AFFECTION DETERMINATION DEVICE FOR DISEASE, AFFECTION DETERMINATION METHOD FOR DISEASE, FEATURE EXTRACTION DEVICE FOR DISEASE, AND FEATURE EXTRACTION METHOD FOR DISEASE

摘要

PROBLEM TO BE SOLVED: To determine the affection of a disease by causing a neural network to learn by using data of an expression level of a biomarker, and to extract a featured biomarker as for the disease by the neural network.SOLUTION: The affection determination device of a disease is configured to acquire sample data in which an expression level of each of a plurality of types of biomarkers is recorded for each person, and to generate a learnt model capable of determining the affection of a disease preliminarily obtained by performing machine learning by using training data, and to input a plurality of sample data with label information indicating whether or not each person is affected by the disease attached thereto to the learnt model for an arithmetic operation, and to digitize significance of a feature of each of the plurality of biomarkers obtained by the learnt model by the arithmetic operation of affection determination for each sample data, and to extract the predetermined number of biomarkers as featured biomarkers related to the disease on the basis of the digitized significance of whole sample data for each biomarker.SELECTED DRAWING: Figure 2