USE OF MACHINE LEARNING ALGORITHMS FOR SUPPORT AND DIAGNOSIS OF ASTHMA: A SYSTEMATIC LITERATURE REVIEW

Introduction: Asthma is a prevalent and heterogeneous chronic disease, making its diagnosis and management challenging. Machine Learning (ML) has emerged as a promising approach to analyze complex clinical data and optimize decision support. Objective: To identify the ML algorithms used in asthma diagnosis and control and evaluate their performance through a systematic literature review. Methods: Following PRISMA guidelines, a systematic search was conducted in PubMed, Scopus, Web of Science, and IEEE databases (2019–2025) using the terms "asthma," "machine learning," and "deep learning." Eight studies featuring empirical experiments of ML models applied to asthma were selected for analysis. Results: Applications covered diagnosis, prognosis, and disease control. Algorithms ranged from classical models (Logistic Regression and SVM) to deep neural networks (DNN, MLP), Natural Language Processing, and Reinforcement Learning (Deep Q-Learning). Data sources included electronic health records, environmental variables, and multi-omics. The models showed high performance and a growing trend toward using Explainable AI (XAI) to enhance clinical interpretability. Conclusions: ML demonstrates significant potential to improve asthma management. However, practical application still requires external validation and integration into medical workflows to translate technological potential into real benefits for patients and healthcare professionals.