 in Institute of Innovation and Technology Flagship Research Project. So welcome to Chris. Thanks for joining us. And we also have Yanis or John Stotis, who's the CTO and partner at Agrinale. I'm not so good at Greek, so apologies for getting the John and Yanis mixed up there. Yanis has been transforming the way food companies use risk assessment and risk prediction. And I say he's a partner in the Agrinale company. Dr. Stotis is an expert in using digital tools for risk assessment of foods. He's worked with several Forbes listed food companies for assessing their ingredients and suppliers and helping them to automate those risks, including the use of artificial intelligence to generate predictive analytics that help companies prevent food safety incidents. Yanis also presents a series of videos explaining how to predict food safety incidents for busy people. And as you can see from the pictures there on screen, I'm Neil Marshall, the managing director of Gov Consulting. Probably most people know me for my 20 plus years working at Coca-Cola. So with that, let's just go step forward and hopefully the screen will respond for me. There we go. So our goal today is try to look at risk assessment and risk prediction using technology and also the expertise of Chris to show how we can help prevent those issues occurring and identify those issues before they happen. Obviously that's the golden triangle that everybody's looking for. But using the software and the system we're gonna show today and partnering with the intelligence and insights from Chris and similar people, we think we have a solution to take as then. So let's just think about the background for a minute. So very long research and I'm sure many people have seen this already. There is generally a low level of trust from consumers in the food safety information that's out there. As Chris can no doubt talk following my brief introduction. You know, there's so much information out there on the internet but it's probably the least trusted by consumers. The food companies and business operators also have lots of information, probably slightly more trusted but always there's a bit of suspicion around that. And then probably consumers trust more trust of the governments and the regulators but again, some indecision still there around that information. So as we can see on the right hand side, you know, the population are worried because there's been so many recalls, risks, food incidents, harm in foodborne illness that over a billion people or 17% of the population say they experienced or seen or been aware of an incident that's happened from eating food in the last two years. So obviously we want to try and prevent that and we want to try to find intelligence from the data and science to help us predict those incidents before they happen. That's what we're trying to go towards. But obviously that's difficult and we need to use technology and intelligence to take us there. So why is risk prediction and prevention more important than ever before? We need to look at what the industry said. We need to try to prevent major incidents. We need to be more proactive, less reactive. And to use one of my phrases, we need to try to see around corners. So see around corners so we know what's coming here. That's difficult, but that's all about prevention. But we're spending a lot of money in that area already. Companies are spending huge amounts of dollars, euros, pounds, wherever you are in the world on trying to prevent incidents or investigate incidents after they've happened. I'm sure Chris will mention it, but one of the points that Chris is trying to raise is, you know, this can be used better on prediction than prevention. And there's so much data out there now and hopefully Yanis can show us that with the tool that we're not using that data efficiently and effectively to help us predict those issues. And ultimately the point, you know, this will make Chris laugh now. The point that we're trying to get to with these tools and intelligence and science is to put people like Chris out of a job so that we're not needed anymore. Now, of course, we always want to have a thousand to give us insights, but the idea is to take us away from prevention. Sorry, take us away from reactive responses to be more predictive. So with that, I'll now pass to Dr. Elliott who will take us through the scientific approach to food safety intelligence. Thank you very much, Neil. Thanks for the introduction. Thanks also for threatening my future career. I'm really appreciative of that. And I really am grateful for the invitation to come and speak at the webinar. Do you know that very, very soon the world's population is going to reach 8 billion people? It's a new landmark, we're getting closer. And the statistic that you showed says that 1 billion people on this planet suffer from foodborne illness in a short space of time. That's only going to get worse. It's just going to get worse. And I'll tell you why. I spend a lot of my time studying the world's food supply system. And that's why I've got no hair, okay? It is unbelievably complicated. Just how we move food from one part of the planet to another. So we've really, we've lost touch of how our food is produced, who's producing it and techniques and so forth. So we have that increasing complexity and with complexity comes lots of issues in terms of people making mistakes. A lot of food safety incidents are about mistakes. Human error. But what sits behind that? Some of the big drivers of food safety that I spend a lot of time looking at. If I was to produce a list of the things that most worry me about food safety going forward, it's our changing climate. Others absolutely no doubt about that. Because with our changing climate what will come lots of risks that we're never seen before in particular geographic regions or with particular types of crops, particular types of food commodities. There's no doubt about that. And again, I work really closely with lots of food companies, locally, nationally and internationally. And I know how much effort is put in to trying to ensure that the food that we all eat close to a billion people is safe to eat. And I often congratulate companies in terms of the efforts that they put in to trying to ensure food is safe. But it is a changing landscape. It's a changing climate, that's for sure. And risks are changing also. And I do, joking apart, spend quite a lot of my time investigating when things have gone wrong. We completed a very large study just a number of months ago to try to understand why hundreds of people became poisoned with food, why some young children died from eating contaminated food. It was a very complex story. But the main reason was our changing climate because toxic agents got into food that nobody thought would ever be there. So things are changing, food safety risks are changing. And I think, again, for a lot of companies, for a lot of businesses, your plans, your risk management plans are actually based on historical plans because this is what we did last year, this is what we did the year before. So very much based on historical data, not on what's happening now and equally important not what's happening in the future. And because of that, I think as Neil referred to, I think a huge amount of money is actually being wasted. So if you've got a sampling and testing program and you're showing that 98% or 99% of everything is okay, my feedback to people is generally, my goodness, aren't you wasting a huge amount of money because you're not actually finding the real food safety risks there. So if you're sitting at this webinar and you look at your last set of data and it was 99% everything's okay, you should be starting to worry. And I think we do live in the era of big data. I mean, I just hear big data day after day after day, big data projects. And there is expression is information is king and knowledge is par. And we have a huge amount of information available to us. And what I try to do is gather together different pieces of information. And we try to think about what the big food safety risks might be or big food fraud risks might be. And we go away and investigate that. I'm firmly of the belief that the real risks that sit within companies or actually sometimes countries aren't really well known or well understood until it's too late. And recently actually in working with Agri, no, I've done quite a lot of thinking about it. And what I would love sitting on my desk is a crystal ball and I call it the digital crystal ball. And if anybody decides to use that, I'm gonna trademark it okay. And if I say it in any article, I'm gonna sue you okay. All right, what is a digital crystal ball is all of this disparate information coming from lots of different sources, okay? Very complex information. But what the art is translating all of that different information, the data into information and then into intelligence. And that intelligence has to be something that you can read really quickly. It has to be a dashboard based system. And we have striven to do this for quite a number of years in terms of collecting all of this different information. And for several years, I have tried to find what I think are the right technology partners for doing this. Myself and my researchers were very good at sampling and testing and doing the analytical chemistry piece, but not the intelligence, gathering and data interpretation. And that's why I really teamed up with Agronome on quite a number of projects now. And what I'm hoping now, Giannis, you're going to tell us okay, how does a digital crystal ball work okay? Sound okay to you? Very happy to do so. Thank you so much, Chris, for the introduction. I think I'm going to steal your word though. You may have to sue me, Chris. I think I like the digital crystal ball. You'll hear from my lawyers. I'm sure, I'm sure. But I think just before we go to Giannis, I think we maybe have a poll, but I took some points there already, Chris, a particularly like your points around human error, the climate change. I think that's something for people to be really thinking about now in every angle. And historical data not being a future proof to 98%. There's many companies who are in that position, I think you think 98% is getting them where they need to be, but they're not. So any information is king, as you say, very, very relevant. So I think maybe we have a poll to do from the side before we just go to Giannis. Yes, there we go, thank you. So if you can, the people from online, if you want to answer and put in your request in the box, and it'll give us a feedback and then we'll transition through to Giannis. I think we've got about a minute if you can just answer now. Am I allowed during that minute need to make a complaint? Sure, go on, Chris. Yeah, I'm not allowed to fill in the poll and I would love. I think I can do it either. So you're not on your own. I think the three of us, please. It's asking the right questions, it really is. Yeah, I think this is some of the things that we've found as you look at the data. You know, there's various different ways to capture data. Different companies are doing it different ways. Many people are still stuck in the Excel spreadsheets approach. But this is just really trying to capture how people are currently capturing their data and analyzing the risks. Okay, I'm not sure if we're gonna show that directly or not, but I think we can go through to Giannis. Over there we go. We've got the results already, so let's have a look. So 40% is still in Excel as we just predicted, I think. 30% between not identifying their risks and not being accurate. 56%, I think I had more being there as well, afraid that unexpected risks will appear in the supply chain that they're not aware of. So hopefully Giannis and maybe Chris afterwards can give us some helpful insights there as well. And what else have we got in there? Evidence-based lack of... So I think that's a good segue through to you, Giannis, to give us your insights now on risk assessment and prediction. Great, that's a very interesting insight from the poll. So just a few words about agronom and what we are doing. So we are the food safety intelligence company that extracts tailored data insights for the global supply chain. What we do is that we collect, translate and unreach global food safety data. And using the technology that really works, the big data technology that can solve real challenges of the industry combined with the data chain. Combined with the data, we help the food safety experts working in the food industry to solve the challenges that they have in their everyday work. So this is the main goal of applying the technology, but also the data, all this data that we are collecting and that was mentioned by Chris and Neil. One of such challenges is how to deal with risk assessment. As already mentioned by Neil, risk assessment is still a manual task. It's time consuming and error prone. And when you are busy and you have many things and many worries, it's not straightforward to deal with the risk as a manual task. Our approach for the risk assessment is based on three steps. The first step is the risk identification. The second step is the risk characterization and ranking. And the first step that goes to the digital crystal ball that Chris mentioned is about risk prediction. So let's go a bit deeper in each of these steps. It was one of the results of the poll that the food safety experts are still worried that they will miss very important and emerging risks that may appear in the global food supply chain. Because food archive provides access to many different data types, to million of data records that is collecting from data sources, such as the websites of national authorities, such as the results of pesticides monitoring programs, valid news websites and media websites. We are delivering real-time incidents and alerting for the supply chain for your supply chain. And this helps to identify emerging hazards for your supply chain. This is something that we will go through during the interactive demo later on. Just some example of such an emerging risk from the recent times. So it was Fipronil case in next in July 2017. It was the ethylene oxide case in sesame seeds in September 2020. It was the Proflora's case in mandarin and oranges in November 2020. It is the ethylene oxide in peppercorns in February 2021. I'm sure that you are thinking, what are the two last ones? Be patient. We will analyze them during the interactive demo. Identifying the risk is the one step. But how can I use the identified risk in order to recalculate and rerun the risk assessment that I'm doing already with very good approaches and very good tools that I have developed. So the Fudakai risk assessment module helps you exactly to do something like that, to automate the risk assessment based on all these food safety incidents that we are collecting from the global supply chain. All these data that are announced by different organizations, by different sites, how you use all these incidents, you integrate them into the risk assessment tool and automatically how you ranking the risk in order to prioritize the preventive actions that you want to activate. So this is exactly what the Fudakai risk assessment is providing. And the food safety experts that we are working with have managed to reduce more than 50% of the time devoted to risk assessment task using the automated risk assessment approach that Fudakai is providing. Risk identification and risk assessment are part of the reactive, of a very good reactive strategy. But we hear from the food safety leaders and from all the experts that we are working with that is very important also to predict food safety risks and food fraud risks. Because in this way, they will be able to prevent recalls. So this is why to address this challenge we have built the Fudakai risk prediction services. It's the live food risk predictions that is based on new risks that are identified that can predict incidents and hazards that are likely to occur. And how we do that in order to build such predictions, we are using something that we call the food safety intelligence question. And this question starts from understanding very well the business problem, the business question. It's based on using the right data and selecting the right data, selecting the right prediction method, algorithm and then visualizing, presenting the prediction outcome in a way that is meaningful to the experts and that will help the food safety experts to make informed decisions. Such predictions can help us to answer critical business questions. Like for example, which are the specifically, which hazards will increase for a specific ingredients for the during the next few months. I hear many times the question, if such predictions could really work in order to prevent the food recall. I will mention an example here. I will go again to the case of ethylene oxide in sesame seeds, more than 900 recalls and border rejections, more than 150 different products and ingredient types affected, 32 countries affected. It was found that this chemical was 1000 times larger than the MRL that we have in the regulation. So could we really prevent a recall in a finished product that includes sesame seeds? Let's go back to November 1st and then I would like to go back even a few months before. So in November, there was a real recall for a food company that is selling instant noodles for the unauthorized substance ethylene oxide that was found in one of the ingredients, guess which. And this was a large recall, but what happened in September, 2020? What the predictions were telling to us in September, 2020? Based on the prediction in September, 2020, it was highlighted that there will be an important increase, a significant increase of incidence for sesame seeds, more than 140% increase. And it was also highlighted that the ethylene oxide, but also salmonella, but mainly the ethylene oxide will be highly one of the emerging and the very increasing risk. So this was explicitly highlighted in the dashboard of global predictions. It was also mentioned, which are automatically with no manual work, which are the products, the finished products, but also which suppliers are affected by these specific emerging risks. And using such an information, the company could include this parameter in the lab test plan to make sure that there is no such chemical hazards found in the ingredient and in the finished product. Ask suppliers who are using this ingredient to provide certificate of analysis and plan remote audits if needed for suppliers in order to check if they are affected by the emerging risk. So in this way, do you believe that such a recall could be prevented if all these measures were taken into, were activated on September, 2020? Let's see how such a digital crystal ball looks like in the interactive demo. So during the interactive demo, we will show you all the services that are provided by Fudakai, but mainly focused on the ones that are in Fudakai, that are focusing to food risk assessment. There are services, as I mentioned, for risk monitoring and risk assessment, services for real-time monitoring and alerting, for hazard analysis, for automated reporting, for supplier evaluation, for automated risk assessment and for risk prediction. Thank you, Yanis. And I think you could see on the last slide the multiple of different offerings from the Fudakai platform, you know, particularly we'll be interested to see in the demo, I think, about that predictive, but also the supplier information. Maybe you can touch on that in the demo as well. I think before that, yeah, we have a poll. Good timing, good reminder for me there. So this time, it's from the information you've just seen and just from the demonstration that Yanis just gave. Can you also click in the box and give us a response, please, for which of the modules you think would be most appropriate for you? Most valuable to help you in your jobs to be more informed. So from the overview that Yanis gave, can you give us a response, please? And then we'll just review the data. We'll just give you a minute and then as we prepare for Yanis to do the demo. There's been a couple of questions in the chat box which I've answered already. Hopefully I've got the right answer and Yanis won't hit me afterwards, particularly around the recording. The session will be recorded and we apologize if people had some access issues gaining to the webinar at the beginning. Not sure why, but hopefully there we go. So the results are, let's work. Global incident monitoring, 49%. Also automating the risk assessment is high on 46, but guess what? Look at the bottom one, risk prediction. Exactly what main Chris is talking about, risk prediction, predicting the future. That's what we wanna go and I think most of you on the call are in agreement. So that's good to know. Thank you for sharing your insights and with that we'll move to the demo Yanis to put you under pressure really now to show a live demonstration of the capabilities of this fantastic tool. So over to you. It's a pleasure, it's always a pleasure to demo the platform. So I'll try to ask you some questions as we go through Yanis to make it a bit interactive for us. Thank you so much Nel for making it more interactive. So I will start and please at any point you can interrupt me and ask clarifications and questions. So with Fudake platform, as I mentioned, you have, this is a web-based solution. So you have your account, you can enter the system and you have access to more than 400,000 food recalls and border rejections that are collected by more than 50 food safety authorities all around the world. In addition to that, you get access to 102 millions of laboratory testing results that are collected and processed from 34 pesticide monitoring programs. So it's very important to have all these diverse data, data sources and to collect, to have access and to identify risks that are announced by all these data sources. But one of the things that we hear a lot is that it's very important to make all this, big data, all this information to make it relevant for my supply chain. I don't know, I don't want to know any risk but I want to know the risks that are relevant to my supply chain. So for this reason, we have a very good customization part in the platform where you can add your ingredients, the ingredients that you want to monitor for emerging and increasing risks, the suppliers that you want to monitor, the finished product for which you want to have automated risk assessment, the suppliers for which you want to have automated risk assessments. So when you do that, you will get the risk monitoring part that will be tailored or made to your supply chain. So this means that you will get alerts that will be relevant to your supply chain that will be linked to your ingredients and your suppliers and also this dashboard that you see here will be focusing on the ingredients, will be showing the incidents for the ingredients that you are interested in, the news for ingredients that you are interested in and also alerts for suppliers that you have in your supply chain or suppliers in your industry that you want to follow. But the most important thing is how you can identify increasing and emerging risk. So here in the dashboard, you can see that there are two very important blocks. The one is the increasing issues blocks where you can see that for instance, we have a significant increase of incidents during the last period in sesame seeds compared to a previous year and also a very significant increase in mandarins in oranges. So if you click on each of this increasing issue, you will get a very good and deep hazard analysis of the last period of the last 24 months. So you can identify which was the origin country of this ingredient that caused the issue. It's India, which was the main issue. It was a chemical problem and you can drill down to specific chemical problem. Like for instance, it's mainly pesticides and it was the ethylene oxide. So you can get such the system, the platform identify such an increasing issues for you automatically. You don't need to do anything more than just visiting the dashboard and identifying such increasing issue. But the most important thing is that you have, you can identify the increasing and emerging risk. So in this block of the risk, you can see which are the ingredients that are at a high risk region, at the medium risk region, at a low risk region. The ingredients that we have added all in the customization. So here, we're very fast in very fast way. We can see that there is an emerging risk with proclosal in mandarins. And you can go and there is also an increasing risk of aflatoxin in almonds. And you can go directly and check what is risk is about if you click on the risk analysis part. So what we are doing here is that we are collecting, we are processing all this data and we are applying a risk assessment approach that is based on the severity and on the frequency of the hazard, specifically now here for the mandarins, but this can be done for thousands of ingredients. And very in a very fast way, this risk characterization and risk ranking is performed in the system. So it gives you very important information such as that there are the new ones, the new risks are the emerging risk are the ones that were reported for the first time during the last 10 years for mandarins. And there is an important emerging issue with the use of proclosal in mandarins from Turkey. But there are also other increasing issues like this chemical, this pesticide and other emerging issues like the chlorperifos, the use of chlorperifos in mandarins. And the same risk assessment, automated risk assessment can be performed for other products, for finished products, like for instance, the case of the noodles where we have some specific ingredients, we can add these ingredients for this product and we can estimate the system, we estimate the risk assessment automatically for us and we'll highlight which are the top risks for this product, like for instance, the highest risk are the salmonella in black pepper and sesame seeds and new risks, guess which is the ethylene oxide in sesame seeds. And the same analysis can be performed also for your suppliers. So you can add the suppliers with the ingredients that you are getting from them and you can have such a very good analysis. Sorry, Janice, can I just interrupt there? So on that piece there, if I gave a name of a supplier, not obviously on the webinar, but you could enter specific names of those suppliers and pull up individual data by supplier. Yes, this is a risk, that's correct. This is a risk that is estimated based on the ingredients that the supplier is providing to you. If you go to the supplier evaluation module, you can use any name, you can search for the supplier and see if the supplier was linked to an incident that was reported by national authorities. And the same can be done, you can save the name of all your suppliers and the system will continuously monitor the suppliers and will provide a risk assessment report for each supplier based on the food safety history of the suppliers. Something that is required by national authorities like the FDA that is required the foreign supplier verification actions to be put in place by the companies that are sourcing ingredients from other regions. So this is something that can help very much to come. Yeah, I see a huge value in that for companies to search on their existing suppliers and also search for future suppliers who you may be considering to use. And the other point, I think maybe I just, another question, I don't know if Chris is in question, but the other point is about the amount of data that you're capturing in the platform. That's a huge amount of data, Yanis. How are you keeping it updated? How is that continually updated? So our system is continuously scanning the very, all the trusted and validated data sources. And this scanning is updated every few minutes. So every few minutes we check for new incidents, we collect the new incidents, we process them. There is a team of human experts that is validating the information or that is validating that two announcements or even more announcements are linked to the same product, to the same ingredient. So they are performing such a validation task in order to deliver the most accurate and updated information about food safety incidents to the user. So that's impressive, because that's almost real-time basically, I've been updating it all the time. This is the idea and the risk is also real-time. So you get the incidents, but automatically you have the risk recalculated and the risk ranking performed. So there are many expectations as we see from the poll about the risk prediction part. So what we are doing here is that we are applying the methodology that I mentioned during the first part of my presentation in all the ingredients that are important for your supply chain. So for instance, if I'm using the first step is that the system has calculated, has used all the historical data to calculate the predictions for the incidents and hazards that may happen in these specific ingredients. So the first thing is that the system highlights to us which are the ingredients for which we will have increased number of incidents. So for instance, for black pepper, for peanuts, in general, for a category like cereals and bakery products, it is predicted that the incidents will go up, but there will be increase in the incidents. So specifically for black pepper, if you click, you can see which is the increase of the incidents, the predicted increase of the incidents, how this prediction will go throughout the next 12 months. But the very important thing is that the system highlights to you which are the hazards that are likely to increase, how much it is anticipated that they will increase. So for instance, here, we see that there is a focus in the problem of salmonella and also different serotypes for salmonella that were identified. And the system also highlights to us which is the emerging hazards. So for instance, this specific case of serotype of salmonella is for the first time reported for black pepper. So it may be something that you are not expecting, but you need to take into consideration. Also, we are transferring this predicted hazards to the risk, to a dissipated risk. So you can see the risk pattern here which are the hazards at high risk region, which are a low or medium risk. So we see here how the risk looks like now and how the risk will look like within the next 12 months. So there is an increase in the case of salmonella. So this is highlighted. And this is also highlighted in the evolution of the risk where the system anticipates that there will be approximately 10% increase in the risk for salmonella in black pepper. The system automatically having the knowledge of all the products and suppliers that are using black pepper, sorry, it will highlight to us which are the affected product and suppliers. And it will also highlight which are other ingredients and products that will be affected by salmonella and using all the knowledge and the history of the data that we have. And if we go for instance, it's also important to see predictions even in the case of ingredients that it is a dissipated to have less incidence in the future because the incidence may be less but we still may have some increasing hazards. So for instance, in the case of oranges that I'm sourcing from Turkey, I can see that the hazards that are likely to increase is the chlorperifos methyl but also progrosal, as I mentioned in the case of mandarins. You can see again here how the pattern of the incident will change. So for instance, of the risk will change. So you see here that in the actual, right now there's no hazard at a medium or high risk region. Whereas it is predicted that the hazard that has to do with chlorperifos methyl will increase and will be at a medium risk. So this is something that you can use as a very important information to take some measures and to design better your lab testing plan. Again, sorry for the case of chlorperifos, you can see how the risk will increase. And one very important part is that the predictions when you are sourcing for a specific region and the system knows about it. So here we have oranges that I'm sourcing from Turkey. So we can also have predictions for the incidents from Turkey. So we know how the risk profile of the specific country will increase or will be stable or will decrease. So it seems that it will decrease a bit, but still it will be at a higher level than the last 12 months for the case of oranges coming from Turkey. So we can see there's no limitation about the predictions. Here you can see for all the ingredients that you have, especially for those that we have data and we have the history and the system, the predictions know how can predict how the hazards or the risk will increase. Very happy to answer the question during the Q&A session or Neil, if you have any other questions that... Maybe there's a couple of questions in the chat we can just try and answer as well, Marianne. One is a simple one, maybe nice, not a simple question, but a question from Manfred. Does the food archive platform offer capability to issue automatic alerts to a user based on user-defined set of criteria? For example, a new high-risk hazard for ingredient from suppliers. This would be very helpful for larger food manufacturers who have a large portfolio of ingredients and suppliers. So he's looking for an automatic alert. Yes, the risk monitoring part provides automatic alerts for any ingredient and any supplier that you have in your supply chain that was linked to a specific incident that was mentioned in a specific incident. So it depends on the preferences that you will use in the customization, but you can get daily or even instant alerts every time that there is a problem with a supplier or you can focus on specific hazards, on specific contaminants and add alerts for the specific contaminants. So if you want only to see alerts for Fipronil, for instance, you can add one preference and get alerts only every time that there is a problem in bananas, Fipronil for peppers that Fipronil was found, sorry. Cool, so I think that answers that perfectly. Thank you. There's another question about sesame seeds from the previous year were also high. What was driving this? Could the tool provide, I can't speak, predictions that the subsequent pesticide issue would occur? I think you had some data to look at the sesame seed prediction already, I think. Yeah, the sesame seed started. It was high in 2020 because it started in September 2020, it was the first signal, but it increased very much until the end of the year and this was predicted because we identified the merge increase and then the prediction algorithm predicted also the increasing issues. So that's why in 2020 it was so high, it was accumulated for all the months, September, October, November, December. Yeah, the idea is that we could, how the predictions can also predict pesticides that will affect another ingredient. So as we mentioned, Proclosal was mentioned initially in mandarins, was reported for mandarins, but we also predicted that it will be affecting oranges because the two ingredients are sourced from the raw materials are sourced from the same region and they use as a practice in the cultivation, they use this kind of pesticide. So this knowledge is modelled in our system and this is provided in the predictions. Cool, thank you. I think one more comment from the chats is does the Food Aki app work on a mobile device or have you got an app yet? And I think that's maybe something you want to just mention, the future work. Yeah, you can use it now from tablets, it is available for, you can use it, the web-based version can be used by smaller screens, but it's not, we don't have a native app, it's something that we are considering because we hear that it would be very helpful at an individual level to be able to follow alerts and the main insights or some very important emerging risks to be available in a mobile app. So this is something that we will definitely take into account for the next developments. Thank you so much. So thank you Yanis for the demo and obviously thank you for Chris for the comments before. As you can see on screen now, this is your opportunity to use the QR code to scan, obviously to get further information from AgroNow about the different solutions that they can offer, particularly if you're interested from what you see and you believe you can use the platform to help you to be more informed about predictive solutions. I'd encourage you to contact the team there to get some more information. So before we just now going to back to summarise and maybe do a little Q&A with Chris and yourself, just want to summarise from what I think I've seen. So some very interesting insights as usual from Chris, particularly around the human error, the risk of climate change and the impact on the food supply and these points around historical data and the digital crystal bowl that I'm going to try and stay off him as a new patron, but that's a good idea. Insightful as always from Chris and then the demo and the insights on food archive that we've just seen from Yanis. Always open to more questions and solutions but please contact Anna and the team if you want more information. But maybe now just to try and wrap up in the last seven or eight minutes or so. So first of all, a question for you Yanis. Just remind me again, I think you said it in the presentation. How many different data sets and sources of data are you including in the platform? So we are including many different types of data like food recalls, border rejections, fraud cases and adulteration cases. I see a question about that. So we also include fraud and adulteration cases. We have data about an information type, data type that is linked to the laboratory testing data. We have country risk data that is used for the fraud reporting and adulteration predicting the adulteration cases. We have sustainability data. We have inspections results that is presented in the food safety profile of each supplier. These are the inspection results that are the results of the inspections that are performed by the different governments, by the different national authorities. And just to mention again, we are collecting all the food safety incidents that are announced by more than 50 national authorities. We are collecting lab test data from 34 countries. We give access to more than 400,000 incidents historically and this data are increasing by 5% every month because we are collecting almost in the real time new data, new incidents. We are integrating batch of new laboratory testing results. So all this information is dynamic and it's continuously increasing. And the data coverage is 196 countries. So it's a very high coverage in terms of origin of the ingredients and raw materials. Thank you, thank you Yanis. So Chris, to you, to your favorite topic of food fraud. How can these predictive tools and solutions be better used out with the food fraud prevention? I think already as Yanis said, you're doing some predictions around food fraud. And whenever myself and my research group, we start to look for potential issues. There's many different data sets that we would use that you currently use. So a lot is about commodity prices, supply, demand, changes in consumers' eating patterns. If I give you an example at the moment, the sales of organic food has increased globally by about 20% during the pandemic period. And the question that I've asked lots of people, where did suddenly 20% more organic food suddenly come from? No, but it's a fair question. So to me, that's my indicator of increased fraud in organic and we will go and look at that. So I think the economic data is really important in terms of predicting fraud as well. Absolutely. And could you also perhaps share, Chris, any thoughts or insights you may have on using your digital crystal ball again? What do you think is the next big issue that's around the corner that's coming towards us? If you had to put money on it, what would you say is the next one? Well, if we call 2020 the year of the pandemic, 2021 is the year we're going to get out of the pandemic, but we're going to have massive issues in our food system caused by a lack of inspections, audits, I believe you're with a sesame seed problems, you can track back to pandemic issues about the redirection of food from one country going into another country. So I think what we will see probably in the second half of 2021 is a lot of the issues that have actually been happening during a lot of the harvesting storage of crops during the pandemic period. Also, when you think about the number of cargo ships that have got stranded in different parts of the world, well, we have a lot more problems with, say, mycotoxins, for instance, in cereals and grains that should have been a trans ship and should have been a nice dry store somewhere. So I think probably over the next 12 months, we'll see a lot of issues arriving because of those massive issues that the pandemic has called. That would be really, in all honesty, when I look at my digital crystal ball, that's number one on my list. I think that's very useful, very insightful. Thank you as well, Chris. I'm just checking the chat room now and just checking our time. Don't see anything obvious on the question. If it's okay, I wouldn't mind asking actually, Janice, a question because I really like all of the data. Now, very soon, I'm going to turn on my television, I'm gonna listen to the news and the thing that I will always listen to at the end of the news is a prediction. It's a weather prediction. What will the weather be like tomorrow, the day after, the day after? Now, I'm pretty sure if the TV tells me it's going to rain tomorrow, but it's going to rain tomorrow, but the further away you move from now, the less accurate that prediction will get. Now, I think in terms of the agronome data sets, have you gone back and looked about how accurate your predictions have been over short-term, medium-term, and long-term? Because they say a week is a long time in politics. Well, a week is even a longer time in the global food supply system. Things move so quickly, things change. So when you're showing a prediction of six months or 10 months, how reliable do you think that prediction really is? Great question, Chris. Thank you, Janice. I'm going to pass that one to you as your area. Oh, why? No, it's a very good point. It's a very good point. And it was one of the questions which we will answer also after the webinar. So the answers will be provided by regarding the accuracy. First of all, the way that we are selecting the best performing prediction algorithm is using the retrospective validation results. So we have many candidate algorithms that can be used. Many candidate prediction methods. But doing this kind of validation, we select the one with the best accuracy for the specific ingredient. So it's very specific to ingredient because generic predictions are not working well. This is something that we have tested. So this is how we have tested, which is the optimum amount of historical data to use how safe is to provide predictions for six of 12 months. So what you see on the system, the 20 years of history data and the 12 months of predictions is the ones, the prediction algorithms that are performing the best for the specific ingredients. So this is something that we do. We would not show something that is not accurate and tested before. And regarding the accuracy, it's more than 75% in general, but it depends on the specific ingredient. Something that would be very interesting is that we can share, of course, this data with the customers and with the ones that are interested in the prediction. So this is something that can be also presented to the users. Okay, thank you, Janice. And I think with the time now, we're actually at time. So being respectful, and as Chris needs to go and watch the six o'clock news in the UK, let's end the call. Thank you, Janice. Thank you again, Chris. Thanks for a great demonstration and thank you for sharing your insights. The presentation will be shared or recorded afterwards. Thanks very much for your time. Thank you so much, Gil, for facilitating. Thank you so much, Chris. Thank you very much. Thanks, Chris. Thank you, everyone, for watching. Bye-bye.