The journey towards making best use of biomedical data often starts with a thorough assessment of the status quo and the target requirements. This includes stakeholder analysis and transformation management. What data is available in what form? What is the functional vision for leveraging this data? How should the data best be assessed, stored, and processed? Who is responsible for which parts of the data at what point? And who is affected by it?

Once a roadmap is set showing which data will be used from which sources, the data has to be integrated, transformed, cleansed, augmented, combined, and analyzed. Much dependent on the data at hand, these steps typically require in depth domain knowledge and iterative exploration: How can text data, structured data, and imaging data be combined in a single analysis? Which elements are most relevant to answer a specific question? How can predictions be made using the available data?

Pulling data together and creating analyses and predictive models is great. But the magic comes from getting your data productive. Designing and building software around the created models and insights and - closing the loop - around data acquisition itself to streamline and improve data quality is key to data utility: How can predictions be used for decision support? How can analyses be integrated into existing workflows? How can new models be evaluated in real world settings?