Thinking Like a Doctor: Conversational Diagnosis through the Exploration of Diagnostic Knowledge Graphs
Jeongmoon Won Seungwon Kook Yohan JoGraduate School of Data Science, Seoul National University {jmwon06231,cacu5615,yohan.jo}@snu.ac.kr · 2026
Read Thinking Like a Doctor: Conversational Diagnosis through the Exploration of Diagnostic Knowledge Graphs by Jeongmoon Won Seungwon Kook Yohan JoGraduate School of Data Science, Seoul National University {jmwon06231,cacu5615,yohan.jo}@snu.ac.kr (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/arxiv-2602-01995.txt.
Description
Jeongmoon Won and colleagues build a chat diagnosis tool exploring a disease knowledge graph, asking questions under incomplete data. Existing methods rely on model knowledge or assume patients give rich concrete information. The system reasons in two steps: forming hypotheses from talk, then checking them through questions until diagnosis. Tests use PatientSim, a persona simulator, with MIMIC-IV profiles, adapted for vague early symptoms. Experiments show better accuracy and efficiency over strong baselines, with doctors supporting simulator realism and question use.
Catalog record
| CREATOR | Jeongmoon Won Seungwon Kook Yohan JoGraduate School of Data Science, Seoul National University {jmwon06231,cacu5615,yohan.jo}@snu.ac.kr |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 70856 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2602.01995