AI for Health

AI for Health connects algorithmic innovation with clinical questions. We develop methods that learn effectively from limited, heterogeneous or imperfect medical data, with applications to diagnosis, prognosis and clinical decision support.

Research directions

• Machine learning and medical image analysis.
• Learning from limited and heterogeneous health data.
• Privacy-preserving collaboration through federated learning.
• Signal processing and deep learning for neuroimaging.

A two-way exchange with clinical practice

Clinical challenges guide the development of new learning methods, while methodological advances are evaluated on real-world health problems. This exchange links the design of algorithms with the evidence needed to assess their usefulness.

Collaboration across institutions

Complementary expertise in medical AI, federated learning, neuroimaging and uncertainty connects health research with the unit’s foundations and systems pillars.