SNOMED-DEX aims to remove one of the barriers to the development of AI in clinical medicine: the lack of data for training. Using advanced language models (LLMs), we generate and validate a unique corpus of annotated textual descriptions for all 368,000 concepts in the international SNOMED CT terminology, in French and English.

"Open Research Data is essential in our view because research in general, and digital health research in particular, cannot progress in isolation. By making our datasets freely accessible, where possible, we not only enable the reproducibility of our work, but above all we provide the global scientific community with the necessary foundations to create more robust and interoperable clinical analysis tools, ultimately benefiting patients."

MOVE-Share aims to clean, structure and make public the data collected at the Kinesiology Laboratory (HUG/UNIGE) as part of several research projects. With the help of a Data Curator, this data will be harmonised according to FAIR principles and MoveD guidelines (for data sharing in the field of motion analysis), then deposited in open repositories such as Yareta or Zenodo. The aim is to make these datasets accessible, reusable and better integrated into the scientific ecosystem.

"Open Research Data helps to boost essential but time-consuming tasks that are undervalued, often relegated to the background or even ignored, despite being mandatory for public funding bodies. Yet they are crucial to advancing research: they maximise the impact of existing data, avoid wasting resources and enable other teams to aggregate, compare and discover more quickly. UNIGE's Open Science strategy is fully in line with this vision of a more open, efficient and collaborative science – a vision to which we at the Kinesiology Laboratory are fully committed."

This project aims to structure, secure and enhance the value of data from animal experimentation through a modern approach to data management. It will improve traceability, scientific quality and data reuse while strengthening the application of the 3Rs principles (Replace, Reduce, Refine).

"Open Research Data is essential to ensure the transparency, reproducibility and robustness of scientific results. In a field as sensitive as animal experimentation, structured data sharing not only improves the quality of research, but also limits unnecessary duplication of experiments and therefore the use of animals. UNIGE's Open Science strategy fully supports this vision by promoting more responsible, collaborative and efficient research for the benefit of the scientific community and society."

This pilot project aims to evaluate the benefits of the open source tool DataLad, which suppports version control, data organisation, reproducibility and collaboration. Tests will be carried out on human neuroscience data, which represents a particular data mangement challenge due to its multimodality, sensitivity and the muliplicity of analytical approaches typically employed. If the tool shows promise, the ultimate goal is to implement this distributed data management tool on UNIGE infrastructure to make it available to all.

"Following Open Science principles has become a sine qua non of forward-looking research institutions. The UNIGE Open Research Data strategy allows researchers to be fully equipped with the resources, tools and practical support they need to ensure systematic and structured data sharing thereby maximising long-term value and utility to other groups. "

The NESTOR project aims to develop automatic extraction tools capable of transcribing scanned medical records from the Geneva Brain Bank (Belle-Idée Psychiatric Hospital, HUG). Given the sensitivity of this information, it will be processed using proven secure infrastructure, including that co-developed by the UNIGE Chair in Digital Humanities. A significant effort will be devoted to anonymisation in order to enable the sharing of data that is rarely available for research. Based on manually transcribed training data, synthetic data will be generated to help train a free, open-source generic mega-model that will be useful to teams in all disciplines (medicine, history, literature, law, etc.).

"The transformation of analogue data into digital data is a major challenge for the preservation and computational analysis of medical archives. Digital humanities have extensive experience in the automatic acquisition and restructuring of heritage data, but the models created are not suitable for medical documents. UNIGE's Open Science strategy is enabling progress in the creation of tools for the digitisation of medical data that can be used by other local and international entities, thanks to a rare collaboration between the Faculties of Medicine and of Humanities."