HiRID
收藏资源简介:
HiRID数据集是由瑞士伯尔尼大学医院的重症医学部门提供的,包含超过33,000名患者的重症监护数据。该数据集与苏黎世联邦理工学院合作开发,包含712个常规收集的生理变量、诊断测试结果和治疗参数。HiRID具有比其他已发布的重症监护数据集更高的时序分辨率,特别是在床边监测方面,大多数生命体征每2分钟记录一次。该数据集旨在通过提供一个全面的机器学习基准,覆盖与ICU相关的广泛任务,从而改善机器学习方法在时间序列数据上的应用。HiRID的应用领域包括预测任务,如警报系统预测不同类型的器官衰竭,以及开发新的方法以支持临床决策。
The HiRID dataset, provided by the Department of Intensive Care Medicine at the University Hospital of Bern, Switzerland, contains intensive care unit (ICU) data from over 33,000 patients. Developed in collaboration with ETH Zurich, the dataset encompasses 712 routinely collected physiological variables, diagnostic test results, and therapeutic parameters. HiRID offers higher temporal resolution than other publicly available ICU datasets, particularly for bedside monitoring, with most vital signs recorded every 2 minutes. This dataset aims to improve the application of machine learning methods on time-series data by providing a comprehensive machine learning benchmark covering a wide range of ICU-related tasks. Application scenarios of HiRID include predictive tasks such as alert systems for predicting different types of organ failure, as well as the development of novel methods to support clinical decision-making.




