Scholarlink

This paper presents a system for Shared Task 1 of the 11th China Health Information Processing Conference (CHIP 2025): Semantic Quality Control for Inpatient Electronic Medical Records (EMRs). Our framework integrates rule-guided strategies, Retrieval-Augmented Generation (RAG), and iterative prompt optimization within a Natural Language Processing (NLP) paradigm to improve semantic-level quality control. The system decomposes complex multi-rule tasks into single-rule subtasks and applies feedback-based prompt refinement for higher logical accuracy. To mitigate limited data, a Large Language Model (LLM) is used for data augmentation. In evaluation, rule-based judgments handle simple cases, while LLM-based reasoning addresses complex ones. The proposed approach achieved a comprehensive score of 0.6484 on the no-fine-tuning track, demonstrating its effectiveness.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Semantic Quality Control of EMR Admission Notes: Integrating Rule Guidance, Prompt Optimization, and RAG

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Yuxin Liu,
  • Yitong Pan,
  • Wusi Zhen,
  • Junwen Duan

摘要

This paper presents a system for Shared Task 1 of the 11th China Health Information Processing Conference (CHIP 2025): Semantic Quality Control for Inpatient Electronic Medical Records (EMRs). Our framework integrates rule-guided strategies, Retrieval-Augmented Generation (RAG), and iterative prompt optimization within a Natural Language Processing (NLP) paradigm to improve semantic-level quality control. The system decomposes complex multi-rule tasks into single-rule subtasks and applies feedback-based prompt refinement for higher logical accuracy. To mitigate limited data, a Large Language Model (LLM) is used for data augmentation. In evaluation, rule-based judgments handle simple cases, while LLM-based reasoning addresses complex ones. The proposed approach achieved a comprehensive score of 0.6484 on the no-fine-tuning track, demonstrating its effectiveness.