Second-Order Chemometric Classification of Candida Species Using Excitation-Emission Matrix Fluorescence...

Microorganisms of the genus Candida are responsible for a wide range of fungal infections and are associated with increased morbidity and mortality rates. Although several clinical methods are available for the diagnosis of candidiasis, they have limitations related to response time and laboratory complexity. In this study, spectroscopic techniques based on excitation-emission matrix (EEM) fluorescence were explored in combination with higher-order chemometric methods, such as PARAFAC and Tucker3, coupled with classification algorithms, including Quadratic Discriminant Analysis (QDA), for the identification of seven Candida classes: C. albicans, C. auris, C. glabrata, C. hemeulonii, C. krusei, C. parapsilosis, and C. tropicalis. The analyses were conducted using a limited data set consisting of seven Candida species measured in ten independent replicates, resulting in 70 data points. PARAFAC-QDA and Tucker3-QDA models were developed for multiclass classification. Although evaluated on a small-scale sample size, both models reached 100% classification accuracy, sensitivity, and specificity for training and test set, highlighting the potential of multiway analysis for early exploration. The selected excitation and emission wavelengths associated with the latent factors were mainly distributed in the excitation region of 260, 270, and 280 nm, and in the emission range of 280 to 350 nm, which are strongly related to aromatic amino acids, such as tyrosine, tryptophan, and phenylalanine, reflecting the structural composition of these microorganisms. These results demonstrate the potential of the proposed approach as a proof-of-concept for the discrimination of reference Candida strains under controlled culture conditions, supporting future validation using independent clinical isolates.