NordDRG-AI-Benchmark

该数据集是首个公开的测试平台，它完整捕捉了诊断相关分组（DRG）的规则集，并评估了大型语言模型（LLM）在多语言诊断、程序和收费标准逻辑推理方面的能力。该数据集包含了定义表、专家手册、变更日志模板以及包含14个病例组合任务的提示包。规模上，它涵盖了20个相互关联的表格，涉及DRG逻辑、国际疾病分类（ICD）和国家临床程序规范（NCSP）代码的各个方面。任务方面，该数据集专注于利用大型语言模型进行临床编码和决策支持。

This dataset is the first publicly available testbed that fully captures the diagnosis-related group (DRG) rule sets, and evaluates the capabilities of large language models (LLMs) in multilingual diagnostic, procedural, and billing standard logical reasoning. It includes definition tables, expert manuals, changelog templates, and a prompt package containing 14 case mix tasks. In terms of scale, it covers 20 interconnected tables covering all aspects of DRG logic, International Classification of Diseases (ICD), and National Clinical Procedure Specification (NCSP) codes. Regarding the tasks, this dataset focuses on leveraging large language models for clinical coding and decision support.