| 2,452 | 4 | 63 |
| 下载次数 | 被引频次 | 阅读次数 |
针对医疗领域大语言模型知识更新滞后、领域适配性不足及个性化服务能力受限的瓶颈问题,设计了一种基于Qwen 2.5大语言模型的智能优化框架。通过融合提示词工程、RAG检索增强算法、LoRA微调,构建医学领域专用问答数据集,实现动态知识融合与轻量化领域适配的协同优化。实验表明,相较于知识图谱驱动模型及基线模型,全模型在医疗问答准确性、术语规范性与推理逻辑性方面均呈现系统性提升。该框架通过本底数据驱动的动态扩展和参数调整机制,突破传统医疗大模型受限于静态数据、部署复杂度高的限制,为垂直领域人工智能应用提供了可扩展的技术范式。
Abstract：To address the bottlenecks of knowledge update lag,insufficient domain adaptation,and limited personalized service capability in large language models(LLMs) for the medical field,an intelligent optimization framework based on the Qwen 2.5 LLM is designed.By integrating Prompt Engineering,Retrieval—Augmented Generation(RAG) algorithms,and LoRA fine—tuning,a specialized medical question—answering dataset is constructed to achieve dynamic knowledge fusion and lightweight domain adaptation through collaborative optimization.Experimental results demonstrate that,compared to knowledge graph—driven models and baseline models,the proposed framework exhibits systematic improvements in medical Q&A accuracy, terminology standardization,and reasoning logic.Leveraging a locally driven dynamic expansion and parameter adjustment mechanism,this framework overcomes the limitations of traditional medical LLMs,which are constrained by static data and high deployment complexity.It provides a scalable technical paradigm for vertical—domain AI applications.
[1]赵飞,兰蓝,曹战强,等.我国人工智能在健康医疗领域应用发展现状研究[J].中国卫生信息管理杂志,2018,15(3):344-349.
[2]颜昕.基于自然语言处理的医疗健康问答系统[J].通讯世界,2018(6):255-256.
[3]任海玉,刘建平,王健,等.基于大语言模型的智能问答系统研究综述[J].计算机工程与应用,2025,61(7):1-24.
[4]冰博客.2024医疗大模型行业纵览[J].互联网周刊,2025(3):32-33.
[5]肖革新,陈善吉,王博远,等.医疗大模型的应用现状与展望[J].中国数字医学,2025,20(2):39-45.
[6]陈悦,刘则渊.悄然兴起的科学知识图谱[J].科学学研究,2005(2):149-154.
[7]SINGHAL A.Introducing the knowledge graph:Things,not strings[DB/OL].Official Google blog,2012-05-16.
[8]KORNGIEBEL M,MOONEY S D.Considering the possibilities and pitfalls of generative pretrained transformer (GPT-3) in healthcare delivery[J].NPJ Digital Medicine,2021,4(1):93.
[9]洪季芳.Transformer研究现状综述[J].信息系统工程,2022(2):125-128.
[10]JUNG K,SEO H,KIM H J,et al.Semantic similarity-based intent analysis using pre-trained transformer for natural language understanding[J].Journal of KIISE,2020,47(8):748-760.
[11]胡振生,杨瑞,朱嘉豪,等.大语言模型在医学领域的研究与应用发展[J].人工智能,2023(4):10-19.
[12]LEWIS P,REREZ E,PIKTUS A,et al.Retrievalaugmented generation for knowledge-intensive NLP Tasks[DB/OL].https://arxiv.org/abs/2005.11401.2020-05-22.
[13]李佳丽.面向儿科常见疾病的智能问答技术研究与应用[D].贵阳:贵州大学,2024.
[14]黄晨.基于提示微调的医疗小样本关系抽取模型[D].上海:华东师范大学,2024.
[15]李莉,时榕良,郭旭,等.融合大模型与图神经网络的电力设备缺陷诊断[J].计算机科学与探索,2024,18(10):2643-2655.
[16]WU P,CHEN L F,FANG S Y,et al.An application programming interface (API) sensitive data identification method based on the federated large language model[J].Applied Sciences,2024,14(22):10162.
[17]李彦宏.10年后全世界50%的工作会是提示词工程[J].中国商人,2023(11):68-69.
[18]宋筱璇,徐丹,张紫蕴,等.提示词重构类型对AI生成内容效果的影响[J].图书馆论坛,2025,45(4):129-138.
[19]王东清,芦飞,张炳会,等.大语言模型中提示词工程综述[J].计算机系统应用,2025,34(1):1-10.
[20]LO S.The CLEAR path:A framework for enhancing information literacy through prompt engineering[J].The Journal of Academic Librarianship.2023,49(4):1-7.
[21]陈冬雷,秦薇.基于检索增强生成的智能问答系统关键技术研究[J].办公自动化,2024,29(19):82-86.
[22]丁宁,宋雨欣,单泽田,等.基于检索增强生成(RAG)技术的医学教学辅助智能问答系统的构建探索[J].中国医学教育技术,2025,39(1):1-5.
[23]黄星晨.基于自适应量化的大语言模型微调方法[J].信息技术与信息化,2024(9):9-12.
[24]韩霄龙,曾曦,刘锟,等.基于Lo RA高效微调通用语言大模型的文本立场检测[J].计算机与现代化,2025(1):1-6.
[25]LIN Y.ROUGE:A package for automatic evaluation of summaries[C]//Text Summarization Branches Out,Barcelona,Spain:Association for Computational Linguistics,2004:74-81.
[26]LEI B,CAO Y,ZHOU T Y,et al.Corpussteered query expansion with large language models[DB/OL].https://arxiv.org/abs/2402.18031.2024-02-28.
[27]PAPINENI K,ROUKOS S,WARD T,et al.Bleu:method for automatic evaluation of machine translation[C]//40th Annual Meeting of the Association for Computational Linguistics,Philadelphia,Pennsylvania,USA:Association for Computational Linguistics,2002:311-318.
基本信息:
DOI：10.13990/j.issn1001-3679.2025.05.015
中图分类号:R319;TP391.3;TP18
引用信息:
[1]杨诗卓,周燕玲,解翔杰.融合RAG检索增强与LoRA微调的医疗问答系统优化研究[J].江西科学,2025,43(05):907-913+1003.DOI:10.13990/j.issn1001-3679.2025.05.015.
基金信息:
江西中医药大学2024年大学生创新创业训练计划项目（202410412211）
2025-04-02
2025
2025-04-17
2025-04-25
2025
1
2025-05-08
2025-05-08
2025-05-08