Development and validation of a noninvasive prediction model for identifying eosinophilic asthma

Area of Science:

• Pulmonology
• Allergy and Immunology
• Biostatistics

Background:

• Eosinophilic asthma (EA) diagnosis via sputum analysis is crucial but challenging.
• Multidimensional assessment (MDA) offers a potential alternative for EA prediction.

Purpose of the Study:

• To develop and validate a predictive model for EA using MDA.
• To assess the diagnostic accuracy and clinical utility of the developed model.

Main Methods:

• Patients undergoing sputum induction were recruited.
• Variables were screened using LASSO and logistic regression to build a nomogram and web calculator.
• Internal and external validation was performed, including subgroup analysis.

Main Results:

• A five-variable model (gender, nasal polyp, blood eosinophils, blood basophils, FeNO) was developed.
• The model demonstrated strong diagnostic performance with C-indices of 0.86 (training) and 0.84 (validation).
• A user-friendly online web calculator was created for clinical application.

Conclusions:

• A validated MDA-based model accurately predicts EA.
• The model is clinically practical and serves as a valuable tool for diagnosing EA.
• This approach may enhance individualized asthma treatment strategies.