A new projects section — work in progress
We have just published a reorganised projects section on the lab website. The previous version grouped projects by funding source — European, national, industrial doctorates — which is a natural way to think about administrative paperwork, but not a particularly useful way to understand what we actually do.
The new structure organises things by research goal:
- Foundational models in health — our primary strategic direction, covering transformer-based models on electronic health records and multimodal models for emergency medicine
- Rare diseases — patient platforms, computational phenotyping, and in silico disease modelling
- Metabolomics — biomarker discovery and annotation methods
- Technology transfer — Xartec Salut and our industrial doctorate projects, where research is developed in direct partnership with hospitals and companies
- Climate change & sensors — AI for climate-health risk, sensor networks for ecosystem monitoring, and air quality research
This framing reflects better where we are heading as a lab. The foundational models goal in particular deserves a clear statement: we think that large models pre-trained on real clinical data — and made available for fine-tuning on specific tasks — are the most transformative opportunity for AI in healthcare right now, and it is where a significant part of our energy is going.
A caveat: the section is still incomplete. Several projects are described only briefly, some lack proper references to papers and software, and a few ongoing collaborations are not yet listed. We will keep filling it in over the coming weeks. In the meantime, the publications page remains the most complete record of what we have done.
Feedback welcome, as always.