Medical information processing apparatus, medical information processing method, and non-transitory computer-readable storage medium storing program

1. A medical information processing apparatus comprising:

processing circuitry configured to:

perform processing using a trained model that is generated by machine learning with a plurality of training data;

acquire first attribute data indicating at least one attribute of an object-to-be-trained related to the plurality of training data that are used for generating the trained model;

acquire second attribute data indicating at least one attribute of an object-to-be-diagnosed; and

perform the processing using the trained model based on goodness of fit that is an index indicating degree of matching between the first attribute data and the second attribute data, wherein

the at least one attribute of the object-to-be-trained comprises a plurality of types of attributes; and

the goodness of fit is determined based on (i) a distribution range of the first attribute data calculated for each of the plurality of types of attributes of the object-to-be-trained and (ii) number of types of attributes for which the second attribute data are outside the distribution range.

2. The medical information processing apparatus according to claim 1 , wherein the processing circuitry is further configured to:

input at least one of a medical image and biological information of the object-to-be-diagnosed into the trained model; and

output an inferred diagnosis result of the object-to-be-diagnosed.

3. The medical information processing apparatus according to claim 2 , wherein the processing circuitry is further configured to:

input input-data of the object-to-be-diagnosed into the trained model and perform the processing using the trained model when the goodness of fit satisfies a fitting condition that fits the trained model; and

perform processing of issuing a warning to a user or other alternative processing when the goodness of fit does not satisfy the fitting condition, the alternative processing being processing that does not use the trained model.

4. The medical information processing apparatus according to claim 3 , wherein the alternative processing includes processing of performing computer-aided diagnosis that is not based on the machine learning.

5. The medical information processing apparatus according to claim 3 , wherein the alternative processing includes (i) processing related to a display suggesting that a user should visually diagnose the medical image or (ii) processing related to a display that introduces another medical institution.

6. The medical information processing apparatus according to claim 1 , wherein the goodness of fit is calculated based on distance between a distribution range of the first attribute data and the second attribute data.

7. The medical information processing apparatus according to claim 1 , wherein types of the at least one attribute of the object-to-be-trained includes at least one of age, gender, body weight, race, area of birth, type of lifestyle, and a blood test value.

8. The medical information processing apparatus according to claim 1 , wherein the processing circuitry is further configured to:

input imaging raw data of the object-to-be-diagnosed into the trained model; and

output a reconstructed image of the object-to-be-diagnosed.

9. A medical information processing apparatus comprising:

processing circuitry configured to:

perform processing using a trained model that is generated by machine learning with a plurality of training data;

acquire first attribute data indicating at least one attribute of an object-to-be-trained related to the plurality of training data that are used for generating the trained model;

acquire second attribute data indicating at least one attribute of an object-to-be-diagnosed; and

perform the processing using the trained model based on goodness of fit that is an index indicating degree of matching between the first attribute data and the second attribute data, wherein

the at least one attribute of the object-to-be-trained comprises a plurality of types of attributes; and

the processing circuitry is configured to perform processing of issuing a warning to a user or other alternative processing when the at least one attribute of the object-to-be-diagnosed is a specific attribute and the goodness of fit of the specific attribute is equal to or less than a predetermined value, the alternative processing being processing that does not use the trained model.

10. A medical information processing method comprising:

performing processing using a trained model that is generated by machine learning with a plurality of training data;

acquiring first attribute data indicating at least one attribute of an object-to-be-trained related to the plurality of training data that are used for generating the trained model;

acquiring second attribute data indicating an attribute of an object-to-be-diagnosed; and

performing the processing on input-data of the object-to-be-diagnosed based on goodness of fit that is an index indicating degree of matching between the first attribute data and the second attribute data, wherein

the at least one attribute of the object-to-be-trained comprises a plurality of types of attributes; and

the goodness of fit is determined based on (i) a distribution range of the first attribute data calculated for each of the plurality of types of attributes of the object-to-be-trained and (ii) number of types of attributes for which the second attribute data are outside the distribution range.

11. A medical information processing method comprising:

performing processing using a trained model that is generated by machine learning with a plurality of training data;

acquiring first attribute data indicating at least one attribute of an object-to-be-trained related to the plurality of training data that are used for generating the trained model;

acquiring second attribute data indicating an attribute of an object-to-be-diagnosed; and

performing the processing on input-data of the object-to-be-diagnosed based on goodness of fit that is an index indicating degree of matching between the first attribute data and the second attribute data, wherein

the at least one attribute of the object-to-be-trained comprises a plurality of types of attributes; and

the method further comprises performing processing of issuing a warning to a user or other alternative processing when the at least one attribute of the object-to-be-diagnosed is a specific attribute and the goodness of fit of the specific attribute is equal to or less than a predetermined value, the alternative processing being processing that does not use the trained model.