Classifying patients with invasive fungal disease: towards a unified case definition?

Management of invasive fungal disease (IFD) is increasingly challenging due to recognition of novel at-risk groups, emergence of new fungal pathogens and antifungal drug resistance. Together with the availability of new diagnostic tests and treatment modalities, robust and broadly applicable IFD case definitions are critical to support research. However, the ability to classify IFDs with the current definitions has decreased, prompting the development of new case definitions. Furthermore, current case definitions rely on a single positive test as mycological evidence, while not considering discordant evidence. We propose to explore the development of a machine learning (ML)-based IFD classification model, which uses algorithms to automatically 'learn' from observed data to consistently and accurately classify IFDs. Although developing and validating an ML-based IFD classification model is a significant undertaking, such an endeavour should be considered a worthwhile investment by the mycology community to standardize and reduce the ambiguity in the diagnosis of non-proven IFD.