Harnessing AI to predict asthma exacerbations: a promising frontier in respiratory health

Abstract

<bold>Background:</bold> Artificial intelligence (AI) and machine learning (ML) offer potential new opportunities to improve the management of asthma exacerbations (AE). <bold>Aim:</bold> To evaluate if AI/ML show the ability to identify reliable predictive parameters of AE in the adult population and to summarize them in a narrative review. <bold>Methods:</bold> A review of literature (2019-2024) was conducted using “Pubmed” and “Google Scholar” using the keywords “artificial intelligence, asthma, machine learning, risk factors, exacerbation”. The selection was limited to interventional studies published in English, excluding those that were irrelevant, concerned with the pediatric population, or lacking abstracts, for a total of 14 studies analyzed. <bold>Results:</bold> The main statistically significant predictive parameters of AE identified by AI/ML are anamnestic features (demographic factors, comorbidities and asthma medications), lung function impairment (FEV1/FVC, FEV1, PEF) and environmental factors (occupational exposure, pollution, air temperature and humidity, pollen and respiratory virus). It also emerged the ability of AI/ML to identify “frequent exacerbators” through the analysis of these predictive parameters. <bold>Conclusion:</bold> AI is a reliable tool in identifying predictive parameters of AE, with results consistent with clinical evidence of the literature. Thus, AI/ML have the potential to improve asthma management through accurate prediction of exacerbations and “frequent exacerbators” by integrating clinical, functional and environmental data. ML models also analyzed data from tele-monitoring systems and wearable devices. In the future, large scale validation is needed to effectively implement these tools in clinical practice.

Ähnliche Arbeiten