 107. value, the second estimated GM-BHQ value, and the estimated FA-BHQ value (S804).
The above estimation calculation process is executed for a plurality of second subjects 107.

Next, the average/deviation calculation processing unit 306 groups the estimated BHQ values obtained in step S804 by age and gender of the plurality of second subjects 107. Then, the average/deviation calculation processing unit 306 calculates the estimated GM-BHQ average value, the estimated FA-BHQ average value, the estimated GM-BHQ standard deviation, and the estimated FA-BHQ standard deviation for each age and gender. The calculation is performed using the first schedule and the second schedule and output to a predetermined storage medium or the like (S805).

Finally, the t-test calculation processing unit 307 calculates the estimated GM-BHQ average value, the estimated FA-BHQ average value, the estimated GM-BHQ standard deviation, and the estimated FA-BHQ standard deviation for each age and gender. The p value for each is calculated and output to a predetermined storage medium or the like (S806).
Then, the series of processing ends (S807).

[First embodiment: intervention evaluation device 101: learning calculation processing unit 304]
FIG. 9 is a block diagram showing details of data processing by the learning calculation processing unit 304. This process corresponds to step S705 in FIG. Note that in FIGS. 9, 10, and 11, an input/output control unit 301 intervenes in data exchange between each functional block and in searches in various tables.

The questionnaire table 311 includes records of a plurality of questionnaire reception dates for a single subject ID. Therefore, records are identified by the questionnaire reception date.
Similarly, the measurement BHQ table 310 includes records of multiple MRI imaging dates for a single subject ID. Therefore, records are identified by the MRI imaging date.
Since this questionnaire reception date is equal to the MRI shooting date of the measurement BHQ table 310, the questionnaire table 311 and the measurement BHQ table 310 are linked by the subject ID specified from the subject master 308 and the questionnaire reception date = MRI shooting date. ing.

The age, gender, etc. of the first subject 102 outputted from the subject master 308 and each response item of the first subject 102 on a specific questionnaire reception date outputted from the questionnaire table 311 are input to the learning calculation processing unit 304 as a feature vector. be done.
The measured GM-BHQ value and measured FA-BHQ value of the first subject 102 on a specific MRI imaging date, which are output from the measured BHQ table 310, are input to the learning calculation processing unit 304 as training data.
The learning calculation processing unit 304 executes learning calculation processing using, for example, a supervised learning algorithm using SVM (Support Vector Machine), and generates or updates the approximate function parameters 106.

[First embodiment: intervention evaluation device 101: estimation calculation processing unit 305]
FIG. 10 is a block diagram showing details of data processing by the estimation calculation processing unit 305. This process corresponds to step S804 in FIG.
First, the input/output control unit 301 searches the intervention table 314 using the intervention ID 1001 that specifies the intervention to be evaluated, and lists the corresponding subject ID 506. At this time, the input/output control unit 301 also extracts the pre-intervention schedule 108 and the post-intervention schedule 110 from the intervention table 314.

In the questionnaire table 311, for a single subject ID 506, there are records of a plurality of questionnaire reception dates: a record of the pre-intervention schedule 108 and a record of the post-intervention schedule 110.
The age, gender, etc. of the second subject 107 outputted from the subject master 308 and each answer item of the second subject 107 in the pre-intervention schedule 108 outputted from the questionnaire table 311 are subjected to estimation calculation processing by the input/output control unit 301. It is input to section 305 as a feature vector.
The estimation calculation processing unit 305 executes estimation calculation processing using a supervised learning algorithm that refers to the approximate function parameter 106, and uses the first estimated GM- as the pre-intervention estimated BHQ 1002 corresponding to the input feature vector. Output the BHQ value and the first estimated FA-BHQ value.

Similarly, the age, gender, etc. of the second subject 107 outputted from the subject master 308 and each response item of the second subject 107 on the post-intervention schedule 110 outputted from the questionnaire table 311 are controlled by the input/output control unit 301. It is input to the estimation calculation processing unit 305 as a feature vector.
The estimation calculation processing unit 305 executes estimation calculation processing using a supervised learning algorithm that refers to the approximate function parameter 106, and uses the second estimated GM- as the post-intervention estimated BHQ 1003 corresponding to the input feature vector. Output the BHQ value and the second estimated FA-BHQ value.

The estimation calculation processing unit 305 outputs the pre-intervention estimated BHQ 1002 and the post-intervention estimated BHQ 1003 to the subject IDs 506 of the plurality of second subjects 107 corresponding to the intervention ID 1001 listed from the intervention table 314.
The above subject ID 506, pre-intervention schedule 108, estimated pre-intervention BHQ 1002, post-intervention schedule 110, and estimated post-intervention BHQ 1003 are stored in the estimated BHQ table 312 by the input/output control unit 301.

In the estimated BHQ table 312, the pre-intervention schedule 108 and the post-intervention schedule 110 are stored as different records in the questionnaire reception date field. To find out which record in the estimated BHQ table 312 corresponds to the pre-intervention schedule 108 or which record in the estimated BHQ table 312 corresponds to the post-intervention schedule 110, search the intervention table 314 using the subject ID and intervention ID. This can be determined from the values in the pre-intervention questionnaire reception date field and the post-intervention questionnaire reception date field of the hit record.

[First embodiment: intervention evaluation device 101: average/deviation calculation processing unit 306 and t-test calculation processing unit 307]
FIG. 11 is a block diagram showing details of data processing by the average/deviation calculation processing unit 306 and the t-test calculation processing unit 307. This processing corresponds to step S805 and step S806 in FIG. 8, respectively.
First, the input/output control unit 301 searches the intervention table 314 using the intervention ID 1001 that specifies the intervention to be evaluated, and lists the corresponding subject ID 506. At this time, the input/output control unit 301 also extracts the pre-intervention schedule 108 and the post-intervention schedule 110 from the intervention table 314.

In the estimated BHQ table 312, for a single subject ID 506, there are records of a plurality of questionnaire reception dates: a record for the pre-intervention schedule 108 and a record for the post-intervention schedule 110.
The input/output control unit 301 extracts the pre-intervention estimated BHQ 1002 of the second subject 107 on the pre-intervention schedule 108 from the estimated BHQ table 312, and calculates the average and deviation together with the age and gender of the second subject 107 extracted from the subject master 308. The information is input to the processing unit 306.

The average/deviation calculation processing unit 306 groups the listed subject IDs 506 corresponding to the intervention ID 1001 according to the age and gender of the second subject 107 output from the subject master 308, and then calculates the pre-intervention estimated BHQ 1002. , calculate the average value and standard deviation of the first estimated GM-BHQ value and the first estimated FA-BHQ value.
This is the pre-intervention estimated BHQ mean value and standard deviation for each age and gender, 1101.

Similarly, the input/output control unit 301 extracts the estimated post-intervention BHQ 1003 of the second subject 107 on the post-intervention date 110 from the estimated BHQ table 312 and averages - Input to the deviation calculation processing section 306.
The average/deviation calculation processing unit 306 groups the listed subject IDs 506 corresponding to the intervention ID 1001 according to the age and gender of the second subject 107 output from the subject master 308, and then calculates the post-intervention estimated BHQ 1003. The average value and standard deviation of the second estimated GM-BHQ value and the second estimated FA-BHQ value are calculated.
This is the estimated post-intervention BHQ mean value and standard deviation for each age and gender, 1102.

The input/output control unit 301 calculates the average value and standard deviation of the first estimated GM-BHQ value and the first estimated FA-BHQ value, and the second estimated GM-BHQ value and The average value and standard deviation of the second estimated FA-BHQ value are input to the t-test calculation processing unit 307.
The t-test calculation processing unit 307 calculates the average value and standard deviation of the first estimated GM-BHQ value and the first estimated FA-BHQ value, and the second estimated GM-BHQ value and the second estimated FA-BHQ. Based on the mean value and standard deviation of the values, a hypothesis test for two corresponding samples is processed to obtain a test statistic. Then, the t-test calculation processing unit 307 converts the t-value, which is a type of test statistic, into a p-value, which is the probability of statistical hypothesis testing. This is the p value of 1103 for the estimated BHQ for each age and gender.

The input/output control unit 301 calculates the estimated pre-intervention BHQ average value and standard deviation 1101 for each age and gender, the estimated post-intervention BHQ average value and standard deviation 1102 for each age and gender, and the p value 1103 of the estimated BHQ for each age and gender. Output to nonvolatile storage 206 or the like. The pre-intervention estimated BHQ average value and standard deviation 1101 for each age and gender, the estimated post-intervention BHQ average value and standard deviation 1102 for each age and gender, and the p value 1103 of estimated BHQ for each age and gender are the interventions corresponding to the intervention ID. used to evaluate the effectiveness of

The p-value calculated and output by the t-test calculation processing unit 307 is obtained by statistically comparing the estimated BHQ values of the plurality of second subjects 107 to whom the intervention is performed before and after the intervention. It is a value. Therefore, in the intervention evaluation device 101 according to the first embodiment of the present invention, the sample of the t-test executed by the t-test calculation processing unit 307 is the group of second subjects 107 identified by one intervention ID. These are two paired samples, the estimated BHQ values before and after the intervention, obtained from the same subject. Then, the t-test calculation processing unit 307 calculates the estimated BHQ values before and after the intervention in the same subject, the group of second subjects 107 identified by one intervention ID. Perform a t-test on samples.

GM-BHQ and FA-BHQ disclosed in Patent Document 1 tend to differ greatly depending on the age and gender of the subject.
In general, gray matter volume and nerve fiber anisotropy in the brain tend to decrease with age. The bias of GM-BHQ and FA-BHQ also differs greatly by gender. If an attempt was made to statistically evaluate the intervention using the second subjects 107 of all ages and genders, the standard deviations of the estimated GM-BHQ values and estimated FA-BHQ values would become too large, making it difficult to statistically evaluate the intervention. The accuracy of the evaluation will drop significantly.
Therefore, in the intervention evaluation device according to the present invention, the second subjects 107 are grouped by age group and gender, and statistical evaluation is performed within these groups, thereby suppressing data dispersion and keeping the standard deviation small. This increases the accuracy of statistical evaluation.

As described above, when the intervention evaluation device 101 according to the first embodiment of the present invention inputs the questionnaire answers before and after the intervention by the second subject 107, the estimated GM-BHQ value and the estimated FA of the second subject 107 before and after the intervention are input. - Statistically significant differences in BHQ values can be obtained. Then, it becomes possible to objectively evaluate the effectiveness of the intervention from statistically significant differences such as p-values.

[Second embodiment: intervention evaluation device 101: overall configuration]
The intervention evaluation device 101 according to the first embodiment of the present invention objectively determined the effectiveness of the intervention by estimating the BHQ value from the questionnaire and calculating a statistically significant difference from the change in the estimated BHQ value. . At the time of the determination, the intervention evaluation device 101 outputs the estimated BHQ value based on the learning algorithm.

Let's assume that the difference value for each answer item between the first and second questionnaires is used as a feature vector, and the difference between the first measured BHQ value and the second measured BHQ value is used as training data to perform a learning calculation. If the processing unit 304 can be made to learn, it can be expected that the effectiveness of the intervention will be determined more accurately.

In order to realize the estimation calculation process of estimating the difference in BHQ value when the difference between the questionnaires of the first questionnaire and the second questionnaire is input, the premise is that the difference between the questionnaires is used as a feature vector for the learning calculation. It is necessary to subject it to processing. Furthermore, it is also necessary to provide the difference between the measured BHQ values as training data to the learning calculation process.

Therefore, in the intervention evaluation device 1201 according to the second embodiment of the present invention, which will be described from now on, in the learning phase of the intervention evaluation device 1201, the first subject who is to provide learning data to the intervention evaluation device 1201 The subject 102 receives the intervention in the same way as the second subject 107 in the estimation phase. Then, on a pre-intervention date 1202 immediately before starting the intervention, an MRI scan is performed and the living environment questionnaire 601 is answered, and on a post-intervention date 1205 immediately after the intervention, an MRI scan is performed again and the living environment questionnaire 601 is completed. Make sure to answer.

FIG. 12A is a schematic diagram showing the learning phase in the pre-intervention stage of the intervention evaluation device 1201 according to the second embodiment of the present invention.
First, the first subject 102 answers the first living environment questionnaire 601 in a pre-intervention schedule 1202, which is a stage immediately before implementing the intervention, and creates a pre-intervention living environment questionnaire result 1203. Then, in this pre-intervention schedule 1202, the MRI apparatus 103 photographs the brain of the first subject 102 and outputs a pre-intervention MRI image file group 1204. That is, the intervention evaluation device 1201 takes in the subject's pre-intervention MRI image file group 1204 and the pre-intervention living environment questionnaire results 1203 in the pre-intervention schedule 1202.
The first subject 102 then performs the intervention from the pre-intervention schedule 1202 to the post-intervention schedule 1205, which is a predetermined period of time.

FIG. 12B is a schematic diagram showing the learning phase in the post-intervention stage of the intervention evaluation device 1201 according to the second embodiment of the present invention.
The first subject 102 answers the second living environment questionnaire 601 on a post-intervention date 1205, which is a stage immediately after implementing the intervention, and creates a post-intervention living environment questionnaire result 1206. Then, on this post-intervention schedule 1205, the MRI apparatus 103 again images the brain of the first subject 102 and outputs a post-intervention MRI image file group 1207. That is, the intervention evaluation device 1201 takes in the subject's post-intervention MRI image file group 1207 and the post-intervention living environment questionnaire results 1206 on the post-intervention schedule 1205.

The intervention evaluation device 1201 performs a learning process using the pre-intervention living environment questionnaire results 1203, the pre-intervention MRI image file group 1204, the post-intervention living environment questionnaire results 1206, and the post-intervention MRI image file group 1207. As a result, difference approximation function parameters 1208 are generated or updated.

In the intervention evaluation device 101 according to the first embodiment shown in FIG. 1, the first subject 102 in the learning phase sends the MRI image file group 104 obtained by MRI imaging and the living environment questionnaire results 105 to the intervention evaluation device 1201, unrelated to the intervention. was provided to.
On the other hand, in the intervention evaluation device 1201 according to the second embodiment, the first subject 102 in the learning phase is in a predetermined state, similar to the second subject 107 in the estimation phase in the intervention evaluation device 1201 according to the first embodiment. receive intervention.

For this reason, the pre-intervention living environment questionnaire results 1203 and the pre-intervention MRI image file group 1204 on the pre-intervention date 1202 immediately before receiving the intervention, and the post-intervention date 1205 after the intervention has been continued for a predetermined period of time. Post-intervention living environment questionnaire results 1206 and post-intervention MRI image file group 1207 are required in the learning phase.

FIG. 13 is a schematic diagram showing the estimation phase of the intervention evaluation device 1201 according to the second embodiment of the present invention.
The intervention evaluation device 1201 takes in the pre-intervention living environment questionnaire results 109 on the pre-intervention schedule 108 and the post-intervention living environment questionnaire results 111 on the post-intervention schedule 110 of the second subject 107, which are different from the time points shown in FIGS. 12A and 12B. , performs estimation processing. As a result, an estimated BHQ difference corresponding to the difference between the pre-intervention living environment questionnaire result 109 and the post-intervention living environment questionnaire result 111 is obtained.

The intervention evaluation device 1201 executes this estimated BHQ difference estimation calculation process for the plurality of second subjects 107, and calculates the estimated BHQ difference average value and estimated BHQ difference standard deviation for each age and gender of the second subjects 107. calculate.
The estimated BHQ difference average value is a set of the estimated GM-BHQ difference average value and the estimated FA-BHQ difference average value for each age and gender of the second subject 107.
The estimated BHQ standard deviation is a set of the standard deviation of the estimated GM-BHQ difference and the standard deviation of the estimated FA-BHQ difference for each age and gender of the second subject 107.

Furthermore, the intervention evaluation device 1201 calculates in advance the estimated GM-BHQ difference average value, the estimated FA-BHQ difference average value, the estimated GM-BHQ difference standard deviation, and the estimated FA-BHQ difference standard deviation for each age and gender. From the estimated GM-BHQ difference mean value and estimated FA-BHQ difference mean value without any intervention, and the estimated GM-BHQ difference standard deviation and estimated FA-BHQ difference standard deviation, the estimated GM-BHQ difference by age and gender is calculated. The p value and the p value of the estimated FA-BHQ difference are calculated and output to the nonvolatile storage 206 or the like.
This is the estimated BHQ difference average value, standard deviation, and p value 1301 for each age and gender.

[Second embodiment: intervention evaluation device 1201: entire software functions]
The hardware configuration of the intervention evaluation device 1201 according to the second embodiment of the present invention is the same as that of the first embodiment shown in FIG. 2, so a description thereof will be omitted.
FIG. 14 is a block diagram showing the overall software functions of the intervention evaluation device 1201 according to the second embodiment of the present invention.

The differences between the intervention evaluation device 1201 shown in FIG. 14 and the intervention evaluation device 101 according to the first embodiment of the present invention shown in FIG. 3 are listed below.
(1) Instead of the MRI image file group 104, the pre-intervention MRI image file group 1204 and the post-intervention MRI image file group 1207 are input to the input/output control unit 1401.
(2) The questionnaire response group 302 includes pre-intervention living environment questionnaire results 1203 and post-intervention living environment questionnaire results 1206 instead of the living environment questionnaire results 105 in the learning mode.
(3) A difference learning calculation processing unit 1402, a difference estimation calculation processing unit 1403, a difference approximation function parameter 1208, an estimated BHQ difference table 1404, and a measured BHQ difference table 1405 are added.

The differential learning calculation processing unit 1402 takes in the pre-intervention living environment questionnaire results 1203 on the pre-intervention schedule 1202 and the post-intervention living environment questionnaire results 1206 on the post-intervention schedule 1205 of the first subject 102, and calculates the difference between the questionnaire results. Then, the difference learning calculation processing unit 1402 executes learning calculation processing using the BHQ difference as teacher data and the questionnaire result difference as a feature vector, and generates or updates the difference approximation function parameter 1208.

In addition, the difference estimation calculation processing unit 1403 takes in the pre-intervention living environment questionnaire results 109 on the pre-intervention schedule 108 of the second subject 107 and the post-intervention living environment questionnaire results 111 on the post-intervention schedule 110, and calculates the difference between the questionnaire results. do. Then, the difference estimation calculation processing unit 1403 uses the questionnaire result difference as a feature vector, refers to the difference approximation function parameter 1208, executes estimation calculation processing, and outputs the estimated BHQ difference.

FIG. 15 is a table showing the field configurations of the estimated BHQ difference table 1404 and the measured BHQ difference table 1405.
Estimated BHQ difference table 1404 has a subject ID field, an intervention ID field, an estimated GM-BHQ difference field, and an estimated FA-BHQ difference field.
The subject ID field is the same as the same name field in the subject master 308.
The intervention ID field is the same as the field with the same name in the intervention table 314.
The estimated GM-BHQ difference field stores the estimated GM-BHQ difference value of the subject in the questionnaire difference estimated by the difference estimation calculation processing unit 1403.
The estimated FA-BHQ difference field stores the estimated FA-BHQ difference value of the subject in the questionnaire difference calculated by the difference estimation calculation processing unit 1403.

The measurement BHQ difference table 1405 has a subject ID field, a pre-intervention MRI imaging date field, a post-intervention MRI imaging date field, a measurement GM-BHQ difference field, and a measurement FA-BHQ difference field.
The subject ID field is the same as the same name field in the subject master 308.
The pre-intervention MRI imaging date field stores the MRI imaging date before intervention, that is, the pre-intervention schedule 1202.
The post-intervention MRI imaging date field stores the post-intervention MRI imaging date, that is, the post-intervention date 1205.

The measured GM-BHQ difference field contains the GM-BHQ value of the first subject 102 on the post-intervention date 1205 to the GM-BHQ value of the first subject 102 on the pre-intervention date 1202, which is calculated by the BHQ difference calculation processing unit described later. The measured GM-BHQ difference value of the first subject 102 after subtracting is stored.
The estimated FA-BHQ difference field contains the FA-BHQ value of the first subject 102 on the pre-intervention date 1202 from the FA-BHQ value of the first subject 102 on the post-intervention date 1205, which is calculated by the BHQ difference calculation processing unit described later. The measured FA-BHQ difference value of the first subject 102 after subtracting is stored.

Note that, as can be seen by comparing the measured BHQ difference table 1405 and the estimated BHQ difference table 1404, the measured BHQ difference table 1405 includes a pre-intervention MRI imaging date field and a post-intervention MRI imaging date field instead of the intervention ID field. ing. These may be replaced with intervention ID fields in the same way as the estimated BHQ difference table 1404. In that case, the difference learning calculation processing unit 1402 links the measurement BHQ difference table 1405 with the intervention table 314 using the intervention ID, and sets the value of the pre-intervention questionnaire reception date field of the intervention table 314 as the pre-intervention MRI imaging date. The value of the post-intervention questionnaire reception date field will be imported as the post-intervention MRI imaging date.

FIG. 16 is a flowchart showing the overall flow of operations in the learning phase of the intervention evaluation device 1201.
When the process is started (S1601), the input/output control unit 1401 first registers an intervention ID in the intervention table 314 for the subject ID to be processed (S1602).
Next, an MRI operator (not shown) performs MRI imaging of the first subject 102 on the pre-intervention schedule 1202 to obtain a pre-intervention MRI image file group 1204 of the first subject 102 (S1603).
Next, the input/output control unit 1401 provides the pre-intervention MRI image file group 1204 of the first subject 102 to the BHQ calculation processing unit 303. The BHQ calculation processing unit 303 calculates the first GM-BHQ and first FA-BHQ of the first subject 102 from the pre-intervention MRI image file group 1204 of the first subject 102, and stores them in the measurement BHQ table 310. (S1604).
In addition, the input/output control unit 1401 converts the pre-intervention living environment questionnaire results 1203 in the pre-intervention schedule 1202 of the first subject 102 into data and stores it in the questionnaire table 311 (S1605).

Next, an MRI operator (not shown) performs MRI imaging of the first subject 102 on a post-intervention schedule 1205 to obtain a post-intervention MRI image file group 1207 of the first subject 102 (S1606).
Next, the input/output control unit 1401 provides the post-intervention MRI image file group 1207 of the first subject 102 to the BHQ calculation processing unit 303. The BHQ calculation processing unit 303 calculates the second GM-BHQ and second FA-BHQ of the first subject 102 from the post-intervention MRI image file group 1207 of the first subject 102, and stores them in the measurement BHQ table 310. (S1607).
In addition, the input/output control unit 1401 converts the post-intervention living environment questionnaire results 1206 of the first subject 102 on the post-intervention schedule 1205 into data, and stores it in the questionnaire table 311 (S1608).

Next, the difference learning calculation processing unit 1402 reads out the second GM-BHQ and second FA-BHQ of the first subject 102 from the measurement BHQ table 310, and sends them to the BHQ difference calculation processing unit 1802 (see FIG. 18). input. The BHQ difference calculation processing unit 1802 subtracts the first GM-BHQ and the first FA-BHQ from the second GM-BHQ and second FA-BHQ of the first subject 102, respectively, to obtain GM-BHQ. The difference and the FA-BHQ difference are acquired (S1609).

Next, the difference learning calculation processing unit 1402 reads out the pre-intervention living environment questionnaire results 1203 and the post-intervention living environment questionnaire results 1206 of the first subject 102 from the questionnaire table 311, and the questionnaire difference calculation processing unit 1801 (see FIG. 18) give to The questionnaire difference calculation processing unit 1801 subtracts each item of the pre-intervention living environment questionnaire result 1203 from each item of the post-intervention living environment questionnaire result 1206 to obtain a questionnaire difference (S1610).

Then, the learning calculation processing unit 304 uses the questionnaire difference of the first subject 102 obtained from the questionnaire difference calculation processing unit 1801 and the age, gender, etc. obtained from the subject master 308 as feature vectors. Using the GM-BHQ difference and FA-BHQ difference of the first subject 102 obtained from the above as training data, the difference learning calculation processing unit 1402 executes learning processing to generate or update the difference approximation function parameter 1208 ( S1611).
Then, the series of processing ends (S1612).

FIG. 17 is a flowchart showing the overall operation flow in the estimation phase of the intervention evaluation device 1201.
When the process starts (S1701), the input/output control unit 1401 first searches the intervention table 314 using the intervention ID that specifies the intervention to be evaluated, and finds the subject ID of the second subject 107 to be evaluated. are extracted and listed (S1702).
Next, the input/output control unit 1401 outputs the pre-intervention living environment questionnaire results 109 of the second subject 107 on the pre-intervention date 108 before the intervention, and the post-intervention living environment questionnaire results 111 on the post-intervention date 110 after the intervention. are converted into data and stored in the questionnaire table 311 (S1703).

Next, the difference estimation calculation processing unit 1403 extracts the pre-intervention living environment questionnaire result 109 and the post-intervention living environment questionnaire result 111 of the second subject 107 from the questionnaire table 311, and provides them to the questionnaire difference calculation processing unit 1801. The questionnaire difference calculation processing unit 1801 subtracts each item of the pre-intervention living environment questionnaire result 109 from each item of the post-intervention living environment questionnaire result 111 of the second subject 107, and outputs the questionnaire difference (S1704).

Next, the difference estimation calculation processing unit 1403 gives the questionnaire difference to the estimation calculation processing unit 305. The estimation calculation processing unit 305 executes estimation processing using the difference approximation function parameter 1208 using the questionnaire difference of the second subject 107 as a feature vector, and calculates the estimated GM-BHQ difference value and the estimated FA-BHQ difference of the second test subject 107. The value is output (S1705).
The above difference estimation calculation process is executed for all the second subjects 107 listed in step S1702.

Next, the average/deviation calculation processing unit 306 groups the estimated BHQ differences obtained in step S1705 by age and gender of the plurality of second subjects 107. Then, the average/deviation calculation processing unit 306 estimates the estimated GM-BHQ difference average value, the estimated FA-BHQ difference average value, and the estimated GM-BHQ difference standard deviation for each age and gender of the second subject 107. The FA-BHQ difference standard deviation is calculated and output to a predetermined storage medium or the like (S1706).

Finally, the t-test calculation processing unit 307 calculates the estimated GM-BHQ difference average value, the estimated FA-BHQ difference average value, and the estimated GM- BHQ difference standard deviation, estimated FA-BHQ difference standard deviation, and pre-calculated estimated GM-BHQ difference mean value, estimated FA-BHQ difference mean value, and estimated GM for each age and gender in the subject group without intervention. -BHQ difference standard deviation and estimated FA-BHQ difference standard deviation, p values for each age and gender are calculated and output to a predetermined storage medium or the like (S1707).
Then, the series of processing ends (S1708).

FIG. 18 is a block diagram showing details of data processing by the differential learning calculation processing unit 1402. This process corresponds to steps S1609, S1610, and S1611 in FIG. Note that in FIGS. 18, 19, and 20, an input/output control unit 1401 intervenes in data exchange between each functional block.
First, the differential learning calculation processing unit 1402 searches the intervention table 314 using the intervention ID 1001 that specifies the intervention to be learned, and lists the corresponding subject ID 506. At this time, the differential learning calculation processing unit 1402 also extracts a pre-intervention schedule 1202 and a post-intervention schedule 1205, which are the questionnaire reception dates, from the intervention table 314.

The questionnaire table 311 includes records of a plurality of questionnaire reception dates for a single subject ID 506. Therefore, the differential learning calculation processing unit 1402 identifies a record in the questionnaire table 311 based on the pre-intervention schedule 1202 or the post-intervention schedule 1205, which is the questionnaire reception date.
Similarly, the measurement BHQ table 310 includes records of multiple MRI imaging dates for a single subject ID 506. Therefore, the differential learning calculation processing unit 1402 identifies a record in the measurement BHQ table 310 using the pre-intervention schedule 1202 or the post-intervention schedule 1205, which is the MRI imaging date.

That is, since the questionnaire reception date is equal to the MRI shooting date of the measurement BHQ table 310, the questionnaire table 311 and the measurement BHQ table 310 are linked by the subject ID 506 specified from the subject master 308 and the questionnaire reception date = MRI shooting date. There is.

The differential learning calculation processing unit 1402 searches the questionnaire table 311 using the subject ID 506, pre-intervention schedule 1202, and post-intervention schedule 1205 listed from the intervention table 314 previously. The differential learning calculation processing unit 1402 then calculates the pre-intervention living environment questionnaire results 1203 conducted on the pre-intervention date 1202 and the post-intervention lifestyle conducted on the post-intervention date 1205 for the subject ID 506 of the first subject 102 from the questionnaire table 311. Obtain environmental questionnaire results 1206. Each answer item of the pre-intervention living environment questionnaire result 1203 and each answer item of the post-intervention living environment questionnaire result 1206 are input to the questionnaire difference calculation processing unit 1801.

The questionnaire difference calculation processing unit 1801 creates a questionnaire difference by subtracting the value of the answer item of the pre-intervention living environment questionnaire result 1203 from the value of the answer item of the post-intervention living environment questionnaire result 1206 for each answer item of the questionnaire. do.

The differential learning arithmetic processing unit 1402 searches the measurement BHQ table 310 using the subject ID 506, pre-intervention schedule 1202, and post-intervention schedule 1205 listed previously from the intervention table 314. Then, from the measured BHQ table 310, the measured BHQ value for the pre-intervention schedule 1202 and the measured BHQ value for the post-intervention schedule 1205 for the subject ID 506 of the first subject 102 are acquired.

The first measured GM-BHQ value and the first measured FA-BHQ value, which are the measured BHQ values on the pre-intervention schedule 1202 for the subject ID 506 of the first subject 102, obtained from the measured BHQ table 310, and the post-intervention schedule 1205 The second measured GM-BHQ value and the second measured FA-BHQ value, which are the measured BHQ values at , are input to the BHQ difference calculation processing unit 1802 .
The BHQ difference calculation processing unit 1802 subtracts the first measured GM-BHQ value from the second measured GM-BHQ value and outputs the measured GM-BHQ difference. Similarly, the BHQ difference calculation processing unit 1802 subtracts the first measured FA-BHQ value from the second measured FA-BHQ value, and outputs the measured FA-BHQ difference.

The differential learning calculation processing unit 1402 acquires the age, gender, etc. of the first subject 102 from the subject master 308. The age, gender, etc. of the first subject 102 and each item of the questionnaire difference output from the questionnaire difference calculation processing section 1801 are inputted to the learning calculation processing section 304 as a feature vector.
The measured GM-BHQ difference and the measured FA-BHQ difference of the first subject 102 output from the BHQ difference calculation processing unit 1802 are input to the learning calculation processing unit 304 as teacher data.
The learning calculation processing unit 304 executes learning calculation processing using, for example, a supervised learning algorithm using SVM (Support Vector Machine), and generates or updates the difference approximation function parameters 1208.

FIG. 19 is a block diagram showing details of data processing by the difference estimation calculation processing unit 1403. This process corresponds to step S1705 in FIG.
First, the difference estimation calculation processing unit 1403 searches the intervention table 314 using the intervention ID 1001 that specifies the intervention to be evaluated, and lists the corresponding subject ID 506. At this time, the difference estimation calculation processing unit 1403 also extracts the pre-intervention schedule 108 and the post-intervention schedule 110, which are the questionnaire reception dates, from the intervention table 314.

The questionnaire table 311 includes records of a plurality of questionnaire reception dates for a single subject ID 506. Therefore, the difference estimation calculation processing unit 1403 identifies a record in the questionnaire table 311 based on the pre-intervention schedule 108 or the post-intervention schedule 110, which is the questionnaire reception date.

The difference estimation calculation processing unit 1403 searches the questionnaire table 311 using the subject ID 506, pre-intervention schedule 108, and post-intervention schedule 110 listed previously from the intervention table 314. Then, from the questionnaire table 311, the pre-intervention living environment questionnaire results 109 conducted on the pre-intervention date 108 and the post-intervention living environment questionnaire results 111 conducted on the post-intervention date 110 are obtained for the subject ID 506 of the second subject 107. Each answer item of the pre-intervention living environment questionnaire result 109 and each answer item of the post-intervention living environment questionnaire result 111 are input to the questionnaire difference calculation processing unit 1801.

The questionnaire difference calculation processing unit 1801 creates a questionnaire difference by subtracting the value of the answer item of the pre-intervention living environment questionnaire result 109 from the value of the answer item of the post-intervention living environment questionnaire result 111 for each answer item of the questionnaire. do.

The difference estimation calculation processing unit 1403 acquires the age, gender, etc. of the second subject 107 from the subject master 308. Then, the age, gender, etc. of the second subject 107 and each item of the questionnaire difference outputted from the questionnaire difference calculation processing section 1801 are inputted to the estimation calculation processing section 305 as a feature vector.

The estimation calculation processing unit 305 executes estimation calculation processing using a supervised learning algorithm that refers to the difference approximation function parameter 1208, and calculates the estimated GM-BHQ difference and the estimated BHQ difference corresponding to the input feature vector. Output the estimated FA-BHQ difference.

Furthermore, the estimation calculation processing unit 305 outputs the estimated GM-BHQ difference and the estimated FA-BHQ difference to the listed subject ID 506 corresponding to the intervention ID 1001.
Through the above processing, the subject ID 506, intervention ID 1001, pre-intervention estimated BHQ 1002, and post-intervention estimated BHQ 1003 are stored in the estimated BHQ difference table 1404.

FIG. 20 is a block diagram showing details of data processing by the average/deviation calculation processing section 306 and the t-test calculation processing section 307. These processes correspond to step S1706 and step S1707 in FIG.
First, the input/output control unit 1401 searches the intervention table 314 using the intervention ID 1001 that specifies the intervention to be evaluated, and lists the corresponding subject ID 506.
The estimated BHQ difference table 1404 includes records of a plurality of intervention IDs 1001 for a single subject ID 506. Therefore, the input/output control unit 1401 identifies a record in the estimated BHQ difference table 1404 using the subject ID 506 and the intervention ID 1001.
The estimated BHQ difference of the second subject 107 with the intervention ID 1001 outputted from the estimated BHQ table 312 is subjected to mean/deviation calculation processing by the input/output control unit 1401 along with the age and gender of the second subject 107 outputted from the subject master 308. The information is input to section 306 .

The average/deviation calculation processing unit 306 groups the listed subject IDs 506 corresponding to the intervention ID 1001 according to the age and gender of the second subject 107 output from the subject master 308, and then calculates the estimated BHQ difference, which is the estimated BHQ difference. The average value and standard deviation of the GM-BHQ difference and the estimated FA-BHQ difference are calculated.
This is the estimated BHQ difference average value and standard deviation 2001 for each age and gender.

The estimated GM-BHQ difference and the estimated FA-BHQ difference average value and standard deviation calculated by the average/deviation calculation processing unit 306 described above are data of the plurality of second subjects 107 who performed the intervention.
Note that the average value and standard deviation of the estimated GM-BHQ difference and the estimated FA-BHQ difference do not have anything to compare with, so the mean/deviation calculation processing unit 306 cannot operate the t-test calculation processing unit 307. .

On the other hand, the measured BHQ table 310 stores measured GM-BHQ values and measured FA-BHQ values of a large number of first subjects 102. These records also include data on the first subject 102 who regularly underwent MRI imaging, regardless of the intervention. That is, the average/deviation calculation processing unit 306 can create a difference between the measured BHQ values of the first subject 102 without intervention.

Therefore, the input/output control unit 1401 selects in advance records without intervention from the measurement BHQ table 310, focusing on records to which the intervention ID 1001 recorded in the intervention table 314 is not attached. Then, the input/output control unit 1401 obtains the first measured GM-BHQ value and the first measured GM-BHQ value for the first schedule corresponding to the pre-intervention schedule 1202 from two records with the same subject ID and different MRI imaging dates. , the second measured GM-BHQ value and the second measured FA-BHQ value related to the second schedule corresponding to the post-intervention schedule 1205. The input/output control unit 1401 then causes the BHQ difference calculation processing unit 1802 to read these measured BHQ value groups.

The BHQ difference calculation processing unit 1802 subtracts the first measured GM-BHQ value from the second measured GM-BHQ value, outputs a measured GM-BHQ difference value, and stores this in the measured BHQ difference table 1405. . Similarly, the first measured FA-BHQ value is subtracted from the second measured FA-BHQ value to output the measured FA-BHQ difference, which is stored in the measured BHQ difference table 1405.
The measured BHQ difference stored in the measured BHQ difference table 1405 is input by the input/output control unit 1401 to the average/deviation calculation processing unit 306 along with the age and gender of the first subject 102 output from the subject master 308.

The average/deviation calculation processing unit 306 groups the measured BHQ differences according to the age and gender of the first subject 102 output from the subject master 308, and then calculates the measured BHQ differences, the measured GM-BHQ difference and the measured FA-BHQ. Calculate the average value and standard deviation of the differences.
This is the measured BHQ difference average value and standard deviation 2002 for each age and gender.
The average value and standard deviation of the measured GM-BHQ difference and the measured FA-BHQ difference are the data of the plurality of first subjects 102 without intervention, and the average value of the estimated GM-BHQ difference and the estimated FA-BHQ difference. and the standard deviation will be compared.

The estimated BHQ difference average value and standard deviation 2001 for each age and gender and the measured BHQ difference average value and standard deviation 2002 for each age and gender are input to the t-test calculation processing unit 307.
The t-test calculation processing unit 307 calculates the average value and standard deviation of the estimated GM-BHQ difference and the estimated FA-BHQ difference, and the average value of the measured GM-BHQ difference and the measured FA-BHQ difference, for each age and gender of the subject. Based on the value and standard deviation, a t-test is performed on two unpaired samples to obtain a t-value. Then, the t-test calculation processing unit 307 converts the t-value into a p-value.

The average value and standard deviation of the estimated GM-BHQ difference and the estimated FA-BHQ difference, the average value and standard deviation of the measured GM-BHQ difference and the measured FA-BHQ difference, and age, calculated for each age and gender. - The p-value of the estimated GM-BHQ difference and the p-value of the estimated FA-BHQ difference for each gender are output to a predetermined nonvolatile storage 206, etc., and are used to evaluate the effectiveness of the intervention corresponding to the intervention ID 1001.

As described above, in the intervention evaluation device 1201 according to the second embodiment of the present invention, when the questionnaires before and after the intervention are input, the estimated GM-BHQ difference and the estimated FA-BHQ difference where the intervention was performed and the estimated FA-BHQ difference where the intervention was performed are displayed. It is possible to obtain a statistically significant difference between the measured GM-BHQ difference and the measured FA-BHQ difference that are not measured. Then, it becomes possible to objectively evaluate the effectiveness of the intervention from statistically significant differences such as p-values.
The intervention evaluation device 1201 according to the second embodiment estimates the difference in BHQ and uses this as an evaluation target. Therefore, since the intervention evaluation device 101 according to the first embodiment compares differences between individuals based on the estimation of BHQ, it is possible to expect higher estimation accuracy since it evaluates intra-individual changes.

Note that although the intervention evaluation device 101 according to the first embodiment of the present invention and the intervention evaluation device 1201 according to the second embodiment employ the t-test for hypothesis testing, other methods such as the Mann-Whitney U test may also be used. The following hypothesis testing method may be adopted.

In the embodiments of the present invention, an intervention evaluation device has been described.
An intervention evaluation device according to an embodiment of the present invention includes an intervention table that links interventions and subjects, and performs statistical processing on the contents of questionnaire responses of subjects who have undergone a certain intervention, based on intervention IDs. Therefore, according to the intervention evaluation device according to the embodiment of the present invention, it is possible to objectively evaluate whether the intervention is effective for brain health.

According to the intervention evaluation device 101 according to the first embodiment of the present invention, when the questionnaire responses of the second subject 107 before and after the intervention are inputted, the estimated GM-BHQ value before and after the intervention and the estimated Statistically significant differences in FA-BHQ values can be obtained. Then, it becomes possible to objectively evaluate the effectiveness of the intervention from statistically significant differences such as p-values.

According to the intervention evaluation device 1201 according to the second embodiment of the present invention, when the questionnaire of the second subject 107 before and after the intervention is inputted, the estimated GM-BHQ difference and the A statistically significant difference can be obtained between the estimated FA-BHQ difference and the measured GM-BHQ difference and the measured FA-BHQ difference without intervention. Then, it becomes possible to objectively evaluate the effectiveness of the intervention from statistically significant differences such as p-values.

Although the embodiments of the present invention have been described above, the present invention is not limited to the above embodiments, and other modifications and applications may be made without departing from the gist of the present invention as described in the claims. include.

101...Intervention evaluation device, 102...First subject, 103...MRI device, 104...MRI image file group, 105...Living environment questionnaire results, 106...Approximate function parameter, 107...Second subject, 108...Pre-intervention schedule, 109 ...Pre-intervention living environment questionnaire results, 110...Post-intervention schedule, 111...Post-intervention living environment questionnaire results, 112...Estimated BHQ average value, standard deviation and p value by age and gender, 201...CPU, 202...ROM, 203... RAM, 204...Display unit, 205...Operation unit, 206...Nonvolatile storage, 207...RTC, 208...Bus, 209...Serial port, 210...NIC, 301...I/O control unit, 302...Questionnaire response group, 303... BHQ calculation processing unit, 304... Learning calculation processing unit, 305... Estimation calculation processing unit, 306... Average/deviation calculation processing unit, 307... T-test calculation processing unit, 308... Subject master, 309... Brain information table, 310... Measurement BHQ table, 311...Questionnaire table, 312...Estimated BHQ table, 313...Intervention master, 314...Intervention table, 402...Nerve fiber anisotropy data group, 501...Gray matter amount calculation section, 502...Nerve fiber anisotropy calculation section , 503... Average value calculation unit, 504... Gray matter volume data group, 505... Nerve fiber anisotropy data group, 507... MRI imaging date, 601... Living environment questionnaire, 1002... Pre-intervention estimated BHQ, 1003... Post-intervention estimated BHQ , 1101... Mean value and standard deviation of estimated BHQ before intervention for each age and gender, 1102... Mean value and standard deviation of estimated BHQ after intervention for each age and gender, 1103... p value of estimated BHQ for each age and gender, 1201... Intervention evaluation device , 1202... Pre-intervention schedule, 1203... Pre-intervention living environment questionnaire results, 1204... Pre-intervention MRI image file group, 1205... Post-intervention schedule, 1206... Post-intervention living environment questionnaire results, 1207... Post-intervention MRI image file group, 1208 ...Difference approximation function parameter, 1301...Estimated BHQ difference average value, standard deviation and p value for each age/gender, 1401...Input/output control section, 1402...Difference learning calculation processing section, 1403...Difference estimation calculation processing section, 1404...Estimation BHQ difference table, 1405...Measurement BHQ difference table, 1801...Questionnaire difference calculation processing unit, 1802...BHQ difference calculation processing unit, 2001...Estimated BHQ difference average value/standard deviation for each age/gender, 2002...Measurement BHQ for each age/gender Difference mean value/standard deviation

Claims (3)

An intervention evaluation device that evaluates the effectiveness of an intervention using objective numerical values,
A subject ID that uniquely identifies the subject receiving the intervention, an intervention ID that uniquely identifies the intervention that the subject receives, and the first living environment questionnaire on the pre-intervention schedule, which is the stage immediately before implementing the intervention. They answered the second living environment questionnaire on the pre-intervention date when the pre-intervention living environment questionnaire results were created, and on the post-intervention date immediately after implementing the intervention. an intervention table that describes the relationship with the post-intervention schedule when the questionnaire results were created;
a subject master in which a relationship between the subject ID, the age of the subject, and the gender of the subject is described;
Based on the estimated BHQ value of the subject derived from the pre-intervention living environment questionnaire results and the post-intervention living environment questionnaire results of the subject or the estimated BHQ difference of the subject before and after the intervention, for each age and gender of the subject. , an average/deviation calculation processing unit that calculates the average value and standard deviation of the estimated BHQ value or the average value and standard deviation of the estimated BHQ difference;
a t-test calculation processing unit that calculates a t-test for each age and gender of the subject based on the average value and standard deviation of the estimated BHQ value or the average value and standard deviation of the estimated BHQ difference, and outputs a p-value; An intervention evaluation device comprising: Furthermore,
Estimation based on the subject ID, the pre-intervention schedule, the pre-intervention estimated BHQ value obtained by estimation calculation processing based on the pre-intervention living environment questionnaire results, the post-intervention schedule, and the post-intervention living environment questionnaire results. and an estimated BHQ table in which a relationship with the post-intervention estimated BHQ value obtained through arithmetic processing is described,
The average/deviation calculation processing unit calculates the above for each age and gender of the subject based on the estimated BHQ value of the subject derived from the pre-intervention living environment questionnaire result and the post-intervention living environment questionnaire result of the subject. It calculates the average value and standard deviation of the estimated BHQ value,
The t-test calculation processing unit calculates a t-test for two paired samples for each age and gender of the subject based on the average value and standard deviation of the estimated BHQ value, and outputs a p value. ,
The intervention evaluation device according to claim 1. Furthermore,
An estimated BHQ difference table that describes the relationship between the subject ID and the estimated BHQ difference obtained by estimation calculation processing based on the difference between the pre-intervention living environment questionnaire results and the post-intervention living environment questionnaire results. death,
The average/deviation calculation processing unit calculates, for each age and gender of the subject, based on the estimated BHQ difference of the subject derived from the pre-intervention living environment questionnaire result and the post-intervention living environment questionnaire result of the subject. The average value and standard deviation of the estimated BHQ difference are calculated,
The t-test calculation processing unit calculates a t-test for two unpaired samples for each age and gender of the subject based on the average value and standard deviation of the estimated BHQ difference, and outputs a p value. ,
The intervention evaluation device according to claim 1.

Images