NoteChat
收藏资源简介:
NoteChat是一个创新的多代理合作框架,利用大型语言模型(LLMs)生成基于临床笔记的患者-医生对话数据集。该数据集包含167000条合成对话,旨在通过结构化的角色扮演和策略提示,提高LLMs在其指定角色中的表现效率。NoteChat的应用领域主要集中在改善临床文档,这是医生职业倦怠的主要原因之一。数据集的创建过程涉及三个模块:规划、角色扮演和润色,确保对话生成既符合临床笔记的事实性,又具有合理的交流逻辑。
NoteChat is an innovative multi-agent collaboration framework that leverages Large Language Models (LLMs) to generate patient-doctor dialogue datasets based on clinical notes. This dataset contains 167,000 synthetic conversations, aiming to improve the performance efficiency of LLMs in their assigned roles through structured role-playing and strategic prompting. The core application scope of NoteChat focuses on enhancing clinical documentation, which is one of the leading causes of physician burnout. The dataset creation process involves three modules: planning, role-playing, and polishing, ensuring that the generated conversations conform to the factual accuracy of clinical notes while maintaining plausible communicative logic.

- 1NoteChat: A Dataset of Synthetic Doctor-Patient Conversations Conditioned on Clinical Notes马萨诸塞大学阿默斯特分校 · 2023年



