Generative artificial intelligence in osteoarthritis: A systematic scoping review of current applications and future directions.

OBJECTIVE

Generative artificial intelligence (GenAI) is a transformative tool capable of producing realistic synthetic data, modeling complex data, and translating large volumes of information into clinically meaningful insights. This systematic scoping review aimed to characterize the current applications of GenAI in OA care and research.

DESIGN

Medline, EMBASE, Pubmed and Scopus were searched electronically to December 2025 using keywords related to OA and GenAI. English-language studies detailing the applications of GenAI in OA were included.

RESULTS

Forty-nine studies were included and grouped into four domains. Firstly, large language models were applied to diagnose OA and/or make guideline concordant treatment recommendations (12 studies, 25%), generate patient-facing materials (15 studies, 31%), and other applications (three studies, 6%). Secondly, GenAI was used to generate synthetic data (eight studies, 16%), primarily imaging datasets, which enhanced model fidelity and improved accuracy of downstream classification by 5-19%. Thirdly, GenAI was applied to analyse imaging data (ten studies, 20%), where generative models achieved strong performance in OA severity grading (70-90% accuracy) and cross-modal translation. Finally, GenAI was used to analyse and integrate complex multi-omic data (one study, 2%), where variational autoencoder-based frameworks uncovered biologically-distinct OA subgroups associated with clinical outcomes.

CONCLUSIONS

While GenAI holds unique potential to transform OA care and research, current studies are small-scale and heterogeneous, with limited exploration of applications beyond imaging data. Future efforts must focus on expansion to larger, more complex datasets, with rigorous validation and integration into multimodal clinical research pipelines.