Abstract Background/Aims Artificial intelligence (AI) involves the use of computational technology to perform tasks that would previously require human intelligence, achieved through techniques such as machine learning, deep learning, and natural language processing. These approaches can interrogate complex data sources, detect patterns, and generate predictive models to support clinical decision-making. Autoimmune connective tissue diseases (CTDs), including systemic lupus erythematosus (SLE), systemic sclerosis, myositis, and Sjögren’s syndrome, are clinically heterogeneous and could benefit greatly from such approaches. While AI has been explored within rheumatology more broadly, there has been no systematic evaluation of its specific use within CTDs. Methods A systematic scoping review was conducted in accordance with PRISMA-ScR guidelines using OVID MEDLINE and Embase databases (1 January 2000 to 23 August 2024). The systematic search strategy combined medical subject headings (MeSH terms) and free-text terms related to autoimmune CTDs and AI techniques (e.g. machine learning, neural networks) to identify studies utilising AI for clinically relevant tasks. Inclusion criteria comprised primary human research, the use of patient-level data, and publication in English. Studies focused solely on basic science or without direct clinical application were excluded. Titles, abstracts and full texts were independently screened by two reviewers, with extracted data including disease focus, clinical purpose (diagnosis, phenotyping, prognostication, treatment decisions), study design, AI technique(s) and data inputs. Findings were synthesised narratively. Results From 1,442 screened records, 203 studies published between 2002-2024 met inclusion criteria. 79% were published since 2021, highlighting the rapid recent growth of AI research in CTDs. SLE accounted for the majority (100/203), followed by systemic sclerosis (31), myositis (29), and Sjögren’s syndrome (23). Most studies used AI to support diagnosis of CTDs, with multimodal data inputs including clinical records, imaging, immunophenotyping, and multi-omics such as metabolomic, genomic, and transcriptomic data. Supervised learning models (such as random forests, support vector machines, and convolutional neural networks) were most frequently used and demonstrate promising accuracy for identifying CTDs and their complications, such as lupus nephritis, neuropsychiatric lupus, and interstitial lung disease. Unsupervised clustering was used to define novel disease subtypes in SLE and systemic sclerosis, supporting data-driven disease stratification. Prognostic models predicted disease activity, flares, and mortality, while models to support treatment optimisation were less common but demonstrated potential for forecasting biologic or stem cell therapy outcomes. Conclusion AI research in autoimmune CTDs is expanding rapidly, with the strongest evidence in SLE and emerging data in systemic sclerosis and Sjogren’s. Translation to clinical practice, however, remains limited, with few models externally validated or compared with either clinician performance or conventional statistical models. Collaborative, multicentre studies with transparent, standardised reporting and external validation are essential to advance AI from proof-of-concept to real-world applications to support precision care in CTDs. Disclosure P. Saha: None. T. Guruparan: None. N. Fuggle: None. J. Tsigarides: Corporate appointments; J.T. has a flexible unpaid position as Chief Medical Officer for Revolve Labs Ltd. Consultancies; J.T. has received consultancy fees from Revolve Labs Ltd for education and virtual reality consulting. Honoraria; J.T. has received honoraria for non-promotional speaking commitments for Novartis. Grants/research support; J.T has received research funding from British Society for Rheumatology, NIHR, EPSRC and Revolve Labs Ltd.
Saha et al. (Wed,) studied this question.