Artificial Intelligence in Asthma: From Diagnosis to Management

Area of Science:

• Pulmonology
• Medical Informatics
• Artificial Intelligence

Background:

• Asthma management can be enhanced by artificial intelligence (AI) applications across the care continuum.
• Evidence on AI's role in asthma diagnosis, classification, monitoring, and treatment requires synthesis.

Purpose of the Study:

• To review and synthesize existing evidence on AI applications in asthma management.
• To evaluate the performance of AI tools in asthma diagnosis, monitoring, and prediction.

Main Methods:

• A narrative review of studies published between 1995 and 2025 was conducted.
• PubMed database was searched for peer-reviewed reports on AI in asthma diagnosis, monitoring, prediction, and treatment.
• Studies were selected based on outcomes including diagnostic accuracy, risk prediction, and clinical decision support.

Main Results:

• AI tools integrating clinical and objective data showed high diagnostic accuracy (up to 98%).
• Automated pulmonary function test interpretation surpassed specialists; acoustic analyses achieved >90% sensitivity/specificity for remote monitoring.
• AI models demonstrated effectiveness in predicting exacerbations (AUC up to 0.85) and identifying asthma phenotypes.

Conclusions:

• Artificial intelligence shows significant potential to improve asthma diagnosis, phenotyping, and monitoring.
• Most current AI studies are proof-of-concept, necessitating external validation and real-world impact assessment.
• Future research should focus on pragmatic trials and implementation to establish clinical and cost-effectiveness.