Artificial Intelligence in Asthma: Current Status, Opportunities, and Pitfalls.

Abstract

Asthma is a complex disease with multiple phenotypes and endotypes. Despite the availability of effective treatments, asthma control remains challenging as misdiagnosis and misclassification, poor efficacy or adherence, and errors in inhalation technique are common. Recent advances in computer science, artificial intelligence (AI), and machine learning (ML) and their integration in clinical practice hold significant promise, optimizing medical decisions and patient care. Ongoing development of huge databases has made AI/ML a valuable instrument in a more accurate classification, diagnosis, risk assessment, and prediction process, being able to assess a large amount of information compared with physicians. The implementation of AI/ML techniques in asthma care, in particular, has increased rapidly, with multiple applications in several aspects of the disease, from early and accurate diagnosis to personalized treatment and prevention of exacerbations, having the potential to transform precision medicine. In this narrative review, we present the most recent literature regarding AI/ML and asthma in the last 3 years, aiming to map existing most up-to-date knowledge and elucidate the road for future research, focusing on asthma future risk identification, screening and diagnosis, patient classification, prediction of asthma exacerbations, and asthma management and guided treatment. Although AI holds significant promise, it will certainly not replace clinicians, but will, with equal certainty, support them in optimizing medical decisions and practices in the imminent future.