KR20220136226A - Method and apparatus for providing medical expectations using artificial intelligence model - Google Patents

Abstract

Provided is a prediction device operated by at least one processor, which includes: a risk factor inference model receiving medical images and outputting at least one inferred risk factor, as an artificial intelligence model trained to infer disease risk factors from input images; and a medical prediction model receiving patient information including the inferred at least one risk factor and outputting a medical prediction including a disease risk.

Description

Method and apparatus for providing medical prediction using artificial intelligence model

The present disclosure relates to artificial intelligence-based prediction technology.

Most of the risk factors influencing disease development have been identified. For example, it is known that breast cancer is affected by factors such as age, race, mammographic density, family history, age at menarche, fertility, and whether hormone therapy is used. Therefore, there is a study for breast cancer risk assessment using risk factors, and risk assessment exists not only for breast cancer but also for other carcinomas such as lung cancer.

Through disease risk assessment, medical staff can designate patients as high-risk groups and induce active testing for early detection. For example, in most guidelines, annual MRI scans are recommended for high-risk groups with a breast cancer risk of 20% or more, and appropriate tests are recommended for moderate-risk groups with a breast cancer risk of 15% to 20% depending on risk factors.

However, the conventional disease risk assessment program receives various variables used for risk assessment from the user and calculates the risk according to a probabilistic model based on an arithmetic expression or logistic regression analysis. Taking the conventional breast cancer risk assessment program as an example, not only patient information such as age, breast density, and fertility history, but also family history information such as mother and sister must be input on the screen. Therefore, the conventional disease risk assessment program has disadvantages in that usability is poor and predictive performance is deteriorated when unknown risk factors or inaccurate risk factors exist.

The present disclosure provides a method and apparatus for inferring risk factors of a disease from an input using an artificial intelligence model, and providing various medical predictions including disease risk using patient information including the inferred risk factors.

An artificial intelligence model trained to infer a risk factor of a disease from an input image as a prediction device operated by at least one processor according to an embodiment, receiving a medical image and outputting the inferred at least one risk factor a risk factor inference model, and a medical prediction model that receives patient information including the inferred at least one risk factor and outputs a medical prediction including a disease risk.

The medical prediction model may be implemented as an artificial intelligence model trained to output a medical prediction from an input or a statistical model that outputs a medical prediction calculated from the input.

The patient information may further include risk factors input from an interface screen provided to the user terminal or obtained from a database.

The patient information may further include disease information inferred from the medical image.

The at least one inferred risk factor may be input into the medical prediction model after being confirmed and/or corrected in the user terminal.

The medical prediction may further include at least one of a recommended test personalized based on the disease risk and the patient information, and a recommended test execution time.

According to an embodiment, a method of operating a prediction device operated by at least one processor, receiving a medical image of a patient, using a first artificial intelligence model trained to infer risk factors from the input image, Acquiring at least one risk factor inferred from the image, and performing a medical prediction including disease risk using patient information including the inferred at least one risk factor.

In the performing the medical prediction, a second artificial intelligence model trained to output a medical prediction from an input or a statistical model that outputs a medical prediction calculated from the input is used to obtain a medical prediction including the disease risk can do.

The method may further include receiving at least one risk factor through an interface screen provided to the user terminal, and adding the received risk factor to the patient information.

The operating method may further include adding at least one risk factor retrieved from a database to the patient information.

The operating method may further include adding disease information deduced from the medical image to the patient information.

In the performing the medical prediction, the medical prediction may be performed by providing the inferred at least one risk factor to the user terminal, and using the confirmed and/or modified risk factor in the user terminal.

The method of operation may further include providing a patient report prepared based on the medical prediction.

The patient report may further include at least one of a recommended test personalized based on the disease risk and the patient information, and a recommended test execution time.

The patient report may further include risk groups classified based on the disease risk.

As a computing device according to an embodiment, when a medical image is received, at least one risk factor is inferred from the medical image using a first artificial intelligence model trained to infer a risk factor of a disease from the input image, and the inferred The processor may include a processor for predicting a disease risk indicating a likelihood of developing a disease by using at least one risk factor.

The processor is a second artificial intelligence model trained to output a medical prediction from an input, or a statistical model that outputs a medical prediction calculated from an input, and inputs the inferred at least one risk factor to predict the disease risk can be obtained.

The processor is configured to include at least one of at least one risk factor input from an interface screen provided to the user terminal, at least one risk factor fetched from a database, and disease information inferred from the medical image together with the inferred at least one risk factor. The disease risk can be predicted using one.

The processor may provide at least one of a personalized recommended test and a recommended test execution time based on the disease risk and the acquired patient information.

The processor may provide the inferred at least one risk factor and/or the disease risk to the user terminal.

According to the embodiment, since a medical prediction is made using risk factors inferred through an artificial intelligence model without the user having to directly input risk factors necessary for medical prediction, user convenience can be improved.

According to the embodiment, even risk factors that the user is not aware of can be inferred through the artificial intelligence model, and medical prediction is made using the artificial intelligence model, so that prediction accuracy can be increased.

According to an embodiment, an inaccurate risk factor or an erroneously input risk factor can be corrected through an artificial intelligence model, and a medical prediction is made using the artificial intelligence model, so that prediction accuracy can be improved.

According to an embodiment, since a medical prediction is made from inferred risk factors through an artificial intelligence-based medical prediction model, it is possible to provide a variety of medical predictions with higher accuracy than existing statistical models.

According to the embodiment, it is possible to make a medical prediction including not only the disease risk indicating the possibility of developing a disease, but also whether to recommend additional tests, when to perform the additional tests, and the like, and a potential benefit for performing the additional tests can also be predicted.

1 is a diagram for explaining a prediction apparatus according to an embodiment.
2 is a diagram illustrating an interworking environment of a prediction apparatus according to an exemplary embodiment.
3 to 6 are diagrams for explaining the configuration of an artificial intelligence model according to an embodiment.
7 is a flowchart of a method for providing a medical prediction using an artificial intelligence model according to an embodiment.
8 is an example of an interface screen provided by a user terminal according to an embodiment.
9 is a hardware configuration diagram of a prediction apparatus according to an embodiment.

Hereinafter, with reference to the accompanying drawings, embodiments of the present disclosure will be described in detail so that those of ordinary skill in the art to which the present disclosure pertains can easily implement them. However, the present disclosure may be implemented in several different forms and is not limited to the embodiments described herein. And in order to clearly explain the present disclosure in the drawings, parts irrelevant to the description are omitted, and similar reference numerals are attached to similar parts throughout the specification.

In the description, when a part "includes" a certain component, this means that other components may be further included, rather than excluding other components, unless otherwise stated. In addition, terms such as “…unit”, “…group”, and “module” described in the specification mean a unit that processes at least one function or operation, which may be implemented as hardware or software or a combination of hardware and software. have.

The apparatus of the present disclosure is a computing device configured and connected to enable at least one processor to perform the operations of the present disclosure by executing instructions. The computer program includes instructions that are described for a processor to execute the operations of the present disclosure, and may be stored in a non-transitory computer readable storage medium. The computer program may be downloaded over a network or sold as a product.

The medical image of the present disclosure may be an image of various regions photographed with various modality, for example, the modality of the medical image may be X-ray, magnetic resonance imaging (MRI), ultrasound, It may be computed tomography (CT), mammography (MMG), digital breast tomosynthesis (DBT), or the like.

A user of the present disclosure may be a medical professional, such as a doctor, a nurse, a clinical pathologist, a sonographer, or a medical imaging expert, or a general public such as a patient or a guardian, but is not limited thereto.

An artificial intelligence model (AI model) of the present disclosure is a machine learning model for learning at least one task, and may be implemented as a computer program executed by a processor. The task that the AI model learns may refer to a task to be solved through machine learning or a task to be performed through machine learning. The AI model may be implemented as a computer program running on a computing device, downloaded over a network, or sold in the form of a product. Alternatively, the AI model can be linked with various devices through a network.

1 is a diagram illustrating a prediction apparatus according to an embodiment, and FIG. 2 is a diagram illustrating an interworking environment of the prediction apparatus according to an embodiment.

Referring to FIG. 1 , a prediction device 10 is a computing device operated by at least one processor. The prediction apparatus 10 may infer risk factors of a disease from an input image by using an artificial intelligence (AI) model. In addition, the prediction device 10 may provide a medical prediction including a disease risk using patient information including the inferred risk factors. Disease risk may indicate the likelihood of developing a disease. The disease risk may be a lifetime risk, or may be predicted, such as a risk of developing within 5 years, a risk of developing within 10 years, and the like.

Therefore, risk factors that the user must directly input for medical prediction can be reduced, and even risk factors that the user is not aware of are used for medical prediction, so prediction accuracy can be increased.

The prediction apparatus 10 may include at least one risk factor inference model 100 and at least one medical prediction model 200 .

The risk factor inference model 100 is an artificial intelligence model trained to infer at least one risk factor from an input image. The input image may be at least one medical image. The risk factor inference model 100 may learn a relationship between an input and an output during a training process, and may output a result inferred from a new input. The inferred risk factors may vary depending on the type of medical image, training data, and training method. The training data may be classified according to carcinoma. For training of the risk factor inference model 100 , training data in which at least some risk factors to be inferred from a medical image are labeled may be used.

During the training process, the risk factor inference model 100 may learn the relationship between the medical image and the risk factors. For example, the risk factor inference model 100 may learn a relationship between a mammography image and a breast density by using training data. Alternatively, in the training process, the risk factor inference model 100 further uses the patient/disease information together with the medical image to learn the relationship between the medical image, the patient/disease information, and the risk factor, thereby converting the patient/disease information into a network parameter. can be made into knowledge. The trained risk factor inference model 100 may infer risk factors from a medical image using network parameters in which patient/disease information is knowledgeable.

The risk factor inference model 100 may infer various risk factors according to training. For example, the risk factor inference model 100 may infer body information such as age and medical information such as breast density from a mammogram image.

Meanwhile, in addition to the factors reported as risk factors, the risk factor inference model 100 may infer any candidate factors expected to contribute to disease risk prediction. Thereafter, the candidate factors may be regarded as risk factors according to the reliability of the future prediction result, so that the risk factor inference model 100 may be utilized to search for risk factors for diseases.

To infer a plurality of risk factors, a plurality of risk factor inference models may be used, or a single risk factor inference model may be used. In the description, it is assumed that the risk factor inference model 100 infers a plurality of risk factors from a medical image.

The medical prediction model 200 is arranged to receive risk factors inferred from the risk factor inference model 100 . The medical prediction model 200 receives various patient information including inferred risk factors, and provides a medical prediction including disease risk. The medical prediction may include a personalized recommended scan based on patient information including disease risk and a recommended test execution time. In addition, the medical prediction may further include a prediction of the cancer diagnosis possibility according to whether a recommended test is performed. A medical prediction for a patient may be provided in the form of a patient-specific report.

The medical prediction model 200 may receive additional information in addition to the risk factors inferred from the risk factor inference model 100 . The additional information according to an embodiment is input from an interface screen provided by the prediction device 10 or information obtained from a database by the prediction device 10 and may be known risk factors related to the patient. For example, known risk factors for breast cancer may be the patient's age, family history, age at menarche, fertility, and whether or not hormone therapy is used. According to another embodiment, the additional information may be disease information (eg, disease risk) inferred from the input image. Also, the medical prediction model 200 may receive an input image used for inferring risk factors as additional information.

Meanwhile, the medical prediction model 200 may use the risk factor input as additional information as it is. However, the risk factor entered as additional information may be an incorrect risk factor or an incorrectly entered risk factor. Therefore, the medical prediction model 200 determines the inaccuracy or error of the risk factor input as additional information based on the risk factor inferred from the risk factor inference model 100, and determines whether the risk factor is inaccurate or incorrectly entered. Arguments may not be used for input. Alternatively, the medical prediction model 200 is based on the risk factor inferred from the risk factor inference model 100, corrects an inaccurate risk factor or an erroneously entered risk factor, and makes a medical prediction using the corrected risk factor. can

The medical prediction model 200 may be implemented as an artificial intelligence model. The medical prediction model 200 may learn a relationship between an input and an output during a training process, and may output a predicted result from a new input. The prediction result may vary depending on the type of input, training data, and training method. Meanwhile, the medical prediction model 200 is trained to predict the disease risk from patient information including risk factors. Therefore, the medical prediction model 200 classifies the patient information in the training data into a high-risk group (High), a medium-risk group (Mid), and a low-risk group (Low), and based on the divided training data for each group, the patient characteristics of the group You can study in depth.

According to another embodiment, the risk factor inference model 100 may be implemented to work in conjunction with a statistical model such as an existing risk assessment model, for example, a logistic regression-based probabilistic model. The existing risk assessment model has a problem in that prediction accuracy is lowered when predetermined risk factors are not input from the user. However, since the existing risk assessment model interworking with the risk factor inference model 100 can use risk factors inferred from the risk factor inference model 100 even if some risk factors are not input from the user, the usability of the existing risk assessment model , and the prediction accuracy can be improved.

As such, the prediction device 10 may provide a medical prediction, such as at least one risk factor inferred from the medical image by the risk factor inference model 100, and the disease risk inferred from the risk factor by the medical prediction model 200. have. In addition, the prediction device 10 may reversely analyze the input risk factors affecting the prediction result of the medical prediction model 200 and find risk factors contributing to the prediction result. The prediction device 10 may extract risk factors affecting the prediction result in the order of contribution, and provide it to the patient report. Through this, a patient or doctor can understand whether the risk factor that has a significant influence on the disease risk is age, family history, breast density, or lesion information detected in the image, and can establish an appropriate treatment plan. have. Also, the prediction apparatus 10 may find a risk factor that affects the likelihood of the disease from among various risk factor candidates based on information obtained through the backward analysis.

Referring to FIG. 2 , the prediction device 10 may be built as a standalone type or may be constructed as a linkage type with other devices. For example, the prediction apparatus 10 may be implemented to interwork with a plurality of user terminals 20 . In addition, the prediction device 10 includes various databases 30 of medical institutions, for example, a picture archiving and communication system (PACS), an electronic medical record (EMR), an electronic health record. (Electronic Health Record, EHR), etc., it is possible to acquire various clinical information of the patient. The user terminal 20 may provide a user interface screen that displays necessary information on the screen in conjunction with the prediction device 10 and the database 30 . The user terminal 20 may display information provided by the prediction device 10 through a dedicated viewer.

The prediction device 10 may be a server device, and the user terminal 20 may be a client terminal installed in a medical institution, and they may work together through a network. The prediction device 10 may be a local server connected to a network within a specific medical institution. The prediction device 10 may be a cloud server, and may work with terminals (medical staff terminals) of a plurality of medical institutions having access rights. The prediction device 10 may be a cloud server, and may interwork with a patient personal terminal having an access right.

The prediction apparatus 10 may receive a medical prediction request for a medical image from the user terminal 20 , and may respond to the medical prediction requested by the user terminal 20 . The prediction apparatus 10 may provide the risk factors inferred from the medical image to the user terminal 20 , and may request the user terminal 20 to confirm and/or correct the inferred risk factors. The prediction apparatus 10 may receive or bring additional information from the user terminal 20 or the database 30 according to the implemented artificial intelligence model. When the medical image is stored in the database 30 , the user terminal 20 may transmit the medical image obtained from the database 30 to the prediction device 10 , and the prediction device 10 may request medical treatment from the user terminal 20 . The image may be retrieved from the database 30 .

3 to 6 are diagrams for explaining the configuration of an artificial intelligence model according to an embodiment.

Referring to FIG. 3 , the artificial intelligence model implemented in the prediction device 10 may include a risk factor inference model 100 and a medical prediction model 200A.

The risk factor inference model 100 is an artificial intelligence model trained to infer risk factors from a medical image, and may receive a medical image and output the inferred risk factors. The inferred risk factors may vary depending on the type of medical image, training data, and training method. For example, the risk factor inference model 100 may infer medical information such as breast density, which is known as a risk factor for breast cancer, from a mammogram image, and may further infer physical information such as age and other risk factors. have.

The medical prediction model 200A is an artificial intelligence model trained to output a medical prediction from the input risk factors, and may receive inferred risk factors and output a medical prediction. The medical prediction output from the medical prediction model 200A may include disease risk, a recommended test, a recommended test execution time, and the like. In addition, the medical predictions may further include a prediction of the possibility of cancer diagnosis according to whether a recommended test is performed.

Meanwhile, the medical prediction model 200A may be implemented as an artificial intelligence model, or as a statistical model for calculating disease risk from input risk factors.

Referring to FIG. 4 , the artificial intelligence model implemented in the prediction device 10 may include a risk factor inference model 100 and a medical prediction model 200B.

As described with reference to FIG. 3 , the risk factor inference model 100 may receive a medical image and output inferred risk factors.

The medical prediction model 200B may receive known risk factors as additional information along with the inferred risk factors, and may output a medical prediction of a designated item from the input risk factors. For example, the medical prediction model 200B may receive known risk factors such as age, family history, menarche age, fertility history, and whether hormone treatment is performed along with breast density inferred from the medical image. The known risk factors input to the medical prediction model 200B may be input from the user terminal 20 through an interface screen provided by the prediction device 10 or may be obtained from the database by the prediction device 10 . In addition, the medical prediction model 200B may receive an input image used for risk factor inference as additional information.

The medical prediction model 200B may be implemented as an artificial intelligence model trained to output a medical prediction of a specified item from an input, or a statistical model that outputs a medical prediction calculated from the input.

Referring to FIG. 5 , the artificial intelligence model implemented in the prediction device 10 may include a risk factor inference model 100 , a medical prediction model 200C, and a disease prediction model 300 . . Here, the disease inference model 300 may be implemented in the prediction device 10 , but may be implemented in a separate device and then interwork with the prediction device 10 .

As described with reference to FIG. 3 , the risk factor inference model 100 may receive a medical image and output inferred risk factors.

The medical prediction model 200C may receive disease information inferred from the medical image as additional information together with risk factors inferred from the medical image. The disease information may be, for example, a disease risk inferred from a medical image. In addition, the medical prediction model 200C may receive an input image used for inferring risk factors as additional information.

The medical prediction model 200C may output a medical prediction of a designated item from the inferred risk factors and the inferred disease information. The medical prediction model 200C may be implemented as an artificial intelligence model trained to output a medical prediction from an input or a statistical model that outputs a medical prediction calculated from the input.

Additional information input to the medical prediction model 200C may be input from a separate disease inference model. The disease inference model 300 is an artificial intelligence model trained to infer disease information from a medical image, and may receive a medical image and output disease information inferred from the medical image.

Referring to FIG. 6 , the artificial intelligence model implemented in the prediction device 10 may include a risk factor inference model 100 , a medical prediction model 200D, and a disease prediction model 300 . Here, the disease inference model 300 may be implemented in the prediction device 10 , but may be implemented in a separate device and then interwork with the prediction device 10 .

As described with reference to FIG. 3 , the risk factor inference model 100 may receive a medical image and output inferred risk factors.

The medical prediction model 200D may receive, as additional information, known risk factors and disease information inferred from the medical image along with the risk factors inferred from the medical image. The disease information may be, for example, a disease risk inferred from a medical image. The known risk factors may be input from the user terminal 20 through an interface screen provided by the prediction device 10 , or may be obtained from the database by the prediction device 10 . Disease information inferred from the medical image may be provided from the disease prediction model 300 described with reference to FIG. 5 . In addition, the medical prediction model 200D may receive an input image used for inferring risk factors as additional information.

The medical prediction model 200D may output a medical prediction of a designated item from the input risk factors. The medical prediction model 200D may be implemented as an artificial intelligence model trained to output a medical prediction from an input or a statistical model that outputs a medical prediction calculated from the input.

7 is a flowchart of a method for providing a medical prediction using an artificial intelligence model according to an embodiment.

Referring to FIG. 7 , the prediction apparatus 10 receives a medical image of a patient ( S110 ).

The prediction apparatus 10 acquires at least one risk factor inferred from the input medical image by using an artificial intelligence model trained to infer risk factors from the input image ( S120 ). The prediction apparatus 10 may provide the risk factors inferred from the medical image to the user terminal 20 , and may request the user terminal 20 to confirm and/or correct the inferred risk factors.

The prediction device 10 uses the patient information including the inferred risk factor to perform a medical prediction including the disease risk ( S130 ). The prediction device 10 may obtain a predicted medical prediction from patient information using an artificial intelligence model trained to output a medical prediction from an input or a statistical model that outputs a medical prediction calculated from the input. The prediction device 10 may manage risk factors received from the user terminal 20 or risk factors retrieved from a database as known risk factors, and may add them to patient information. The prediction apparatus 10 may add disease information inferred from the medical image, for example, the disease risk inferred from the medical image, to the patient information.

The prediction device 10 provides a patient report prepared based on the medical prediction (S140). The medical prediction may include disease risk, including disease risk/probability of developing disease. The medical prediction may include a personalized recommended test based on patient information including disease risk, a recommended test execution time, and the like. In addition, the medical prediction may further include a prediction of the cancer diagnosis possibility according to whether a recommended test is performed.

8 is an example of an interface screen provided by a user terminal according to an embodiment.

Referring to FIG. 8 , the prediction apparatus 10 may provide a screen 21 for inputting a medical image, which is a medical prediction source, to the user terminal 20 . The user terminal 20 may display the medical image input screen 21 and transmit the medical image selected by the user to the prediction apparatus 10 . The user terminal 20 may receive a selection of a mammography image of a patient.

The prediction apparatus 10 receives a medical image of a patient and acquires at least one risk factor inferred from the medical image by using an artificial intelligence model trained to infer risk factors from the input image. The prediction apparatus 10 may provide the user terminal 20 with risk factors inferred from the medical image to the user terminal 20 . For example, the prediction apparatus 10 may infer risk factors including the patient's age and density from the medical image.

The user terminal 20 may display the inferred risk factors on the screen 22 and request confirmation of the inferred risk factors. On the screen 22 , if the risk factors are correct, the user may press the OK button, and after correcting the risk factors, press the OK button.

The prediction device 10 receives confirmation information about the risk factors inferred from the user terminal 20 , and uses the patient information including the identified risk factors to perform a medical prediction for the patient.

The user terminal 20 may provide the patient report prepared based on the medical prediction in the prediction device 10 to the screen 23 . The patient report may include predictable disease risk (eg, breast cancer risk) from the input image. Disease risk can be expressed as a risk score. The patient report may include information about high-risk, mid-risk, and low-risk groups classified according to the predicted disease risk score. Since it is difficult for the user to determine whether the disease risk is high based on the disease risk score alone, the risk group classified in the patient report may be provided so that the user can intuitively know the disease risk. In addition, the patient report may include a personalized recommended examination, when the recommended examination is performed, and the like. In addition, the patient report may further include a prediction of the possibility of cancer diagnosis according to whether a recommended test is performed.

Meanwhile, the patient report may be processed in a document form based on a medical prediction including risk factors and disease risk inferred by the prediction device 10 . The patient report may be prepared by the prediction device 10 , and may be provided to the user terminal 20 after a separate device prepares it based on information received from the prediction device 10 , or A viewer may display the information in the form of a document based on the information received from the prediction device 10 .

Patient reports can provide recommendations for testing intervals according to risk groups. For example, the patient report may provide information that recommends frequent tests at a frequency higher than a standard for high-risk patients and recommends frequent tests for low-risk patients with a frequency lower than the standard. Through this, low-risk patients can reduce the cost of unnecessary tests.

Patient reports may include personalized treatment plans, recommended tests, recommended treatments, and the like. For example, a patient report of a patient with a predicted higher than baseline ovarian cancer risk score would be “high risk with a risk score of 0.9, indicating the likelihood of developing ovarian cancer. You can recommend an ovarian cancer test through the phrase “Get an additional ovarian cancer test.”

In addition, as described in FIGS. 5 and 6 , when the prediction device 10 performs a medical prediction using the disease risk inferred from the image, the patient report specifically specifies the information used for the disease risk prediction. can For example, a patient report states, "The probability of cancer risk according to image analysis is predicted to be 50%, but when other risk factors are taken into account, the cancer risk probability is predicted to be as high as 80%, so caution is required", " Taking into account image analysis, family history, and breast density, the breast cancer risk score was 0.2, which is low-risk." It is possible to inform the information that contributed to the medical prediction and the prediction result according to the phrase.

Patient reports may include key risk factors that affected the patient's disease risk. The main risk factor is a risk factor that has a great influence on the prediction result among the input risk factors of the medical prediction model 200 , and may be extracted through reverse analysis. A patient or a doctor can understand whether a risk factor that has a significant influence on the disease risk is age, family history, breast density, or lesion information detected in an image, and can establish an appropriate treatment plan.

9 is a hardware configuration diagram of a prediction apparatus according to an embodiment.

Referring to FIG. 9 , the prediction device 10 is a computing device operated by one or more processors 11 , and includes a processor 11 , a memory 13 for loading a computer program executed by the processor 11 , and a computer It may include a storage 15 for storing programs and various data, a communication interface 17 , and a bus 19 connecting them. In addition, the prediction apparatus 10 may further include various components. The computer program may include instructions that, when loaded into the memory 13 , cause the processor 11 to perform methods/operations according to various embodiments of the present disclosure. That is, the processor 11 may perform methods/operations according to various embodiments of the present disclosure by executing instructions. A computer program consists of a series of computer readable instructions grouped by function, and refers to being executed by a processor. The computer program may include a risk factor inference model trained to infer a risk factor of a disease from an input image, and a medical prediction model that outputs a medical prediction including a disease risk from patient information including the input risk factor.

The processor 11 controls the overall operation of each component of the prediction device 10 . The processor 11 includes at least one of a central processing unit (CPU), a micro processor unit (MPU), a micro controller unit (MCU), a graphic processing unit (GPU), or any type of processor well known in the art of the present disclosure. may be included. In addition, the processor 11 may perform an operation on at least one application or computer program for executing the method/operation according to various embodiments of the present disclosure.

The memory 13 stores various data, commands and/or information. The memory 13 may load one or more computer programs from the storage 15 to execute methods/operations according to various embodiments of the present disclosure. The memory 13 may be implemented as a volatile memory such as RAM, but the technical scope of the present disclosure is not limited thereto.

The storage 15 may non-temporarily store a computer program. The storage 15 is a non-volatile memory such as a read only memory (ROM), an erasable programmable ROM (EPROM), an electrically erasable programmable ROM (EEPROM), a flash memory, a hard disk, a removable disk, or well in the art to which the present disclosure pertains. It may be configured to include any known computer-readable recording medium.

The communication interface 17 supports wired/wireless Internet communication of the prediction device 10 . In addition, the communication interface 17 may support various communication methods other than Internet communication. To this end, the communication interface 17 may be configured to include a communication module well known in the art of the present disclosure.

The bus 19 provides communication functions between the components of the prediction device 10 . The bus 19 may be implemented as various types of buses, such as an address bus, a data bus, and a control bus.

The embodiments of the present disclosure described above are not implemented only through the apparatus and method, and may be implemented through a program for realizing a function corresponding to the configuration of the embodiment of the present disclosure or a recording medium in which the program is recorded.

Although the embodiments of the present disclosure have been described in detail above, the scope of the present disclosure is not limited thereto, and various modifications and improved forms of the present disclosure are also provided by those skilled in the art using the basic concept of the present disclosure as defined in the following claims. is within the scope of the right.

Claims (20)

A prediction device operated by at least one processor, comprising:
An artificial intelligence model trained to infer risk factors of a disease from an input image, a risk factor inference model that receives a medical image and outputs at least one inferred risk factor, and
A medical prediction model that receives patient information including the inferred at least one risk factor and outputs a medical prediction including the disease risk
Prediction device comprising. In claim 1,
The medical prediction model is
A prediction device, implemented as an artificial intelligence model trained to output a medical prediction from an input, or a statistical model that outputs a medical prediction calculated from the input. In claim 1,
The patient information is
Input from the interface screen provided to the user terminal, or further comprising risk factors fetched from the database, prediction device. In claim 1,
The patient information is
Prediction apparatus further comprising disease information inferred from the medical image. In claim 1,
The at least one inferred risk factor is confirmed and/or corrected in the user terminal, and then is input to the medical prediction model. In claim 1,
The medical prediction is
The prediction device further comprising at least one of a recommended test personalized based on the disease risk and the patient information, and a recommended test execution time. A method of operating a prediction device operated by at least one processor, the method comprising:
receiving a patient's medical image;
obtaining at least one risk factor inferred from the medical image by using a first artificial intelligence model trained to infer risk factors from the input image; and
performing a medical prediction including disease risk by using the patient information including the inferred at least one risk factor
A method of operation comprising a. In claim 7,
The step of performing the medical prediction is
An operating method of acquiring a medical prediction including the disease risk by using a second artificial intelligence model trained to output a medical prediction from an input or a statistical model that outputs a medical prediction calculated from the input. In claim 7,
receiving at least one risk factor through an interface screen provided to the user terminal, and adding the received risk factor to the patient information
Further comprising, the method of operation. In claim 7,
adding at least one risk factor retrieved from a database to the patient information;
Further comprising, the method of operation. In claim 7,
adding disease information inferred from the medical image to the patient information
Further comprising, the method of operation. In claim 7,
The step of performing the medical prediction is
An operating method of providing the inferred at least one risk factor to a user terminal, and performing the medical prediction by using the checked and/or modified risk factor in the user terminal. In claim 7,
providing a patient report prepared based on the medical prediction;
An operation method further comprising a. In claim 13,
The patient report
The method of claim 1, further comprising: a recommended test personalized based on the disease risk and the patient information, and at least one of a recommended test execution time. In claim 13,
The patient report
Further comprising a risk group classified based on the disease risk, the operating method. When a medical image is input, at least one risk factor is inferred from the medical image using a first artificial intelligence model trained to infer a risk factor of a disease from the input image, and a disease using the inferred at least one risk factor A computing device, comprising: a processor for predicting a disease risk indicative of a likelihood of developing an illness. 17. In claim 16,
the processor
A second artificial intelligence model trained to output a medical prediction from an input or a statistical model that outputs a medical prediction calculated from the input, inputting the inferred at least one risk factor to obtain the predicted disease risk , computing devices. 17. In claim 16,
the processor
At least one of at least one risk factor input from an interface screen provided to a user terminal, at least one risk factor retrieved from a database, and disease information inferred from the medical image together with the inferred at least one risk factor is used to predict the disease risk, computing device. 17. In claim 16,
the processor
A computing device that provides at least one of a personalized recommended test and a recommended test execution time based on the disease risk and acquired patient information. 17. In claim 16,
the processor
and providing the inferred at least one risk factor and/or the disease risk to a user terminal.

Images