 I'm going to talk about the primary use of data, and in government and health care, the primary use of data is running public services. And I'm going to start by talking about some of the work we're doing on NHS.uk. We're building the next generation of digital services that can connect people to the care that they need. A little plug, if you go on digital.nhs.uk you can have a look at some of the work we've been doing to prototype what the future of NHS services could be online. Now the opportunity in that is more than just about buying more technology or putting paper based services online. Consumer expectations have been radically transformed by the digital revolution that we've heard about already. Services like Airbnb and Netflix and Amazon have raised expectations and people are now expecting the same of public services. They expect us to get rid of the lumps and the bumps like when you're trying to book an appointment, when you're trying to pick up your prescriptions. And there's a big opportunity to reduce the burden on professionals as well to make sure that they can spend time seeing patients rather than dealing with paperwork. In short, what we're trying to do is build an NHS that's fit for the internet age. But doing all of that is impossible with the data infrastructure that we've got. So today I want to ask the question, what kind of data infrastructure do we need to deliver modern public services? I'm going to share with you three challenges of many related to data in healthcare, as you can imagine, but just three that we've been grappling with. The first one is who owns our data and when should it be shared? So Andrew's already touched on some of this. When we go out and speak with real people in the real world, as we do every other week when we're building NHS.uk, people ask us, why isn't information shared between different bits of the NHS and why am I repeatedly asked the same questions? Now there's a simple answer to that, which is that the NHS isn't a single thing. It's actually 9,000 organisations and over one million professionals. But in the same breath, patients ask us, they sometimes express concerns about where the data is stored and who can access that. And we know that, as we've heard already, beyond the purposes of direct care, there's huge benefits to using data and collecting data and using data. For things like healthcare planning, for things like medical research, this thing can save lives. So it's not just interesting, it's not just good for academic research, it's good to advance medicine and it can help save lives. But what we struggle to distinguish between in the kind of policy debate is the contradiction between data for the purpose of direct care and which requires transactional permission-based access and the bulk data that we need to do healthcare research. Now these are really difficult, ethical and social issues. You know, who do I trust? How sensitive are different parts of my healthcare data? Is pseudonymisation anonymised enough? What am I willing to give up in exchange for what kind of services? Now many have made the case for giving patients direct control of their data, but that's easier said than done. We've had decades of patient notes being written for the benefit of clinicians, but not for patients. So it's going to take a massive cultural shift and not just a bit of IT to reorientate those healthcare data around patients rather than the NHS organisations that we've got today. So that's the first challenge and that's a big one. The second challenge that we've got is even if we agree data can be shared, how can we do it in practice? So I met a GP at the weekend who was telling me that in her local area they've decided to move all of the GP practices to the same IT system from the same supplier. Now from a kind of competition perspective and from a kind of flexibility perspective, that is quite depressing. But it's a pragmatic response to one of the biggest challenges that we've got in healthcare data, which is interoperability. We've got systems that talk different languages. We've got data locked up behind complex systems and we've got no APIs to get data in or out. And we've got organisations, 9,000 of them plus social care, that have very similar but different rules about how you manage data. So that makes joining up those silos even harder. Now we've come up with ways over the decades to try and tackle some of these problems. So we've now got an industry in data standards, especially in healthcare, but implementing those data standards is really hard. It can take years to implement the full specification of a healthcare data standard in a system. Take one small example, even the, well one small simple example actually, the NHS number. Everybody in theory in this room has an NHS number and it's now law that your health, any kind of care records that are stored by NHS organisations are tagged with your NHS number. Except the foreigners, newly born patients, unidentified patients, even the Scottish don't have an NHS number. So that's an issue of policy but practically that's an issue that affects people delivering care every day. And you've also got issues around technology. I know of at least one IT system that can't use the NHS number because it doesn't allow numbers longer than 10 digits. Practical problems that are actually stopping us from using data in the right way. And the third final issue challenge that we've got, and this is one that frustrates a lot of people who are dealing directly with data on a day-to-day basis. It's the basics. It's some of the real basic stuff. I'm not talking about healthcare record data. I'm not talking about open data or big data. I'm talking about boring data. I'm talking about the kind of data that everyday services rely on. So things like lists of hospitals, opening hours, operations, medicines, pharmacies, addresses. I try and do a mash-up today of the data from CQC and the data about hospitals that you get from the NHS. I've tried to do that and it's quite hard because they don't even use the same consistent identifiers for hospitals. If we can't even do that, how can we deliver health, join up healthcare? Now those are the kind of issues that people trying to deliver services and also do research are wrangling with every single day. So they're having to roll their own solutions because the data that's already there isn't good enough, it's not accessible enough and they can't work with it. Now this stuff is hard and it's not particularly exciting and it's even harder when we've got a healthcare system that's changing constantly and restructuring constantly. But I think what we found in doing the work that we've done is that developers and researchers alike want data that's usable, predictable, reliable and comprehensive. That's the very minimum we need to deliver great services and advanced medicine. So those are a small snapshot of the challenges that we've got. Now in healthcare we've had well over decades of effort to deliver a paperless NHS. So we've now got trillions of rows of data sat in database systems and spreadsheets. Some of it's published, some of it's replicated several times over, some of it's usable, some of it's not. None of this was the result of any grand design. It's just a series of pragmatic decisions made by lots of people over a number of decades. But that kind of data infrastructure really isn't good enough anymore. We're failing to meet the needs of the users of the data. That means the staff that have to grapple with crap IT systems every day. That means the patients who can't get access to services online. And that means those on the outside as well in research who want to advance medicine and advance our understanding of how to generate efficiencies and make our NHS better. But they can't do that with the data infrastructure that we've got. Quite simply we can't have an internet age healthcare service without a new data infrastructure. So my kind of short prognosis, short kind of prescription for this is that we need to start treating data infrastructure as a public asset. We need the same kind of ambition with our data that Basil Jett had when he built the Victorian sewage system in London. That's going to require long term thinking and new patterns for granting permission to different kinds of data. The solution is not more centralisation and it's not more detailed standards that are hard to implement. We tried all of those things before. It is about an open joined up approach. It's about some common principles for how we organise the stuff. And it's about a whole effort across the NHS, across social care, but also academia as well. But building this new data infrastructure for a 21st century NHS is going to be really hard work, but it'd be worth it. Thank you.