Multimodal Clinical Benchmark for Emergency Care (MC-BEC)

MC-BEC是由哈佛医学院生物医学信息学系和斯坦福大学计算机科学系共同创建的综合性基准数据集，用于评估急诊医学中的基础模型。该数据集包含2020至2022年间超过10万次持续监测的急诊科访问记录，涵盖了从分钟到天的时间尺度上的临床相关预测任务，如患者恶化、处置和急诊科再访。数据集包括广泛的详细临床数据，如分诊信息、先前诊断和药物、持续测量的生命体征、心电图和光体积描记波形、就诊期间下达的订单和给药、影像学研究的自由文本报告以及急诊诊断、处置和后续再访的信息。MC-BEC旨在鼓励研究人员开发更有效、可泛化和可访问的多模态临床数据基础模型，以改善患者预后并推进真实世界电子健康记录数据的分析。

MC-BEC is a comprehensive benchmark dataset jointly created by the Department of Biomedical Informatics at Harvard Medical School and the Department of Computer Science at Stanford University, designed for evaluating foundation models in emergency medicine. It contains over 100,000 continuously monitored emergency department visit records from 2020 to 2022, covering clinically relevant prediction tasks across timescales ranging from minutes to days, such as patient deterioration, disposition, and emergency department revisits. The dataset includes a wide range of detailed clinical data, including triage information, prior diagnoses and medications, continuously measured vital signs, electrocardiogram (ECG) and photoplethysmogram (PPG) waveforms, orders and medications administered during the visit, free-text reports of imaging studies, as well as information on emergency diagnoses, dispositions and subsequent revisits. MC-BEC aims to encourage researchers to develop more effective, generalizable and accessible multimodal clinical foundation models, so as to improve patient outcomes and advance the analysis of real-world electronic health record (EHR) data.

搜集汇总

背景与挑战

背景概述

MC-BEC是一个包含10万次急诊科访问记录的多模态临床数据集，涵盖多种时间尺度的预测任务和详细的临床数据，旨在推动急诊医学基础模型的研究。