Advancing Rheumatology Practice: Systematic Review of Natural Language Processing Applications

Background With the advent of large language models (LLM), such as ChatGPT, natural language processing (NLP) is revolutionizing healthcare. We systematically reviewed NLP’s role in rheumatology and assessed its impact on diagnostics, disease monitoring, and treatment strategies. Methods Following PRISMA guidelines, we conducted a systematic search to identify original research articles exploring NLP applications in rheumatology. This search was performed in PubMed, Embase, Web of Science, and Scopus until January 2024. Results Our search produced 17 studies that showcased diverse applications of NLP in rheumatology, addressing disease diagnosis, data handling, and monitoring. Notably, GPT-4 demonstrated strong performance in diagnosing and managing rheumatic diseases. Performance metrics indicated high accuracy and reliability in various tasks. However, challenges like data dependency and limited generalizability were noted. Conclusion NLP, and especially LLM, show promise in advancing rheumatology practice, enhancing diagnostic precision, data handling, and patient care. Future research should address current limitations, focusing on data integrity and model generalizability.