English 
Deutsch 
NFDI4Health is already cooperating with the Mælstrom Research group on different levels e.g. through direct cooperation of core NFDI4Health members or re-using software and methods developed by Mælstrom. In addition, Dr. Isabel Fortier supports NFDI4Health as member of the NFDI4Health’s Scientific Advisory Board (SAB). To benefit more intensively from the experience of the Maelstrom Research Group and to strengthen the cooperation, a team from NFDI4Health visited the Maelstrom Group in Canada during the week from December 5th to December 9th 2022.
Both groups spent the first two days presenting their core activities and the status of developments - in terms of software development, collection and cataloguing of studies as well as the network projects and application areas where joint data analyses are planned and data harmonization is necessary.
Mælstrom Research gave an extensive overview of their annotation procedure, beginning with the collection of metadata for research projects or single studies and the consecutive cataloguing of study documents using their in-house classification system, the Mælstrom taxonomy. The Software MICA [2], a development of the OBiBa group in collaboration with Maelstrom [3], makes studies and variables discoverable but also allows filtering and comparing studies based on the Mælstrom taxonomy. Of note, Maelstrom fully relies on a centralized approach to study cataloguing and variable annotation to ensure a high homogeneous standard. NFDI4Health received a brief tutorial on the annotation procedure, so that selected data dictionnaries could be annotated and made discoverable on the Central Health Study Hub [4]. A pilot for further annotation support by the Maelstrom Group was agreed on.
NDI4Health presented the current status of the Central Health Study Hub and the integration of the MICA software within the Central Health Study Hub. Projected functional software extensions on both sides were discussed to align development plans. Also, the exchange between groups was intensified when the Metadata Annotation Workbench developed by NFDI4Health was presented and the user experiences with the tool were discussed. Further potential areas of co-operations and the possibility of common scientific publications were also addressed. A first focus will lie on improving automatisation of Mælstrom taxonomy cataloguing. Both groups have preliminary results and agreed to work closely together.
Furthermore, Mælstrom research introduced in detail their data harmonisation procedures [5] and R package [6] they developed to support the process, limit errors and improve documentation. The multistep method offered great insight into their complex approach of successfully harmonising variables from different studies to ensure comparability. Based on the NFDI4Health use cases ‘Nutritional Epidemiology’ and ‘Epidemiology of Chronic Diseases’ with over ten participating studies, appropriate strategies for harmonisation were discussed. The methods of both teams were compared and suggestions for possible improvements were offered by both teams.
Overall, the meeting was a very fruitful and inspiring. It illustrated the need to plan well ahead to ensure a high cataloguing standard for the NFDI4Health Central Study Hub. Future collaboration with the Mælstrom Research group will support this. The two groups will continue to collaborate to promote the discoverability and sharing of personal health data internationally to boost exploitation of these valuable datasets.
NFDI4Health Team participating: Prof. Dr. Carsten Oliver Schmidt (Uni Greifswald), Prof. Dr. Juliane Fluck (ZB MED), Johannes Darms (ZB MED), Julia Fürst (ZB MED), Sophie Klopfenstein (BIH/Charite), Dr. Carolina Schwedhelm (MSC Berlin), and Dr. Timm Intemann (BIPS).
[1] https://www.maelstrom-research.org/
[2] https://www.obiba.org/pages/products/mica/
[3] Open Software for Epidemiology (OBiBa)
[4] https://csh.nfdi4health.de/
[5] Fortier, I et al. (2022) Life course of retrospective harmonization initiatives: key elements to consider. J Dev Orig Health Dis. 1-9. DOI: 10.1017/s2040174422000460
[6] Fabre, G et al. harmonizR (unpublished R package)
