Researchers from the DQBM Krauthammer Lab contributed to a Scientific Reports study developing an early-warning system for antifungal resistance. The approach combines routinely acquired MALDI-TOF mass spectra with drug information and machine learning to predict whether fungal pathogens are resistant or susceptible.
Using 658 pathogen spectra and 3,046 resistance measurements, the researchers found that a neural-network model combined with principal component analysis achieved the strongest performance. In a retrospective assessment, the system identified resistance in 29% of cases where species-based guidelines could have led to treatment with an ineffective antifungal. The authors highlight the need for broader external and prospective validation before clinical implementation.
Publication: https://doi.org/10.1038/s41598-026-53519-y