Medical information processing apparatus
A medical information processing apparatus according to an embodiment includes a processing circuitry. The processing circuitry is configured: to register first relevance information relevant to a first trained model to be newly generated; to calculate, with respect to each of a plurality of second trained models being existing trained models, a similarity degree between second relevance information relevant to the second trained model and the first relevance information; to calculate a data quantity required to generate the first trained model with respect to each of the plurality of second trained models, on the basis of the similarity degrees each corresponding to a different one of the plurality of second trained models; and to output the data quantity required to generate the first trained model with respect to each of the plurality of second trained models.
1. A medical information processing apparatus, comprising:
processing circuitry configured to
store, in an electronic memory, a first medical information training set that was used to generate a first trained machine-learning model, the first medical information training set including first medical image data obtained from scanning performed using a medical imaging apparatus, the first trained model being trained machine-learning to identify an object in the first medical image data;
calculate, with respect to each second trained machine-learning model of a plurality of second trained machine-learning models being previously trained machine-learning models, a similarity degree between (1) a second medical information training set having been used in training the second trained machine-learning model, and (2) the first medical information training set, the second medical information training set including second medical image data obtained from scanning performed using the medical imaging apparatus, the second trained machine-learning model being trained to identify the object in the second medical image data;
calculate, for each second trained machine-learning model of the plurality of second trained machine-learning models, a data quantity of the first medical information training set that causes, a precision of the second trained machine-learning model to achieve a target precision value by using data extracted from the electronic memory and indicating a correlational relationship between similarity degrees and corresponding data quantities that each achieves the target precision value when training is performed using image data having a corresponding similarity degree, wherein each second trained machine-learning model of the plurality of second trained machine-learning models can be retrained to become a new trained machine-learning model, based on the calculated similarity degree corresponding to the second trained machine-learning model;
display, on a display device, a list of the calculated data quantity for each of the plurality of second trained machine-learning models;
receive selection, by a user, of a particular one of the plurality of second trained machine-learning models displayed in the list of the calculated data quantity for each of the plurality of second trained machine-learning models; and
in response to the selection by the user, re-train the particular one of the plurality of second trained machine-learning models selected by a the user to generate the new trained machine-learning model, using the corresponding calculated data quantity of the first medical information training set as training data.
2. The medical information processing apparatus according to claim 1 , wherein the processing circuitry is further configured to
estimate a correlational relationship between the data quantity and an accuracy rate of an output result from each of the plurality of second trained machine-learning models, and
output information indicating the estimated correlational relationship.
3. The medical information processing apparatus according to claim 2 , wherein the processing circuitry is further configured to
calculate the data quantity based on the estimated correlational relationship, and
output the calculated data quantity.
4. The medical information processing apparatus according to claim 1 , wherein the processing circuitry is further configured to
designate a category of the first trained machine-learning model in deep learning,
from trained model information keeping the second trained machine-learning models, categories thereof, and the second medical information training set in correspondence with one another, extract one or more of the second trained machine-learning models that belong to the designated category, and
calculate the similarity degree with respect to each of the one or more extracted second trained machine-learning models.
5. The medical information processing apparatus according to claim 4 , wherein the processing circuitry is further configured to
register the first medical information training set including first label data that is information identifying the first medical information, and
designate the category based on the first label data.
6. The medical information processing apparatus according to claim 5 , wherein the processing circuitry is further configured to calculate the similarity degree based on the first label data of the first medical information training set and second label data that is information identifying the second medical information training set.
7. The medical information processing apparatus according to claim 1 , wherein the processing circuitry is further configured to calculate the similarity degree between the first medical information training set, which includes first image data for evaluating a capability of a first apparatus that generated the first image data used for generating the first trained machine-learning model and the second medical information training set, which includes second image data for evaluating a capability of a second apparatus that generated the second image data used for generating each of the second trained machine-learning models.
8. The medical information processing apparatus according to claim 1 , wherein the processing circuitry is further configured to calculate the similarity degree between the first medical information training set, which includes a first obtaining condition of image data used for generating the first trained machine-learning model and the second medical information training set, which includes a second obtaining condition of image data that was used for generating each of the plurality of second trained, which includes models.
9. The medical information processing apparatus according to claim 1 , wherein the processing circuitry is further configured to
extract an item of obtaining conditions that improves the similarity degree between the first medical information training set indicating an obtaining condition of image data used for generating the first trained machine-learning model and the second medical information training set indicating an obtaining condition of image data that was used for generating each of the plurality of second trained machine-learning models, and
output the extracted item.