LINS: A general medical Q&A framework for enhancing the quality and credibility of LLM-generated responses
Large language models can lighten the workload of clinicians and patients, yet their responses often include fabricated evidence, outdated knowledge, and insufficient medical specificity. We introduce a general retrieval-augmented question-answering framework that continuously gathers up-to-date, high-quality medical knowledge and generates evidence-traceable responses. Here we show that this approach significantly improves the evidence validity, medical expertise, and timeliness of large language model outputs, thereby enhancing their overall quality and credibility. Evaluation against 15,530 objective questions, together with two physician-curated clinical test sets covering evidence-based medical practice and medical order explanation, confirms the improvements. In blinded trials, resident physicians indicate meaningful assistance in 87.00% of evidence-based medical scenarios, and lay users find it helpful in 90.09% of medical order explanations. These findings demonstrate a practical route to trustworthy, general-purpose language assistants for clinical applications.
Authors
- Yi Zhao (ORCID: https://orcid.org/0000-0001-6046-8420)
- Hongjie Liu
- Wei Huang (ORCID: https://orcid.org/0000-0002-0541-8612)
- Lianhe Zhao (ORCID: https://orcid.org/0000-0002-7736-2443)
- Mengtong Xu (ORCID: https://orcid.org/0009-0007-6753-9931)
- Jianjun Luo (ORCID: https://orcid.org/0000-0002-5550-9889)
- Runsheng Chen (ORCID: https://orcid.org/0000-0001-6049-8347)
- Jingbo Sun (ORCID: https://orcid.org/0000-0002-9380-6147)
- Xiaoxi Zeng (ORCID: https://orcid.org/0000-0001-9946-4291)
- Xirun Wan (ORCID: https://orcid.org/0000-0002-1688-298X)
- Wei Zhang (ORCID: https://orcid.org/0000-0001-9298-4497)
- Yang Wu (ORCID: https://orcid.org/0000-0001-6547-9524)
- Zhiyuan Yuan (ORCID: https://orcid.org/0000-0002-9367-4236)
- Zhaohui Yang (ORCID: https://orcid.org/0000-0002-4475-589X)
- Hongjia Zhang (ORCID: https://orcid.org/0000-0002-5737-4084)
- Jingjia Liu (ORCID: https://orcid.org/0000-0001-6490-3135)
- Shu Wang (ORCID: https://orcid.org/0000-0002-8651-4795)
- Kang Zhang (ORCID: https://orcid.org/0000-0002-4549-1697)
- Tao Wang (ORCID: https://orcid.org/0000-0003-4451-2721)
- Yang Lu (ORCID: https://orcid.org/0000-0002-0383-2289)
- Fangyuan Zhao (ORCID: https://orcid.org/0009-0009-9623-7945)
- Jianjun Zheng (ORCID: https://orcid.org/0000-0002-6177-9607)
- Ming Gong (ORCID: https://orcid.org/0000-0001-5137-836X)
- Ben Wan (ORCID: https://orcid.org/0000-0001-8388-0215)
- Sheng Wang (ORCID: https://orcid.org/0009-0009-0537-905X)
- Dan Bu
Institutions
- Macau University of Science and Technology (MO)
- Capital Medical University (CN)
- Chinese Academy of Sciences (CN)
- Chinese Academy of Medical Sciences & Peking Union Medical College (CN)
- Peking University (CN)
- Sichuan University (CN)
- Fudan University (CN)
- West China Medical Center of Sichuan University (CN)
- Peking Union Medical College Hospital (CN)
- West China Hospital of Sichuan University (CN)
- Institute of Computing Technology (CN)
- Ningbo No. 2 Hospital (CN)
- Institute of Biophysics (CN)
- Henan Institute of Technology (CN)
- Beijing Anzhen Hospital (CN)
- Peking University People's Hospital (CN)
- Henan Communication Science and Technology Research Institute (China) (CN)
- Henan Institute of Science and Technology (CN)
- Ningbo No.6 Hospital (CN)
- Shanghai Center for Brain Science and Brain-Inspired Technology (CN)
- University of Chinese Academy of Sciences (CN)
- Institute of Science and Technology
Publication Details
- Journal
- Nature Communications
- Published
- 2025-10-13
- DOI
- https://doi.org/10.1038/s41467-025-64142-2
- Citations
- 5
- Primary Topic
- Topic Modeling
- Type
- article
- Field-Weighted Citation Impact
- 12.40
Funders
- National Natural Science Foundation of China