 Sean Astley, thank you so much for being here today, she's a very, quite big, has strong knowledge about food composition and so on because of, she right now works at Eurofield, you know, it's the biggest consortium we have about food composition and so on, so please open your mind to her presentation here and take notes for the different questions you can have because we are very lucky to have her today to have this piece of information. Thank you very much. I too am going to go over here. My ears need to be bigger. What were the microphone on one and the glasses on the other? My ears need to be bigger. Right, let's see if I can manage to press the right buttons. Okay, they said come and talk about current and future developments in food information resources. I said you're giving me 15 minutes, I can't do that. And I can't, so I'm not going to. What I'm going to do is talk about some of the problems, talk about some of the solutions and we'll see how far we get in that 15 minutes. So when we talk about food, we're not actually talking about food, we're actually talking about food systems. So we're talking about everything that makes up the food that we ultimately put in our mouths. And it is a complex, adaptive, multi-actor, multi-level, multinational, multifunctional networks that are joined up in informal and informal ways and there are trade-offs and synergies and that's just the reality that we live with. And of course that doesn't sit in isolation out there in society, it's impacted by politics, by technology, by the environment, by the economies and social factors including the choices we make in our lifestyle and also in terms of our cultural norms. So this is where we are. It's a little bit complicated. So you've got primary production at this end, you might have health and disease at this, you've got the issues around researchers which are publishing an assessment, you've got things about manufacture, retail, preparation, losses and waste. And then you've got, and this is not all the projects, these are the ones I just had logos available to put into this slide. But you've got all of the projects that have ever been done. I go back to Framework 4, I think that makes me a dinosaur. But we've got networks, we've got clouds and we've got research infrastructures, we've got global agencies as well as regional activities and regional agencies. We don't speak all the same language even when we're all speaking English. We don't all mean the same thing when we talk about the things we're talking about. So the European Union decided it would try to tackle some of those issues through the Green Deal and indeed food is a major issue for food, that's that one, they're the affordable foods but there's also this thing down, and that cascades down to farm to fork and ultimately to particular actions, one example being the nutritional labeling. However, there's also this here, this future proofing of jobs and skills. When you look at the European population, two in five adults don't have any digital skills, one in three lack basic digital, sorry, two in five do not have sufficient, one in three lack even basic digital skills. There is still a low representation of women in the tech sector, they make up only one in six ICT specialists and about one in three STEM. But actually if you look at natural science researchers or indeed medics or indeed any of the people that are contributing that very complex, they also lack digital skills, albeit at a higher level. They lack confidence in using digital tools and services and they lack trust in the outcomes or their interpretation. Similarly though, when you talk to ICT specialists, they don't feel fully equipped to understand what's going on with the natural sciences and they lack confidence therefore in the developing of the tools and indeed trust in their own tools that they're producing the information that they're expecting. So if we're going to start to address some of the issues that we've identified, we've got to start to be able to talk the same language and use the same information. And again the European Commission decided we sort of had a little exploration of it during Horizon 2020. Horizon Europe data has to be fair. But when we talk about fair data, we're only actually talking about this column. We talk about it being findable, accessible, interoperable and reusable. But there's a whole bunch of foundations that you've got to have in place for any data to be fair. Things like, see now I can't do it either. Get the right column. There's a whole bunch of things. So in findable, something's got to have a permanent identifier and you've got to have it in some sort of repository and you've got to be able to find your way around that repository. And there are other issues that underpin the accessible, interoperable and reusable. But when we talk about fair, we've also got a three-way pull. We've got the pull of confidentiality, the GDPR and the privacy. You've got the integrity, integrity of the data in the sense that it is whole and completed informative, but you've also got human integrity. Are we doing the right thing with the data? And then we've got availability which gets down to the open science and the fair data. So the human integrity, to get that out of the way, it's about whether there is a need for that data that you're collecting. If I am purchasing a television, you do not need to know my date of birth. It has zero relevance. You might want to know my year of birth on the basis it allows you to put me in some sort of category, but you don't need to know that it's March and you don't need to know that it's the 20th. It's irrelevant. So you don't need to collect that data. There's a whole bunch of things around capturing data about food and about humans. Currently, if you're talking about personalized nutrition, they're not regulated to any great extent anywhere around the globe. There are issues that touch on the declaration in Helsinki. There are issues around the Hippocratic Oath. There are issues around minimizing risk and burdens to the individual and maximizing their benefit. They're kind of covered by things around genetic testing, GDPR, food laws such as the nutrition and health claims for products that companies are producing. And it's widely recognized. There's a lack of understanding about both between providers and consumers about false and obstantiated risks, consent, risk, and ownership. And there's an interesting dichotomy. If you're delivering personalized nutrition, you can test for ApoE. If you're delivering medical services, you can't test for it because it has to come under genetic testing because whilst ApoE impacts your risk for cardiovascular disease, it also impacts your risk for dementia. And we can't do anything about the dementia and therefore it comes under genetic testing. The only reason for personalized nutrition, the only need for personalized nutrition actually falls under consent, that you want to know that information. That may change as it becomes more mainstream and moves down the cascade of actors. So data just are. People, we talk a lot about data being fair. It isn't fair. It isn't unfair. It isn't anything. It's just data. And data is anything that can be stored, read, or moved about by machines. You do not engage with data. You are a human. You might pick up a USB stick and move it, but that doesn't count as moving data. It's machine moving. And it might be audio. It might be visual. It might be genetic. So it can be text, figures, images, or symbols. And the commonists we're used to is binary, ones and zeros. Data is, you can let it out in the whole world and it's fine because it's completely meaningless without metadata. Actually it's the metadata, the data about other data that underpins interoperability and also fair. And it might be descriptive, structural, administrative, reference, statistical, or legal. And the point is it allows you to interpret the data. So when you're talking about systems that allow people to share fair data, you've got to know what data you've got. You've got to protect it in place. How many people have left a USB in the toilet? Usually with defense information on. I don't know how you do that. No idea. But they do. You need to implement physical and logical controls. In other words, passwords and usernames. You need to keep it up to date. You need to define who and what can access it. Oh, and you've got to pay for it. So if the European Union, the European Commission is saying to us as researchers, you must share your data. You must put it out in the world. Well, that's fine. That's lovely. But who's paying for that? Because these computer things are not free. Nobody's given them away. Similarly, you've got to control who accesses them. You've got to have a log on. Check out what cloud services are doing. They're not in the cloud. Anybody wants to know? They're not in clouds. They're in massive great buildings with dirty great big servers. So what are they doing for making sure that your data is secure on those? You've got to understand your responsibilities. You've got to back up your data, the three, two, one, one. So that's three copies, two devices, one off site, one offline. If that's what you're really doing with your data and you mean it. Executables from trusted sources. And I love this, sanitizing your solid media. What it means is actually take a hammer to it because you can't actually scrub it. The data is still there. Oh, and have a think about your printers and all the other things that you put data into because they're also holding your data. And of course, there are issues associated with this because those dirty great big units of servers impact on climate. Then there's confidentiality, GDPR in European Union, but actually about three quarters of countries around the world have some form of protection for data. And those are the reasons for it. And I'm not going to get into them. Personal data, anything that allows you to be identified. It begins at your point of birth. It ends at your point of death. If you read the legislation, it says legal capacity. That is incorrect. It is an argument I would love to have with a GDPR lawyer because a person who has lost legal capacity, they might have dementia, for example, still is covered by the GDPR. That GDPR does not cease until they die. And after they die, privacy and confidentiality still apply. And the joy is that if you are a European citizen or someone sitting in Europe who is not a European citizen, I am no longer a European citizen, don't get me started on that. My rights are protected all around the globe. I mean, I have to say for a change, the European Commission played a blinder when they did GDPR legislation, which is why it's European law, but the Americans have to follow it. I'm going to leave my biases there. So it's anything that allows you to be identified as an individual. And if you're talking about sensitive personalised data, so that's personalised nutrition or ancestry information, there are additional protections that you are entitled to. So that means we have to have our consent in place. We've got to deal with the encryption. We've got to think about data protection, right of access, right to be forgotten and the right to be told that they're holding that data about you. It doesn't get in our way. We can collect data. You just have to deal with it properly. And then we get into anonymisation and pseudo-anonymisation. Anonymisation is exactly what it sounds like. You make it impossible for an individual to be identified. There are degrees to which you can achieve that. You can use redacting, so removing information, re-identifying blurring, which is giving ranges rather than specific information, and a whole bunch of others that allow you to anonymise it. Pseudo-anonymisation means that there is a code, a key, that allows you to understand who those individuals are. And when you are thinking about is my data anonymous or is it pseudo-anonymous, you have to work on what's called a motivated intruder test. In other words, if there is any way, however unlikely, that that individual can obtain that key that will allow them to identify the individuals in the pseudo-anonymised data, your data is not safe. Most sensitive data is pseudo-anonymised. There are exceptions for research, but you can't cause substantial damage or distress as a result of it being released, and of course that impacts interoperability. If you're taking data sets from different sources around the world and trying to bring them together, but those are anonymised or pseudo-anonymised, it's really, really hard to work out how you can do that successfully and make sense of that data even before you start to do the research. Open science, so I said that the commission has gone down the route we are having open science, so that's focusing on getting the shareable knowledge out as soon as possible, and there's a whole element to that. It's not just about open access publications now, it's about fair, it's about open science, cloud, it's about new generational metrics. That's rewarding researchers who are participating in open science in the same way that they currently award researchers for publishing. At the moment, the research assessment exercise virtually anywhere where you are stops at when you publish, it doesn't think about the fact that you're participating in open science. No opting out of that, and data must be managed under fair principles, but there's always the but, open as possible, closed as necessary. And what does that mean? Well that's a very good question, nobody's come up with a definition yet. So in terms of peer review publication, that's about open access publication in terms of data and metadata, that's about getting it out there under license, and we'll talk about that a bit more in a little while. The only reasons for not engaging in open science are risk to exploitation, so if you've got a patent, national security, or it is not compatible with protecting personal data, but those are not excuses for not doing this to the extent that you can, it just means you have to work a bit harder at protecting the data. So creative commons license is one way of doing this. I think we've all engaged at some point in a shared data agreement, it keeps the lawyers paid, but unless your organisation is prepared to go to court and protect that, and trust me, most academic organisations are not interested in going to court to protect data sharing agreements, it offers you no comeback. So that we have a very false sense of security around data sharing agreements. I'm not sure that licences are actually any better in the sense of protecting, but they are a bit more public, they cost a lot less, and they're a lot easier to achieve and you don't need a lawyer. You can have the discussion about the value of that. Creative commons have a tool, and it would be nice then that you could actually feed that into the assessment exercise that would allow it to be noted. This is one particular set of data we have, which is an average EU fish and seafood nutrients. I created it, my colleagues created it in Seafood Tomorrow, and then Fish EU Trust, which is another project wanted it, and I'm like, yeah, I want you to have it, but I need to stick it under a licence, because we put quite a lot of effort into that. So I went on to Creative Commons and I selected the international licence that it has, which essentially, you can't see it, but it says that we have to be given credit. That seems reasonable. I want it recognised that you referred to this work. You can distribute it, don't mind, you can mess about with it, you can adapt it, you can build on it, that's all fine, but you have to give us credit, and you have to say where it originally came from before you started doing that, and you can't use it for commercial reasons. That seems reasonable. Literally took me two minutes on Creative Commons, no lawyers. By the way, I have friends who are lawyers. It's okay. Yeah, that's fine. So FNS Cloud is one example of the kinds of projects that are dealing with these issues. So the aim with FNS Cloud was to try and overcome some of this fragmentation, specifically looking at food nutrition security, and to improve the fair data that was available around there. These are the beneficiaries, there's 35, oh my God, the paperwork, you should see it. You're not terribly interested in that. What I am interested in is IT specialists, FNS researchers, social scientists, lawyers, science communicators, and science multipliers. Problem, we don't speak the same language when we speak English. I said that at the beginning. Here's one specific and tangible example. Sitting at lunch, I discovered that when IT specialists talk about implementing something, they mean going away and coding it. I'm an FNS researcher. I mean going and seeing if it works. No wonder we'd been arguing over something over two months. We weren't talking about the same thing. So out of curiosity, having discovered that, I asked them what benchmarks meant. Six people at the table, seven definitions. Make sure you understand what you're saying to each other when you're working across disciplines. So we had a number of existing data that we had developed, a search engine. We had a look at how you can use food labeling data concentrating on branded foods, which is surprisingly rare out there in the wild. How we can make access to total diet study data, both for consumers but also a wider group of professionals, not just risk assessors. Food intake, consumer behavior and lifestyle, how you map, merge and look at the data quality and the usability of that. And also specifically looking at could we build some algorithms based on the massive data sets that are out there, focusing particularly on type 2 diabetes but also on blood pressure. I'm betting riboflavin wasn't in there. Need to go back and talk to them. And then we looked at emerging data, looking at issues around those for whom English is not a native language but are living in a specific member state, looking at the elderly, meal planning, using public recipe databases, looking at healthy diets for healthy microbiome and also food drug interactions. Some people know that you can't eat grapefruit when you are a calcium channel inhibitor but there are millions of other food drug interactions that on a daily basis we don't take much notice of. And we've now got demonstrators that bring these together under the headers of agri-food, nutritional lifestyle, NCDs and microbiome. And we've also got a community of practice because as Mary said, not only have we got to educate ourselves, we have to educate the other actors in the system so that we all know what we're talking about when we talk about implementation or benchmarking. We're not alone in this and I said that we started here while all we're doing at the moment is sort of knocking off bits of these to try and reduce the number of resources that people are going to look at. So Blue Cloud is the big sister of FNS Cloud. That's been around for over a decade and it's focused on marine research, which includes fish. And we have played the tour, but yeah, I literally, this is my last one. We have played the game of exchanging data between FNS Cloud and Blue Cloud and it does work. Metrophood is an emerging S3 research infrastructure and it's about how you measure stuff in food. So we've got our catalogue, you've got the S3 catalogue which tells you all about this and other research infrastructures. You've got Zanodo which is just one trusted repository, trusted to be defined. And you've got eventually the mothership, which is still being built, which is slightly disconcerting. So you've got EOS, hopefully at some point in the future you'll type something in there. You won't find it in here, but what will happen is you are signposted to FNS Cloud or Blue Cloud or Zanodo to find the information that you're looking for. So thank you so much for being on time. And as I promised you, you open your mind and now your mind is full of data. We need to extract the data from here to try to unravel something, to extract something from there. Any question here or comment or do you prefer the questions during the coffee break? Yes, please. Questions from the internet? No, they're still all going. Let me check. No, we have a hand. Can we have a microphone, please? One brave attendee. Yeah, get the microphone, male microphone. I hear you talking about data and I hear it everywhere. I always wonder if I got all this amount of data, what could I possibly do with it? So I mean, it's the same thing for any enterprise or whatever who wants to collect data, but what are they doing with it? Yeah, and I think the intention with the demonstrators in things like FNS Cloud or Blue Cloud is to show you what you might do with it, what is out there. So for example, with the AgriFood demonstrator in FNS Cloud, we look at the food chains that are coming from agriculture into the food chain system. We look at the metrology, how do you measure the stuff in that food, and then we look at composition consumption so that you get an idea of how you might, well, okay, we'll take that data set, that data set, and that data set and that we can in some way get those to be interoperable in order to ask a particular research question. So a research question might be, for example, how does animal husbandry impact the nutrient content of milk? And what you would need are the food chains for milk, the composition for milk, and you could bring those together to get some idea of how animal husbandry impacts the nutritional status of milk. It's not so much about the statistics, it's about pulling down those data that are fair, so they're findable, available, they're interoperable, and they allow you to ask your research question, because if you're talking about any of those food systems, you're going to know the food data that you need. It's just that hopefully we can sign post you to where it is and how you might bring it together. I would like to say, well, we can go for a coffee break if you would like to answer these questions now, or like we can leave them for, you can... Okay, hello Anonymous. Why speak of so open science if tools for Europe are is to pay. Right, okay, something you need to understand about Europe. Yes, we do ask that members pay, however, you can go and get the national food composition tables for free anywhere online. The difficulty is that for historical reasons, the food composition data sets around Europe and indeed globally have all been developed independently for different reasons, and they are not interoperable. That is fact. So if you want to search for riboflavin in Switzerland, Thailand, New Zealand, and the US, you can, but first you'll have to process that data so that those various data sets are interoperable. Now, you can do that and it's free. Okay, what your effort does is we work with the compilers and we take that data and we do that work to make it interoperable. I would love to be able to make our data sets free, but somebody has to pay for my data science, a scientist now because that's the fact, that's the world we live in, that someone has to pay someone to do that work to make it easier to interrogate those food composition data sets. If I can find a way to do that for free, then I will make those data sets free, because they are free, because they're funded nationally, but that's the reality. So when I talk about open science, I am conscious of that duality that we pay for, but over the last decade, we've reduced our costs for membership and you can now join as an individual for 50 euros a year. Yeah, I'm aware it's not fair, yeah, I'm aware that we're still asking for 50 euros and not everybody has 50 euros, but somebody can find a way of paying my data scientists to make them all interoperable, then I promise you I'll open it up completely. FNS Clouds, yes, they will be. We're nearly there. September 2023 is the launch party, 12th and 13th of September in this building. Thank you for the pre-advert. You can come, you can play, we won't just talk about them, we'll actually let you play with them. We did a dry run in March in Athens and I had lots and lots of demonstrator leads who are like, I don't believe you're making me do this, please don't make me do this, I don't want to do this, all that work, that was nice. So I'm hoping they're going to come on Brussels in September and go, actually, this is okay, I quite like doing this. Letting my tool out there in the wild, it's fun. It's nice. So please, a big applause for the champion. Thanks very much.