IMBI Freiburg: AI-enhanced Support for Collaborative Research Centers

The Working Group Research Data Management at the Institute of Medical Biometry and Statistics (IMBI) at the University Medical Center Freiburg is involved in several institution-level projects covering mostly biomedical research. These are (transregional) collaborative research centres (CRC/TRR) and excellence clusters, involving large numbers of scientists in several universities and other research institutions. For these, IMBI provides generic consultation in the area of research data management as well as specifically customised services using a combination of proprietary and open-source software. Centrepiece of the IMBI research ecosystem is its own platform, the Freiburg Research Data Tool (fredato). It provides a cloud application for file sharing among project members and version-controlled data storage. Publications, datasets and other data items can be annotated according to project-specific data schemata that the IMBI develops in collaboration with scientists from the partner projects. All metadata are stored in an open file format (JSON) and can be exported to public data repositories. Automated import of bibliographic metadata is also supported, e. g. from PubMed. We foresee to implement a connection between fredato and the electronic lab notebook software eLabFTW which the IMBI offers as a hosted service. fredato is built from well-established open-source software components. The cloud file hosting software NextCloud and the version control system software GitLab are used for file sharing and access management, respectively, with GitLab also providing advanced file versioning functionality. Metadata handling is covered by the OpenSearch engine which also allows for the creation of custom dashboards in project-specific fredato instances. The working group is currently building on institutional knowledge at IMBI to harness Large Language Models (LLM) for minimizing the documentation effort in research data management within the fredato system. LLMs will be trained on information from various sources to facilitate metadata enrichment and assist researchers in the annotation of datasets. As a basis for this research, the IMBI is pursuing the path of a local deployment of various LLMs in order to meet data protection law requirements for knowledge enrichment.