UroLlmEvalSet
收藏资源简介:
UroLlmEvalSet数据集由约翰内斯古腾堡大学美因茨大学医学中心的研究团队创建,旨在评估开源大语言模型在德国泌尿科医生笔记中的肿瘤文档处理能力。该数据集包含149条经过注释的文本片段,涵盖了151个肿瘤诊断,其中84个标注了首次诊断日期。数据来源于匿名化的泌尿科医生笔记,经过严格的隐私保护处理,确保患者信息的匿名性。数据集的应用领域主要集中在医疗文档自动化处理,特别是肿瘤诊断的自动编码和信息提取,旨在提高肿瘤文档处理的效率和准确性,减少医疗文档人员的工作负担。
The UroLlmEvalSet dataset was developed by a research team from the University Medical Center Mainz, Johannes Gutenberg University, with the aim of evaluating the tumor document processing capabilities of open-source large language models (LLMs) in German urologist clinical notes. This dataset consists of 149 annotated text segments, covering 151 tumor diagnoses, 84 of which are labeled with their initial diagnosis dates. The data is derived from anonymized urologist clinical notes, and has undergone rigorous privacy protection procedures to guarantee the anonymity of patient information. The primary application areas of this dataset lie in automated medical document processing, specifically automatic coding and information extraction for tumor diagnoses, with the objectives of enhancing the efficiency and accuracy of tumor document processing and alleviating the workload of medical documentation personnel.




