Development of a Prognostic Nomogram for Nonneutropenic Invasive Pulmonary Aspergillosis Based on Machine Learning.

BACKGROUND

The incidence of invasive pulmonary aspergillosis (IPA) is progressively rising in the nonneutropenic population, but studies investigating relevant prognostic factors remain scarce.

METHODS

Participants who were hospitalized at Zhongshan City People's Hospital from January 2018 to May 2023 and diagnosed with nonneutropenic deficient IPA were included in this study. The least absolute shrinkage and selection operator (LASSO) regression and multivariate Cox regression methods were used to select variables for constructing the predictive model. The performance of the predictive model was evaluated using the concordance index (C-index), calibration curve, time-dependent receiver operating characteristic (T-ROC) curve, area under the curve (AUC), and decision curve analysis (DCA). Finally, prognostic risk stratification was performed for nonneutropenic IPA patients, transforming the nomogram into a risk-stratified prognostic model.

RESULTS

A total of 210 participants were included in this study, divided into training and validation cohorts at a ratio of 7.5:2.5. Lasso regression identified seven potential predictive factors, including age, comorbid bacterial pneumonia, pleural effusion, neutrophil to lymphocyte ratio (NLR), lactate dehydrogenase (LDH), invasive mechanical ventilation and ICU treatment. Multivariate Cox regression analysis showed age (HR=1.02), comorbid bacterial pneumonia (HR=3.36), NLR (HR=1.02), LDH (HR=1.001), and invasive mechanical ventilation (HR=4.86) as independent predictive factors and constructed nomogram. The calibration curves show that the nomogram performs well in terms of consistency between predictions and actual observations. The T-ROC curves and DCA of the nomogram model show that the recognition ability of the nomogram model was outstanding. Participants could be classified into high and low-risk groups based on the final score of this nomogram, demonstrating the excellent risk stratification performance of our model.

CONCLUSION

The nomogram model developed in this study is an effective tool for predicting mortality risk in nonneutropenic IPA patients, aiding clinicians in identifying high-risk patients and optimizing early treatment strategies.