MEDEVAL
收藏资源简介:
MEDEVAL是一个多层次、多任务、多领域的医疗基准数据集,旨在促进医疗领域语言模型的发展。该数据集包含来自多个医疗系统的数据,覆盖35个人体部位和8种检查模式,共有22,779个收集的句子和21,228份报告。数据集提供多层次的专家标注,支持广泛的医疗任务。MEDEVAL的应用领域包括医疗报告的自动分类和生成,旨在提高医疗决策的效率和准确性。
MEDEVAL is a multi-level, multi-task, multi-domain medical benchmark dataset designed to advance the development of language models in the medical field. This dataset contains data from multiple medical systems, covering 35 human body parts and 8 imaging modalities, with a total of 22,779 collected sentences and 21,228 medical reports. It provides multi-level expert annotations to support a wide range of medical tasks. Application scenarios of MEDEVAL include automatic classification and generation of medical reports, aiming to improve the efficiency and accuracy of medical decision-making.

- 1MedEval: A Multi-Level, Multi-Task, and Multi-Domain Medical Benchmark for Language Model Evaluation加州大学圣地亚哥分校 · 2023年



