ARTIFICIAL INTELLIGENCE IN PEDIATRIC ASTHMA MANAGEMENT: PREDICTIVE MODELING FOR EXACERBATION RISK AND PERSONALIZED TREATMENT OPTIMIZATION

Abstract

Background: Asthma is the most common chronic disease of childhood, with exacerbation prediction and treatment personalization remaining key clinical challenges. Objective: To develop and validate machine learning models for predicting acute asthma exacerbations and optimizing preventive treatment in children. Methods: A prospective cohort of 6,847 children with physician-diagnosed asthma was followed for 24 months. Random Forest and neural network models were trained on clinical, environmental, and digital health sensor data. Results: The exacerbation prediction model achieved AUC-ROC of 0.88 (sensitivity 83.2%, specificity 87.4%) at 14-day prediction horizon. Environmental feature integration (air quality, allergen levels, viral surveillance data) improved model performance by 12.3% over clinical data alone. AI-guided treatment optimization reduced exacerbation rate by 34.7% compared to standard care. Conclusion: AI integrating clinical, environmental, and digital health data substantially improves pediatric asthma exacerbation prediction and treatment personalization.