Disease prediction method, apparatus, and computer program

An apparatus for predicting disease, the apparatus comprising:

• a memory storing one or more instructions;
• and
• a processor configured to perform, by executing the stored one or more instructions,
• an operation of constructing a disease prediction model by learning ribosome data for training and training data including disease information,
• an operation of acquiring ribosome test data of an examinee, and
• an operation of predicting disease information of the examinee from the ribosome test data using the disease prediction model,
• wherein the ribosome data for training and the ribosome test data include data related to expression rates of ribosomal proteins,
• wherein constructing the disease prediction model comprises:
• constructing a temporary disease prediction model by learning the training data;
• generating second ribosome data samples by changing at least part of expression rates of ribosomal proteins in first ribosome data samples constituting the ribosome data for training;
• acquiring a first prediction value by entering the first ribosome data samples into the temporary disease prediction model and acquiring a second prediction value by entering the second ribosome data samples into the temporary disease prediction model;
• detecting an expression pattern of ribosomal proteins associated with the disease information based on a difference between the first prediction value and the second prediction value;
• assigning sample weights to data samples constituting the training data using the detected expression pattern;
• and
• constructing the disease prediction model by re-learning the training data based on the sample weights,
• wherein the training data are stored in a memory of the computing device, and
• wherein constructing the disease prediction model is executed by a processor of the computing device performing numerical operations on the training data stored in the memory.

A computer program stored in a computer-readable recording medium, the computer program, being combined with a computing device, comprising:

• constructing a disease prediction model by learning ribosome data for training and training data including disease information;
• acquiring ribosome test data of an examinee;
• and
• predicting disease information of the examinee from the ribosome test data using the disease prediction model, wherein
• the ribosome data for training and the ribosome test data include data related to expression rates of ribosomal proteins,
• wherein constructing the disease prediction model comprises:
• constructing a temporary disease prediction model by learning the training data;
• generating second ribosome data samples by changing at least part of expression rates of ribosomal proteins in first ribosome data samples constituting the ribosome data for training;
• acquiring a first prediction value by entering the first ribosome data samples into the temporary disease prediction model and acquiring a second prediction value by entering the second ribosome data samples into the temporary disease prediction model;
• detecting an expression pattern of ribosomal proteins associated with the disease information based on a difference between the first prediction value and the second prediction value;
• assigning sample weights to data samples constituting the training data using the detected expression pattern;
• and
• constructing the disease prediction model by re-learning the training data based on the sample weights,
• wherein the training data are stored in a memory of the computing device, and
• wherein constructing the disease prediction model is executed by a processor of the computing device performing numerical operations on the training data stored in the memory.