Novel Machine Learning Can Predict Acute Asthma Exacerbation.

Area of Science:

• Pulmonary Medicine
• Data Science
• Health Informatics

Background:

• Asthma exacerbations pose significant health and economic burdens.
• Predicting asthma exacerbations remains a challenge in clinical practice.

Purpose of the Study:

• To evaluate the efficacy of machine learning (ML) models in predicting asthma exacerbations.
• To assess the utility of large-scale outpatient electronic health record (EHR) data for predictive modeling.

Main Methods:

• Analysis of EHR data from 60,302 asthma patients (2010-2018).
• Development of three ML models (logistic regression, random forests, gradient boosting) to predict nonsevere exacerbations, ED visits, and hospitalizations.
• Inclusion of demographic data, comorbidities, lab values, and medication history as covariates.

Main Results:

• The light gradient boosting machine model demonstrated the best predictive performance.
• Area under the ROC curve for predicting nonsevere exacerbation, ED visits, and hospitalizations were 0.71, 0.88, and 0.85, respectively.
• Key risk factors identified include age, long-acting beta-agonist use, and inhaled/oral glucocorticoid therapy; low FEV1 and FEV1/FVC ratio were significant in a subgroup analysis.

Conclusions:

• ML models utilizing real-world EHR data can accurately predict asthma exacerbations.
• These predictive models have the potential to be integrated into clinical decision support tools.
• Enhanced outpatient care and prevention of adverse asthma outcomes are possible through ML-driven predictions.