Univ. Lille, CHU Lille, ULR 2694 - METRICS: Évaluation des Technologies de Santé et des Pratiques Médicales, Lille, France.
InterHop, Rennes, France.
J Clin Monit Comput. 2023 Apr;37(2):461-472. doi: 10.1007/s10877-022-00898-y. Epub 2022 Aug 6.
This paper describes the development and implementation of an anesthesia data warehouse in the Lille University Hospital. We share the lessons learned from a ten-year project and provide guidance for the implementation of such a project. Our clinical data warehouse is mainly fed with data collected by the anesthesia information management system and hospital discharge reports. The data warehouse stores historical and accurate data with an accuracy level of the day for administrative data, and of the second for monitoring data. Datamarts complete the architecture and provide secondary computed data and indicators, in order to execute queries faster and easily. Between 2010 and 2021, 636 784 anesthesia records were integrated for 353 152 patients. We reported the main concerns and barriers during the development of this project and we provided 8 tips to handle them. We have implemented our data warehouse into the OMOP common data model as a complementary downstream data model. The next step of the project will be to disseminate the use of the OMOP data model for anesthesia and critical care, and drive the trend towards federated learning to enhance collaborations and multicenter studies.
本文描述了里尔大学医院麻醉数据仓库的开发和实施。我们分享了一个十年项目的经验教训，并为该项目的实施提供了指导。我们的临床数据仓库主要由麻醉信息管理系统和医院出院报告收集的数据提供。数据仓库存储历史和准确的数据，行政数据的准确性水平为每天，监测数据的准确性水平为每两天。数据集市完成了架构，并提供了二级计算数据和指标，以便更快、更轻松地执行查询。在 2010 年至 2021 年间，为 353152 名患者整合了 636784 份麻醉记录。我们报告了在项目开发过程中的主要关注点和障碍，并提供了 8 个技巧来处理这些问题。我们已经将我们的数据仓库实现到 OMOP 通用数据模型中，作为补充的下游数据模型。项目的下一步将是传播麻醉和重症监护 OMOP 数据模型的使用，并推动联合学习的趋势，以增强协作和多中心研究。
