基于大语言模型的糖尿病智能问答系统

Intelligent Diabetes Question Answering System Based on Large Language Models

摘要: 为探索大语言模型在医学知识问答领域的应用，满足糖尿病患者的健康管理知识需求，本研究构建了糖尿病智能问答系统。方法上，以权威糖尿病防治指南为核心，结合爬虫采集的科普信息构建含32,104个实体、31,242个关系的本地知识图谱，采用LangChain4J框架、RAG技术及前后端分离架构开发系统。结论为：该系统可精准理解患者口语化提问并提供专业通俗的解答，支持多轮对话，还整合了健康记录、饮食推荐、用药提醒等功能，在提高患者自我管理能力的同时，也能缓解医疗资源紧张的问题，具有一定的应用前景和社会价值。

Abstract: To explore the application of large language models (LLMs) in the field of medical knowledge question answering and meet the health management knowledge needs of diabetic patients, this study constructed an intelligent diabetes question answering system. Methodologically, guided by authoritative diabetes prevention and management guidelines and combined with popular science information collected via web crawlers, a local knowledge graph containing 32,104 entities and 31,242 relationships was constructed. The system was developed using the LangChain4J framework, RAG technology, and a front-end/back-end separation architecture. The conclusion is that this system can precisely understand patients’ colloquial queries and provide professional yet easy-to-understand answers, supporting multi-turn dialogues. It also integrates functions such as health record tracking, dietary recommendations, and medication reminders. By enhancing patients’ self-management capabilities, it also helps alleviate the strain on medical resources, demonstrating certain application prospects and social value.