Rheumatoid arthritis (RA) is a chronic systemic inflammatory disease that can cause severe joint damage and disability. The management of RA patients has evolved significantly over the past few decades due to improved detection of early disease progression and the initiation of targeted advanced treatments. However, several gaps remain, including disease risk assessment, early diagnosis, phenotypic identification, and risk of treatment failure. These gaps could potentially be addressed by the application of artificial intelligence (AI), defined as the ability of a machine to mimic intelligent human behavior, using machine learning and deep learning models. By analyzing various types of data, including clinical, laboratory, and imaging data, omics, demographics, and data from sensor applications or wearable technologies, it may be possible to improve the management of RA patients, as demonstrated by several studies. However, limitations related to interindividual variability, small datasets, study designs, and intrinsic model bias remain, making it difficult to generalize findings to larger cohorts of RA patients. The aim of this narrative review is to discuss the potential uses and limitations of AI-based algorithms in RA patients in light of the most recent scientific evidence.
Talotta et al. (Mon,) studied this question.