 So good evening everyone, I am Eduardo Jorgensen and today I will be speaking about predictive algorithms to detect future risk of chronic pathologies such as diabetes, but I will also talk about healthcare innovation. So I need to give you a little bit of context first. I studied medicine with a strong focus on neurosurgery. I'm currently the CEO at MediXen, a health tech company that I co-founded in 2015 and thanks to which I have managed to gather some recognition from key opinion leaders such as being the Spanish innovator of the year 2017 by BMIT or 430 under 30 in 2018. And I am passionate about innovation, especially in the healthcare industry and oriented towards improving quality of life with technological tools. And my path in innovation started in medical school while I was still a student when a little girl with diabetes rejected the insulin treatment in the hospital consult where I was arguing that she hated needles. She had to inject at least three times per day, which means over 1,000 injections per year. And on top of that, she never knew if her glucose was going to be safe or not. So she had to make uncertain decisions and calculations around insulin administration. So she couldn't have a normal life. It was kind of a horrible life indeed. And she rejected the treatment. So with all that information, I managed to gather a team of four amazing founders, two medical doctors. I studied with Patricia in medical school and two telecom engineers with whom I studied from low school and middle school, including playing in basketball together. So we really knew each other, which was very nice to start a company together. But we also shared that same passion to solve the problems of that little girl and improve the quality of life. And we were able to come up with an elegant solution and noninvasive artificial pancreas for diabetes, which may sound very fancy, but in the end, it's just a hardware device to deliver insulin to the patient and a software part to calculate how much insulin does the patient need in each moment. So in the end, the cool part about the hardware technology is that we were able to create a needle-free smart patch for painless drug delivery through the skin. We invented a small piece of technology that can open micropores in the skin. And it works generating harmless sand waves that increase the size of the natural pores in the skin temporarily, pushing big drugs through and closing again once we stop the device. And we built that piece into a wearable smart patch, completely free of needles. It can work with multiple different drugs, like insulin, heparin, antibodies, vaccines. And in a typical use case, they usually refills the device with traditional drug syringes or presentations, places it on the arm or the abdomen. And when activated, it takes less than one minute to fully deliver the required amount. It is completely safe and painless to use. And it has two components, a durable part that includes the electronics in charge of generating the skin permeability, which is there to last for over three years. And then it has a disposable part that includes the drug reservoir and adhesive, which is there to be changed after each use. So what the user would do is with a normal syringe injects insulin in the cartridge and then inserts the cartridge in the device and places the device on the arm or the abdomen, activated and gets insulin. This is a significant improvement from what's already out there because currently syringes are not only painful, but they also generate social rejection. Insulin pumps are really uncomfortable because you have to insert a catheter in your body and keep it there for 24 hours a day. That catheter can get obstructed, so it tends to fail frequently in the end as well, and these are expensive devices. And inhaled insulin is not better because it increases the risk of lung cancer. And then we have alternative innovations for drug delivery, like jet injectors, which are not wearable, micro needles, which is still invasive, and implants, which aren't safe for drugs like insulin because in the end you cannot trust that a dangerous drug like insulin, you cannot trust to have a dose for one week inside your body without any control on that dosage. You need to have some way of knowing the amount that's going to be released, activated at its time, and that's something that cannot be done with those implants and smart drugs. So the future is around ultrasound smart patches. And in part because it will also allow to focus on specific verticals, we won't have to regulate, like a new drug would have to pass five years of trials. If we want to have a smart patch for antibodies, heparin or insulin, we just have to do very easy tests to adapt those molecules, to adapt the reservoirs for those molecules and be able to release it to the market. So it also allows to sign future agreements with pharmaceutical stakeholders. So far we have been able to validate it in vivo, proving that when we inject insulin with our smart patch, it reduces glucose on alive picks exactly at the same level and rate as when we injected with a traditional injection. So that has provided us with enough evidence to follow on and complete animal trials and start human trials next year in order to reach the C mark by 2022 and get to the market to eliminate needles by 2023. But this advice that will reach the market in 2023 is a manual smart patch. This means that the user will have to manually select the dosage that they want whenever they want it. But our ultimate goal is to be able to automate that device with a software that can autonomously decide when the patient needs what amount of insulin. And that takes a lot of time. We're expecting it for 2025, but the regulation is much stronger. We require not only to develop the software and have it accurately enough, but we also need to go through new trials with the connected pieces. So we're expecting it for 2025. In the meantime, we were required to develop the software in order to create that non-invasive artificial pancreas, which is that combination of the software and the hardware. So we build a glucose predictor and lifestyle manager for the patient that there are patients currently using this system. It is used through an app that's pretty easy to use. We've made it, we've kept it very simple in which they can either with voice commands or with text messages and tell a chatbot interface about their data, any data input that they want, either meals, ate a Caesar salad 30 minutes ago, or any additional data point that cannot be tracked automatically, because we can also connect to third party sensors and gather data from them in order to inform the system on how the glucose levels of the users are going or alternative data points such as the physical activity. And with all that information, we're able to provide customized recommendations, not only building a complete lifestyle plan with meals and physical activities that are suitable for the user and will account for all the weekly needs of a diabetic patient, but also with awesome functionalities such as the patient can ask, what happens if I drink a Coke in 30 minutes? Or what if I go for a 30 minute run right now? And the system will forecast the glucose levels for the next two hours in order to tell them, hey, if you're going to run for 30 minutes, you should eat some peanuts before because otherwise you're going to have a hypoglycemic event and this is how we can avoid it. So that's a real value that the user can see in our technology, reducing all their uncertainty and they can do that thanks to our predictive algorithms that are created based on a massive understanding of the physiological behavior of diabetes thanks to our medical team and through deep machine learning knowledge, because we have strong team members that have created awesome models in there. What we do is we analyze a time series of glucose, which is in the end glucose measurements, and we're able to augment that data in order to understand the future glucose predictions and also understand it, not only comparing with other patients, but with that same patient in the past. And this gives our patients more information. It gives them the ability to eliminate those calculations and uncertain decisions from their life. Now they know they have to eat some peanuts before going for a 30 minute run because alternatively they will suffer a hypoglycemic event, so they can feel safe, they can make safer decisions and that will lead to improved treatment outcomes. And this is also beneficial for healthcare providers, insurances and hospitals because in the end they need to get the patients to have better outcomes, but when trying to get to business agreements with these companies, we faced a couple barriers. The thing is that even though these healthcare providers generate huge amounts of data, they typically don't have access to the results of a blood test. They can only know that the patient actually took that blood test and that's for data protection measures. And in the end, most of the data ends up being rotten, unused, lost and not being taking all the value that could be out of them. So it's impossible for the healthcare provider to know which patient is more likely to suffer a chronic pathology. Is it Maria with risk of type 1 diabetes or is it Paul with risk of type 2 diabetes? It's impossible for them to know who will bring higher costs of care, so they have to make uncertain decisions on where to invest, where and how to invest time and money. So that's what our latest piece of technology is solving. We understood that our predictive algorithms were very good for the patient, but if we wanted to make some business with healthcare providers, we needed to adapt the technology a little bit. So we transformed it into a future diabetes or diabetes complications predictor. And what it does is we get the medical act. That's the data that virtually any healthcare provider in the world has available, because it's not sensitive. It's only you don't need to know the results of that blood test. You only need to know that the patient took it. So we will end up when we see the database data scattered all around, we will end up having this timeline of medical events that the patient had, medical interactions that the patient had with a healthcare provider. He went to the cardiologist, he took the blood test, went for an x-ray. And within that information, we're able to find hidden information. We're able to extract additional variables by creating an enriched layer that can add new variables such as frequency between blood tests or number of times that the patient went to see a cardiologist or total events that are associated to endocrinology. And those are new variables that were not in there before. And thanks to the analysis of massive amounts of data, of massive amounts of patients, we're able to predict which ones have risk of ending up in diabetes and what type of diabetes. So is it a 70% risk in three years of being an autoimmune patient? So you require an urgent care or is it a less riskier path and associated to a cardiovascular risk? This allows the doctors to understand the risk patterns of their individual patients, but also of their group of patients because they can end up saying, hey, which is the group of patients that associates more risk? And he will see, hey, patients over 65 years living in rural areas that don't have nutritionists associated to them. And in the end, this will also help the manager understand the potential return of investment by simply dragging and dropping variables on the dashboard to, I don't know, link risks and costs. He will be able to understand the ranking of riskier hospitals and which ones would benefit from a bigger risk reduction in case they hire an additional nutritionist or in case they change their policies of attending, offering three nutritional visits to patients per year to four nutritional visits on patients for a year. He will be able to understand the return on investment on those decisions, how much the risk will be reduced because we have all the previous data to back it up. So in the end, it's a predictive healthcare analytics and visualization tool to improve decision making, which is easy as dragging and dropping variables in a dashboard. Youthful as being able to predict future risks and costs and compliant with any regulation because we don't access sensitive patient data. We only need to know that the patient took a blood test, not the results. And it's highly scalable to alternative solutions, to alternative diseases because the minute we start looking for risky paths in medical acts for cardiology events or for lupus or for obesity, we can add additional layers of information on risk for the healthcare provider and our system will keep growing. But if we want to do it, we're definitely going to face some barriers, which are in the end the barriers that are facing virtually any medical device company. I would focus initially on one of the hardest ones, which is to really center the technology on the patient. We all like to say it, but it's a real struggle to find focus groups, be able to gather relevant information and feedback from them, transform that information into real actionable development, and then re-testing and re-engineering all around the process while the environment changes around you. Of course, if all the money in the world, all the resources in the world were available, everyone will test with their patients. I believe no one says, hey, I want to just, I think patient testing is not needed, just keep through with this technology. So despite being something that everyone recognizes, it's something that we still have to invest huge amounts of time and resources on securing. And it's mainly because there's a lot of security around medical data. First of all, around GDPR. This is a very important regulation, but it's also a very important barrier because, well, look at the technology that I just presented. This technology only makes sense because patient data is protected. If healthcare providers were able to access patient data whenever they wanted, for whichever reason they wanted to use it, our system with predicting future diabetes will probably won't have any kind of use for these healthcare providers because they will be able to build their own risk-predictive models and even say, hey, I can really know if this patient is a diabetic patient so I can update their policies and everything. So data protection is there for a reason. And while it's a barrier that we have to account for, it's still important. This is just a small note on something very silly, but true. It's a tremendous news. Hey, hackers can actually make your pacemaker or your insulin pump kill you. And everyone would say, oh, wow, we need to protect this. And it's true. We need to build the security layers in order to make sure that this cannot happen. But technological progress involves risks. And zero risk is not possible in the medical industry. So these are going to be some of the challenges of the next few years, especially when automating this technology, because if you have an automated insulin drug delivery device that you don't have to worry about, it's making self-decisions. And all of a sudden, someone can connect and activate it. Problems can happen. So it's a complex environment. And it's also a highly competitive environment because there are huge stakeholders with billions in revenue trying to solve these same problems. Big companies are typically slow giants, but they're also pushing forward in the market. So one can try to partner with them. And it's typically good to establish those relationships. But one needs to be careful as well. And not only on the information that discloses, but also on the time that one invests in those operations, because it can end up consuming you if you need to adapt to their life cycle, to their time cycles. So it's definitely an additional risk in this aspect. One of the biggest ones is around the fragmented legal framework. Look, medical device regulation is even changing and it's being updated next May. And there are no clear guidelines on how to regulate certain medical devices. So there are a lot of companies and technologies that are really struggling to build their market plans, the market entry plans, because they can never know what sort of clinical validation they're going to need and how long it's going to take to get it or enter into the market. And all of these with limited resources, especially in small companies, which in the end is creating a horrible cycle of I need resources to get validation, but I need validation to get resources. And in the end, it's extremely difficult to break the cycle, be able to get out of it and secure that validation in a bootstrapped way or through partners or by public funding. So one is to create creative strategies to in order to secure the resources and validations required to enter the market. Probably the most difficult one to solve because it's likely to require time is the attitude of the individuals in the healthcare providers and public administrations when it comes to integrating new technology because despite the change in the trend and it's true that we're seeing more acceptance, we still see reluctance when integrating new technology, not even integrating, also when testing it. So I believe we still need to work on decreasing that risk aversion and being more hands on trying new technology that could benefit patients. And I think that the only way of doing it that is through innovation. And innovation in the healthcare, it's beautiful because in the end, you're trying to solve problems for patients and it's always offering a high return on investment. As long as you're creating value and the technology is going to be used, it's offering a high return on investment. So every one of us can become champions in our own organizations to lead that transformation movement and create or build new processes, technologies, do things differently that in the end are generating a cost-effective benefit for anyone, for someone, which is in the end what innovation is after. So if it's inside all of us and all of us can become the champions of our own organizations, we have to do it. I think that the first thing that you're doing very good is attending these kind of presentations because we're not going to reinvent the wheel in these places. We're not going to tell you or disclose extreme secrets, but what you're probably going to find is some good examples of innovation and ways to innovate and cultivate it. So if I could send one final message about what any one of us could do is to center technology on the patient. Not only say it, but invest real time and money on doing it. We're facing those challenges with a wonderful team with over 15 professionals, including engineers that come from NASA and MIT or a healthcare team that grew up in Phillips and Rush. Our business team may be relatively young, but we've been able to secure sales and we're advised by a strong team of key opinion leaders such as Young Brooks, the former CEO of the Justine Diabetes Center in Boston. And if you want to help us, you can always introduce us to healthcare providers like insurances or hospitals in order to perform some pilots with them. We have the software completely ready to test it and prove the value that it's offering. But we can also pilot with the hardware, with pharmaceutical companies. We can demonstrate them that with our smart pads, their drug can go through the skin with no pain. So, well, and always, we're always accepting funding and resources. So if you know anyone, let us know. So we really hope that you join us to shape the future of healthcare. I know that my presentation has been a little bit shorter than the others. I tend to talk a little bit fast, but I think it will be interesting to have more questions from the audiences available. I can always answer questions on the technology as well. So if anyone wants to ask around that, I'm happy to answer. Thank you very much. Thanks a lot. We have a first question here from the audience. They are asking about the possible effects in insurance companies. You are working with very sensitive data. You are going to be able to discover possible effects in the future. Do you think that the insurance industry will be affected in any sense with all this stuff that you and other companies are working with? How do you feel about this? Well, of course. I mean, partially the trend in the insurance companies industry is towards being able to understand an overall risk score or, yeah, score of the patient. In the end, they want to see, they don't want to see, hey, I paid this much of things for this patient this year. They want to try to understand how much are they going to have to keep paying for that patient, and this can impact in a variety of ways. Of course, there are wrong ways of using this technology in the means that you would say, hey, if they know that I will have diabetes in 50 years, they can always charge me more. Yeah, but they can always offer you that if you treat yourself better, charge you less, give you a discount on that, which is something that there are already some companies even in Spain starting to do it. If you sleep this amount of time, you will get this amount of discount. If you walk or do this exercise every day, you will get a certain discount at the end of the period. I think that it's one trend, but it's also offering one very important thing. One is that healthcare providers will be able to really understand risk and cost drivers and what has a real impact on their organization. It's something that they need and they want, but there's an additional real important point, which is it will be good for the patient. They will be able to have a more personalized access to healthcare, and additionally, they will be, I don't know if the audience knows that, but at least in Spain, if you have diabetes, it is extremely hard for you to get a private insurance to offer your policy and ensure you. It's not that you can pay whatever you want to know, it's that they don't take the risk. Our assumption in there was they don't take the risk because they don't know the risk. If we can make them know the risk, they will probably be able to ensure diabetic patients. That's our final appreciation on this technology. Another question is about how affordable this will be. Are you working in and trying to make these solutions you are working in affordable for everyone? Do you think that this could be solutions available all over the world, even in the third world, even in the asking specifically for some places in Africa or in the pure areas in Asia? Are you working in this and making this affordable? The technology in the end is relatively cheap to produce, so we'll be able to offer an affordable price, especially compared to the solutions that are today. I mean patients can be paying from 3,000 to 30,000 euros for an insulin pump. Our technology will be for the user. We plan to sell it on a business-to-consumer side, the smart pads drug delivery device for 100 euros a month in a subscription way patient will receive all the disposable cartridges. That's on the hardware and on the software. We plan on giving the glucose predictor free for patients until we are able to link it and integrate it with the smart pads. And regarding the healthcare provider's technology, the one that predicts future diabetes, that one will be as a software, as a service, depending on the databases of the healthcare provider on the sizes. And if it can be worldwide level, all of our technology can be worldwide level, but the smart pads itself, it can also, it doesn't need a phone to be controlled. You can also access it with buttons. So it can be included in programs like buy one for yourself and buy one for someone in need. So we're thinking about a bunch of strategies in order to make it accessible for everyone. The last question, I'm not a surprising one because there's a lot of tech people in the audience is of course about technology. So how is this technology, if you could give the facts some more details about the hardware, about the software, about how are you working from a tech side? Nice question. I didn't include it because it's a typical thing that the audience may hate, but I know that they ask. So the smart pads has a really cool technology. It generates ultrasounds with a very small piece and part of the innovation is on the size of the piece because up until today, you needed very big machines and connected through cables because they require huge amounts of power in order to be able to generate the required ultrasound or ultrasonic waves. So that's the first thing that is innovative. We've been able to reduce the size of the ultrasound generator. And what it does is that those waves hit the skin such as boom, boom, boom, boom, boom, boom, boom, boom, boom. And that way we're able to loosen or dilate a little bit the pores that are naturally there on the skin. And when we loosen them, we create enough space for these big drugs to go through. So that's basically the way that the pads work. We've been three years doing research on the optimal setting of waves in order to make sure that all the drugs that we put there can go through and everything. And now we've been able to validate it in vivo with insulin and we have validation in vitro with alternative molecules. So we'll jump into animal trials with other molecules soon. And regarding the predictive algorithms, so what we do is I can get into a small detail about this because most of that is know-how. But the process of the healthcare providers algorithm, what it does is we get to a database that is typically a mess. I mean, there are healthcare providers that have good databases, but typically you find something that is hard to use. So when we get there, it would typically be required a lot of human work. And what we do is we have created a bunch of algorithms that can automatically understand those databases, not all of the content, but the relevant content, and create that enriched layer that's saying all those aggregated variables, combining that we can create up to 2,000 different variables. But some of them don't have any kind of interest, some of them are extremely interesting. And then we make sort of an engine between those variables and our predictive algorithms. Those predictive algorithms in the end are machine learning models that need to be adapted to the specific needs of the healthcare provider, but we know which options. We can use something like support vector machines. We can also use, I mean, it's not needed to enter into strong deep learning. Generally, if you have a complex environment, it might be required, but typically we've been able to sort out these problems without any kind of deep learning. So in the end, we produce that number that tells us if the patient is in risk of diabetes or not. And the number comes in a model that's associated to a certain timeline. That's it. So thanks a lot, Eduardo, for this great presentation. Enjoy the Iceland. Enjoy it in a reef. And that's amazing weather. Thank you very much. Have a wonderful day. Bye.