 Why do we need data and what do we need it for? Give examples of where I've seen data actually help improve lives and say yes to Matthew's challenge there and then outlining actually if we're going to make that happen and happen on a consistent basis what are the things that need to be in place across the country in order for that to happen well. I'm going to talk about research and statistics but clearly there's a huge amount of gain from using data for other purposes. I mean a recent report suggested that across the UK somewhere between 16 and 33 billion pounds can be saved to the public sector from use of data better and that's kind of through things like reducing fraud and proving efficiency as well as avoiding any harms from not sharing information across public services that we see sadly too frequently. So start off by saying something about the key challenge and for me it's about public finance. So our government expenditure revenue statistics that were published in August showed that over the last year public spending in Scotland is higher than the tax that we brought in and that's happened for the last couple of years and while there are moves to reduce that deficit across the UK it's unclear exactly how long that's going to take. What this means is there's ongoing financial challenge to the public sector and this though is fantastic that people are living longer the financial situation becomes more challenging as people get older and need higher expensive services. So I think it brings two opportunities for data for me. First one is the need to use data to drive performance within existing services, improve the efficiency and understand what works and invest in that. And then the second thing is the need to use data to transform the way that public services are delivered. The other element of the public finance constraints and challenges is the significant pressure that that puts on us collecting and analysing and using data. So we've seen earlier on during the year we had a consultation about reducing the sample size of the Scottish household survey for example. In the end we decided not to reduce the size because we managed to make the case for that but I think it illustrates that there's ongoing pressure on budgets. I think if the Secretary of State or Cabinet Secretary for Finance was here he would say if I'm making choices between investing in public services and investing in surveys or things like that I can't tell doctors that I'm shutting things but that we've got a cracking Scottish health survey or household survey. And that will be an ongoing thing. So using data to improve the quality and efficiency of services where I'm at, that's what I'm trying to achieve. I just wanted to run through three kind of examples of doing that, in particular a specific example around predictive analytics. This is using data about people, people or organisations but in this case people to try and identify those that could benefit from preventative services predicting those most likely to have quite negative outcomes or really good outcomes and then putting services in place upstream in order to avoid those negatives or enhance those and bring out the positives. Before I worked here in Scottish Government I worked in the Department of Health and I worked on services for people with long-term or chronic conditions. What we found in the analysis of data is that 5% of people that use hospital services, which is a really small number, account for half of all the use of hospital services. And that ends up being generally a few years of ill health and then all of a sudden there's a real chronic need for hospital care for a year and then that recedes. So the question was could we actually predict these very high intensity users and could we put in preventative services to avoid that? Well actually there was a good preventative service to avoid that, a sort of case management service that had been shown to reduce hospital use by 20%. What we then looked at is what's currently happening and we found that GPs have a way of trying to identify these people and they weren't actually all that good at identifying these people that would over the next year need intense hospital services. They got it right about one in three times, which meant they put this case management service into the wrong people and while it benefited them it wasn't necessarily benefiting the people that needed it the most. And so what we did is we got a collaboration across a number of different universities across the Kings Fund research think-tack in England and they got data about people's health history or the history of using hospital services and said, oh actually could we predict those people that will have high intensity use? Now actually we found that we could instead of getting that decision right a third of the time just by looking at the data by alone you could get that right two thirds of the time and that is a potentially massive saving, a saving of £100 million across England and that was essentially rolled out to the NHS, that there was a predictive analytics model that people around the country put their data into and they changed the kind of nature of the people that were offered that service. And in fact what they found is not only if they didn't just look at the data alone but they brought in a combination of data and professional advice they would not get it right a third of the time, not get it right two thirds of the time but get right about three quarters of the time. And so what this meant is people living longer, people having a higher quality of life and a significant saving to the NHS and that was all by using data in a different way. Well actually there's so much more potential there with just looking at the health service. Can we put in other preventative things upstream and I know that there's already quite a lot of history in doing something similar here in Scotland. But that's just one service. We're looking at a range of other things, actually still within the health service, could you predict for people that arrive in hospital, how long they're going to be in hospital, whether they're going to need social care services when they leave and if so what social care services in order so that we can put those services in place sooner and avoid people hanging around in hospital unnecessarily. Could we do that with people leaving prison? Could we identify what are the mix of interventions that work for a particular person by bringing together data about their offending history, their housing options, their employment history and education history. So these are some of the projects that we're looking at at the moment to say well actually we know that they can work in the health in a particular aspect. Can we use predictive analytics elsewhere to have an impact upon the quality and efficiency of services to transform our public services. In fact we're using quite a lot of data linkage now across different services to understand what works because the data often across public services is quite siloed so you go to school and you get a record about you at school. You go to hospital or a GP and you get a record there but actually there's two bits of information about you that if you brought them together they can tell you something quite interesting. So we're looking at a whole range of different things there so for example looking at the health services used by people that are homeless and asking the question actually are people that are in that situation are they getting appropriate health care services and if they're not why not and what services do they need. We're bringing together energy efficiency program data so which houses or properties have got which energy efficiency measures on them and combining that with smart meter data and home assessment data to understand actually which energy efficiency measures work for what kind of properties in which parts of the country. So data linkage can really help us to understand what works across the public services and then the last thing that would say about areas where we can make a significant difference is by using data to empower communities by making sure that our data is available to people that need it in order for them to set goals within their areas to secure resources and to measure success and when I was again when I was at the Department of Health a person experienced the transformation of a community in Cornwall and local people were empowered to say look actually love living here but where we're living is not good enough and what they what we're able to do is ensure that people had the money sorry had had the the data on crime on unemployment on the health of the population so that they could seek money and they secured quite a lot of money to transform that community and what they found because we could we could use the data to measure success was the actually as a result crime did reduce unemployment reduced there's improved breastfeeding people reported that their community was a different place to live and we only know that because of the data that we had was available at a relatively small area level and it was yeah it was it was a part but it was a really quite an important part of transforming that community I said that at the beginning what I was going to say was in order to do all of those things the predictive analytics the data linkage the empower communities with data there needs to be a set of things in place some of those things happened because of the circumstances at the time that it was there was a particular economic pressure there was somebody with the skills in the right place at the right time but I think what's important for me if data innovation is going to happen widely if it's going to have the impact that I think all of us here are hoping and hoping that it will do I think there needs to be some things in place that ensure that ensure that it can happen regularly so I'd argue there's four things one is that there's accessible good quality data that's well managed there's trust in the way that data is handled that we've got great skills around data engineering and putting that data together and the analysis of the data and that we've got a range of the organizations particularly I'm thinking for me in the public sector but organizations have data driven business processes so they're changing the way that they work and that forms the basis of our vision for Scotland's data vision it's about moving from a situation where organizations are overly cautious about data protection where data is held by organizations are not shared for that public good to the situation where citizens are aware and own their own data and where it's shared securely for the benefit and the impact that we were describing earlier on so the first of these is about access to good quality data well managed and we can start with the problems that we want to solve the opportunities that we want to solve or we can start with kind of what data we've got so that which might drive new innovative ways of thinking but either one of those we've got to know what data we've got and you know I think that metadata that discoverability of the data which is why the work the UK data service does is really important and we've got a similar piece of work in the Scottish Government making the data that we've got searchable the metadata searchable so that researchers across Scotland across the world can see actually you've got some fantastic data in Scotland that there's going to be secure process for for accessing that or there'll be a mention something about that later but that we've got we at least know what data is out there and I think the other part of this is about widening the sets of data that we've got available for research and I'm sure Bono may say something about that later on but knowing what data we've got is one thing accessing it's something else and we there's an important piece of legislation that's going through at the moment which is called the digital economy bill and we're pursuing this with vigour because what this should do is free up some significant UK public sector data sets for research in particular those data sets from the HMRC and DWP around people's tax and the benefits that folk are getting and companies tax situation and clearly that's really sensitive data and there's going to be very strong processes for accessing that but without access to that we can tell whether somebody comes out of prison and they then they go back into prison but what we can't tell is whether they go into work they can't tell whether they're receiving benefits and therefore we can't put in the most appropriate services for for that group of people the same for education people leaving university knowing the the the payback on the investment that we make in schools in higher education further education apprenticeships and so on in terms of the the kind of work that people get and the income that they're getting this is why this digital economy bill I think is going to be really important for us and again having the legal power I think is not quite enough people have got to have convenient ways of accessing and using data and there are different aspects to that so we've launched recently our statistics.gov.scot open data platform to publish all official statistics or access to the the statistics that underpin all official statistics about Scotland and Bill may say a little bit about this but it's a significant contribution to a Scotland wide open data strategy making that data available to people. I think stepping down from that because that's about data that's nondisclosive. What we're also looking at is exploring products that allow users to create statistically disclosure controlled tables from underlying data sets. We're looking at the business case at the moment for investing in that but it's got the potential to significantly open the the lots of data sets that we've got. I appreciate the UK data service has got a tool that allows some of that what we're looking at is potentially because it's disclosure controlled tables allowing access to all the data sets that we've got but that's some way off and we're looking at that business case at the moment or we will be soon. Then lodging data sets with the UK data service. This is vital resource I think for researchers and we're an important contributor but also an important user of that and then the final thing is where it's really super disclosive data that's needed. We've invested alongside the NHS information services alongside the National Records of Scotland and with ESRC and MRC funding a safe haven for data at which is situated at the buy a quarter in Edinburgh and that's a place where there's expertise that can bring together make available individual data sets, make available linked data sets to researchers across Scotland. I can't underestimate the importance of that as a research platform. So that's about access to data and good quality data. There's trust in handling the data so having the legal powers to receive process analyse and publish that data isn't enough here we need to have trust in those processes. The Department of Health in England where I used to work had those powers around getting data from GPs but was forced to stop its care.data project following public concerns. Because of situations like this organisations are often cautious in handling and sharing data. There's mixed advice about sharing. Generally in data linkage projects that we've had where we are trying to link data from local authorities for example got 32 local authorities in Scotland. Typically what would happen is 10 local authorities would come back and say yeah go on link the data that would be great 10 would come back and say no our lawyers have said no we definitely can't link the data that's not right and then 10 would say this is all too difficult we haven't got lawyers in any way that can advise us on that. So I think there's something about the expertise that's needed in data law in order to make consistent decisions around how we can put data to use legally. We also we've got now systems in place for secure data sharing and handling consideration of ethics and assurance to people about the trustworthiness of those handling of the way that we handle data and all of that has got to be vital without that we can't do any of the things that we're talking about today. But I think there's something else and that's ensuring that people know what data's available about them and what's done with it. I don't think we're talking about people having complete control about the data but increasing the level of control I think is important. And all of this is really tricky stuff actually. We're considering about how we take this forward across Scottish government and with partners across Scotland but this is big and it's something we need to get right because it's slowing everything down at the moment. Almost the last thing I'll say is about skills, having the right skills. Clearly skills in this field they're important, they're precious and very much valued by employers. Now skills in putting together managing data, making it available for reuse but it's also skills in bringing new insights by linking data sets and using techniques like predictive analytics that I mentioned earlier on. And finally it's about skills in communicating what that data is telling us. And those are all traditional statistics, analytical, IT type skills, some people call them data science these days. But I think there's one more skill that's important and that's the skill of using data to make decision making and driving improvement. So I think what we're trying to do in Scottish government is tackle skills at those three levels. The data science expert, the analytical enthusiast and that intelligent customer, that decision maker who needs to know when could data make a difference to me and if so when. And I think we're doing lots of work in conjunction with Gillian and the data lab on building skills and being really supportive of us and organisations across Scotland and in getting those examples so that we can tell a story to people that can use data differently to change the business processes that we're working with. Because I think that's it for me demonstrating the value to those people of how we can use data differently using the examples that I've talked about today and that I'm sure others will talk about today. But also talking to people about the practicalities. What are the things that, if you're the leader of a hospital trust or the leader of a local authority or the leader of business, what are the things that you need to practically do and think about in order to make decisions in order to make that organisation data driven. So I think that this is the step that I think that we need to do in order for data to have impact. So yeah, we can't solve any of this alone. We need to be working together. We need to do those four things that I was saying. Ensure that we've got accessible, shareable, good quality data that we're trusted in handling that data, that we've got the skills and so of decision makers in using that data and that the businesses and organisations change in order to deliver that impact for people. If we do all of that, then very soon we'll be using data to improve the lives of people of Scotland. Thank you very much.