Related Experiment Video
Updated: May 26, 2026
12:04
The bm12 Inducible Model of Systemic Lupus Erythematosus (SLE) in C57BL/6 Mice
Published on: November 1, 2015
Performance of Large Language Models in Differentiating Systemic Lupus Erythematosus From Mimicking Conditions Using
Mirza Zaka Pratama1, Bagus Putu Putra Suryana1, Cesarius Singgih Wahono1
1Rheumatology Division, Internal Medicine Department, Dr Saiful General Hospital, Brawijaya University, Malang, Indonesia.
International Journal of Rheumatic Diseases
|May 25, 2026
Summary
Gemini demonstrated high accuracy in distinguishing Systemic Lupus Erythematosus (SLE) from similar conditions, showing potential for clinical decision support. Further real-world validation is needed before widespread adoption.
Area of Science:
- Artificial Intelligence in Medicine
- Rheumatology
- Diagnostic Accuracy
Background:
- Systemic Lupus Erythematosus (SLE) diagnosis is challenging due to varied symptoms and overlap with other autoimmune diseases.
- Large Language Models (LLMs) show promise in aiding clinical decision-making.
- This study assessed LLM performance in differentiating SLE from mimicking conditions.
Purpose of the Study:
- To evaluate the diagnostic accuracy of four LLMs in distinguishing SLE from other rheumatologic conditions.
- To compare the performance of Gemini, ChatGPT 4.0, Claude Sonnet 4, and Deepseek in SLE diagnosis.
- To quantify the potential of LLMs as clinical decision support tools for SLE.
Main Methods:
- A retrospective study included 100 patients (50 SLE, 50 non-SLE controls).
- Four LLMs (Gemini, ChatGPT 4.0, Claude Sonnet 4, Deepseek) were evaluated using the 2019 EULAR/ACR criteria.
- Diagnostic accuracy, PPV, NPV, and AUC were calculated.
Main Results:
- Gemini achieved the highest performance: 96% accuracy, 94% sensitivity, 98% specificity, and 0.960 AUC.
- ChatGPT 4.0 and Claude Sonnet 4 showed comparable accuracy.
- Deepseek had the lowest performance score.
Conclusions:
- Gemini shows significant potential to aid clinicians in differentiating SLE from mimicking conditions.
- Prospective validation in real-world clinical settings is essential.
- LLMs may become valuable tools in rheumatologic diagnostics.
