JP2022164013A - Health check system, health check result processing program, and health check result processing method - Google Patents

Abstract

To provide a health check system that predicts a future risk of a subject using existing health check information without conducting special measurement and presents it in a way that helps the subject themselves assume their future image.SOLUTION: Provided is a health check system 101 for creating a report in which a predicted physical activity level and an action plan are correlated and presenting it to a subject, said system comprising an inspection information acquisition unit 3, a future prediction unit 4, an action plan creation unit 5, an action plan database 6, and a report creation unit 7. The inspection information acquisition unit 3 reads out the health diagnosis inspection result of prescribed items of the subject. The future prediction unit 4 predicts the physical activity level. The action plan creation unit 5 creates an action plan to be exercised by the subject based on a future prediction result obtained by the future prediction unit 4.SELECTED DRAWING: Figure 1

Description

The present invention relates to a system that presents risk prediction results based on the results of physical examinations of subjects and countermeasures against them.

With the rapid progress of aging in recent years, there is a social problem that the number of elderly people who cannot walk independently due to locomotor disorders is increasing. For example, due to age-related decline in muscle strength, diseases of the joints and spine, osteoporosis, and the like, the function of locomotive organs may deteriorate, and people may become bedridden or require nursing care. Also known are locomotive syndrome (hereafter referred to as locomotive syndrome), which indicates a high risk of weakening of locomotory functions, and sarcopenia, which indicates a state in which muscle mass decreases with age and general muscle strength declines. In recent years, in order to prevent locomotive syndrome and sarcopenia, risk estimation is performed by combining biometric information obtained from various measuring devices and individual lifestyle information.

In Patent Document 1, the subject's posture (“head”, “neck”, “waist”, “foot” coordinates) and external factors (“walking speed”, “step length”, “step distance”, A system has been proposed that acquires the characteristics of the walking angle) and presents the risk of falling or the risk of developing dementia in numerical values to the subject.

JP 2020-146435 A

In Patent Document 1, the subject's posture (“head”, “neck”, “waist”, “foot” coordinates) and external factors (“walking speed”, “step length”, “step distance”, "Walking angle") is described as being collected at regular medical examinations, but no specific collection method is described. Therefore, we can extract the coordinates of the “head”, “neck”, “waist”, and “legs” by shooting a video of the subject walking, and It is presumed that special measurements such as calculating "" and "walking angle" must be performed at regular checkups. Since such measurements are not performed in a normal health checkup, if they are to be performed in a normal health checkup, it is necessary to newly provide a special measurement item.

In addition, in order to prevent locomotive syndrome and sarcopenia, it is necessary for subjects to pay attention to their daily behavior and lifestyle. In order to avoid this, it is important to clearly communicate the actions to be taken in daily life (exercise, gymnastics, attention to nutrition, etc.) and to motivate them to carry out the actions. However, in the case of a system that expresses risk prediction results numerically as in Patent Document 1, it is difficult for subjects to intuitively grasp the risk, and they cannot receive information about the next action to be taken. Therefore, there is a demand for a technique that enhances the effect of reducing the future risk of falls and the risk of certification by making use of prediction results in daily life.

An object of the present invention is to predict the future risk of a subject using existing health checkup information without special measurements, and present it in a way that makes it easy for the subject to imagine their own future image. to do.

In order to achieve the above object, the present invention provides an examination information acquisition unit that reads examination results of predetermined items of a subject from a storage device storing the results of health examinations, and a future physical activity level of the subject. To provide a health checkup system having a future prediction part for predicting, an action plan creation part for creating an action plan to be taken by a subject, and a report creation part for creating a report to be presented to the subject. The future prediction unit includes a learned learning model, inputs test results of predetermined items to the learning model, and receives the subject's future physical activity level predicted by the learning model. The action plan creation unit includes an action plan database in which a plurality of action plans are stored in advance for each of a plurality of physical activity levels, and selects an action plan corresponding to the future physical activity level of the subject predicted by the future prediction unit. do. The report creation unit creates a report that displays the physical activity level predicted by the future prediction unit in association with the action plan selected by the action plan creation unit.

According to the present invention, it is possible to predict the future of the subject using existing medical examination information without performing special measurements, and present the future image to the subject at a level that is easy to assume, such as the physical activity level, In addition, since a specific action plan can be proposed to the subject, the subject can intuitively grasp his/her risk.

1 is a functional block diagram of a health checkup system according to an embodiment of the present invention; FIG. 1 is a block diagram showing the hardware configuration of a health checkup system according to an embodiment of the present invention; FIG. 4 is a flow chart showing the flow of processing of the health checkup system of Embodiment 1. FIG. 4 is a flow chart showing the flow of processing of the health checkup system of Embodiment 1. FIG. 4 is a flow chart showing the flow of processing of the health checkup system of Embodiment 1. FIG. 4 is a flow chart showing the flow of processing of the health checkup system of Embodiment 1. FIG. 4 is a flow chart showing the flow of processing of the health checkup system of Embodiment 1. FIG. FIG. 4 is an explanatory diagram of icons indicating physical activity levels used in a report of the health checkup system of Embodiment 1; It is an example of a table in the action plan database 6 of the health checkup system of the first embodiment. 4 is an explanatory diagram showing an example of a report of the health checkup system of Embodiment 1; FIG. 9 is a flow chart showing the flow of processing of the health checkup system of Embodiment 2; 9 is a flow chart showing the flow of processing of the health checkup system of Embodiment 2; 9 is a flow chart showing the flow of processing of the health checkup system of Embodiment 2; 9 is a flow chart showing the flow of processing of the health checkup system of Embodiment 2; 9 is a flow chart showing the flow of processing of the health checkup system of Embodiment 2; FIG. 10 is an explanatory diagram showing an example of a prediction result of a spinal curvature prediction model of the health checkup system of Embodiment 2; FIG. 11 is an explanatory diagram of icons showing locomotive syndrome onset risk levels used in a report of the health checkup system of Embodiment 2; It is an example of a table in the action plan database 6 of the health checkup system of the second embodiment. FIG. 11 is an explanatory diagram showing an example of a report of the health checkup system of Embodiment 2; 14 is a flow chart showing the flow of processing of the health checkup system of Embodiment 3. FIG. FIG. 11 is an explanatory diagram showing an example of a report of the health checkup system of Embodiment 3;

An embodiment of the present invention will be described below with reference to the drawings.

FIG. 1 shows the functional configuration of a health checkup system 101 of this embodiment. FIG. 2 shows the hardware configuration of the health checkup system 101. As shown in FIG.

As shown in FIGS. 1 and 2, a health checkup system 101 of this embodiment includes a future prediction/presentation system 1 and a risk prediction model learning system 2 . The future prediction and presentation system 1 and the risk prediction model learning system 2 are connected to an in-facility network 103 as shown in FIG. , are connected to a medical information system 104 that manages them. Also, the in-facility network 103 is connected to devices for examining medical examination items, such as a blood/urine test device 125, an X-ray CT device 126, an MRI device, an ultrasonic imaging device, and other imaging devices. In addition, the health checkup system 101 is connected via an external network 120 to a large-scale health checkup result database 121 in which past health checkup results of many subjects are accumulated, and to a subject terminal 122 such as a smartphone. It is

First, the functions of the future prediction and presentation system 1 and the risk prediction model learning system 2 will be described with reference to FIG.

The future prediction and presentation system 1 includes an examination information acquisition unit 3 , a future prediction unit 4 , an action plan creation unit 5 , an action plan database 6 and a report creation unit 7 . The risk prediction model learning system 2 includes a learning data collection unit 8 , a learning database 9 and a model learning unit 10 .

In the future prediction and presentation system 1, the test information acquisition unit 3 reads the test results of predetermined items of the subject from the medical information storage 105 storing the health checkup results, and sends the test results to the future prediction unit 4. input.

The future prediction unit 4 uses the learned learning model generated by the risk prediction model learning system 2 to predict at least the future physical activity level as the subject's future risk prediction. Specifically, the future prediction unit 4 inputs test results of predetermined items acquired by the test information acquisition unit 3 into the learning model, and receives the subject's future physical activity level and the like predicted by the learning model. .

The action plan creating unit 5 creates an action plan to be taken by the subject based on the future prediction result obtained by the future prediction unit 4 . The action plan database 6 stores in advance a plurality of action plans for each physical activity level. The action plan creation unit 5 selects from the action plan database 6 an action plan corresponding to the future physical activity level of the subject predicted by the future prediction unit 4 .

That is, the report creating unit 7 creates a report to be presented to the subject. The report generation unit 7 generates and outputs a report in a predetermined format that displays the prediction result of the physical activity level predicted by the future prediction unit and the action plan selected by the action plan generation unit 5 in association with each other. do.

In the risk prediction model learning system 2, the learning data collection unit 8 presents test results of predetermined items of a large number of subjects from the large-scale medical examination result database 121, physical activity levels of those subjects, etc. Information (learning data) is collected and stored in the learning database 9 . The model learning unit 10 uses the information stored in the learning database 9 to learn a learning model and generates a trained learning model to be used by the future prediction unit 4 .

Hardware configurations of the future prediction and presentation system 1 and the risk prediction model learning system 2 will be described with reference to FIG.

The future prediction and presentation system 1 and the risk prediction model learning system 2 include a CPU 106, a main memory 107, a storage device 108, a learning database 9, a communication device 115, a common bus 116, a display memory 109, and a display. It comprises a device 110 , a controller 111 , a mouse 112 , a keyboard 113 and a printing device 114 .

The storage device 108 contains a learning data collection program 123, a model learning program, a future prediction program 117, an action plan creation program 118, a report creation program 119, an operating system (OS), and device drives for peripheral devices. programs are stored in advance.

CPU 106 is connected to other components by common bus 116 . The CPU 106 reads the learning data collection program 123, the model learning program, the future prediction program 117, the action plan creation program 118, and the report creation program 119 from the storage device 108, and develops them in the main memory 107. The functions of the learning data collection unit 8, the model learning unit 10, the future prediction unit 4, the action plan creation unit 5, and the report creation unit 7 are realized by software.

The CPU 106 also controls the operation of each component of the main memory 107 , storage device 108 , display memory 109 , controller 111 , keyboard 113 , printing device 114 and communication device 115 . The main memory 107 is used as a storage area for control programs or as a work area during program execution. The display device 110 displays an image on the display device 110 based on the data received from the display memory 109 . Mouse 112 and keyboard 113 are input devices for the operator. The controller 111 outputs a signal including the position of the mouse pointer of the mouse 112 to the CPU 106 to enable the operator's input operation using the mouse 112 . Printer 114 outputs the report. The communication device 115 is connected to the in-facility network 103 and the external network 120 .

The medical information system 104 is an existing system such as an electronic medical record or an image management system, and includes electronic medical record data, image data captured by an imaging device such as an X-ray CT device, and test results of a blood/urine test device. Health checkup result data such as data is received and stored in the medical information storage 105 .

[[Embodiment 1]]
Next, Embodiment 1 of the present invention will be described with reference to flowcharts of FIGS. 3 to 7 and FIGS. 8 to 10. FIG. FIG. 8 shows icons for intuitively presenting the physical activity level in a report, and FIG. 9 shows an example of a table stored in the action plan database 6. FIG. FIG. 10 shows an example of a report.

First, the processing of the future prediction and presentation system 1 will be described using FIG.

<<Step S1>>
In step S<b>1 , the test information acquisition unit 3 acquires test information such as the subject's height, weight, BMI, blood test results, and urine test results from the medical information storage 105 . Blood test results and urine test results use the bone metabolism markers serum BAP, serum P1NP, serum NTX, serum CTX, serum TRACP-5b, serum ucOC, urine DPD, urine NTX, and urine CTX ("Osteoporosis Prevention and Treatment Guidelines 2015”). The items acquired as the inspection information by the inspection information acquisition unit 3 are not limited to the above. good too.

<<Step S2>>
In step S2, the future prediction unit 4 makes a future prediction using the examination information acquired in step S1. Embodiment 1 predicts a subject's future physical activity level and cumulative cost. As for the physical activity level, as shown in FIGS. 8 and 9, it is predicted which of the four levels "bedridden", "wheelchair", "needs a cane", and "able to walk". As the accumulated cost, the accumulated cost of medical expenses and/or nursing care expenses required in the future of the subject is predicted. These will be described in detail later with reference to FIG. 4 and the like.

<<Step S3>>
In step S3, the action plan creating unit 5 creates a specific action plan for the subject from the physical activity level predicted by the future prediction unit 4 in step S2 and the prediction result of the total cost. An optimum action plan is obtained from the action plans stored in advance in the action plan database 6 of FIG.

In the action plan database 6, for example, as shown in FIG. 9, action plans are stored in advance in the form of a table or the like for each combination of accumulated cost and a plurality of physical activity levels. The action plan creating unit 5 selects an action plan by selecting from the table of FIG. 9 an action plan corresponding to the combination of the subject's future cumulative cost and physical activity level predicted by the future prediction unit 4 . As an action plan, in order to improve the future physical activity level and reduce the cumulative cost, it is a plan that recommends exercise and nutritional intake suitable for the subject. Exercises once a day, muscle training three times a week, etc. are set.

<<Step S4>>
In step S4, the report creation unit 7 creates a report that displays the physical activity level and cumulative cost predicted by the future prediction unit 4 in steps S2 and S3 in association with the action plan selected by the action plan creation unit 5. to create

FIG. 10 shows an example of the report. In the example of FIG. 10, the physical activity level is intuitively displayed by the icons and letters shown in FIG. 8 to indicate which of "bedridden", "wheelchair", "cane required", and "ambulatory". ing. Also, an action plan corresponding to the physical activity level and cumulative cost is displayed as a "recommended action plan". In addition, the results of the blood and urine tests of the physical examination are also displayed.

This report may be output as a paper medium by printing with the printing device 114, or may be output by sending it as electronic data to a terminal (smartphone, tablet, PC, etc.) 122 owned by the subject. good too.

<<Future Forecast>>
Detailed processing of the future prediction in step S2 of FIG. 3 will be described with reference to FIG.

<Step S21>
In step S21, the future prediction unit 4 uses the learning model for predicting the physical activity level generated by the risk prediction model learning system 2 to predict the physical activity level.

<Step S22>
In step S<b>22 , the future prediction unit 4 uses the cumulative cost prediction learning model generated by the risk prediction model learning system 2 to predict cumulative costs.

<< Generate learning model >>
Here, a process of generating a physical activity level prediction learning model and a cumulative cost prediction learning model by the risk prediction model learning system 2 will be described with reference to FIG. 5 . The physical activity level prediction learning model and the cumulative cost prediction learning model are generated by pre-learning a learning model such as a known neural network using a supervised learning method among machine learning algorithms.

<Step S201>
First, in step S201, the learning data collection unit 8 acquires the same test items as the test items acquired by the test information acquisition unit 3 in step S1 of FIG. Predetermined data indicating the condition and predetermined data for calculating the actual medical and nursing care costs for the subject in the future (for example, the period from the medical examination to 10 years later) are stored on a large scale. From the health checkup result database 121 , the data for many people are collected and stored in the learning database 9 .

<Step S202>
In step S202, the learning data collection unit 8 processes the collected information into a format that can be used for learning. Specifically, the units of the numerical values of the inspection information are unified, and the numerical values are normalized.

In addition, the physical activity level is determined from predetermined data indicating the physical condition of the subject in the future (for example, 10 years from now), and the processing is divided into four levels: bedridden, wheelchair, cane required, and ambulatory. It is performed based on a predetermined calculation.

Furthermore, from the predetermined data for calculating the actual medical expenses and nursing care expenses in the future of the subject (for example, the period from the medical examination to 10 years later), the medical expenses and nursing care expenses are calculated in advance. It is calculated based on the specified formula.

<Step S203>
At step S203, the model learning unit 10 generates a physical activity prediction model. For the training of the physical activity prediction model, we use an existing supervised machine learning algorithm that performs classification. Specifically, the model learning unit 10 uses the test item data collected in step S201 as input data for the learning model, and uses the physical activity level obtained in step S202 (bedridden/wheelchair/cane required) as correct data.・The model is constructed so that the correct physical activity level (bedridden/wheelchair/cane required/ambulatory) is output when the test item data is input and the model is learned using the model's ability to walk. .

<Step S204>
In step S204, the model learning unit 10 generates a cumulative cost prediction model. For the learning of the cumulative cost prediction model, an existing supervised learning algorithm that performs regression is used. Specifically, the model learning unit 10 uses the test item data collected in step S201 as input data for the learning model, and learns using the amounts of medical expenses and nursing care costs obtained in step S202 as correct data. Then, when the inspection item data is input, the model is constructed so that the correct medical and nursing care costs are output.

<<< Prediction of physical activity level >>>
Next, details of the physical activity level prediction by the future prediction unit 4 in step S21 of FIG. 4 will be described with reference to FIG.

(Step S211)
In step S211, the future prediction unit 4 reads the physical activity level prediction model learned in step S203 of FIG.

(Step S212)
In step S212, the future prediction unit 4 inputs the test item data of the subject acquired by the test information acquisition unit 3 in step S1 of FIG. 3 to the physical activity level prediction model.

(Step S212)
In step S213, the future prediction unit 4 receives the physical activity level output from the physical activity level prediction model (bedridden/wheelchair/cane required/walkable) and outputs it as a prediction result.

<<< Cumulative Cost Forecast >>>
Details of cumulative cost prediction by the future prediction unit 4 in step S22 of FIG. 4 will be described with reference to FIG.

(Step S221)
In step S221, the future prediction unit 4 reads the cumulative cost prediction model learned in step S204 of FIG.

(Step S222)
In step S222, the examination information of the subject is input to the model.

(Step S223)
In step S223, the future prediction unit 4 outputs the cumulative costs (medical costs, nursing care costs) output by the cumulative cost prediction model.

As described above, in the health checkup system of the first embodiment, the future image of the subject, such as the physical activity level and the cumulative cost, is presented in a manner that is easy for the subject to imagine. , a concrete action plan for reducing cumulative costs can also be proposed. Therefore, the subject himself/herself can grasp the risk intuitively, which leads to improvement of the subject's motivation to execute the action plan.

[[Embodiment 2]]
Next, Embodiment 2 will be described with reference to flowcharts of FIGS. 11 to 15 and FIGS. 16 to 19. FIG. FIG. 16 shows an image of the vertebral body region and the curved line of the spine predicted by the future prediction unit 4 . FIG. 17 is an icon for intuitively presenting the level of locomotion syndrome onset risk in a report. FIG. 18 is an example of a table showing correspondence between locomotion syndrome onset risks (hereinafter referred to as locomotion onset risks) stored in the action plan database 6 and action plans. FIG. 19 shows an example of a report.

In the health checkup system 101 of the second embodiment, the examination results of the subject's predetermined items read by the examination information acquisition unit 3 from the medical information storage 105 include a two-dimensional or three-dimensional image of a predetermined region of the subject. include. The examination information acquisition unit 3 calculates, from the acquired image, a feature amount of a predetermined configuration of internal organs and/or tissues that constitute the subject. The future prediction unit 4 predicts whether the physical activity level corresponds to one of four levels of "bedridden", "wheelchair", "requires a cane", and "able to walk". Predict the shape of the curvature of the spine and which of a plurality of predetermined levels of locomotive syndrome onset risk corresponds. A specific description will be given below.

First, the processing of the future prediction and presentation system 1 will be described with reference to FIG. 11 .

<<Step S11>>
In step S11 of FIG. 11, the examination information acquisition unit 3 acquires an X-ray CT image of a predetermined range (for example, from the clavicle to the hip joint) including the chest and abdomen captured in the chest CT examination of the subject as medical information. Acquire from the storage 105 . An example of obtaining an X-ray CT image will be described here, but any two-dimensional or three-dimensional image of the subject, such as an MRI image or an ultrasound image, may be used.

<<Step S12>>
In step S12, the examination information acquiring unit 3 uses the image acquired in step S11 to determine the vertebral body region, the subject's muscle at a predetermined position, and the subject's fat region at a predetermined position. are extracted in two or three dimensions. Here, the examination information acquisition unit 3 extracts the vertebral body region and automatically extracts the muscle and fat regions (three-dimensional data) at the navel position. For the extraction, a known calculation method using machine learning or the like is used.

<<Step S13>>
In step S13, the examination information acquisition unit 3 determines whether or not the automatic extraction result in step S12 needs to be corrected. Specific examples that require correction include cases where the subject's vertebral body has a different shape and bone quality than normal due to osteoporosis, etc., and is not automatically extracted as a vertebral body region, or when the subject's muscle mass is small and muscle are not extracted or erroneous extraction occurs.

<<Step S14>>
If it is determined in step S13 that correction is necessary, the inspection information acquiring unit 3 displays a message prompting the operator to manually correct the extraction region in step S14, and accepts manual correction by the operator.

<<Step S15>>
In step S15, the examination information acquiring unit 3 uses the three-dimensional data of the vertebral body region extracted in step S12 to calculate the bone mineral density (BMD) of each vertebral body. An existing quantitative CT method (Quantitative computer tomography) is used to measure the BMD value. Young Adult Mean (YAM), which is the diagnostic criteria for osteoporosis, is calculated from the BMD value. % is calculated by

Further, the examination information acquisition unit 3 calculates the current curvature line of the spine from the three-dimensional data of the vertebral body region extracted in step S12, and creates an image in which the vertebral body region and the curved line are drawn.

<<Step S16>>
In step S16, the examination information acquiring unit 3 obtains the CT value of at least one of the vertebral body region, the subject's muscle at a predetermined position, and the subject's fat region at a predetermined position from the X-ray CT image. to extract That is, the examination information acquiring unit 3 uses the muscle and fat regions extracted in step S12 to calculate the area of visceral fat and subcutaneous fat, the area of the erector spinae muscle and the psoas major muscle, and the average CT value. The CT value is a value expressing the X-ray absorption value of a substance photographed under the condition that water is set to 0 and air is set to -1000 as a relative value to water at the origin.

<<Step S17>>
In step S<b>17 , the test information acquisition unit 3 acquires predetermined test information such as height, weight, BMI, blood test results, urine test results, etc. of the subject from the medical information storage 105 . As blood test results and urine test results, serum BAP, serum P1NP, serum NTX, serum CTX, serum TRACP-5b, serum ucOC, urinary DPD, urinary NTX, and urinary CTX are used (see "Prevention of osteoporosis and Treatment Guidelines 2015”). The items acquired as the inspection information by the inspection information acquisition unit 3 are not limited to the above. good too.

<<Step S18>>
In step S18, the future prediction unit 4 performs future prediction using the information calculated in steps S15 and S16 and the examination information acquired in step S17. Here, the future prediction unit 4 predicts whether the physical activity level corresponds to one of the four levels of "bedridden", "wheelchair", "requires cane", and "able to walk". , shape of future spinal curvature, and which of a plurality of predetermined levels of locomotive syndrome onset risk corresponds. These will be described in detail later with reference to FIG. 12 and the like.

<<Step S19>>
In step S19, the action plan creating unit 5 creates a specific action plan for the subject from the results predicted in step S18. In the action plan database 6, as shown in FIG. 18, action plans corresponding to multiple levels of locomotive syndrome onset risk, and actions corresponding to combinations of accumulated costs and multiple physical activity levels as shown in FIG. A plan is stored in advance.

The action plan creation unit 5 creates an action plan corresponding to the future locomotive syndrome onset risk level of the subject predicted by the future prediction unit 4, and an action plan corresponding to the combination of the cumulative cost and the physical activity level, as shown in FIG. Each is selected from FIG.

<<Step S20>>
In step S20, the risk of locomotive syndrome predicted by the future prediction unit 4, the shape of the future spinal curvature, the physical activity level, and the cumulative cost predicted in steps S18 and S19 correspond to the action plan selected by the action plan creation unit 5. to create a report to display.

FIG. 19 shows an example of the report. In the example of FIG. 10, the risk of developing locomotive syndrome is displayed intuitively and easily with the icons and characters shown in FIG. The future curvature of the spine is displayed in an easy-to-understand image. As in the first embodiment, the physical activity level can be intuitively understood by the icons and letters shown in FIG. is displayed. Also, an action plan corresponding to the physical activity level and cumulative cost is displayed as a "recommended action plan". In addition, an image showing the current curved shape of the spine used for prediction, bone density, areas of fat and muscle, and YAM values used as diagnostic criteria for osteoporosis are also displayed.

<<Future predictions>>
Details of the future prediction in step S18 of FIG. 11 will now be described with reference to FIG.

<Steps S181 to 184>
The future prediction unit 4 predicts the spinal curvature in step S181, predicts the risk of developing locomotive syndrome in step S182, predicts the physical activity level in step S183, and predicts the cumulative cost in step S184. This is done using a learning model.

<< Generate learning model >>
Here, a learning model generation process by the risk prediction model learning system 2 will be described.

The processing for generating the physical activity level prediction learning model used in step S183 and the accumulated cost prediction learning model used in step S184 is the same as the processing in FIG.

With reference to FIG. 13, the method of learning the spinal curvature prediction learning model used in step S181 and the learning method of the locomotive syndrome onset risk prediction learning model used in step S182 will be described.

<<Step S1801>>
First, in step S1801, the learning data collection unit 8 acquires the image of the vertebral body/muscle/fat region acquired by the examination information acquisition unit 3 in step S12 of FIG. The acquired muscle/fat information and the same items as the examination information acquired in step S17 are collected from the large-scale health checkup result database 121 for a large number of people, and stored in the learning database.

In addition, the learning data collection unit 8 can grasp information that can extract the curvature of the spine from the health checkup data of the subject in the future (for example, 10 years later) and whether or not locomotive syndrome has developed. Such predetermined information is collected from the large-scale health checkup result database 121 and stored in the learning database 9 .

<<Step S1802>>
In step S1802, the learning data collection unit 8 processes the collected information into a format that can be used for learning. Specifically, the units of the numerical values of the inspection information are unified, and the numerical values are normalized. If the information acquired in step S1801 is an X-ray CT image, the examination information acquisition unit 3 extracts the vertebral body/muscle/fat regions, calculates vertebral body information, etc. in steps S16 and S17. Extracted by the same process as the process described above.

Also, a curvature of the spine is generated from predetermined data indicating the curvature of the spine in the future (for example, 10 years from now) of the subject. Furthermore, it determines whether future subjects have developed locomotive syndrome and the risk of developing locomotive syndrome, and divides the processing into four levels of risk of developing locomotive syndrome: very high, high, low, and very low. based on

<<Step S1803>>
In step S1803, the model learning unit 10 generates a spinal curvature prediction model. For the training of the spinal curvature prediction model, a machine learning algorithm that generates images is used among existing supervised learning. Specifically, the model learning unit 10 uses the image of the vertebral body region, the vertebral body information, the muscle/fat information, and the examination information collected in step S1802 as input data for the learning model, and uses the correct data as correct data for the future model. Learning is performed using the vertebral body region (for example, 10 years later) and the curved line of the spine. As a result, when an image of the vertebral body region of the subject, vertebral body information, muscle/fat information, and examination information are input, the vertebral body region and the curve of the spine 10 years later of the subject are output. A curvature prediction model is constructed. FIG. 16 shows an example of output vertebral body regions and curvature lines of the spine.

<<Step S1804>>
In step S1804, the model learning unit 10 generates a locomotive syndrome onset risk prediction model. To learn the locomotive syndrome risk prediction model, we use an existing supervised learning machine learning algorithm that performs classification. Specifically, the model learning unit 10 uses the vertebral body information, muscle/fat information, and examination information collected in step S1802 as input data for the learning model, and uses the future ( For example, 10 years later), four stages of locomotive syndrome onset risk levels (see FIG. 17) are used for learning. As a result, when the image of the vertebral body region of the subject, vertebral body information, muscle / fat information, and examination information are input, the locomotive syndrome development risk prediction model that outputs the level of the risk of locomotive syndrome development after 10 years of the subject. is constructed.

<<<Spine Curvature Prediction>>>
Next, details of spinal curvature prediction by the future prediction unit 4 in step S181 of FIG. 12 will be described with reference to FIG.

(Step S1811)
In step S1811, the future prediction unit 4 reads the spinal curvature prediction model learned in step S1803 of FIG.

(Step S1812)
In step S1812, the future prediction unit 4 adds the vertebral body image, vertebral body information, muscle/fat information of the subject acquired by the examination information acquisition unit 3 in steps S15 to S17 of FIG. Enter test information.

(Step S1813)
In step S1813, the future prediction unit 4 receives and outputs the prediction result of the future vertebral body region and the curvature line of the spine of the subject predicted by the spinal curvature model.

<<< Prediction of locomotive syndrome onset risk >>>
Next, details of locomotive syndrome onset risk prediction by the future prediction unit 4 in step S182 of FIG. 12 will be described with reference to FIG.

(Step S1821)
In step S1821, the future prediction unit 4 reads the locomotive syndrome onset risk prediction model learned in step S1804 of FIG.

(Step S1822)
In step S1822, the future prediction unit 4 adds the vertebral body information, muscle/fat information, and examination information of the subject acquired by the examination information acquisition unit 3 in steps S15 to S17 of FIG. 11 to the locomotive syndrome onset risk prediction model. input.

(Step S1823)
In step S1823, the future prediction unit 4 receives and outputs the prediction result of the locomotive syndrome onset risk level (FIG. 17) output by the locomotive syndrome onset risk prediction model.

Note that the prediction of the physical activity level and the prediction of the cumulative cost in steps S183 and S184 of FIG. 12 are the same as those in the first embodiment, so the description is omitted.

According to the second embodiment described above, future risks can be estimated at the time of medical examination without performing special measurements by using existing images or the like obtained by chest CT examination. In addition, it is possible to simultaneously estimate risks related to muscles and bones, which are important elements of the locomotor system.

In the above, an example using the chest and abdomen images captured by the chest CT examination has been described, but the images of each region captured by the whole body CT examination can also be used. In this case, for example, it is possible to acquire muscle and bone information corresponding to parts such as upper body and lower body, and to present a future prediction (for example, 10 years later) for each part. Based on this, it becomes possible to create an action plan according to the difference in body parts. As a result, the accuracy of the action plan presented to the subject can be improved, and the effect of preventing locomotive syndrome in the subject can be further enhanced.

In this embodiment, the examination information acquisition unit 3 is configured to calculate the feature amount of the internal organs and structures from the image of the subject. may be configured to receive the

[[Embodiment 3]]
Next, the health checkup system of the third embodiment will be described with reference to the flow chart of FIG. 20 and the report diagram of FIG.

In Embodiment 3, the future prediction unit 4 is configured to store the predicted physical activity level and the like of the subject in a connected storage unit (for example, the medical information storage 105 or the like). As a result, after predicting the physical activity level, etc. based on the results of the current health checkup, the prediction results calculated based on the results of the previous health checkup for the same subject are read out from the connected storage unit, and Calculate the difference in physical activity level and total cost between

<<Step S31>>
In step S31 of FIG. 20, the future prediction unit 4 reads the prediction result of the previous physical activity level and accumulated cost of the subject from the storage unit (for example, the medical information storage 105 or the like).

<<Step S32>>
In step S32, the future prediction unit 4 acquires test information such as the subject's current height, weight, BMI, blood test results, and urine test results, as in step S1 of FIG. 3 of the first embodiment.

<<Step S33>>
In step S33, the future prediction unit 4 uses the examination information acquired in step S32 to predict the physical activity level and the accumulated cost in the future, as in step S2 of FIG. 3 of the first embodiment.

<<Step S34>>
In step S34, the future prediction unit 4 calculates the difference between the accumulated cost obtained in the future prediction in step S33 and the accumulated cost obtained from the past accumulated cost obtained in step S31.

<<Step S35>>
In step S35, the action plan creation unit 5 prepares a specific action plan for the subject based on the prediction result of the physical activity level and cumulative cost predicted in step S32, as in step S3 in FIG. 3 of the first embodiment. create.

<<Step S36>>
In step S36, the report creation unit compares the current risk prediction result with the previous risk prediction result, and creates a report such as that shown in FIG. 21 that clearly presents the changes. In FIG. 21, the physical activity level of the previous time and this time is presented in comparison, the cumulative cost is presented in comparison, and the difference between them is also displayed.

According to the present embodiment 3, risk prediction is performed again after a certain period of time or the like, and changes over time from the previous time are presented, so that the subject can be expected to take active preventive actions and thereby reduce risks.
Although Embodiments 1 to 3 of the present invention have been described above, the present invention is not limited to these.

1 presentation system, 2 risk prediction model learning system, 3 test information acquisition unit, 4 future prediction unit, 5 action plan creation unit, 6 action plan database, 7 report creation unit, 8 learning data collection unit, 9 learning database, 10 model learning unit, 101 health checkup system, 103 facility network, 104 medical information system, 105 medical information storage, 107 main memory, 108 storage device, 109 display memory, 110 display device, 111 controller, 112 mouse, 113 keyboard, 114 printer, 115 communication device, 116 common bus, 117 future prediction program, 118 action plan creation program, 119 report creation program, 120 external network, 121 large-scale medical examination results database, 122 subject terminal, 123 learning data Acquisition program, 125 urinalysis device, 126 X-ray CT device

Claims (11)

a test information acquisition unit that reads the test result of a predetermined item of the subject from the storage device in which the health checkup result is stored;
a future prediction unit that predicts the future physical activity level of the subject;
an action plan creation unit that creates an action plan to be taken by the subject;
a report creation unit that creates a report to be presented to the subject;
The future prediction unit includes a learned learning model, inputs test results of the predetermined item into the learning model, receives the future physical activity level of the subject predicted by the learning model,
The action plan creation unit includes an action plan database in which a plurality of action plans are stored in advance for each of the plurality of physical activity levels, and corresponds to the future physical activity level of the subject predicted by the future prediction unit. Choose an action plan
The health checkup system, wherein the report creation unit creates a report displaying the physical activity level predicted by the future prediction unit and the action plan selected by the action plan creation unit in association with each other. The health checkup system according to claim 1,
The test information acquisition unit reads a blood test result and/or a urine test result as the test result of the predetermined item of the subject,
The future prediction unit predicts which of four levels of "bedridden", "wheelchair", "requires a cane", and "able to walk" corresponds to the physical activity level. diagnosis system. 3. The health checkup system according to claim 2, wherein said future prediction unit predicts a cumulative cost of medical expenses and/or nursing care expenses required in the future of said subject. 3. The health checkup system according to claim 2, wherein an action plan is stored in advance in the action plan database for each combination of the cumulative cost and a plurality of the physical activity levels,
The health checkup system, wherein the action plan creation unit selects an action plan corresponding to a combination of the future accumulated cost and the physical activity level of the subject predicted by the future prediction unit. The health checkup system according to claim 2,
The examination result of the predetermined item of the subject read from the storage device by the examination information acquisition unit includes a two-dimensional or three-dimensional image of a predetermined area of the subject,
The examination information acquiring unit obtains, from the image, a feature amount of a predetermined configuration of internal organs and/or tissues that constitute the subject, or causes an external computing device to calculate the feature amount. to receive the operation result,
The future prediction unit determines whether the physical activity level corresponds to one of the four levels of "bedridden", "wheelchair", "needs a cane", and "able to walk", as well as the future A health checkup system characterized by predicting the shape of the curvature of the spine and/or which of a plurality of predetermined levels of locomotive syndrome onset risk corresponds. 6. The health checkup system according to claim 5, wherein the action plan database stores action plans corresponding to the plurality of levels of the locomotive syndrome onset risk, and combinations of the cumulative cost and the plurality of physical activity levels. action plans are stored in advance,
The action plan creation unit includes an action plan corresponding to the level of the future locomotive syndrome onset risk of the subject predicted by the future prediction unit, and an action plan corresponding to the combination of the cumulative cost and the physical activity level. A health checkup system characterized by selecting each. 6. The health checkup system according to claim 5, wherein the examination information acquisition unit extracts the vertebral body region, the subject's muscle at a predetermined position, and the subject's fat region at a predetermined position from the image. A health checkup system characterized by extracting at least one of them two-dimensionally or three-dimensionally. 8. The health checkup system according to claim 7, wherein the previous image is an X-ray CT image, and the examination information acquiring unit extracts the vertebral body region and the subject's muscle at a predetermined position from the X-ray CT image. and extracting a CT value of at least one fat region of a subject at a predetermined position. 2. The health checkup system according to claim 1, wherein the future prediction unit stores the predicted physical activity level of the subject in a connected storage unit, After estimating the physical activity level, the physical activity level calculated based on the previous health checkup result for the same subject is read from the connected storage unit, and the difference between the current and previous physical activity levels is calculated. A health checkup system that calculates and outputs to the report creation unit. to the computer,
a first step of reading test results of predetermined items of the subject from a storage device in which the results of physical examinations are stored;
a second step of predicting the subject's future physical activity level;
a third step of creating an action plan for the subject to take;
A health checkup result processing program for executing a fourth step of creating a report to be presented to the subject,
In the second step, the test result of the predetermined item is input to the learned learning model, the future physical activity level of the subject predicted by the learning model is received,
In the third step, an action plan corresponding to the predicted future physical activity level of the subject is selected from an action plan database in which a plurality of action plans are stored in advance for each of the plurality of physical activity levels,
A health checkup result processing program, wherein in the fourth step, a report is created that displays the physical activity level predicted in the second step and the action plan selected in the third step in association with each other. a first step of reading test results of predetermined items of the subject from a storage device in which the results of physical examinations are stored;
a second step of predicting the subject's future physical activity level;
a third step of creating an action plan for the subject to take;
and a fourth step of creating a report to be presented to the subject, wherein
In the second step, the test result of the predetermined item is input to the learned learning model, the future physical activity level of the subject predicted by the learning model is received,
In the third step, an action plan corresponding to the predicted future physical activity level of the subject is selected from an action plan database in which a plurality of action plans are stored in advance for each of the plurality of physical activity levels,
In the fourth step, the physical activity level predicted in the second step and the action plan selected in the third step are generated in a corresponding manner to display a report.

Images