gms | German Medical Science

62. Jahrestagung der Deutschen Gesellschaft für Medizinische Informatik, Biometrie und Epidemiologie e. V. (GMDS)

Deutsche Gesellschaft für Medizinische Informatik, Biometrie und Epidemiologie

17.09. - 21.09.2017, Oldenburg

Integrating relational data into clinical information model based data repositories

Meeting Abstract

Search Medline for

  • Erik Tute - Peter L. Reichertz Institut für Medizinische Informatik der Technischen Universität Braunschweig und der Medizinischen Hochschule Hannover, Hannover, Deutschland
  • Birger Haarbrandt - Peter L. Reichertz Institut für Medizinische Informatik der Technischen Universität Braunschweig und der Medizinischen Hochschule Hannover, Hannover, Deutschland

Deutsche Gesellschaft für Medizinische Informatik, Biometrie und Epidemiologie. 62. Jahrestagung der Deutschen Gesellschaft für Medizinische Informatik, Biometrie und Epidemiologie e.V. (GMDS). Oldenburg, 17.-21.09.2017. Düsseldorf: German Medical Science GMS Publishing House; 2017. DocAbstr. 149

doi: 10.3205/17gmds145, urn:nbn:de:0183-17gmds1451

Published: August 29, 2017

© 2017 Tute et al.
This is an Open Access article distributed under the terms of the Creative Commons Attribution 4.0 License. See license information at http://creativecommons.org/licenses/by/4.0/.


Outline

Text

Introduction: A Clinical data warehouse (CDW) is a means for making routine data available for secondary use e.g. for supporting research. A clinical information model (CIM) based CDW has potential merits [1], [2]. Hannover Medical School (MHH) has a CDW based on a relational data schema and a CIM based CDW as part of medical informatics research activities [3]. This work addresses the feasibility of a generic method for the conversion of relational data into a CIM based representation.

Methods: In an agile process, we developed a generic method and simple supporting tools, followed by a pragmatic evaluation regarding manual efforts and performance in given technical environment. Data was extracted from a Micosoft SQL-Server 2012 and loaded into an openEHR based EHR platform (Think!EHR). We tested the approach using six different clinical contents. For brevity, we describe only results from evaluation with mechanical ventilation data.

Results: The first step for converting the data is to build the content models (e.g. openEHR Archetypes and Templates). Second step is to create a SQL-Statement extracting all relevant data from the relational database. We developed a Web-App for mapping the columns of the SQL-Statement’s result to the content model’s paths. We define the mappings by declaring instructions like “if column fulfills condition then execute actions with the following parameters”. Three different action types suffice (create composition, create subtree, create entry). The Mapping-Tool returns an instruction set, which is processed by a self-developed generic SQL Server Integration Services (SSIS) C#-Script-Task, which converts the instance data to JSON-representations matching the content model. Finally, the data is sent to the target database using REST calls. Based on the experience gained from mapping mechanical ventilation data from two patient data management systems, the most labor-intensive task was to gain a correct understanding of the contents of the data (stored in an Entity–attribute–value model) to create an accurate mapping. This work also entails dependency of local experts knowing the source data. To stay reasonable, we mapped many attributes on generic data fields. Manual creation of the mapping included approximately 110 attributes and 75 different actions to perform and took less than a day. Conversion of approximately 180 million data rows (circa 12 years of data) took less than a day.

Discussion: We developed a generic method and supporting tools fulfilling the task with an acceptable performance. Findings from [1] and [2] support our finding of performance and technical resource requirements to be uncritical at least in a one-clinic setting. Understanding the contents of clinical data is a bottleneck because it is labor-intensive and entails dependencies of local experts. Findings from other work [4], [5], [6], [7] support that finding. Evidently, our pragmatic evaluation with few different clinical contents cannot exclude requirements arising from other data, which cannot be met with our approach. Source data verification and support of incremental data loading are must-have functionalities that are currently not supported by our unsophisticated prototypical tools. Nevertheless, we made concepts for that and intend to add these functionalities.



Die Autoren geben an, dass kein Interessenkonflikt besteht.

Die Autoren geben an, dass kein Ethikvotum erforderlich ist.


References

1.
Haarbrandt B, Gerbel S, Marschollek M. Einbindung von openEHR Archetypen in den ETL-Prozess eines klinischen Data Warehouse. In: GMDS 2014. 59. Jahrestagung der Deutschen Gesellschaft für Medizinische Informatik, Biometrie und Epidemiologie e.V. (GMDS). Göttingen, 07.-10.09.2014. Düsseldorf: German Medical Science GMS Publishing House; 2014. DocAbstr. 230. DOI: 10.3205/14gmds227 External link
2.
Marco-Ruiz L, Moner D, Maldonado JA, Kolstrup N, Bellika JG. Archetype-based data warehouse environment to enable the reuse of electronic health record data. Int J Med Inform. 2015 Sep;84(9):702-14. DOI: 10.1016/j.ijmedinf.2015.05.016 External link
3.
Hannover Medical School Translational Research Framework. Peter L. Reichertz Institute for Medical Informatics, [cited 2016 Dec 27]. Available from: https://plri.de/en/forschung/projekte/hannover-medical-school-translational-research-framework-hamstr External link
4.
Turley CB, Obeid J, Larsen R, et al. Leveraging a Statewide Clinical Data Warehouse to Expand Boundaries of the Learning Health System. eGEMs. 2016;4(1):1245. DOI: 10.13063/2327-9214.1245 External link
5.
Oniki TA, Zhuo N, Beebe CE, Liu H, Coyle JF, Parker CG, Solbrig HR, Marchant K, Kaggal VC, Chute CG, Huff SM. Clinical element models in the SHARPn consortium. J Am Med Inform Assoc. 2016 Mar;23(2):248-56. DOI: 10.1093/jamia/ocv134 External link
6.
Haarbrandt B, Wilschko A, Marschollek M. Modelling of Operative Report Documents for Data Integration into an openEHR-based Enterprise Data Warehouse. Stud Health Technol Inform. 2016;228:407-11.
7.
Haarbrandt B, Marschollek M. Modeling and Integration of Intensive Care Data Into an openEHR-based Enterprise Data Warehouse. Curr Ther Res Clin Exp. 2016;78:Supplement 8-9.