Methods for creating training data, methods for identifying microorganisms, analysis equipment, programs

Abstract

To improve the quality of learning data of a machine learning model for discriminating microorganisms applying mass spectrometry, and in accordance therewith, to provide a machine learning model of high discrimination accuracy.SOLUTION: A method for creating learning data of a machine learning model includes: a step S2 for creating collection data containing a mass spectrum obtained by executing mass analysis of microorganisms; a step for determining whether a specific condition is satisfied with regard to the mass spectrum contained in the collection data; and a step S8 for creating learning data using a mass spectrum satisfying the specific condition. The step for determining whether a specific condition is satisfied includes: a step S4 for determining whether a first condition, in which a concordance rate of m / z corresponding to peaks of multiple mass spectra obtained by applying mass analysis multiple times to the same microorganism is a first numerical value or more, is satisfied; and a step S6 for determining whether a second condition, in which a peak of a specific pattern not reflecting features of microorganism classification is not included, is satisfied.SELECTED DRAWING: Figure 4

Images