Winner of the Bpifrance call for projects ‘Digital Commons for Generative AI’, PARTAGES aims to develop AI tools serving practical use cases for healthcare professionals. The GCS HOURAA will contribute to this ambitious initiative over two years with dedicated funding.
Natural Language Processing (NLP) offers numerous strategic use cases in healthcare: automatic transcription of consultation exchanges, automated summarization of medical records, and even support for clinical decision‑making or medico‑administrative tasks. Overall, these new tools can improve both the quality of work for medical staff and the patient–caregiver relationship, ultimately delivering significant benefits for the public health system.
To address these challenges, the Health Data Hub and its partners have developed ‘PARTAGES’ (Advanced develoPment of digital commons for Generative ArtificIal Intelligence in Health) as part of the France 2030 call for projects ‘Digital commons for generative artificial intelligence’. Operated on behalf of the State by Bpifrance, this initiative aims to accelerate the creation and availability of digital commons across the entire generative AI value chain, ensuring broad usability and fostering the emergence of innovative products and services.
An ambitious, multi‑partner project involving the GCS HOURAA.
This project brings together 32 French stakeholders, including research laboratories (CNRS, INRIA, universities), public and private healthcare institutions, as well as deep‑tech companies. This collective effort aims to design tools tailored to the concrete needs of healthcare professionals and to democratize the use of generative AI in healthcare.
explains Audrey Sokolo, Executive Director of the GCS HOURAA.PARTAGES is an initiative that reflects our commitment to developing innovative solutions through a use‑case‑driven approach. Thanks to this collaboration with the Health Data Hub and other key partners, the GCS HOURAA is actively engaged in co‑developing artificial intelligence tools that will address concrete challenges. This project represents a unique opportunity for the university hospitals (CHUs) in our region to contribute to the creation of sustainable, open solutions designed around the needs of healthcare professionals.
And concretely, how is this going to work for the GCS HOURAA?
The GCS HOURAA will receive funding for two years in order to recruit the necessary resources to evaluate the algorithms that will be provided for three use cases (that is, to specialise the pre‑trained language models).
Explains Jean‑Christophe Bernadac, Director of Digital Services at the Hospices Civils de Lyon.Operationally, the PARTAGES project will enable us to assemble a cross‑functional team across the four university hospitals (CHUs). This team will be responsible for organising and carrying out the work of data extraction and annotation (such as medical reports). These data will be extracted from the Easily EHRs of the four institutions in order to assess the performance of the AI algorithms developed within the framework of this project for each identified use case.
The three defined use cases:
- Automatic pseudonymisation of textual health data for any secondary use of health data across the other use cases.
- Structured summarisation of medical reports for healthcare professionals to save time.
- Assistance with medical coding to facilitate the classification of hospital services and reimbursement by the national health insurance system.
Clarifies Professor Alexandre MOREAU‑GAUDRY, specialist in Digital Public Health and in evaluating the service provided by Technological Innovation at the Grenoble Alpes University Hospital.PARTAGES makes it possible to initiate and fine‑tune our regional scientific and real‑world data operations within our EDS Hub, which is currently being deployed across the four university hospitals participating in the GCS HOURAA. The goal is not only to contribute to national developments of new digital health tools involving AI—drawing on the richness of our hospital data—but also to locally adopt and assess the value provided by these new tools for users. This evaluation of the actual service provided will, naturally, be carried out in close interaction with users, following established methodological standards in the field. It will be a key factor in determining whether these new tools achieve lasting acceptance.
