Using artificial intelligence and machine learning algorithms for interstitial lung diseases (ILD): A systematic review

Introduction: Interstitial lung diseases (ILDs) are a complex group of diseases characterized by inflammation and scarring in the lung interstitium [۱]. Machine learning algorithms show potential in disease prediction, outcome evaluation, diagnosis, and prognosis [۲]. This study explores the use of these algorithms in ILD management. Methods: We searched PubMed and Scopus for articles up to February ۲, ۲۰۲۵, following the PRISMA guideline [۳]. Keywords included machine learning, computer vision, knowledge, deep learning, expert systems, NLP, neural networks, interstitial lung diseases, and diffuse parenchymal lung disease. Four researchers reviewed titles, abstracts, and full texts, then collected data in accordance with the study's aim. Results: The study included ۸۱ articles, mainly using machine learning and deep learning for ILD management in diagnosis (۷۶ articles, ۶۳.۳۳%), follow-up (۲۶, ۲۱.۶۶%), and prevention (۱۷, ۱۴.۱۶%). The most common algorithms were CNN (۶۸, ۵۱.۹۰%), SVM (۸, ۶.۱۰%), and RF (۷, ۵.۳۴%). Conclusion: This study highlights the role of machine learning and deep learning in managing ILD, including diagnosis, follow-up, prevention, and treatment. It shows that these algorithms are mostly used for diagnosis, especially CNN. A structured approach to validate and explain these models would help clinicians develop and implement them more effectively. Keywords: ILD, Machine Learning, Diagnosis, Artificial Intelligence.

کلیدواژه ها:

ILD ، Machine Learning ، Diagnosis ، Artificial Intelligence