A clinician-oriented machine learning model for adult asthma exacerbation prediction: comparative analysis of nine algorithms.

Background: Asthma exacerbations significantly contribute to morbidity and healthcare burden, yet accurate risk prediction remains challenging. This study aimed to develop and validate a machine learning (ML)-based model to predict the exacerbation risk in adult asthma patients. Methods: This study analyzed data from the National Health and Nutrition Examination Survey (NHANES) collected between 2007 and 2012, comprising a cohort of 1,480 adult participants diagnosed with asthma. A total of 37 candidate features were assessed, and feature selection via least absolute shrinkage and selection operator (LASSO) identified seven key predictors. Nine ML models were developed and evaluated using the area under the curve (AUC) to determine the optimal model. The best-performing model underwent further validation using calibration curves, precision-recall (PR) curves, and decision curve analysis (DCA). Shapley Additive Explanations (SHAP) was applied to interpret the model's predictions. Results: A light gradient boosting machine (LightGBM) model showed the best predictive performance, with an AUC of 0.902 in the training set and comparable discrimination in the validation set. Model performance in the validation cohort showed consistent results across calibration analysis, PR curves, and DCA. Model interpretability was examined using SHAP and a web-based calculator was developed to support individualized risk estimation. Conclusions: This ML-based model demonstrated strong predictive accuracy and could serve as a valuable tool for risk assessment of adult asthma in clinical practice.