BMC medical informatics and decision making · 2026
A distributed analysis approach for pharmacovigilance data from electronic medical records in German university hospitals: the POLAR_MI ETL Pipeline.
СтатьяНазвание и аннотация: Перевод в очереди
Обработка статьи · этапы, ошибки и управление
Загрузка состояния…
Ошибки и ограничения
История · было → стало
Состояние обновляется каждые 5 секунд, пока блок открыт. Процент показывается только там, где известен объём работы. Пауза останавливает продолжение этапа; текущий запрос к источнику может завершиться.
Kesselmeier Miriam, Thalheim Torsten, Schmidt Florian, Peschel Thomas, Palm Julia, Strübing Alexander, Medek André, Przybilla Jens, Wermund Anna Maria, Maas Renke, Härterich Steffen, Redeker Louisa, Federbusch Martin, Steinbach Daniel, Gewehr Jan, Wurlitzer Marcus, Riedel Andrea, Meineke Frank, Neumann Daniel, Scherag André, Loeffler Markus, POLAR_MI
Аннотация
BACKGROUND: The cooperative project "POLypharmacy, drug interActions and Risks" (POLAR_MI) of the Medical Informatics Initiative Germany (MII) aimed at detecting medication-related risks attributed to polymedication in adult patients from German university hospitals. Here, we report technological challenges and solutions to undertake this large-scale multicentre project relying on routine healthcare data stored and processed by the data integration centres, which were recently established at the German university hospitals.
METHODS: We developed and implemented a two-step, privacy-preserving, distributed analysis approach to analyse clinical routine healthcare data relying on the internationally balloted MII HL7® FHIR® core data set specifications (version 1.0). In this approach, without direct data access for the data analysts, a local data aggregation step comprising data extraction, transformation (including statistical analyses) and loading (ETL) at each university hospital's data integration centre was followed by a central random-effects meta-analysis.
RESULTS: Using an iterative procedure between data integration centres and the cross-institutional analysis team, we overcame many challenges and established the "POLAR_MI ETL Pipeline". These challenges originated from the heterogeneity of the data integration centres and the IT infrastructure of the related university hospitals including their local hospital information system. Applying our pipeline, we analysed data from ten centres on nearly 800,000 encounters from about 500,000 patients.
CONCLUSIONS: For the first time within the MII infrastructure, we demonstrated that a project on routine healthcare data is feasible using a distributed analysis approach based on the recently established network of data integration centres in Germany. We describe an approach to obtain a valuable and insightful overview of health risks in routine healthcare and share the related code. Moreover, we propose improvements to the ETL process for future distributed analyses. Finally, our data-related challenges and solutions can be adapted to other healthcare settings (in other countries and initiatives, respectively) as long as data integration centres equivalents and a common data format/model are available.
TRIAL REGISTRATION: POLAR_MI was registered on 27/11/2020 in the "HMA-EMA Catalogues of real-world data sources and studies" (EU PAS number: EUPAS36582).
METHODS: We developed and implemented a two-step, privacy-preserving, distributed analysis approach to analyse clinical routine healthcare data relying on the internationally balloted MII HL7® FHIR® core data set specifications (version 1.0). In this approach, without direct data access for the data analysts, a local data aggregation step comprising data extraction, transformation (including statistical analyses) and loading (ETL) at each university hospital's data integration centre was followed by a central random-effects meta-analysis.
RESULTS: Using an iterative procedure between data integration centres and the cross-institutional analysis team, we overcame many challenges and established the "POLAR_MI ETL Pipeline". These challenges originated from the heterogeneity of the data integration centres and the IT infrastructure of the related university hospitals including their local hospital information system. Applying our pipeline, we analysed data from ten centres on nearly 800,000 encounters from about 500,000 patients.
CONCLUSIONS: For the first time within the MII infrastructure, we demonstrated that a project on routine healthcare data is feasible using a distributed analysis approach based on the recently established network of data integration centres in Germany. We describe an approach to obtain a valuable and insightful overview of health risks in routine healthcare and share the related code. Moreover, we propose improvements to the ETL process for future distributed analyses. Finally, our data-related challenges and solutions can be adapted to other healthcare settings (in other countries and initiatives, respectively) as long as data integration centres equivalents and a common data format/model are available.
TRIAL REGISTRATION: POLAR_MI was registered on 27/11/2020 in the "HMA-EMA Catalogues of real-world data sources and studies" (EU PAS number: EUPAS36582).