Processing device, processing program, processing method, and processing system

1 . A processing device, comprising:

an interface configured to communicate with a camera, the camera being configured to capture a plurality of subject images of a subject, the plurality of subject images of the subject including a part of an oral cavity of a user;

a memory configured to store computer readable instructions, the plurality of subject images, which are obtained via the interface, a trained determination image selection model, and a trained determination model; and

a processor configured to execute the computer readable instructions so as to:

perform a first machine leaning on a plurality of learning images having label information to learn a selection pattern for selecting a plurality of determination images from the plurality of subject images such that the trained determination image selection model is obtained, the plurality of learning images including a part of an oral cavity of a plurality of persons, the label information indicating whether an image, among the plurality of learning images, is used as one of the plurality of determination images, the plurality of determination images being used to determine whether the user contracts a predetermined disease;

perform a second machine learning on a plurality of learning determination images having disease positive label information, as the label information, to learn a positive rate determination pattern with respect to the plurality of determination images, the plurality of learning determination images including the part of the oral cavity of the plurality of persons;

obtain a plurality of candidate images by inputting the plurality of subject images into the trained determination image selection model, the plurality of candidate images having potential to become the plurality of determination images;

select the plurality of determination images from the plurality of candidate images based on a predetermined condition;

determine a possibility of contracting the predetermined disease of the user by inputting the plurality of determination images into the trained determination model to obtain a determination result; and

output the determination result to a display to display the possibility of contracting the predetermined disease of the user.

2 . The processing device according to claim 1 , wherein the part of the oral cavity is pharynx.

3 . The processing device according to claim 1 , wherein the part of the oral cavity is tonsils.

4 . The processing device according to claim 1 ,

wherein the processor is further configured to obtain the plurality of determination images by selecting the images, from the plurality of candidate images, having a potential value that is lower than a threshold of a predetermined value, as the predetermined condition, when compared with other images among the plurality of candidate images.

5 . The processing device according to claim 1 ,

wherein the processor is further configured to:

calculate a predetermined feature quantity by inputting the plurality of determination images to a trained feature quantity extractor for extracting the predetermined feature quantity from the plurality of determination images; and

determine the possibility of contracting the predetermined disease based on the predetermined feature quantity and the trained determination model.

6 . The processing device according to claim 1 ,

wherein the processor is further configured to:

acquire at least one of medical interview information and attribute information of the user; and

determine the possibility of contracting the predetermined disease based on the trained determination model, the plurality of determination images, and at least one of the medical interview information and the attribute information.

7 . The processing device according to claim 1 ,

wherein the processor is further configured to:

acquire at least one of medical interview information and attribute information of the user;

determine a first possibility of contracting the predetermined disease based on the trained determination model and the plurality of determination images without using either the medical interview information or the attribute information to obtain a first sub-determination result;

determine a second possibility of contracting the predetermined disease based on the trained determination model, the plurality of determination images, and at least one of the medical interview information and the attribute information to obtain a second sub-determination result; and

determine the possibility of contracting the predetermined disease based on the first sub-determination result and the second sub-determination result.

8 . The processing device according to claim 6 , wherein the attribute information is acquired from the plurality of subject images.

9 . The processing device according to claim 1 ,

wherein the processor is further configured to:

acquire external factor information related to the predetermined disease; and

determine the possibility of contracting the predetermined disease based on the external factor information, the trained determination model, and the plurality of determination images.

10 . A processing system, comprising:

the camera;

the display; and

the processing device according to claim 1 connected to the camera through a wired or wireless network.

11 . A computer program product embodying computer readable instructions stored on a non-transitory computer-readable storage medium for causing a computer to execute a process by a processor so as to perform the steps of:

storing a plurality of subject images, a trained determination image selection model, and a trained determination model into a memory, the plurality of subject images being obtained by capturing a subject by a camera, the plurality of subject images of the subject including a part of an oral cavity of a user;

performing a first machine leaning on a plurality of learning images having label information to learn a selection pattern for selecting a plurality of determination images from the plurality of subject images such that the trained determination image selection model is obtained, the plurality of learning images including a part of an oral cavity of a plurality of persons, the label information indicating whether an image, among the plurality of learning images, is used as one of the plurality of determination images, the plurality of determination images being used to determine whether the user contracts a predetermined disease;

performing a second machine learning on a plurality of learning determination images having disease positive label information, as the label information, to learn a positive rate determination pattern with respect to the plurality of determination images, the plurality of learning determination images including the part of the oral cavity of the plurality of persons;

obtaining a plurality of candidate images by inputting the plurality of subject images into the trained determination image selection model, the plurality of candidate images having potential to become the plurality of determination images;

selecting the plurality of determination images from the plurality of candidate images based on a predetermined condition;

determining a possibility of contracting the predetermined disease of the user by inputting the plurality of determination images into the trained determination model to obtain a determination result; and

outputting the determination result to a display to display the possibility of contracting the predetermined disease of the user.

12 . A processing method for causing a processor to execute a process, the processing method comprising executing on the processor the steps of:

storing a plurality of subject images, a trained determination image selection model, and a trained determination model into a memory, the plurality of subject images being obtained by capturing a subject by a camera, the plurality of subject images of the subject including a part of an oral cavity of a user;

performing a first machine leaning on a plurality of learning images having label information to learn a selection pattern for selecting a plurality of determination images from the plurality of subject images such that the trained determination image selection model is obtained, the plurality of learning images including a part of an oral cavity of a plurality of persons, the label information indicating whether an image, among the plurality of learning images, is used as one of the plurality of determination images, the plurality of determination images being used to determine whether the user contracts a predetermined disease;

performing a second machine learning on a plurality of learning determination images having disease positive label information, as the label information, to learn a positive rate determination pattern with respect to the plurality of determination images, the plurality of learning determination images including the part of the oral cavity of the plurality of persons;

obtaining a plurality of candidate images by inputting the plurality of subject images into the trained determination image selection model, the plurality of candidate images having potential to become the plurality of determination images;

selecting the plurality of determination images from the plurality of candidate images based on a predetermined condition;

determining a possibility of contracting the predetermined disease of the user by inputting the plurality of determination images into the trained determination model to obtain a determination result; and

outputting the determination result to a display to display the possibility of contracting the predetermined disease of the user.