ConfiDx
收藏资源简介:
ConfiDx 是一个由明尼苏达大学计算健康科学部创建的疾病诊断数据集。该数据集包含来自 MIMIC-IV 和 UMN-CDR 数据库的 24884 条临床笔记,涵盖了内分泌、心脏病学和肝病学三个专科。数据集包含各种程度的诊断模糊性,并经过精心注释。ConfiDx 的创建旨在解决临床决策中诊断不确定性的问题,通过提供准确的诊断和解释,以提高自动诊断系统的可靠性。
ConfiDx is a disease diagnosis dataset developed by the Department of Computational Health Sciences at the University of Minnesota. It contains 24,884 clinical notes sourced from the MIMIC-IV and UMN-CDR databases, covering three medical specialties: endocrinology, cardiology, and hepatology. The dataset features varying degrees of diagnostic ambiguity and has been meticulously annotated. ConfiDx was developed to address the challenge of diagnostic uncertainty in clinical decision-making, aiming to enhance the reliability of automated diagnostic systems by providing accurate diagnoses and their corresponding rationales.




