This presentation addresses a central bottleneck in Artificial Intelligence (AI) in medicine: the limited accessibility and usability of health data. While large volumes of clinical and biomedical data exist, they are often fragmented, difficult to discover, and constrained by regulatory and organisational barriers, which significantly limits their effective use in research. We present the concept of cloud-ready health data platforms as a foundational solution to this challenge. These platforms integrate data, metadata, computational environments, and governance processes into secure, scalable infrastructures that enable compliant and reproducible analysis. By addressing not only technical but also legal and regulatory aspects, they provide a practical pathway towards FAIR (Findable, Accessible, Interoperable, Reusable) data in real-world settings. The approach is illustrated through concrete implementations developed within NEXUS Personalized Health at ETH Zurich, including the Biomedical Informatics Platform (BMIP), the KSB Data Platform, and the GECO platform at the Ente Ospedaliero Cantonale (EOC). These examples demonstrate how such infrastructures can support clinical data reuse, translational research, and application development across institutional boundaries. Overall, the work highlights that AI readiness in medicine depends fundamentally on the maturity of data platforms, and positions these infrastructures as a key enabler for scalable, reproducible, and impactful data-driven research.
Daniel Stekhoven (Thu,) studied this question.