Abstract
BACKGROUND: Aspergillus fumigatus sensitized asthma (AFSA) is associated with severe exacerbations and progressive lung damage; however, diagnosis remains challenging in resource-limited settings owing to limited access to Aspergillus-specific IgE (A. f-sIgE) testing. We aimed to develop a clinical prediction model using routinely available biomarkers for AFSA identification. METHODS: This retrospective study enrolled 92 adult patients with asthma at The First Hospital of Qinhuangdao between 2023 and 2025. Participants were classified into AFSA and non-AFSA groups. Candidate predictors (demographics and hematological parameters) were analyzed using Least Absolute Shrinkage and Selection Operator (LASSO) regression, with subsequent multivariable logistic regression. Performance was validated via receiver operating characteristic (ROC) curves, calibration plots, and decision curve analysis (DCA). RESULTS: Among 92 patients (mean age 56.5 ± 12.8 years; 60.9% female), 44.6% (n = 41) had AFSA. LASSO selected five predictors: sex, monocyte percentage, monocyte absolute count, lymphocyte percentage, and total IgE (TIgE). Final model retained male sex (Odds Ratio [OR] = 10.688; 95% Confidence Interval [CI]: 1.661-152.999) and TIgE (OR = 1.006; 95% CI: 1.003-1.011). The model achieved excellent discrimination: training cohort (Area Under the Curve [AUC] = 0.96, sensitivity = 0.93, specificity = 0.92); validation cohort (AUC = 0.88, sensitivity = 0.75, specificity = 1.00). Sex-specific TIgE cutoffs (527.5 IU/mL [males], 906.1 IU/mL [females]) yielded 79.2% accuracy. CONCLUSIONS: The developed prediction model using gender and TIgE provides a practical, accessible tool for AFSA screening, overcoming diagnostic barriers in settings lacking A. f-sIgE testing. However, this model remains exploratory and requires multicenter external validation before widespread clinical implementation.