LLMs in SLE: From Patient Education to Clinical Decision Support Save

The emergence of large language models (LLMs) represents a promising shift in how complex diseases such as systemic lupus erythematosus (SLE) are communicated and interpreted across both patient and clinical settings. Two EULAR 2026 abstracts highlight the evolving role of LLMs across this spectrum, from patient-facing education to diagnostic reasoning.

In OP013-PARE, Kremer et al. evaluate the performance of contemporary LLMs—ChatGPT-5, Claude 4.0 Sonnet, and Gemini 2.5 Pro—against traditional Google Search in answering patient-generated questions across connective tissue diseases, including SLE. Notably, both patients and rheumatologists consistently rated LLM-generated responses as highly accurate, empathetic, and comprehensible, with minimal variation across models in SLE-specific queries. Gemini 2.5 Pro was most frequently ranked highest overall, while Google Search ranked lowest, despite maintaining reasonable medical correctness. Importantly, LLMs demonstrated added value in clarity and emotional alignment, key elements often lacking in conventional search tools. These findings suggest that LLMs may meaningfully enhance patient health literacy in SLE, particularly in areas such as disease understanding, treatment expectations, and psychosocial impact.

Complementing this patient-centered perspective, POS0160 assesses the performance of LLMs as potential clinical decision support tools in autoimmune rheumatic diseases, including SLE. In a blinded evaluation using real-world case vignettes, diagnostic accuracy varied across models, with GPT-4.1 achieving the highest overall accuracy (83%) and strong performance in SLE (87%). Importantly, higher-performing models were associated with low hallucination rates and minimal dangerous clinical errors, highlighting their potential for safe integration into clinical workflows. However, variability across models and the persistence of errors—particularly in lower-performing systems—underscore the need for physician oversight and careful implementation.

Together, these studies illustrate that LLMs are beginning to bridge critical gaps in SLE care—from improving patient education to augmenting diagnostic reasoning. While still evolving, their role will likely expand as tools for precision communication and clinical augmentation, provided their use remains guided by expert clinical judgment.