EyeRAG: graph retrieval-augmented generation for safe and accurate clinical dialogue in ophthalmology
Kaikai Zhao, Daohuan Kang, Tao Yu, Rui Yao, Wenzheng Han, Ke Liang, Rupesh Agrawal, Andrzej Grzybowski, Kai Jin · 2026
Read EyeRAG: graph retrieval-augmented generation for safe and accurate clinical dialogue in ophthalmology by Kaikai Zhao, Daohuan Kang, Tao Yu, Rui Yao, Wenzheng Han, Ke Liang, Rupesh Agrawal, Andrzej Grzybowski, Kai Jin (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/pmc-13562699.txt.
Description
Kaikai Zhao and colleagues built EyeRAG, a question-answering system meant to make eye-care conversations safer. Large language models can explain eye conditions to patients, but they invent facts and lack specialist knowledge. EyeRAG grounds its answers in a knowledge graph, OphthaKG, built only from clinical guidelines, and retrieves relevant guidance before replying. The 2026 study tested it on 120 clinical scenarios across glaucoma, diabetic retinopathy, and cataract.
EyeRAG beat both plain language models and standard retrieval systems in rankings by automated judges and eye specialists. Experts ranked it first on average, and its hallucination rate fell to 3.3 percent from a 30 percent baseline. The authors stress it is a counselling aid under clinician supervision, translating findings into plain language, not a tool for making clinical decisions.
Catalog record
| CREATOR | Kaikai Zhao, Daohuan Kang, Tao Yu, Rui Yao, Wenzheng Han, Ke Liang, Rupesh Agrawal, Andrzej Grzybowski, Kai Jin |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 63 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |