Evaluation device, learning device, prediction device, evaluation method, program, and computer-readable storage medium

A learning device comprising:

• at least one processor;
• and
• at least one memory including computer program code, where the at least one memory and the computer program code are configured, with the at least one processor, to cause the learning device to:
• acquire:
• learning data that is information about cancer in an unspecified subject;
• and
• training data that is information about an effect of an anticancer agent obtained by administering the anticancer agent to cells collected from the unspecified subject, and
• generate a prediction model for making predictions related to therapy using the anticancer agent by causing a learning model to perform supervised learning for a corresponding relationship between the learning data and the training data acquired by the learning information acquisition unit,
• wherein the information about the effect of the anticancer agent includes information whether or not the anticancer agent is effective against the cancer cells, information indicating whether the cancer cells have acquired resistance to the anticancer agent, and a result of performing human evaluation from a result of the administering the anticancer agent.

A prediction device comprising:

• at least one processor;
• and
• at least one memory including computer program code, where the at least one memory and the computer program code are configured, with the at least one processor, to cause the prediction device to:
• input information that is information about cancer in a subject serving as a prediction target;
• and
• make related predictions to therapy using an anticancer agent with the input information and a prediction model,
• wherein the prediction model is a model for making the prediction related to the therapy using the anticancer agent generated by causing a learning model to perform supervised learning for a corresponding relationship between learning data that is information about cancer in an unspecified subject and training data that is information about an effect of the anticancer agent obtained by administering the anticancer agent to cells collected from the subject, and
• wherein the information about the effect of the anticancer agent includes information whether or not the anticancer agent is effective against the cancer cells, information indicating whether the cancer cells have acquired resistance to the anticancer agent, and a result of performing human evaluation from a result of the administering the anticancer agent.

An evaluation method of evaluating an anticancer effect, the evaluation method comprising:

• acquiring learning data that is information about cancer in an unspecified subject and training data that is information about an effect of an anticancer agent obtained by administering the anticancer agent to cells collected from the unspecified subject;
• generating a prediction model for making a prediction related to therapy using the anticancer agent by causing a learning model to perform supervised learning for a corresponding relationship between the learning data and the training data acquired by the learning information acquisition unit;
• storing the prediction model generated by the learning unit;
• acquiring input information that is information about cancer in a subject serving as a prediction target;
• and
• making a prediction related to the therapy using the anticancer agent with the input information and the prediction model,
• wherein the acquiring training data comprises acquiring the information about the anticancer effect obtained by administering the anticancer agent to a three-dimensional cell structure including cancer cells collected from the unspecified subject and cells constituting a stroma as the training data;
• wherein the acquiring learning data and training data comprises:
• acquires the state information indicating the cancer state in the unspecified subject, omics information of the cells collected from the unspecified subject, drug information about the anticancer agent, and administration performance information about the anticancer agent administered to the three-dimensional cell structure as the learning data, and
• acquires drug effectiveness information that is a result of determining whether or not the anticancer agent administered to the three-dimensional cell structure is effective as the training data:
• wherein the generating the prediction model comprises generating the prediction model for predicting an effect of the anticancer agent acting on cancer cells of the cancer patient on the basis of the state information in the cancer patient and the omics information of the cells collected from the cancer patient;
• wherein the drug effectiveness information includes a result of determining whether or not the cancer cells of the unspecified subject have acquired resistance to a predetermined first anticancer agent;
• wherein the administration performance information includes information indicating whether or not a second anticancer agent different from the first anticancer agent administered to the cancer cells of the unspecified subject has been administered after the cancer cells of the unspecified subject acquired resistance to the predetermined first anticancer agent;
• wherein the acquiring input information comprises acquiring subject information including the state information about the cancer patient serving as the prediction target and the omics information of cells collected from the cancer patient as the input information;
• wherein the making the prediction comprises predicting an effect of the second anticancer agent acting on the cancer cells of the cancer patient after the cancer cells of the cancer patient acquired the resistance to the first anticancer agent;
• and
• wherein the evaluation method further comprises outputting a prediction result indicating that an anticancer agent whose anticancer effect in administration after acquisition of the resistance by the unspecified subject is greater than or equal to a predetermined threshold value is an anticancer agent effective at the time of administration after the cancer cells of the unspecified subject acquire the resistance.