Speaker 1: Hello, I'll start this session now. Hopefully everyone can hear me okay. I'm Peter Bradley from Public Health England. Despite what it says on the slide, my job title is Director of Health Intelligence. What I'd like to share with you today is why data is becoming increasingly important in the work that we do in Public Health England. My background's actually medical, so it's quite interesting. Why is somebody like me here today? I want to give you examples of the reasons why we do the work. We do some knowledge about how we work with our audiences, some examples of analysis and just finish up with a couple of challenges that we have and particularly working in the public sector. So today's session, first of all, so we're We're going to cover the examples that I've just talked about and I'll just start with an explanation of what Public Health England is really about. So first of all, we have a number of duties. One of them is to keep the population safe, so if there's a chemical incident or something goes wrong in the environment or there's an infectious disease outbreak, we'll be there. We're also there to try to improve health in the population generally, and we do that so people obviously live happier lives, but also so they're less dependent on services like the NHS. We are particularly concerned about the health inequality gap, so here we're talking about the number of years that people have living a healthy life, and depending on where you grow up, that can vary by 19 years. So some people get ill in their 50s, other people in their 70s, and that has a profound impact on the final section, which is the economy. So a healthier society is obviously very helpful to our economy and to the general standard of living. And that's what Public Health England sees its role as being. Data and evidence is crucial to what we do, but it's not the only thing. At the end of the day, we need people to use that data and evidence, and we guide our work by the use of this knowledge-to-action cycle, which allows us to use data to define our priorities. What's the problems? Which are the biggest ones? Evidence to decide, out of the many approaches we have, which are going to be the most successful. We implement something, and then we evaluate it, and so the data comes back in again. And that allows us to make sure that we're constantly focusing on health outcomes. And in the development of our technology, we always have to think about what is really going to do that job. So another thing that we recently are very clear about is when we are working with the the many audiences that we do to improve health, we need to be very conscious of what they need. And user design, user-led design is becoming very important for us. And in the production of our data outputs, our intelligence outputs, we're beginning to use this more and more and working up prototypes and testing them and using the traditional digital design methodologies. So, an example here would be that we are, we need to ensure that every output that we have is produced in a very timely way for our, and in an appropriate way for our audiences. And we've begun to understand the way that information flows across organizations. So if we took an example here of maybe a local authority. We have on the left-hand side, we've got people a bit like me. We're data lovers. We love the outputs that we have. But in order for us to actually have a bit of an impact on action, we need to work through the people in the middle of this diagram who we refer to as the translators, and they are the people who are able to bridge what we do to the decision makers, and these decision makers are people who are influenced by many factors, so it could be a chief executive in a local authority, or somebody who's doing a similar job in the NHS, and they are influenced by many things above all, probably the politics. So we need to provide our information for these translators in an appropriate way. In order to do this, we've started to develop profiles of the types of people we're trying to target. So here we have Fola, she's a Commissioner of Drugs and Alcohol Services, and the kind of thing she'd need to do is understand exactly what the trends were, the issues for alcohol and drugs in her area, and also know what the solutions to solving those problems might be. At the same time as this, obviously the data that we're able to access and use is changing all the time. So now we have a volume of data that we've never experienced before terabytes of data. The data is changing in its timing, going from data that where we had a big time lag, maybe an annual survey or a monthly return on hospital episode statistics, to data in real time, streamed data. We have a variety of data, so on our main portal, which is called Fingertips, which you might like to look at. We have lots of profiles. And on there, we have at least 100 sources of data to create those profiles. We need to look at the data quality and its accuracy, its veracity, and also then work out all of these possibilities. What is the actual value to improving a public health outcome? So we have to do all that for FOLA, too. And what we've learned from our experiences of working with people like Fola is that they do indeed need data that's provided together with research evidence. So it tells you the problem and the solutions. We need to produce it in a way that is going to be really easy for her to just pick up and take to the decision makers. And we need to do it at the right time. So it might be giving information at the right time in a commissioning cycle, for example. So at the end of the day, if we don't do that, whatever we do with the technology is never going to hit home and we won't get the outcomes that we are looking for. But as well as that, of course, we have got this big drive in technological change. So first of all, there's the data, which, as I say, is coming from many more sources than we have previously experienced. Lots of new, exciting types of data. The technologies that we have to analyze it, derived from data science, mostly, and also rapid growth in other areas, which I know you've heard from other speakers, but the NHS has launched its artificial intelligence lab, for example. We had an announcement from Secretary of State in a recent Green Paper about predictive prevention, which you may have heard of, but this is really data-driven, potentially individualised approaches to try to prevent health problems. We've seen the emergence of UK alliances between the public sector, commercial sector and the academic sector, such as Health Data Research UK. And all of this is providing a very different environment to the one that we've had previously. The data itself is now available in many forms. We have data on the problems that people have, the phenotype, what they're exposed to, what We have this in large geographies and smaller geographies, and we also have behavioral data from wearable devices. And currently, we have the majority of that data, but not all of it. So some of the things that we don't have would be the data from wearable devices. But I can imagine that very soon, we would aspire to join that data up and create very different profiles than the ones that we have today and it is about the linkage of that data that is the very exciting prospect for us. It's going to have major implications for our data systems and data platforms and I know some of the other speakers have been talking about that and it's exactly the same for us in the public sector. So I just want to move on now and talk, give you some examples of the kind of analysis that we have done and that we are hoping to do. I think it's fair to say that so far we've based a lot of our analysis on what's already happened. And what we're trying to do is move to a position where we are trying to predict more of future health problems so we can take action before it actually happens. But sometimes, even now, the retrospective analysis is very meaningful. So the first example I want to give you is one where we looked at a range of indicators, a very broad range of indicators, all concerning health, and we used supervised and unsupervised machine learning to look at a pattern of health in England to try and see if there were patterns that we hadn't really understood before. And we found that there was. And we've called it the London effect. And so there's an animation over on the right-hand side of the slide. And here it compares your expected lifespan with the level of deprivation which is on the bottom axis. And the animation looks and analyses this in five quintiles, so starting with the poorest areas and moving up to the final part of the animation, which are the more affluent areas. And what we found was that London has a different pattern. Nearly everywhere else in the country, poverty is the thing that drives or deprivation drives poorer health. But in London, that relationship isn't as clear. Actually, if you're in the poorest parts of London, your health is probably slightly worse, if anything, than the rest of the country. Well, when we come to look at the more affluent areas, even in the middle of society, we find that Londoners' health is far, far better. And this is really important to us because we have the start of an idea of how we might be able to improve health in other regions if we can investigate further. So I can talk to anyone about that after the talk if they're interested. So the next area that we'll want to look at is to just look at text analysis and how that's helping us target information to certain groups. So the example I'm going to give you is around vaccination. So I'm sure you know vaccination has been controversial and a lot of this was based on a false connection between the link of one particular vaccine, the measles, mumps, rubella vaccine and autism, and it was basically scientific fraud, but the concern around that has continued for decades now. So what we're able to do is look at the Twitter feed, in this case we looked at every Twitter feed that had vaccine, nearly every one, and we can begin to understand the sentiment behind these Twitter feeds, the types of people who are showing concern and it allows us to target messaging so that the appropriate scientific advice is given to groups to counteract this myth that has persisted in society around vaccination. So next example that we're trying to look at is moving more into the predictive elements where we're actually trying to predict the future so that we can use our resources more appropriately particularly and make sure that services are geared up to deal with the demand that will come to them. So this example is from NHS and it's to do with the number of attendances you would expect in an accident and emergency department, and we used R here to create a model that predicts the number of attendances. And what we found is that this model is very accurate. We've obviously tested it with the real data as the statistics have come through. And we now feel we'll be able to predict the number of accidents and emergency by the hour for several years in advance. And this is really important obviously for NHS planners and making sure that the demand meets the supply. So moving on a little bit further and just to think about direct public communication, so I've talked a lot about how we work with stakeholders, but this is more about how we we might work with the public directly. So for a while now in Public Health England, we've been using micro-targeting in our marketing activities. And through the creation of profiles, a bit like I was talking about Fola and making sure that the messages are going to the right people in the right way so that it fits in with their lives. And one of the examples from this is smoking. So the accuracy of targeting for smoking has increased enormously in about eight year span from 20% to about 92%. And this is a much, much more efficient use of our resources, too. Obviously, smoking is a very big problem for public health, generally. So the more good work we can do there, the better. But we're now moving towards a possibility of having a much more personalized approach where we can improve health by using the data that's coming through, big data potentially, and making sure that we are using all those possibilities that we have in terms of using data from smartphones and citizen-generated data in all its forms. So what we need to do to bring this forward is obviously to try to create an environment where we are really focusing on empowering the public to use their own data so that they they can take full responsibility for their own health and allow them to share that data with their health professionals, to help the frontline healthcare workers access that data, because I'm sure many of you have relatives and sometimes when you use the health service it's actually quite difficult to get the data there in real time. And obviously, that data, when it's accumulated, can inform decision makers about how they develop their services for the future. So the sorts of things we are thinking about are maybe making the health check, which is a check that's given to all people in England between 40 and 74. And the purpose of the check is to try to prevent problems like heart disease or stroke or kidney disease, that sort of thing. But offering in the future, this test as a digital option. So we've got an example here of Paul, I don't know whether you can read the text, but Paul might be offered his check, first of all, digitally. He would be able to input his own data, so it might be things about his weight, or maybe he's had a blood pressure check at the local pharmacy, and then maybe the digital service would identify him as low risk, but my advice that he did a bit more exercise, he began to swim or something like that. And then he would be able to receive prompts and nudges to encourage him to be very active. He would get feedback on his heart rate and things like that during the activity. And he would see over time that his fitness had improved, hopefully, and it would encourage him to do more. So he would be able to take a lot more responsibility. And of course, if any problems occurred, they would be notified and they would be shared with this GP. So that's the kind of thing we wonder we might be able to move to in the future. So there are inevitably some challenges with this approach, and if this is going to become a reality, we are going to have to think very closely about what is it that the public really want to share in terms of their data and what do they want to use it for. We work already very closely with the National Data Guardian and other groups that are concerned about information governance and this is a really big area for us. We can't currently link a lot of our data because it's not deemed appropriate. So the need for that debate is really crucial. some of the data is quite difficult. So I think that's another area that we really need to think about. Artificial intelligence and related activities can introduce bias. So the way that we train our data is very important. We don't want to be prejudicing people who we are most trying to help. If I go back to that point about health inequalities. We sometimes need to also just say that actually the innovative methodologies don't help. It's about choosing when to use them so that they actually produce those best public health outcomes. Sometimes we can actually rely on traditional analysis. Keeping up with things, that's really tricky. I don't think government is known for being super speedy, if I may say that, and I think for all of us, probably, that is quite a challenge. So obviously, the other challenges I'd mention are the transparency. So when we're using algorithms, actually understanding what lies behind them is very difficult to explain to others. So we're having the skills for people to be able to address that is particularly important to us. So I think I'll just move on to, and there are other challenges other speakers have mentioned, such as the need for us to be really clear about our information governance and the legislation and issues of that nature. So I suppose my final points would be the The health data environment is actually quite complex and it does involve us using many data types. I think as we move forward, collaboration with the academic and the commercial sector is going to be crucial. I can't imagine we're going to be able to do this on our own. We do need to find ways that we are really keeping up with the latest approaches, but at the same time, even though that technology is really exciting, I have to keep a focus on the fact that whatever we do is really going to help the population's health and particularly those people whose health is most at risk. need to balance the benefits for individuals as against the needs for us to benefit the population. So I'll just finish with those thoughts and in the few minutes left I'm very happy to take any questions.