Disease risk determination system using fundus image, machine learning model generation device, disease risk determination device, and disease risk determination method

Abstract

To provide a disease risk determination system capable of determining disease risk from a fundus image.SOLUTION: A disease risk determination system comprises a machine learning model generation device that generates a trained model for determining disease risk, and a disease risk determination device that determines disease risk using the generated trained model. The machine learning model generation device includes: a learning data acquisition section; a first trained model generation section that generates a first trained model; and a second trained model generation section that generates a second trained model. The disease risk determination device includes: a determination data acquisition section; a health checkup data prediction section that inputs a determination fundus image and determination health checkup data to the first trained model and outputs predicted health checkup data; a disease risk prediction section that inputs the predicted health checkup data and the determination fundus image to the second trained model and outputs predicted disease risk data; and a prediction result presentation section.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present invention relates to a disease risk determination system, and particularly to a disease risk determination system, a machine learning model generation device, a disease risk determination device, and a disease risk determination method for determining a disease risk from a fundus image using a learned model by a neural network.

Background Art

[0002] Conventionally, a system for analyzing the health status of a subject using a fundus image has been proposed. Specifically, the fundus image is analyzed to output the presence or likely progression of a specific medical condition.

[0003] Such an analysis system uses a fundus image to evaluate the risk of a subject for a health event or to evaluate overall health. The system uses a fundus image processing machine learning model to explain the basis for the prediction generated by the system, that is, to generate a specific prediction. The machine learning model processes a plurality of fundus image data and the corresponding health status of the subject to generate a learning model for the subject, and then inputs any subject fundus image data into the learning model to generate health analysis data from the model output.

[0004] However, the conventional system has needed to use data such as a blood test in addition to the fundus image in order to obtain health analysis data regarding a specific disease.

[0005] As a system for analyzing the health status of a subject using fundus images, for example, in Patent Document 1, in a "system that can generate health analysis data for a patient and, optionally, other patient data from an input including one or more fundus images of the patient", "to generate health analysis data for a given patient, the system uses a fundus image processing machine learning model to process one or more fundus images and, optionally, other patient data to generate a model output for the patient, and then generates health analysis data from the model output" is disclosed.

[0006] Further, as a system for analyzing symptoms associated with infectious diseases and signs of exacerbation, among other aspects of the health status of a subject, using eye image data such as fundus images, for example, in Patent Document 2, a medical system is disclosed that "includes a data acquisition unit that acquires at least two types of data among blood oxygen data, auscultation sound data, eye image data, and eye blood flow data from a patient, and a data processing unit that processes the at least two types of data acquired by the data acquisition unit to detect changes in the cardiovascular system state associated with an infectious disease".

[0007] However, neither Patent Document 1 nor Patent Document 2 mentions achieving the same accuracy as that of performing invasive tests such as blood tests without performing invasive tests on the body such as blood tests.

Prior Art Documents

Patent Documents

[0008]

Patent Document 1

Patent Document 2

Summary of the Invention

Problems to be Solved by the Invention

[0009] Therefore, the present invention provides a disease risk determination system that solves the above problems and can determine the disease risk only from the data obtained by non-invasive examinations without performing a blood test. That is, without performing a blood test, the health examination data obtained by a blood test is predicted from a fundus image and used as virtual blood test data, thereby providing a disease risk determination system capable of making a prediction equivalent to that of having performed a blood test. It is expected that this will also contribute to reducing the physical and mental burden on the subject due to the examination and promoting the opportunity to know the disease risk.

[0010] Further, according to the present invention, it is possible to predict unacquired health examination data from the fundus image of a subject using a learned model, and a disease risk determination system is provided that can determine the disease risk from the fundus image of the subject and the health examination data predicted by the learned model using a different learned model.

[0011] Furthermore, in one aspect of the present invention, there is provided a disease risk determination system that can present disease risk data to a user terminal and provide prediction health examination data corresponding to a predicted disease based on the predicted disease risk data to the user terminal.

Means for Solving the Problems

[0012] In order to solve the above problems, in the present invention, there is provided a disease risk determination system, comprising: a machine learning model generation device that generates a learned model for determining the risk of a disease; and a disease risk determination device that determines the risk of a disease using the learned model generated by the machine learning model generation device. The machine learning model generation device includes: a learning data acquisition unit that acquires learning fundus images, learning health examination data, and learning disease risk data from a learning database; a first learned model generation unit that inputs the learning health examination data as correct answer data and the learning fundus images as observation data into a neural network to generate a first learned model; and a second learned model generation unit that inputs the learning disease risk data as correct answer data and the learning fundus images as observation data into a neural network to generate a second learned model. The disease risk determination device includes: a determination data acquisition unit that acquires determination fundus images and determination health examination data from a determination database; a health examination data prediction unit that inputs the determination fundus images and the determination health examination data into the first learned model to output predicted health examination data; a disease risk prediction unit that inputs the predicted health examination data output from the health examination data prediction unit and the determination fundus images into the second learned model to output predicted disease risk data; and a prediction result presentation unit that outputs at least the predicted disease risk data to a user terminal.

[0013] In the disease risk determination system according to an aspect of the present invention, the prediction result presentation unit further outputs, to the user terminal, predicted health examination data corresponding to a predicted disease based on the predicted disease risk data.

[0014] In addition, the present invention provides a machine learning model generation device, comprising: a learning data acquisition unit that acquires learning fundus images, learning medical examination data, and learning disease risk data from a learning database; a first learned model generation unit that uses the learning medical examination data as correct answer data, inputs the learning fundus images as observation data into a neural network, and generates a first learned model; and a second learned model generation unit that uses the learning disease risk data as correct answer data, inputs the learning fundus images as observation data into a neural network, and generates a second learned model.

[0015] In addition, the present invention provides a disease risk determination device, comprising: a determination data acquisition unit that acquires determination fundus images and determination medical examination data from a determination database; a medical examination data prediction unit that inputs the determination fundus images and the determination medical examination data into a first learned model and outputs predicted medical examination data; a disease risk prediction unit that inputs the predicted medical examination data output from the medical examination data prediction unit and the determination fundus images into a second learned model and outputs predicted disease risk data; and a prediction result presentation unit that outputs at least the predicted disease risk data to a user terminal. The first learned model is generated by inputting the learning medical examination data as correct answer data and the learning fundus images as observation data into a neural network, and the second learned model is generated by inputting the learning disease risk data as correct answer data and the learning fundus images as observation data into a neural network.

[0016] In a disease risk determination device according to an aspect of the present invention, the prediction result presentation unit further outputs, to the user terminal, predicted medical examination data corresponding to a predicted disease based on the predicted disease risk data.

[0017] In addition, in the present invention, there is provided a disease risk determination method, including steps where a determination data acquisition unit acquires a determination fundus image and determination health examination data from a determination database; a health examination data prediction unit inputs the determination fundus image and the determination health examination data into a first pre-trained model and outputs predicted health examination data; a disease risk prediction unit inputs the predicted health examination data output from the health examination data prediction unit and the determination fundus image into a second pre-trained model and outputs predicted disease risk data; and a prediction result presentation unit outputs the predicted disease risk data and the predicted health examination data corresponding to the predicted disease based on the predicted disease risk data to a user terminal. The first pre-trained model is generated by inputting learning health examination data as correct data and learning fundus images as observation data into a neural network. The second pre-trained model is generated by inputting learning disease risk data as correct data and learning fundus images as observation data into a neural network.

[0018] There is provided a program characterized by causing a computer to execute each step of the above method.

[0019] In the present invention, "disease risk" refers to the risk related to contracting a disease, including the risk of already having contracted a disease and the risk of contracting a disease in the future.

[0020] In the present invention, "determining" the "disease risk" means obtaining data (such as scores, numerical values, indicators, ranks, etc.) indicating the disease risk derived by the system, apparatus, or method of the present invention, unless otherwise specified below, and does not include the medical act of actually diagnosing a disease by a doctor or the like.

[0021] In the present invention, "health examination data" refers to the basic health examination data of the examinee measured in a health check or the like. The health examination data includes, for example, gender, age at examination, height, weight, abdominal circumference, BMI, systolic blood pressure, and diastolic blood pressure. The health examination data includes health examination items and numerical data of those health examination items.

[0022] In the present invention, "disease risk data" refers to data indicating the risk of a disease, and specifically refers to data obtained by scoring the risk according to the type of the disease. The scored data is typically represented by numerical values. For example, when the disease is "depression", numerical values such as those of the PHQ-9 (Patient Health Questionnaire-9) used for the diagnosis of depression may be used.

[0023] In the present invention, "fundus image" refers to a photograph taken of the bottom of the eye of a subject. The bottom of the eye refers to the inner surface of the eye on the opposite side of the lens. The retina and the optic disc are also included in the bottom of the eye.

Advantages of the Invention

[0024] According to the present invention, it is possible to determine the disease risk only from the data obtained by non-invasive examinations without performing a blood test. That is, without performing a blood test, the health examination data obtained by a blood test can be predicted from the fundus image and used as virtual blood test data, so that predictions equivalent to those obtained by performing a blood test can be made. Therefore, it is expected to contribute to reducing the physical and mental burden on the subject due to the examination and promoting the opportunity to know the disease risk.

[0025] According to the present invention, it is possible to predict unobtained health examination data from the fundus image of a subject using a pre-trained model, and to determine the disease risk from the fundus image of the subject and the health examination data predicted by the pre-trained model using a different pre-trained model.

[0026] Furthermore, according to the present invention, it is possible to present disease risk data to a user terminal and provide prediction health examination data corresponding to the predicted disease based on the predicted disease risk data to the user terminal. Other objects, features, and advantages of the present invention will become apparent from the following description of the embodiments of the present invention with reference to the accompanying drawings.

Brief Description of the Drawings

[0027]

Figure 1

Figure 2

Figure 3

Figure 4

Mode for Carrying Out the Invention

Example

[0028] Figure 1 is a schematic diagram showing the overall disease risk determination system according to the present invention. The disease risk determination system 1 includes a machine learning model generation device 20 that generates a learned model for determining the risk of a disease, and a disease risk determination device 30 that determines the risk of a disease using the learned model generated by the machine learning model generation device 20.

[0029] The machine learning model generation device 20 includes a learning data acquisition unit 201, a first learned model generation unit 202, and a second learned model generation unit 203. The learning data acquisition unit 201 acquires learning fundus images, learning health examination data, and learning disease risk data from the learning database 41. The first learned model generation unit 202 inputs the learning health examination data as correct answer data and the learning fundus image as observation data into a neural network to generate a first learned model. The second learned model generation unit 203 inputs the learning disease risk data as correct answer data and the learning fundus image as observation data into a neural network to generate a second learned model. The first learned model is a learned model for predicting health examination data from a fundus image or the like. The second learned model is a learned model for predicting disease risk from a fundus image and predicted health examination data.

[0030] The disease risk determination device 30 includes a determination data acquisition unit 301, a medical examination data prediction unit 302, a disease risk prediction unit 303, and a prediction result presentation unit 304. The determination data acquisition unit 301 acquires determination fundus images and determination medical examination data from the determination database 42. The medical examination data prediction unit 302 inputs the determination fundus images and the determination medical examination data into the first trained model M1 and outputs predicted medical examination data. The disease risk prediction unit 303 inputs the predicted medical examination data output from the medical examination data prediction unit and the determination fundus images into the second trained model and outputs predicted disease risk data. The prediction result presentation unit 304 outputs at least the predicted disease risk data to the user terminal 50.

[0031] The learning database 41 stores learning data used for generating a trained model in the machine learning model generation device 20. The learning database 41 stores, as learning data, learning fundus images, learning medical examination data, and learning disease risk data. The learning fundus images are captured by an external inspection device such as a fundus image capturing device. The learning fundus images, the learning medical examination data, and the learning disease risk data are collected in advance from medical institutions, medical examination institutions, and the like. The learning database 41 may be constructed on a server or in the cloud.

[0032] The determination database 42 stores determination data used for determining the disease risk in the disease risk determination device 30. The determination database stores, as determination data, determination fundus images and determination health examination data. The determination fundus images are captured by an external examination device such as a fundus image capturing device. The determination fundus images are the fundus images of the subject to be determined for the disease risk, and are the fundus images captured using a fundus image capturing device or the like in a medical institution, a health examination institution, or the like. The determination health examination data is the health examination data of the subject obtained when examinations and health examinations are performed in a medical institution, a health examination institution, or the like. The determination health examination data is data obtained by an external examination device such as a height meter, a weight scale, a body fat meter, a body composition meter, a blood pressure monitor, etc. Typically, the determination health examination data is, for example, gender, age at visit, height, weight, abdominal circumference, BMI, systolic blood pressure, and diastolic blood pressure, etc. The determination health examination data is preferably data that can be obtained without performing a blood test. Another advantage of the present invention is that the disease risk can be determined only from the data obtained by non-invasive examinations. The determination health examination data may include interview data, and the interview data may be data obtained by scoring the answers to the interview items, and the scored data may be represented by numerical values. The learning database 41 may be constructed on a server or in the cloud.

[0033] The examination device is an external examination device such as a fundus image capturing device for capturing the fundus image of the subject, an external examination device such as a height meter, a weight scale, a body fat meter, a body composition meter, a blood pressure monitor, etc. for measuring the height, weight, body fat percentage, blood pressure, etc. of the subject, or an external instrument such as a tape measure for measuring the abdominal circumference of the subject. The examination device may be an existing medical device used for capturing fundus images in medical treatment, health diagnosis, etc. in a medical institution. The learning fundus images captured by the examination device and the learning health examination data obtained by the examination device are stored in the learning database 41. The determination fundus images captured by the examination device and the determination health examination data obtained by the examination device are stored in the determination database 42.

[0034] The user terminal 50 may be a mobile terminal such as a smartphone or a tablet in addition to a personal computer. The determination result output from the prediction result presentation unit 304 of the disease risk determination device 30 is displayed on the user terminal 50. At least the predicted disease risk data predicted by the disease risk prediction unit 303 is output to the user terminal 50 as the determination result of the disease risk. The user terminal 50 may further output prediction health examination data corresponding to the predicted disease based on the predicted disease risk data.

[0035] FIG. 2 is a diagram showing the generation process of the learned model in the machine learning model generation device according to the present invention. The machine learning model generation device 20 includes a first learned model generation unit 202 and a second learned model generation unit 203. The first learned model generation unit 202 is for generating the first learned model M1, and the second learned model generation unit 203 is for generating the second learned model M2. The first learned model M1 is a learned model for predicting health examination data from a fundus image or the like. The second learned model M2 is a learned model for predicting disease risk from a fundus image and prediction health examination data.

[0036] The first learned model generation unit 202 uses the learning fundus image and the learning health examination data acquired by the learning data acquisition unit 201 from the learning database 41 as learning data. The first learned model generation unit 202 inputs the learning health examination data as correct answer data and the learning fundus image as observation data into the neural network, and generates the first learned model M1.

[0037] The learning fundus image is used when the first learned model generation unit 202 generates the first learned model M1. The learning fundus image is captured using an external inspection device such as a fundus image capturing device, and images captured in advance at medical institutions, health examination institutions, etc. are collected and stored in the learning database 41.

[0038] The learning medical examination data is the medical examination data used when the first trained model generation unit 202 generates the first trained model M1. The learning medical examination data refers to the basic medical examination data of the examinee measured in a health check or the like. The medical examination data is, for example, gender, age at examination, height, weight, abdominal circumference, BMI, systolic blood pressure, and diastolic blood pressure, etc. The medical examination data includes a medical examination item and numerical data of the medical examination item.

[0039] The neural network used in the first trained model generation unit 202 may be composed of a first neural network and a second neural network different from the first neural network. In this case, the output from the first neural network may be used as the input to the second neural network.

[0040] The second trained model generation unit 203 uses the learning fundus image and the learning disease risk data acquired by the learning data acquisition unit 201 from the learning database 41 as learning data. The second trained model generation unit 203 inputs the learning disease risk data as correct answer data and the learning fundus image as observation data into a neural network, and generates a second trained model.

[0041] The learning fundus image is also used when the second trained model generation unit 203 generates the second trained model M2. Similar to that used by the first trained model generation unit 202, the learning fundus image is captured using an external inspection device such as a fundus image imaging device, and those captured in advance in a medical institution, a health check institution, etc. are collected and stored in the learning database 41.

[0042] The learning disease risk data is the disease risk data used when generating the second learned model M2 in the second learned model generation unit 203. The learning disease risk data may be represented by different indicators or scored data according to the type of disease. The scored data is typically represented by numerical values. For example, when the disease is "depression", the learning disease risk data may be the numerical value of PHQ-9 (Patient Health Questionnaire-9) used for the diagnosis of depression. The numerical value of PHQ-9 is a numerical value from 0 to 27.

[0043] The neural network used in the second learned model generation unit 203 may be composed of a first neural network and a second neural network different from the first neural network. In this case, the output from the first neural network may be used as the input to the second neural network.

[0044] The first neural network used in the first learned model generation unit 202 and the second learned model generation unit 203 may be a neural network that takes the output of a layer before a certain layer as input. The first neural network may be, for example, a so-called DenseNet.

[0045] The second neural network used in the first learned model generation unit 202 and the second learned model generation unit 203 may be a neural network having one intermediate layer between the input layer and the output layer. The second neural network may be, for example, a so-called Shallow Net.

[0046] FIG. 3 is a diagram showing prediction processing in the disease risk determination device according to the present invention. The disease risk determination device 30 includes a health examination data prediction unit 302 and a disease risk prediction unit 303. The health examination data prediction unit 302 is for predicting health examination data using the first trained model M1, and the disease risk prediction unit 303 is for predicting predicted disease risk data using the second trained model M2.

[0047] The health examination data prediction unit 302 uses the determination fundus image and the determination health examination data acquired by the determination data acquisition unit 301 from the determination database 42 as input data. The health examination data prediction unit 302 inputs the determination fundus image and the determination health examination data into the first trained model M1 and outputs predicted health examination data.

[0048] The disease risk prediction unit 303 uses the determination fundus image acquired by the determination data acquisition unit 301 from the determination database 42 and the predicted health examination data predicted by the health examination data prediction unit 302 as input data. The disease risk prediction unit 303 inputs the predicted health examination data output from the health examination data prediction unit 302 and the determination fundus image into the second trained model M2 and outputs predicted disease risk data.

[0049] The predicted disease risk data is the disease risk data predicted by the disease risk prediction unit 303 using the second trained model M2. The predicted disease risk data may be represented by different indicators or scored data according to the type of disease. The scored data is typically represented by a numerical value. For example, when the disease is "depression", the predicted disease risk data may be the numerical value of PHQ-9 (Patient Health Questionnaire-9) used for the diagnosis of depression. The numerical value of PHQ-9 is a numerical value from 0 to 27. The correlation between the fundus image and depression has been recognized in previous studies in the medical field and the like.

[0050] In other examples, the predicted disease risk data may indicate the risk of each of a plurality of diseases as a probability. For example, it may be a percentage indicating the probability of having a disease risk for each disease, such as "depression: 20%, diabetes: 5%, glaucoma: 3%, cardiovascular disease: 1%".

[0051] The diseases for which the risk can be predicted by the configuration of the disease risk determination system 1 of the present invention are not limited to depression. For any disease for which a correlation is recognized between the fundus image and the disease, the risk of that disease can be predicted by the configuration of the disease risk determination system 1 of the present invention. That is, for a disease in which some symptom, sign, or change appears in the fundus image when a person has the disease or is at risk of developing the disease in the future, the risk can be predicted by the configuration of the disease risk determination system 1 of the present invention. For example, in addition to depression, diabetes, glaucoma, heart disease, vascular disease, etc. have also been recognized to have a correlation with fundus images in conventional research in the medical field and other fields, and the risk can be predicted by the configuration of the disease risk determination system 1 of the present invention.

[0052] The prediction result presentation unit 304 outputs at least the predicted disease risk data predicted by the disease risk prediction unit 303 to the user terminal 50. The prediction result presentation unit 304 may further output prediction health examination data corresponding to the predicted disease based on the predicted disease risk data to the user terminal 50.

[0053] In an example where the predicted disease risk data is represented by different indicators or scored data according to the type of disease, the scored predicted disease risk data may be output to the user terminal 50. For example, when the disease is "depression", the numerical value of PHQ-9 (for example, a numerical value from 0 to 27) may be displayed on the user terminal 50. Further, the prediction health examination data corresponding to "depression" may be output to the user terminal 50.

[0054] In an example where the predicted disease risk data indicates the respective disease risks of a plurality of diseases in terms of probability, the probability of the disease risk for each disease may be output to the user terminal 50. For example, it may be output to the user terminal 50 in the form of a table or graph showing the probability of having a disease risk for each disease, such as "depression: 20%, diabetes: 5%, glaucoma: 3%, cardiovascular disease: 1%" in percentage. Further, the predicted medical examination data corresponding to each disease may be output to the user terminal 50. For example, for the data of "diabetes: 30%", the "HbA1c: 6.3%" of the predicted medical examination data may be output together.

[0055] In this way, by displaying the predicted medical examination data together, when a doctor or the examinee himself / herself actually judges the disease risk with reference to the prediction result, not only the types of diseases with a risk of contracting, but also the relevant medical examination items and their predicted values can be known, so there is an advantage that the efficiency of judgment can be improved.

[0056] Next, the disease risk determination method according to the present invention will be described. FIG. 4 is a flowchart showing the flow of the disease risk determination process according to the present invention. The disease risk determination process in the disease risk determination method of the present invention is roughly divided into a step of obtaining determination data (S401), a step of determining the disease risk by a second learned model (S402), and a step of presenting the prediction result of the disease risk (S403).

[0057] The disease risk determination method of the present invention is executed by a disease risk determination device 30 including a determination data acquisition unit 301, a health examination data prediction unit 302, a disease risk prediction unit 303, and a prediction result presentation unit 304. The disease risk determination method of the present invention includes steps in which the determination data acquisition unit 301 acquires a determination fundus image and determination health examination data from a determination database 42; the health examination data prediction unit 302 inputs the determination fundus image and the determination health examination data into a first learned model M1 and outputs predicted health examination data; the disease risk prediction unit 303 inputs the predicted health examination data output from the health examination data prediction unit and the determination fundus image into a second learned model M2 and outputs predicted disease risk data; and the prediction result presentation unit 304 outputs the predicted disease risk data and the predicted health examination data corresponding to the predicted disease based on the predicted disease risk data to a user terminal 50.

[0058] In the disease risk determination method of the present invention, the first learned model M1 is generated by inputting learning health examination data as correct answer data and learning fundus images as observation data into a neural network. In the disease risk determination method of the present invention, the second learned model is generated by inputting learning disease risk data as correct answer data and learning fundus images as observation data into a neural network. The first learned model M1 and the second learned model M2 are generated by the machine learning model generation device 20 of the present invention.

[0059] Further, a program for causing a computer to execute each step of the above-described disease risk determination method of the present invention may be provided. The program of the present invention may be stored in a computer-readable recording medium. Further, the program of the present invention may be stored in a memory within the disease risk determination system, or may be stored externally or on a cloud that cooperates with the disease risk determination system.

[0060] According to the present invention, it is possible to determine the disease risk only from the data obtained by non-invasive examination without performing a blood test. That is, without performing a blood test, the health examination data obtained by a blood test can be predicted from a fundus image and used as virtual blood test data, so that a prediction equivalent to that obtained by performing a blood test can be made. Therefore, it can be expected to contribute to reducing the physical and mental burden on the subject due to the examination and promoting the opportunity to know the disease risk. That is, even when the subject is not aware of the disease risk, there is an advantage that the disease risk can be known by a less burdensome examination from the stage of pre-disease before suffering from the disease. By being able to obtain sufficient information from a less burdensome examination, the opportunity to undergo the examination increases, and it can be expected to contribute to preventive medicine for avoiding the disease risk at the pre-disease stage.

[0061] Further, according to the present invention, it is possible to predict unobtained health examination data from a fundus image of a subject using a learned model, and to determine the disease risk from the fundus image of the subject and the health examination data predicted by the learned model using another different learned model. Therefore, there is an advantage that the same accuracy as when using the health examination data can be achieved without obtaining the health examination data.

[0062] Furthermore, according to the present invention, it is possible to present disease risk data to a user terminal and provide the user terminal with prediction health examination data corresponding to a predicted disease based on the predicted disease risk data. Therefore, when the user is a doctor or the like, when actually diagnosing and advising a subject with reference to the prediction result of the disease risk obtained by the disease risk determination system 1 according to the present invention, not only the disease risk data but also the prediction health examination data for that disease can be known at the same time, and the efficiency can be improved when making a diagnosis or giving advice. Also, when the user is the subject himself / herself, even without medical knowledge, it is possible to know the diseases with a high risk of suffering and the items of the health examination data that should be noted for that disease, which can be used for self-health management or can serve as a motivation to undergo a detailed examination or a medical examination by a doctor or the like. The above description has been made with respect to embodiments, but it is obvious to those skilled in the art that the present invention is not limited thereto and various changes and modifications can be made within the scope of the principle of the present invention and the scope of the appended claims.

Explanation of Signs

[0063] 1 Disease Risk Determination System 20 Machine Learning Model Generation Device 201 Learning Data Acquisition Unit 202 First Trained Model Generation Unit 203 Second Trained Model Generation Unit 30 Disease Risk Determination Device 301 Determination Data Acquisition Unit 302 Health Examination Data Prediction Unit 303 Disease Risk Prediction Unit 304 Prediction Result Presentation Unit 41 Learning Database 42 Determination Database 50 User Terminal M1 First Trained Model M2 Second Trained Model

Claims

1. A disease risk determination system comprising: a machine learning model generation device that generates a learned model for determining the risk of a disease; and a disease risk determination device that determines the risk of a disease using the learned model generated by the machine learning model generation device, wherein the machine learning model generation device comprises a learning data acquisition unit that acquires learning fundus images, learning health examination data, and learning disease risk data from a learning database; a first learned model generation unit that inputs the learning health examination data as correct data and the learning fundus images as observation data into a neural network to generate a first learned model; and a second learned model generation unit that inputs the learning disease risk data as correct data and the learning fundus images as observation data into a neural network to generate a second learned model, and wherein the disease risk determination device comprises a determination data acquisition unit that acquires determination fundus images and determination health examination data from a determination database; a health examination data prediction unit that inputs the determination fundus images and the determination health examination data into the first learned model and outputs predicted health examination data; a disease risk prediction unit that inputs the predicted health examination data output from the health examination data prediction unit and the determination fundus images into the second learned model and outputs predicted disease risk data; and a prediction result presentation unit that outputs at least the predicted disease risk data to a user terminal. A disease risk determination system, characterized in that it comprises the above.

2. The disease risk determination system according to claim 1, characterized in that the prediction result presentation unit further outputs, to a user terminal, the predicted health examination data corresponding to the predicted disease based on the predicted disease risk data.

3. A machine learning model generation device comprising: a learning data acquisition unit that acquires learning fundus images, learning health examination data, and learning disease risk data from a learning database; a first learned model generation unit that inputs the learning health examination data as correct data and the learning fundus images as observation data into a neural network to generate a first learned model; and a second learned model generation unit that inputs the learning disease risk data as correct data and the learning fundus images as observation data into a neural network to generate a second learned model. A machine learning model generation device, characterized in that it comprises the above.

4. A disease risk determination device comprising: A determination data acquisition unit that acquires a determination fundus image and determination health examination data from a determination database; A health examination data prediction unit that inputs the determination fundus image and the determination health examination data into a first learned model and outputs predicted health examination data; A disease risk prediction unit that inputs the predicted health examination data output from the health examination data prediction unit and the determination fundus image into a second learned model and outputs predicted disease risk data; A prediction result presentation unit that outputs at least the predicted disease risk data to a user terminal; Comprising; The first learned model is generated by inputting learning health examination data as correct data and learning fundus images as observation data into a neural network; The second learned model is generated by inputting learning disease risk data as correct data and the learning fundus image as observation data into a neural network, characterized by a disease risk determination device.

5. The disease risk determination device according to claim 4, wherein the prediction result presentation unit further outputs the predicted health examination data corresponding to the predicted disease based on the predicted disease risk data to a user terminal.

6. A disease risk determination method, comprising: A step in which a determination data acquisition unit acquires a determination fundus image and determination health examination data from a determination database; A step in which a health examination data prediction unit inputs the determination fundus image and the determination health examination data into a first learned model and outputs predicted health examination data; A step in which a disease risk prediction unit inputs the predicted health examination data output from the health examination data prediction unit and the determination fundus image into a second learned model and outputs predicted disease risk data; A step in which a prediction result presentation unit outputs the predicted disease risk data and the predicted health examination data corresponding to the predicted disease based on the predicted disease risk data to a user terminal; Including; The first learned model is generated by inputting learning health examination data as correct data and learning fundus images as observation data into a neural network; The second learned model is generated by inputting learning disease risk data as correct data and the learning fundus image as observation data into a neural network, characterized by a disease risk determination method.

7. A program, characterized in that a computer is caused to execute each step of the method according to claim 6.

Images