Application of Raman spectroscopy and machine learning for Candida auris identification and characterization

Candida auris, an emerging fungal pathogen characterized by multidrug resistance and high-mortality nosocomial infections, poses a serious global health threat. However, the precise and rapid identification and characterization of C. auris remain a challenge. Here, we employed Raman spectroscopy combined with machine learning to identify C. auris isolates and its closely related species as well as to predict antifungal resistance and key virulence factors at the single-cell level. The average accuracy of identification among all Candida species was 93.33%, with an accuracy of 98% for the clinically simulated samples. The drug susceptibility of C. auris to fluconazole and amphotericin B was 99% and 94%, respectively. Furthermore, the phenotypic prediction of C. auris yielded an accuracy of 100% for aggregating cells and 97% for filamentous cells. This proof-of-concept methodology not only precisely identifies C. auris at the clade-specific level but also rapidly predicts the antifungal resistance and biological characteristics, promising a valuable medical diagnostic tool to combat this multidrugresistant pathogen in the future.