Multimodal AI
Across Scales
Trustworthy AI methods and open tools that bridge biological scales — from molecules to cells, tissues, and populations. A Research Programme at Human Technopole‘s Computational Biology Research Centre.
What we do
Modern biology generates data at every scale — molecules, cells, tissues, patients. We build the AI methods that make sense of all of it together, and the open tools that put those methods in the hands of working scientists.
Bridge the scales
Multimodal latent representations that connect microscopy, omics, and health records — so molecular features can be read alongside cellular phenotypes and clinical outcomes.
Know what we don’t know
Posterior models that return distributions of plausible answers, not single guesses — and calibrated uncertainty estimates scientists can actually act on.
Methods become tools
Research-grade methods are only useful when they’re installable, documented, and FAIR. We invest in research software engineering so our work reaches the bench.
Explore
Our Goals →
The vision driving the programme and the methodological pillars it’s built on.
Projects →
What we’re working on right now — denoising, image splitting, spectral unmixing, RNA, cross-modal disease risk.
Solutions →
Open-source tools and methods you can install today.
People →
The team behind MAIAS.
Related initiative at HT
Scientific AI Flagship at Human Technopole
A four-year, institute-wide AI initiative at HT, co-coordinated by Florian Jug and Francesca Ieva. Independent of MAIAS — but closely related, sharing people, projects, and a common commitment to trustworthy AI for the life sciences.