Scholarlink

Lymphoma is a malignant tumor originating from the lymphatic system, with a high global incidence. However, due to the numerous subtypes of lymphoma, each with unique clinical features and diagnostic criteria, accurately coding lymphoma remains challenging. This paper proposes a method for automatically generating ICD codes using large language models. By introducing Automatic Prompt Engineering (APE), we enable the model to generate high-quality prompts autonomously. We further break down the automatic coding process into three stages: disease diagnosis, disease matching, and disease mapping, and apply reasoning and action steps within the Reasoning and Act (ReAct) framework. Our method has shown excellent results across ICD-10, ICD-O-M, and ICD-O-H, ultimately ranking first with a comprehensive score of 92.96%.

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

Automatic ICD Code Generation for Lymphoma Using Large Language Models

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Yu Song,
  • Bohan Yu,
  • Pe ngcheng Wu,
  • Wenhui Fu,
  • Xia Liu,
  • Chenxin Hu,
  • Kunli Zhang,
  • Hongying Zan

摘要

Lymphoma is a malignant tumor originating from the lymphatic system, with a high global incidence. However, due to the numerous subtypes of lymphoma, each with unique clinical features and diagnostic criteria, accurately coding lymphoma remains challenging. This paper proposes a method for automatically generating ICD codes using large language models. By introducing Automatic Prompt Engineering (APE), we enable the model to generate high-quality prompts autonomously. We further break down the automatic coding process into three stages: disease diagnosis, disease matching, and disease mapping, and apply reasoning and action steps within the Reasoning and Act (ReAct) framework. Our method has shown excellent results across ICD-10, ICD-O-M, and ICD-O-H, ultimately ranking first with a comprehensive score of 92.96%.