# Artificial Intelligence Applications in the Prediction and Management of Pediatric Asthma Exacerbation: A Systematic Review

## TL;DR

This paper reviews how artificial intelligence can predict and manage asthma attacks in children, finding that AI models can outperform traditional methods.

## Contribution

The study systematically evaluates AI applications for pediatric asthma exacerbation prediction and management, highlighting performance and challenges.

## Key findings

- AI models outperformed traditional methods in predicting pediatric asthma exacerbations.
- Multimodal data improved AI performance in asthma prediction.
- Most studies had low risk of bias, but real-world impact remains understudied.

## Abstract

Pediatric asthma exacerbations remain a significant global health challenge due to their unpredictable nature and potential for severe morbidity. While artificial intelligence (AI) shows promise in improving prediction and management, the evidence base is fragmented. This systematic review synthesizes current literature on AI applications for pediatric asthma exacerbation prediction and management, evaluating model performance, clinical utility, and methodological quality. Following Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 guidelines, we searched PubMed, Scopus, Elsevier, Web of Science, and Excerpta Medica Database (Embase) (2020-2025) for studies applying AI/machine learning (ML) to pediatric asthma exacerbations. Eight studies met the inclusion criteria after screening 431 records. Data were extracted on study design, AI models, input features, outcomes, and performance metrics. Risk of bias was assessed using Risk Of Bias In Non-randomized Studies of Interventions (ROBINS-I) for non-randomized studies and the Cochrane Risk of Bias 2 (RoB 2) tool for randomized trials. Eight studies demonstrated AI’s effectiveness in predicting pediatric asthma exacerbations, outperforming traditional methods. Performance varied, with multimodal data yielding the best results. Some models faced limitations from data biases or small samples. Most studies had a low risk of bias. AI showed potential to improve clinical workflows, but real-world impact needs more research. AI shows strong potential for pediatric asthma exacerbation prediction, particularly with multimodal data. Key challenges include algorithmic bias mitigation, prospective validation, and standardization of outcome metrics. Future research should prioritize equitable model development and clinical integration.