Discriminant analysis and neural networks for the identification of phytopathogenic fungi by isothermal microcalorimetry

This study introduces an innovative method for identifying phytopathogenic fungi through the application of discriminant analysis based on thermokinetic parameters derived from integrated heat flow-time curves obtained via isothermal microcalorimetry. By merging machine learning techniques with multivariate analysis, the research develops a reliable automated tool for fungal identification based on thermogenic analysis. The canonical discriminant analysis effectively distinguishes among the genera Colletotrichum, Penicillium, and Alternaria, providing discriminant canonical variables that served as the foundation for training various machine learning models. Likewise, the neural network model achieved an impressive 95 % fit to the training data, with a low misclassification over around 10 %. The study also discusses the criteria for discrimination and proposes a microcalorimetric database aimed at enhancing future machine learning systems for the identification of phytopathogenic fungi. This methodology represents a pioneering approach that integrates microcalorimetric data analysis with advanced computational techniques, paving the way for more sophisticated and automated diagnostic tools in the field of plant pathology.