
浏览全部资源
扫码关注微信
1.邯郸市第一医院药学部,河北 邯郸 056003
2.东软集团有限公司,河北 邯郸 056003
副主任药师。研究方向:人工智能药学。E-mail:zhengwangwen@krae.edu.kg
副主任医师,硕士。研究方向:消化道恶性肿瘤、人工智能药学。E-mail:bsg_ly@hotmail.com
收稿:2025-08-13,
修回:2025-12-24,
录用:2025-12-25,
纸质出版:2026-01-30
移动端阅览
温正旺,王嘉莹,杨文月,等.基于AI幻觉抑制的药学智能问答平台的构建与效能验证 [J].中国药房,2026,37(02):226-231.
WEN Zhengwang,WANG Jiaying,YANG Wenyue,et al.Construction and efficacy verification of an intelligent pharmaceutical Q&A platform based on AI hallucination-suppression[J].ZHONGGUO YAOFANG,2026,37(02):226-231.
温正旺,王嘉莹,杨文月,等.基于AI幻觉抑制的药学智能问答平台的构建与效能验证 [J].中国药房,2026,37(02):226-231. DOI: 10.6039/j.issn.1001-0408.2026.02.16.
WEN Zhengwang,WANG Jiaying,YANG Wenyue,et al.Construction and efficacy verification of an intelligent pharmaceutical Q&A platform based on AI hallucination-suppression[J].ZHONGGUO YAOFANG,2026,37(02):226-231. DOI: 10.6039/j.issn.1001-0408.2026.02.16.
目的
2
构建低“人工智能(AI)幻觉”的药学智能问答平台,提升用药咨询的准确性、一致性与可追溯性。
方法
2
利用Python代码对药品说明书进行批量结构化整理并构建本地药学知识库,基于大型语言模型实现检索与问答流程设计,并在Dify平台完成系统集成与本地化部署。通过设计典型临床用药问题,从达峰时间、半衰期检索及肾功能减退患者剂量调整方案推理等维度,将药学智能问答平台的输出结果与在线版DeepSeek进行对比验证,评估其检索和推理结果的准确性与可靠性。
结果
2
基于本地药品说明书构建的药学智能问答平台在达峰时间、半衰期及剂量调整方案的检索和推理准确率均为100%。相比之下,在线版DeepSeek在3个维度方面的准确率分别为30%(6/20)、50%(10/20)和38%(23/60)。
结论
2
构建的药学智能问答平台能够根据临床提问精准检索并提炼本地知识库信息,能避免AI幻觉的出现,为医务人员提供可靠的用药决策支持。
OBJECTIVE
2
To construct an intelligent pharmaceutical Q&A platform for precision medication with low “artificial intelligence (AI) hallucination”, aiming to enhance the accuracy, consistency, and traceability of medication consultations.
METHODS
2
Medication package inserts were batch-processed and converted into structured data through Python programming to build a local pharmaceutical knowledge base. The retrieval and question-answering processes were designed based on large language models, and system integration and localized deployment were completed on Dify platform. By designing typical clinical medication questions and comparing the output of the intelligent pharmaceutical Q&A platform with the online version of DeepSeek across dimensions such as peak time retrieval, half-life, and dosage adjustment reasoning for patients with renal impairment, the accuracy and reliability of its retrieval and reasoning results were evaluated.
RESULTS
2
The intelligent pharmaceutical Q&A platform, constructed based on local drug package inserts, achieved 100% accuracy in retrieval and reasoning for peak time, half-life, and dosage adjustment schemes. In comparison, the online version of DeepSeek demonstrated accuracies of 30%(6/20), 50%(10/20), and 38%(23/60) across these three dimensions, respectively.
CONCLUSIONS
2
The constructed intelligent pharmaceutical Q&A platform is capable of accurately retrieving and extracting information from the local knowledge base based on clinical inquiries, thereby avoiding the occurrence of AI hallucinations and providing reliable medication decision support for healthcare professionals.
汪瑾 , 冷锴 , 陆慧 . “互联网+”视域下智慧医疗服务模式创新研究 [J ] . 南京医科大学学报(社会科学版) , 2020 , 20 ( 1 ): 84 - 87 .
肖鹏 , 李新辰 , 朱斌 , 等 . 基于互联网患者咨询的药师与人工智能回答一致性分析 [J ] . 临床药物治疗杂志 , 2025 , 23 ( 5 ): 46 - 49,55 .
国家卫生健康委办公厅 . 国家卫生健康委办公厅关于印发卫生健康行业人工智能应用场景参考指引的通知 [EB/OL ] .( 2024-11-14 )[ 2025-08-09 ] . https://www.nhc.gov.cn/guihuaxxs/c100133/202411/3dee425b8dc34f739d-63483c4e5c334c.shtml https://www.nhc.gov.cn/guihuaxxs/c100133/202411/3dee425b8dc34f739d-63483c4e5c334c.shtml .
PENG Y F , MALIN B A , ROUSSEAU J F , et al . From GPT to DeepSeek:significant gaps remain in realizing AI in healthcare [J ] . J Biomed Inform , 2025 , 163 : 104791 .
DREYER J . China made waves with DeepSeek,but its real ambition is AI-driven industrial innovation [J ] . Nature , 2025 , 638 ( 8051 ): 609 - 611 .
裴进涛 , 李昊东 , 刘广敏 , 等 . 基于RAG的人工智能模型在高血压药品信息智能问答中的应用 [J ] . 电脑与信息技术 , 2025 , 33 ( 5 ): 43 - 50 .
查英华 , 郭朝霞 , 鞠慧光 . 基于大语言模型的智能学习助手设计与实现 [J ] . 现代信息科技 , 2025 , 9 ( 3 ): 50 - 55 .
齐翌辰 , 王森淼 , 赵亚慧 . 基于倒排索引的问答系统的设计与实现 [J ] . 教育教学论坛 , 2018 ( 35 ): 68 - 69 .
DENG J W , ZUBAIR A , PARK Y J , et al . The use of large language models in medicine:proceeding with caution [J ] . Curr Med Res Opin , 2024 , 40 ( 2 ): 151 - 153 .
ZHANG H J , DENG H R , OU J , et al . Mitigating spatial hallucination in large language models for path planning via prompt engineering [J ] . Sci Rep , 2025 , 15 : 8881 .
PAL A , UMAPATHI L K , SANKARASUBBU M . MedMCQA:a large-scale multi-subject multi-choice dataset for medical domain question answering [EB/OL ] . [ 2025-08-09 ] . https://arxiv.org/abs/2203.14371 https://arxiv.org/abs/2203.14371 .
CALIEBE A , LEVERKUS F , ANTES G , et al . Does big data require a methodological change in medical research? [J ] . BMC Med Res Methodol , 2019 , 19 ( 1 ): 125 .
李威威 , 夏新男 . DeepSeek与Ollama协同的图书馆本地知识库构建 [J ] . 现代信息科技 , 2025 , 9 ( 21 ): 148 - 154 .
薛陶蓓 , 郑浩东 . 基于DeepSeek和RAGFlow的智能项目推荐客服系统部署实践 [J ] . 现代信息科技 , 2025 , 9 ( 18 ): 110 - 114 .
0
浏览量
36
下载量
0
CSCD
关联资源
相关文章
相关作者
相关机构
京公网安备11010802024621