EvoPatient Dataset

该数据集包含了超过20,000个来自真实医疗记录的独特医疗案例，重点关注包括肝癌、阑尾炎和胰腺病变在内的多种疾病。此外，该数据集整合了经过编辑的患者信息，用于评估模拟标准化病人情境中的问题和答案。该数据集的任务是模拟医疗教育中的培训目的，以模拟患者病例。

This dataset contains over 20,000 unique medical cases sourced from real medical records, focusing on a variety of diseases including liver cancer, appendicitis, and pancreatic lesions. Additionally, the dataset integrates curated patient information for evaluating questions and answers in simulated standardized patient scenarios. The core task of this dataset is to simulate patient cases for medical education training purposes.

搜集汇总

背景与挑战

背景概述

EvoPatient数据集是一个用于医疗诊断培训的多智能体模拟系统，基于论文'LLMs Can Simulate Standardized Patients via Agent Coevolution'。该系统通过病人智能体和医生智能体的协同进化，模拟真实的医患交互场景，包含RAG检索系统和评估框架，支持动态专科医生招募和危机场景训练。