Project background The healthcare systems of the future must harness data effectively to support clinicians, allowing them to focus on patient care while leveraging AI to detect patterns beyond human perception, enhance diagnostic accuracy, optimise workflows, improve risk assessment and communication. Developing AI models that address these needs is particularly urgent in ageing societies, where rising patient numbers coincide with increasing workforce constraints.
To do so, we are developing the AI for Science Instrumentation Gym, which is designed to bridge this gap by placing data-driven hypothesis generation at the center of its mission. It introduces a critical intermediate step: the tokenization and cartography of scientific data. Through tokenization, complex data is transformed into coarse-grained, interpretable units. Through cartography, these units are organized into latent spaces that can be explored as structured landscapes. In this way, machine learning becomes a tool...