Post by Nacim Rahal

Today we're releasing our first open-source model on Hugging Face. The work on this model, DoctoBERT, was led by Bofeng Huang and our research team. It's a step towards what we believe AI in healthcare should look like: rigorously evaluated, and useful in real clinical contexts. DoctoBERT is most useful for anyone who needs a machine to read and understand French medical text, whether that's for automating clinical workflows, improving patient data management, or advancing medical AI research. Why release it openly? Because trust in medical AI cannot be built behind closed doors. Open science is not a side project for us. It is part of how we operate at Doctolib Research, alongside our partnerships with Inria and the CHU de Nantes. This release is one small piece of a larger ambition: building clinical AI agents with the highest medical reliability, to help diagnose, treat and prevent disease more effectively. Expect more in the months to come. And if you’re interested in pushing the boundaries of AI in healthcare with us, let's have a chat. Huge credit to Bofeng Huang, to the entire Doctolib Research team, and to GENCI for their support. Doctolib on Hugging Face : https://lnkd.in/embc3yzD