Revolutionizing Health Care with AI: Integrating Multi-Modal Data for Personalized Medicine
Abstract
The integration of multi-modal biomedical data represents a cornerstone for the next generation of personalized medicine. This talk presents CSEM’s AI for Life Sciences approach to combining heterogeneous data modalities—ranging from high-content imaging and omics to clinical and patient metadata—into predictive and interpretable AI systems. Emphasis is placed on biologically informed model architectures and explainable AI methods that preserve mechanistic insight while maintaining high predictive performance. Two flagship use cases are discussed: multimodal phenomics for functional profiling of cellular aging and biologically informed neural networks for disease severity prediction, including acute kidney injury. Together, these examples demonstrate how integrating prior biological knowledge with modern machine learning enables robust, scalable, and clinically meaningful AI solutions. The work highlights how such approaches can accelerate diagnostics, improve patient stratification, and support translational research at the interface of academia and industry.
Publication Reference
TMG Follow-Up Event, University of Zurich (UZH), Zurich, Switzerland
Year
2025