Healthcare is ever more challenging. New technologies allow us to acquire phenotypic and genotypic data in volumes that exceed current capabilities to organise, maintain and analyse it all. From what we can analyse, we learn more about conditions and their effective treatments. This, in turn, makes it possible to do more, differently in clinical practice. But this virtuous cycle can only proceed with scaleable and robust systems architectures to support it. Having an advanced machine learning system to perform (say) feature identification for brain imaging is an impressive engineering feat but it is of little practical value unless it is supported by a robust infrastructure that supplies the right data, at high quality, with the ability to support long term care pathways and longditudinal studies involving federated, heterogeneous data. To add to complexity, these new systems must be built in the context of large and already complex existing healthcare systems at a time when the demands on healthcare systems, including their cost, continue to increase. This is especially true for chronic diseases, as demographics change. Taking examples from Health Data Research UK, I will explain why (from the viewpoint of computation and data) systems engineering is a major (but not insuperable) challenge for computing science itself. I will focus on the practical challenges of data governance, personal data and human-centred care, and give a personal view of why we should have optimism for the future of data intensive healthcare in practice.
