Electronic health records improve the diagnosis and prognosis of pulmonary fungal infections in lung cancer patients

Abstract Pulmonary fungal infections (PFIs) represent a life‐threatening complication in lung cancer patients. However, their clinical management remains challenging due to limitations in current diagnostic and prognostic methods. Electronic health records (EHRs), which systematically capture a broad spectrum of clinically relevant data, offer a valuable resource for delineating complex disease phenotypes. Therefore, we conducted a retrospective and prospective study involving 645 patients to develop diagnostic and prognostic tools based on EHR data. A diagnostic model incorporating seven key features achieved an area under the curve (AUC) of 0.959 (95% confidence interval: 0.937‒0.980), while a five‐feature prognostic model yielded an average time‐dependent AUC of 0.800. Notably, the integration of these models into the clinical decision‒support workflow proves particularly beneficial in terms of cost efficiency and time savings (>80%) for patients susceptible to excessive testing with conventional workflow. Collectively, these findings demonstrate that EHR can improve the clinical management of PFIs in lung cancer patients.