Artificial intelligence in rheumatology: a cross-sectional Scopus-based analysis

Artificial intelligence (AI) is increasingly used in clinical medicine. Rheumatology is well-suited to AI applications due to diagnostic complexity of rheumatic diseases, variable disease presentations, and multisystem involvement. This cross-sectional study examines global AI research in rheumatology through bibliometric analysis of Scopus data. The Scopus database was searched on May 5, 2026, using the terms "artificial intelligence" AND "rheum*" in the title, abstract, and keyword fields. No temporal restrictions were applied, and all English documents were analyzed. Bibliometric data, including publication year, country, institution, author, journal, and keywords, were extracted. Retracted articles were identified using Scopus tags and were manually verified through related journal editorial notices. Disease-specific publication counts were determined through independent searches for rheumatic diseases. Survey-based studies were identified through manual review. Linear regression analysis was conducted to assess temporal trends. A total of 1,057 publications were identified. Annual output remained below 10 until 2018, then rose sharply to a peak of 282 in 2025. Linear regression confirmed a significant upward trend (p < 0.001). The United States produced the most publications (n = 249), followed by the United Kingdom (n = 141) and India (n = 138). Harvard Medical School was the leading institution (n = 33), and Knitza, J., was the most prolific author (n = 17). Rheumatology International published the most articles (n = 27). Osteoarthritis (n = 1,017) and rheumatoid arthritis (n = 646) dominated the field, while systemic vasculitis (n = 40), Behçet disease (n = 29), and familial Mediterranean fever (n = 16) were underrepresented. Eleven survey-based studies were identified, mostly published between 2024 and 2025. Two articles were retracted due to peer review irregularities. AI research in rheumatology has increased significantly, with research outputs stemming from a limited number of countries, institutions, and disease categories. Osteoarthritis and rheumatoid arthritis are the most frequently explored diseases in the field. The increasing number of survey-based studies indicates heightened attention to clinician and patient perspectives. Future research should focus on expanding international collaboration, addressing gaps in AI research on underrepresented diseases, and strengthening ethical and methodological standards to facilitate broader AI adoption in rheumatology.