Collective Pandemic Intelligence
4:33PM Jul 21, 2020
afternoon and welcome to AI for Good Global Summit webinar series. I don't appear from ITU International Telecommunication Union, I have the privilege of introducing Jays
AI to use
flow using price to cheese.
Bear for quick google exercitation partnership 361 sister agencies, ACM booking Switzerland The goal of this summit is agenda applications, sustainable goals and solutions for double impact. Now most of the world as time has gone digital online programming, allowing us to reach even more people throughout 2020. We're pleased today to host today's webinar on Russia. We have a distinguished panelists today but as always, we're counting on you the purpose of this truly helps us in engaging discussion.
We have a q&a now the
chat function. Now using chat function, please make sure to stay up
just about trucks. Now to try out this functionality.
This is your first time here with us.
And let us know your country or state. Please make sure to like
so let's see.
Okay, are we so we have around the world now please be sure to meet up everyone my office. So that further ado, I'm
over here, Andrew.
Thank you Ida. And thank you everybody for joining today. I'd like to welcome you today to today's webinar, where we're going to discuss how to leverage AI technologies to develop database solutions that bring rapid relief to today's COVID-19 situation, and hopefully, long lasting solutions to pandemics in the future. My name is Andrew towered reporting to you from Dallas, Texas. I'm the VP of partnerships and strategic engagements at XPrize. And I'm joined today by a group of a steam brain trust that I will use traduced you in a second here. The purpose of the webinar here is to really introduce these topic areas or themes as we're calling them, in which we're seeking project submissions from the public. As part of the air for good collective pandemic intelligence breakthrough track, a team of thought leaders is braintrust is has leverage their expertise to develop AI, develop data and AI systems to help drive global initiatives in the area of area of COVID-19, response and solutions. So before we dive into those themes, and go through them, I'd like to first introduce the brain trust themselves. I'll introduce each of you ask you to introduce yourselves and we'll go on to the next. So Thomas, maybe you can start.
Hey, everyone, pleasure to be here. Tom Osborne. I'm a medical doctor and cmo at a large healthcare system for the VA in Palo Alto, mostly director for National Center for collaborative healthcare Innovation Center.
Thank you, Tom. Steve. Can you?
Hi, everyone. It's a pleasure to be here today. I'm Steve Griffis coming to you from Abu Dhabi and United Arab Emirates. And I'm the Senior Vice President research development and also a professor of practice at the University of Science and Technology.
Thank you, Steve. Amanda.
everybody, my name is Amanda pournelle. I'm a psychologist by training. I'm the senior innovation fellow. I live here in St. Louis, Missouri, and I work for the VA.
Thank you, Amanda. Next up nada Ross.
Hi, everyone. This is nagaraja grapecity. I'm the Deputy CEO of Life Sciences Queensland from down under Australia.
Thank you Now garage, Cody.
Everybody, Cody Sims here. I'm a partner and Senior Vice President at TechStars. One of the most active early stage venture investors in the world. coming to you today from He thousand feet up in the mountains in Colorado. At the moment though I live in Los Angeles.
Thank you, Cody. And last but certainly not least, Maria.
Hi, everyone. Hello from Boston, Massachusetts. My name is Maria filippova. And I lead innovation for anthem. We take care of one of every three Americans when it comes to their health care benefits.
Thank you, Maria. Okay, so let's dive into it. We probably should start off with really defining what the collective pandemic intelligence breakthrough track is, and what are some of the goals behind it. So for that, I'll ask Amanda to help give us some color on that.
Thank you. So this group of individuals we all got together and discussed what we saw based on our own experience, what we've read, who we've interacted with, and where we see problems or challenges and kind of iterating Several times on a series of what we saw is the most key pressing issues co developing this over time deciding what might be the central issues that we could offer to this group to help leverage the incredible insight and creativity of those who might join or be interested. And we're excited to share those with you.
Thank you, Amanda. And to add to that, maybe you can give us a sense as to what what you've been up to, I mean, over the last little while with you and the rest of the brain trust members.
We've had a series of meetings kind of over time, where we saw so many problems, overlapping problems and tried to think what are the core themes that we could pull together that would generate excitement, enthusiasm, and creativity? What we know is that we don't always know what the best solutions or the best opportunities are to help understand develop new ideas that might help drive insights and help my drive engagement and might help drive solutions. They could come from a variety of fields, right at different places around the world. And so we tried to generate language that would be as broad as possible, while still clearly describing a core problem that would help people be inspired. And we hope that you are inspired by what we've come up with. And I very excited to hear what my people might do with our core issues. Core games.
Excellent. Excellent. Thank you. Well, that's a that's actually a great segue into the themes themselves. So if I could ask you to put up the first chart with the first theme. So the first theme is defined as follows. How can we incentivize data collection and sharing to help build solutions that are equitable and fair when it comes to responses to communities affected by pandemics, such as COVID-19 So to unpack this theme, and really give us a little more color around it, I'll I'll hand it over to Thomas to take us there.
Thanks, Andrew. So, yeah, this, this is the first theme. And I think there's some key words here that help frame it up. And that is looking for solutions that are fair and equitable. And as it comes to or how it relates to our responses to the communities, and you know, so I think this topic is really foundational, in a lot of ways. It's foundational, because the success of our data informed decisions is is foundationally reliant on the robustness of the data. And this, quite frankly, really applies to any rigorous approach, not just AI, but perhaps, perhaps especially AI. You know, having the correct data is critical to accurately understanding what's going on and therefore being able to do Something that's useful or impactful. Um, we've also found that some of the communities that are disproportionately impacted by this disease, have sparse data, missing data. And sometimes the data is unreliable. So there's gaps. And so for example, we know that some communities domestically and around the world are disproportionately affected by this pandemic. And we also know that these same communities, on average, have greater financial struggles than others. So you know, what's, you know, obviously, what's going on. And there's a variety of potential causes that have been proposed. Some have suggested that it has to do with service, job related exposures, transportation options, living conditions, access to or availability of health care, nutritional few food, you know, chronic stress and so There could be these problems, there could be others. And you know which one of these is really this, disproportionately driving this illness? is, you know, we, we obviously are interested in you know, which one of these? Is it some of these Is it a combination of them is it things that we haven't even thought of yet, it's a hard hypothesis to prove or even understand when much of the data that we have is unreliable or or just not available. And if this, if this absent information is not included in the models, if we don't understand the relative impact of the interactions, then the subsequent results that we produce are going to misguide us. We're not going to be able to be empowered with the right tools to create the right solutions at the right time. And I think as a bigger picture, it's really important and I think perhaps a little bit exciting to sort of keep this in perspective that this is an issue That's not only relevant for this very important and tragic pandemic, but also with other healthcare problems. You know, there's other disparities that are common in certain communities. And so this makes this topic all the more important, the advances that we make here, and understanding foundationally what data is important and missing, can have far reaching positive impact in a multitude of other conditions.
Thank you, Thomas. This is actually quite interesting. You certainly hit on some of the kind of key areas of that theme and what why they're important. If I could just come back to one area that you didn't touch on as much but maybe you can elaborate also, as we're talking about incentivizing data collection and sharing. And so you've talked about of course, we it's very sporadic, and it's often you know, the impact of the pandemic is not equitable. But any thoughts on on data sharing as well on things that the community can think about in putting projects forward that could involve more data sharing to address these issues you're talking about.
Yeah, thanks, Andrew. That's, I think that's really good point. You know, there's there's an absence of data for a variety of reasons. Either it's difficult to get ahold of, it's not well curated, or it's not easily shared. And sometimes the data that we're looking for can be sensitive data in the, in the eyes of some communities, especially if they they have a different understanding about what it's been used for. For example, if you're interested in knowing what kind of work environments are important for this pandemic, and if some people are just working their butts off, all with many different jobs and not all of them are being acknowledged, then they might be less than open, about sharing all those different jobs that they have just to make ends meet. Or perhaps living conditions, you know, people doing their best to stay together with people that they love and care about. And sometimes that's crowded living conditions. And so in order to get to some of that information, and for people to be comfortable sharing that information, it would require a degree of engagement that we haven't seen before.
Absolutely, absolutely. Thank you. Lastly, they just kind of close off this theme. Can you give the sense for those that are on the webinar as to some of the types of things that we're looking for, they should keep in mind when they're submitting projects under the theme?
Yeah, thanks, Andrew. So, you know, of course, there's some specific ideas about how to potentially address these challenges that we've thought about. But you know, quite frankly, we none of us pretend to have all the answers or ideas and so some of the things that we thought about are just sort of stimulate other ideas. And, quite frankly, we encourage We're excited to hear about what other people will bring to the table. I mean, things we've never thought about. So because of that reason, we don't really wish or intend to constrain the topic with our preconceived notions. But, you know, proposals may come in many different forms in many different directions. You know, some of them, you know, I think we've talked about would be, you know, how to aggregate important publicly available data for researchers, different novel techniques, perhaps remote sensing, a portal on app, book, education, sensors, you know, we would like to consider, you know, importantly, we talked earlier about how to engage people within these communities, and empower them to participate to be part of the solution and help drive knowledge and understanding.
Great, thank you very much for that. So maybe we'll move on to theme number two. So it if you could cue that slide. Thank you very much. So tsunamis, number two state says, How to create predictive models that support informed policymaking and mitigation of improper resource allocation during a pandemic. So, maybe I think we'll hand it over to Steve and Steve can help to elaborate specifically on what on this particular thing?
Sure. Thanks, Andrew. This is a really interesting theme. To me. It's an area that I've looked at AI model development quite a bit. And so we thought about this as a as a braintrust. Thinking about where we want to go with the theme to the first component was really understand the challenge that the theme would be addressing. And so, with COVID-19 in this pandemic, what we've seen is the ability to provide adequate and targeted medical and financial resources to at risk populations that have elaborated or alluded to just previously, it's been a real challenge. It's been something that we haven't had the ability to do as effectively as we'd like to and so as we thought about the need It's out there. And we're to address this challenge, really thinking about how to mitigate effectively impacts of global pandemic by COVID-19. The ability to predict it and really understand where the disease is going, and how it may propagate is very important. And where do you want to base your predictions? To some extent demographics, we understand that based on personal people's genetics and what have you, there may be predisposition to diseases, but we also want look at socio economics. And I think Tom said, you know quite a bit about the fact that, besides demographic, demographic information, socio economic information is also quite important. And secondly, are lastly on the need, when you're looking at those most at risk populations, and you're looking at how they can be predicted. There needs to be adequate planning that can be taken from these predictions, so that we can see where medical and financial resources can be best targeted. The most effective way possible. So it's really it's looking at trying to see how you can bring the right resources to at risk populations. And addressing a challenge. It's not yet been solved. If we look at the COVID-19 situation and what's really been going on there and why we've had challenges prior to this pandemic, and you think about this in the sense of artificial intelligence, we really haven't had a lot of historical data around pandemics to train models to undertake such predictions. And therefore the problems that we'd like to see solved, are the disease spreading rapidly and impacting communities that are ill prepared for frontline support and medical supply? We don't want to see that happen. Again, we'd like to see rapid response. It's very effective and very targeted. Likewise, when financial assistance has been distributed, this time in this pandemic, it was necessary to go out very quickly and aggressively financial assistance, but it was a bit shotgun approach. In some cases, it wasn't as targeted. Could be. And if we could make the financial assistance more targeted, based on understanding of the populations that are most at risk, we could probably alleviate some of the various significant impacts we've seen in the in the spread of COVID-19. And we could also perhaps mitigate the spread, because we could provide assistance to those people that may be at risk and had to undertake activities that they otherwise wouldn't have if they had been able to benefit from some assistance before they were exposed to disease. So for a number of these reasons, I've mentioned that I feel this is a very important topic and it it does really get at the heart of artificial intelligence, the model says,
Thank you for that. You know, we as a as a brain trust to talk about a lot when done a lot of different paths of exploring where where we could go and how we can structure the themes. Can you give us a sense as to how you arrived at this one?
Well, so this is a good question. When you're looking at artificial intelligence. You're one of the things we're looking for developing The idea for where we're headed with this is we want to be able to look at all the at risk populations, they'll be out there during a pandemic, and they said, tried to find the best resource allocations and frontline support for these people through in this case, it's going to be a model, it's going to be an artificial intelligence model, most likely probably a software product. Now, that said, I would like to see that this model is going to be one that's going to have explainability to it, you know, that artificial intelligence these days, the challenge of black box approach in many cases, although it is statistically based AI, in many cases, we don't get to see the insights of what's happening in the model. We'd like this to be different from those. Those those models that have been less insightful for the policymakers will use them have explainability and bring a risk bound to it. So we'd like to know that there is always a spread of potential solutions within these models. They are based on statistics and so trying to understand that Spread. So policymaker can make an informed decision around the potential outcomes, the model, which we know are stochastic, and they're not necessarily point based, and then allowing that to help us get the social determinants of health, the inputs, and having the data from that allow us to see our culture, job status, financial status, all the things that Tom did due to health care access even would make transmission dynamics mortality, something that we can, we can stop the maximum extent, and then having course data availability. So we can prove out some of these models, at least on a on a proof of concept basis. So I think we're trying to really bring together what I was mentioned previously as the areas we want to address into a model that is usable by policymakers because it does have insight that can be gained you gain from it. In a sense, it's not clear clearly a black box and there's a sufficient data, but make this model one minute, Beast. testable so that we can see that the solutions actually worked.
Actually, that's very helpful. I'm glad you glad you went there, you know, where you were just talking to is almost potential examples of types of projects that we didn't vision coming in, or hope to receive on this. Could you? Are there others that you could elaborate on that or that in terms of what types of things we'd be looking for as outcomes with this?
Well, ideally, we'll have the complete solution. I mean, it would be absolutely fantastic. If we had a predictive model that would allow us to be able to allocate optimally to the most at risk populations, the the best levels of financial and medical resources, you know, frontline support, do that and explain away. It's totally risk bounded, and of course, then have the data for that. I think in many cases, we may find projects that can bring components of this, you know, if you're able to harvest the data that gets us to a, you know, at least a model that is able to address a large number of these potential areas that could have great impact for pandemics and Future, not just COVID-19, but learning COVID-19 the data it's able to generate, and then be able to look at how we can impact future pandemics. That'd be really good. So I think as a, having a way to perhaps bring the data would be one potential project if you if you can't get the full solution, but if you can identify the most robust ways to securely and directly access the data, so it can come into a model and start to be able to make predictions, if not holistically around all the errors I talked about, you know, certainly we can address some of those would be fantastic. I know it's a very challenging problem. I also think there could be projects where the models developed and I think this is actually something we'd seek would be able to not just be trained with an ability to know what's gonna happen in a pandemic very, very similar to COVID-19. But in Artificial Intelligence, that progression you may be able to do to use transfer learning models that are trained in one Context a amenable to another context, that also really great if we can show that show how this is going to be something that's useful for future pandemics. That'd be a really good project. So I'd say on theme to, it was a little bit more software oriented, although I'd say if you know, trying to get toward where this overall initiative is going in theme to if we could do at least some of the steps, if not all of them to get toward an AI based solutions can help future pandemics, it would be really great progress.
Next, thank you. Thank you very much for that clarity. I'll also remind everybody who is dialed in here, that there will be ample time at the end for QA if there are other questions about hypothetical projects or areas of focus within each of these themes. But I think for now, we'll go on to theme three. So thanks again, Steve. So if I could cue the slide for theme three. Theme three is defined as a How can data and evidence based AI models help avoid misinformation and panic while minimizing economics impact and maximizing health outcomes. So to elaborate on this a bit more about what this is, this one's all about, I'll hand it over to Cody and let him talk us through it. Great, thank
you, Andrew. I'm
with Team three here. You know, a big part of what is coming together here is the notion that public health isn't just about the health. It's also about the public side, how is the public? How are we communicating with the public? How is the public getting information about what is happening? And how can that information help inform both individual decisions as well as move into informing policy decisions? I recently read a really, I think, insightful quote that said, when you mix politics plus science, you tend to get politics. And so the idea here is how can we help inform both the populace and policymakers and create technologies that ensure that the right information is getting out in the right way. There are a few areas that we Specifically dove into with this theme that we feel are important to develop technologies around. One is how can data and evidence based AI models help avoid misinformation? and and you know, excuse me, that's the whole theme. But how can we look at projects that eliminate subjective biases? When it comes to economic and health responses? How can we ensure that information is appropriately being understood in terms of how whether this disease or future pandemics are affecting underrepresented minorities or affecting people of color or affecting populations that may not be in charge of creating the bulk of media that's being distributed? So how can technologies look for those types of biases that are out there in the media and in the information that's, that's being distributed? Another key area that we're trying to find innovation around is really Looking for projects that identify the actual seeds of misinformation themselves. So we've seen particularly a lot of algorithmic social media can can generate quickly, incredible support for ideas that may not be correct, whether that's potential over the counter treatments, whether that's, you know, ways to interact socially with people what's safe, what's not safe, whether that's how you should go get treated, you know, based on how you're symptomatic or not symptomatic, etc. So how can seeds of information be identified quickly? And what should be done about that? And, you know, we all know that things that may show up as algorithmic sort of social media driven trends can turn into mainstream articles and mainstream information very quickly, in this day and age. And then finally, a big topic that we're excited about and hoping to find solutions about is what are the actual proper proper channels for sharing information so what things should be distributed directly to the public, whether via government or via policymakers, what things what types of information should be distributed to health systems? What type of information should be distributed to Chamber of Commerce, as it's relevant to business in the economy? What type of information should be distributed to places where people congregate in mass, like local transit authorities, or religious institutions, etc? How are the right channels of information designated and how is the right type of information being distributed accordingly?
Thank you What
a couple of related follow on questions to this. I guess first and foremost, this is a bit obviously all related, very related to pandemic very relevant but also a bit different from the other two themes. So he says, Can you give us a sense as to how we landed on this one?
Sure. I mean, obviously, we're dealing with a virus that has the potential to touch hundreds of millions of people and is already causing causing deaths in Sydney. numbers. That notion of how individuals should react to what's happening out there and where they're getting information is critical to stopping the spread, controlling how people behave and engage. We've seen, you know, a politicized media world that is, you know, sharing different kinds of information all over the place. How can we help? How can technology help actually identify root cause issues that are actually creating endangerment or Panic of the population as a whole?
Thank you, that is very relevant. And following on from that. Can you talked about obviously, it's very, very relevant here now. And you implied also that this would be applicable post hopefully, when we're in a post COVID-19 world. Any thoughts as to where else you would hope that we could apply to projects and the learnings that come out of this to other other either other pandemics, of course or other domains?
Yeah, of course. I mean, I think you know, again, different sort of subtracts that I talked about you think about something like identifying seeds of misinformation. I mean, that's critical not just to this horrible disease, it's incredibly important to this horrible disease right now, while we're all trying to manage our own both personal and government level policy responses. But it's an incredibly, incredibly important aspect for understanding any type of future, whether it's public health issue or just general threats to democracy. So understanding how people actually share information clearly and intelligent with intelligently with one another and understand when information is biased, we're seeing that impact all aspects of our life today. And technologies that can be applied to the COVID solution potentially have the ability to actually apply to broader solutions in this world.
That's definitely food for thought when, when we have folks submitting projects that of course, the goal here being that this project is not just something that is impactful in the current situation, but we can we can take it forward. I know not Roger, you were going to potentially add in a bit more on some of these. Any any thoughts that just in general to add to what Cody had, whether the types of example projects you could envision or just your views on this particular category?
Cody, thank you, Andrew Cody has touched upon it very quite well in terms of what this theme is covering. But I just thought I'll just add a couple of points is basically we know Korea does impact it globally. And Vegas data is represented and interpreted both in the developed countries and in the developing countries is different and that goes for various reasons. And how people are accustomed to receiving information or how it's channeled and it's called infodemic is becoming quite evident. And a purpose within this record see is how can we fast check and have relational data disseminated. And decisions are made, which can avoid problematic material being circulated. And that would be something quite important. And the channels of communication exchange, there's definitely a gap in between countries and within cities and within regions and how that can be integrated. And automated. Again, by keeping it to reliability and can be used for multi user data, I think that would be quite important. And that's what we would be looking for.
I think you touched on also a very key point that is worth mentioning as well, as you know, you talked effectively about equity in terms of how different communities are represented, represented or treated in this domain. And so there is some overlap, we recognize between some of these themes. So we do call on the audience as you do think about your projects. There's there's no right or wrong in terms of what category you pick or what theme you pick, just pick the one that is most represented. But it is interesting to see that issues such as equity do apply across the board, no matter what, what lens we're using to look at this. So thanks. Thank you for that Naga Raj. So to move on, and again, we'll have time later for any other specific thoughts and questions from the those dialed in here about these themes. Let's move on. I'll ask Maria to weigh in here and really give us a sense across all of these themes. And as people are thinking about how they could co create a project submit a project within one team or another. If you Marie, if you have any, some principles for them that they should be considering when they're prepping for that.
Sure, and thank you, Andrew. Hello everyone again, as I mentioned, in my role at andsome Where we take care of about 100 million members in the US within our Blue Cross Blue Shield network, I spend a lot of my time thinking about what does it take to achieve impact at scale? And, and so when we think about some of the themes that we outlined, so far, I would love for us to anchor on what is the impact and the first principle that I'd like to offer up to you is really the notion to start with the outcome, where scientists, we get enamored or excited about our predictive power of our of our algorithms, the area under the curve or anything else that we could improve on another algorithm. At the end of the day, I really encourage all the teams who are submitting their projects to think about the why, why are we doing this? What is the outcome that change the impact that we want to see in our communities in the patient with the project that we're proposing this this group, this brain trusted being very thoughtful and framing of the three major challenges that we see within data collection within ways of informing policy decisions and targeting, resource allocation and combating misinformation. And so we've geared you with steered you towards areas of impact. Please take a moment to think about the impact that you want to see before we start modeling and, and throwing data at things. The second principle that comes to mind is connect the dots and connect the dots both at the data level where we could be looking at data that combines not only the public health data of new cases but also economic data of unemployment rates of purchasing patterns, mobility data of where people are spending their time outside of home, is it the airport is it a coffee shops in the grocery stores, fooding security risk? So we we found in on Monday Have these connections between social drivers of health and clinical outcomes. These are adults that are not, quote unquote, traditionally connected in clinical practice. So far, the this pandemic has made it more evident and more urgent for us to be adult connecting across different parts of our life of our health, and the data that could inform some of these insights. We have seen some examples of great connections so far, there is some there is a kind of an aggregator of insights called the C 19. Explore that we have helped develop with some of our alliance partners. And they really do a good job at not only thinking about the insides are coming out that are retrospective, but also thinking about how these insights can help us be a little bit more proactive and predictive in our response. The final piece of kind of a principle that I'd like The teams to focus on his question the commonly accepted methods of business as usual. That pandemic has really faced us with an unprecedented situation and there is really no playbook to follow in in solving for it. And when there is no playbook people like me get excited, frankly, because I get to ride the playbook. And so take that on the you know, the innovators in this community. You have a lot of supporters eating on the the this brain trust and on others who are looking to solve these problems in question the status quo and traditional, traditionally accepted principles. Be creative with your solutions. So I'll see these three principles up and open for questions.
Thank you for that. Marina. Let me build on Well, first, I'll just build on your excitement. And it's very clear that you can see what the potential is here. You highlighted some key points here. First off, I mean, we've got an amazing brain trust here. And we're assembling a group of experts that will help to take any of these projects that are submitted, and really make them a reality put them on the on the world stage at the global summit, which I'll talk to later. But Maria, you, of course, are pivotal with the XPrize pandemic Alliance, which we formed in partnership with anthem. So and you've seen the results of things like that, where we can create projects and I'm having impact what some of you maybe you can share some of your excitement of what you what your view is to the work of the pandemic Alliance and how that can, can connect to this and and further the work that comes out of this project.
Yeah, thank you, Andrew. Um, the pandemic is actually and I touched on this a little bit, it has surfaced a series of problems that are so pervasive that they expand beyond the resources or the purview of any one single organization or agency or individual For that matter. And so what we what we did with the XPrize foundation back in February was to bring together a group of like minded global leaders who believe that there are certain problems and certain solutions that need to be available at a global level at the global community is almost as a public service. Because these solutions were not necessarily there. They didn't need to be there to solve a market need, they needed to be there to solve for our future of humanity. And so with that, we founded the Express pandemic Alliance. It's a nonprofit voluntary organization at this stage that was intended to do two things. Number one, bring the best and brightest minds to help us so for and find the problems that would be the what I call the fulcrum for solutions. If we solve one of those problems, our it would disproportionately impact our ability to solve for the pandemic overall. And we asked the global coming immunity of infectious disease experts have machine learning experts of former regulators and heads of state. And we asked them which which problems which one, which were these this problems for to for us to focus on. So we created a short list of that some of the thinking that you're seeing today that have made it that has made it to this three themes that we called out is a direct result of the advice that we have gotten from the Alliance experts. Some our problems that come to mind are problems around rapid, reliable, affordable testing, that we could deploy at scale quickly. problems around and we pulled it out already, disproportionate access to resources and issues around kind of data reliability, as we as we called it out. So that was the first problem statement that we wanted to solve for the first mission of the Alliance and the second was really to Surface up opportunities where existing Alliance members are working on solutions to solve for these problems. And we didn't want to duplicate efforts, we wanted to be able to channel resources together collectively to So for that, and in that, in that ethos, we created a Alliance exchange, where members get to submit their ideas or ongoing projects, and they could contribute to each other's projects, and also taking full advantage of the data collaborative as part of the Alliance. And the reason that's important is because it really offers a whole new model of problem solving, and collaboration that is really driven and anchored on the handful of problems that we really need to solve for as a global community.
So thank you for that. I think that's actually a great context to come to really talk about what what can you do now? So thank you, Maria. And as Maria pointed out, I mean, we've seen such impactful results out of the pandemic alliance that knowing the community that surrounds AI for Good. We have high expectations that we can do the same here and have organizations like the Alliance get behind that as well. So how do you get involved? We have a process. So we'd like you can queue up the next slide. We've opened up a web page here, you can see the URL, as well as the QR code for this web page, where you can submit projects, this is open today. So please come join and submit your ideas that the the submission period will be open until August 14. So that's the deadline. And then thereafter, our brain trusts and the experts that we're putting around it will be both reviewing the submissions will be giving opportunities to mentor teams as well. So as you have project ideas, and you need some guidance from some of these very people that are on this On this webinar, as well as the rest of our experts, there'll be great opportunities for that to cultivate the idea and really position it for the the the AI for Good Global Summit. So going into the AI for Good Global Summit, we will pick the top three of those projects submissions, this will be held on September 21, to 25th. It's a virtual event, and and really present them to on a world stage and really give them both the visibility of all those that surround the AI for Good Summit, the guidance of the input from all of our great braintrust members and experts and in the hopes that basically a we can give it the support they need that we can actually launch these initiatives into real projects that have real impact going forward. So really, it's a great opportunity if you can, if you have thoughts that relate to these domains, obviously it's very relevant to what we're all dealing with right now with the pandemic. So it can be both obviously rewarding on many Different levels. I'm sure there may be some additional questions here about either the process or more the point the themes or any of these related topics you'd like to ask of our brain trust members. So to take us through a brief q&a period, I'll introduce Kristin Lawson. Kristin Lawson is with anthem innovation. I'll let her introduce herself. She will lead you through the q&a period. And I just want to thank you for listening to us. And thank you for your time. So Kristin, you want to take it over?
Or Thanks so much, Andrew, and it's nice to meet you all. Thanks for joining us. My name is Kristin Lawson, as Andrew just introduced me, and I'm joining you all from sunny Boston, Massachusetts. I have been looking at the q&a in the chat. So thank you. For those who have posed some questions. If you do have any more, feel free to send them over. We don't get a chance to answer them. Now. We can always follow up with you guys. What so the first question that I want to talk to you guys about and share is how can we create a collective response to pandemics while data is not equally shared in various communities or regions. And this originally was posted in the chat and Amanda had first responded. So I'd like to pass that question over to her. I think Maria did a great job of touching on the pandemic Alliance and how we can share a collective goal and mission. But I know that a lot of the work that the guys speaks to how we're leveraging data. So Amanda, that question would be for you.
Hello, you know, I, I really appreciated this idea of us making use of data that's already available. So one of the things that you know is true is that there's information that's out there, but it's not resourced. Collectively, there's no mechanism for sharing data, or pulling data together in creative ways. The VA absolutely has a lot of really good data we have to be thoughtful and creative and part about protecting privacy, and being thoughtful about how we set those kinds of ideas up. We're very open and interested in solutions. This is absolutely talked about as the brain trust is that we want to take advantage of information that's already out there, and how might be gathered or collected information that's already out there available that could be centralized for more meaningful use by a variety of different people.
That's a great question or a great answer. Excuse me. I'm wondering do any of the other panelists maybe you have any thoughts on that
or any of the other panelists
Okay. All right. So the next question that I, that the that has come through is what are the challenges of deploying artificial intelligence models in a rapidly evolving pandemic? And I think I'd like to pose this one to Steve. Steve, you have any thoughts on sort of how we're managing this from a rapidly evolving pandemic perspective, the challenges of employing artificial intelligence?
Well, I mean, it's a very, very good question, because with artificial intelligence, first of all, I'd like to say artificial intelligence isn't all about machine learning. But many people do associate machine learning models with artificial intelligence. And so if we go in that direction, and we think about it pandemic, pandemic is, such as COVID-19 has little precedence in the shape and form it takes. And so when you start looking for how you're going to leverage artificial intelligence, if you go into machine learning approach, trying to find the data to get a model that's going to be correctly able to give you prediction and give you insight into what's happening is very difficult. And so the benefit of both there is a benefit of having experienced COVID-19 pandemic is that we're now collecting lots of information about how this pandemic evolves, how as we said, some of the maybe the latent variables that may be involved in determining susceptibility of populations, to the virus and the impact on these populations. And these people were able to now if we can wisely get the data and we can collect data that's now becoming historical, and use this to gain future insights, we could probably do a better job of being able to deal with a pandemic as it's evolving. So the ultimate goal would be to, over time learn about how the pandemic is evolving and continually update the knowledge and then use that for your for your prediction and your income. sights. And so that's I think, right now the challenge we're facing in a way artificial intelligence has not been as effective as we'd like it to be in the future is simply there just isn't that much data to rely upon for the current situation that we're in. So if we can change that if we can use this initiative, to think proactively about how we will be able to get these data, you know, I thought about it, you know, this new models, new mode, new modes of using, for instance, federated learning, where you can have some centralized training and you have deployment models to nodes and perhaps, you know, be able to provide a broad benefit to train AI models throughout the world. That'd be fantastic. These are approaches that are evolving now AI is it's you know, it's really moving fairly quickly in technical technological sense. But being able to, to bring the data as we said, here, to this awareness and being able to use that practically, and usefully, I think, is something that we would want to improve upon for the future.
Do any of the other panelists have anything to add there?
Yes, Steve, you said a great did a great job for him, you know. And Thanks, Chris. And I think that, you know, data changes, we know this, right. And so when you have dynamic data, over time, the model is going to tell you information that is relevant for a different period of time. So, you know, strategies to keep up with that to make it. So the insights that we're coming up with are relevant for the time that we're dealing with now, and not solving problems today based on insights from yesterday or last week, because, quite frankly, medications are changing strategies are changing that the demographics of those impacted are changing. I mean, every time we look in the news, it's different, right? So these types of ideas are really important for us to make good informed decisions. So it's a great question and we need ways to rapidly iterate, to rapidly train and test and tune these models in order for us to keep up with it.
Are there any other comments from the panelists? Otherwise we can jump to the next question?
All right. So the next question is How should we manage induced polarization of information on social media? People are following people follow who tend to share the information in the way they like. So if you think about their political affinities, that's what they're following. And that's what they're sharing. I want to turn this one over to Cody. He also commented in the q&a section it looks like he has some thoughts on that.
And this is a core core part of the the third track of this challenge, which is all about how do we help with misinformation spread and ensure that information that is being spread to the public is not Panic inducing. Obviously, within social media, none of us well, I don't know who may be on this call, but I presume very few of us actually control the algorithms that distribute information across social media. So, you know, you have to assume that that's going to be a difficult task to accomplish. But I think that there are potentially other ways to try to solve that. Are there different? You know, bots, you can build inside social media that help reply to sources of misinformation? Are there ways to identify seeds of misinformation before they start to spread so that you can try to head off the public's awareness of these things are bubbling early on. And, you know, really, that's the core of this challenge of this third part of this challenge that we're putting to, you know, everyone on here who is trying to solve these incredibly hard problems are what are some potential solutions here? We don't have the answers necessarily. We just know that this is a Really important problem that needs to be addressed.
Any other thoughts from the panelists? Okay,
we can move on to one of our last couple of questions here. This one has to do with data interoperability. So I'm going to pass this one to Maria. And the question is, how are we looking at data interoperability as a problem in the times of pandemic, or a lot of problems with EMR and EHR being isolated? Also for risk stratification model for any AI algorithm? data is important. And so are the related features. How are we looking at integrated data for modeling?
Hi, everyone. Great question. And thank you, Kristen.
We see data interoperability is actually an opportunity rather than a problem statement. It's a critic. core components for us to be able to get the right data to train the right models on the data so we could have actionable insights out of our analysis. So there's a couple of things that are already in place that allow us to mine data sets with the level of integrity of data and connectivity that allow us to do that. Steve already mentioned federated learning. I think that's a critical breakthrough for the ability to actually rather than move the data around moving the algorithm or the the training of the algorithm and being able to keep the data on let's say, the user's phone and then be able to, you know, be able to learn in a federated way is really what's going to allow us to come up with higher quality algorithms. So, one, that would be one aspect of it. There are industry wide initiatives that you Courage, data sharing and standardization of data protocols for that some organizations that have taken on that mantle, there's a World Health Organization initiative where one of our alliance partners with the commons project, Paul Meyer, has actually taken on that work to make sure that the data that comes from all the COVID triage and COVID risk assessment apps has the same API back end piping, if you will, that allows it to go to a shared repository within the World Health Organization. And so to me, the models are twofold. The strategy here is twofold. One, let's take let's take full advantage of the breakthrough in cryptography in federated learning. There's some very exciting blockchain solutions that allow us to share data in many to many formats rather than by a multiple third parties. So let's take full advantage of what the typical knology could offer us. And the second one is it's truly a governance question. And that's where some of those initiatives that we're seeing, they're trying to make a dent. In the meantime, experts has done a fabulous job putting together a data collaborative. They are immensely national level, at least in the US. Regulations coming up in 2021. Where healthcare players, providers and insurance alike would be mandated to provide data and share data with the members for the entire duration of their of their lifetime as a patient. So lifetime member records is going to be kind of the norm. And so between regulatory conditions coming in and kind of encouragement, if you will, coming in between government governance wide initiatives that are focused on standardization and allowing for for data to be shared, as well as the technology breakthrough. We're seeing a lot of positive movement in that direction.
It's a great answer any of the other panelists have any comments on that?
I think we're about running into at a time but I think Tom, you're going to add one thing, so maybe you can add something and then we'll, we'll wrap up there.
But you're muted.
So this is gonna agree with Maria. I mean, this is a real opportunity. The next wave of impactful insights is going to come from multimodal data from many different sources instead of having one sliver of insight, you have a broader picture and understanding of what the problem is, and some you know, healthcare systems are working on this You know, we happen to you know, have a traditional EHR where we bring data in and it's curated from the clinicians perspective we're working on and others are working on ways to bring in information from a patient's perspective, self reporting, wearable data, other data from social media will be difficult. But that's a really important thing to bring in as well to have a better holistic understanding of what's going on. So it's a fantastic question. And I think it's really important moving forward.
I think it's, that's also obviously a good, good call to the to the crowd here to see we're open to your ideas and your submission. So to take that forward, and really touch on one of the last questions that came in the QA was really about how do you get involved. So there is that form that we put up there actually, it you if you could quickly flash that chart which has that URL again, and, you know, or go to the AI for Good web page and you'll be able to navigate to there, but please submit your projects is a great opportunity to get them to get guidance. them to give them focus etc. And, and when any other way that you want to get involved. So, so please use that. with that. I think we'll we'll hand it back to Ida at ITU, I want to again, thank you for participating thank, of course, all of our brain trust for taking the time to get us to this point. And and Kristin as well for help on on the moderation. So either. Thank you.
Thanks, Andrew. Can you hear me okay? This time? Okay. Thank you very much, Andrew, and for everyone joining today. I think we're all looking forward to seeing the projects that come out of this and hearing about everyone's great ideas. So before we wrap up, I'd like to highlight a few things that may be of interest to you. Together with WHO the World Health Organization iq for the past two years, has been convening the focus group on AI for health. This is an effort to create a benchmarking framework for the accuracy of AI for health diagnostic aids participation is open to everyone. And we are looking for a wide range of experts to work on this with us. Experts from topic areas such as retinopathy, tuberculosis, and many others. You can find a link to the Focus Group and AI for health in the chat for more information. Now for upcoming webinars, we are excited to host a number of them this week, as well. So this Thursday the 24th of July, we will have the latest installment of the global dialogue on eSports. And then the next day on Friday, the 25th of July is the innovation factory is live pitching session, so be sure to tune in. And again, we have information in the chat. One other very cool thing I'd like to point out is another iq Focus Group and this is on AI and machine learning in 5g networks challenge and this is a great way to get involved if this is your area of expertise. So tomorrow, Wednesday, 22nd of July, Dr. Sally are good from Turkcell. We'll be here virtually explaining the problem statement on predicting radiation Failure using weather information and network data. And then the next day on Friday the 24th of July, yes, or 25th of July sorry, we will have Professor law from the Indian Institute of Technology, explaining three problem statements on the use of AI tools to build computer vision techniques for providing immerse and assistive services in telecommunication. We are also posting the links to these in the chat and you can find out more about any of the things that we have mentioned over the past hour or so on AI for Good dot ITU dot IMT. And with that, we have reached the end of the webinar and would like to thank everybody involved for making it a success for the panelists, the participants, our partners, sponsors and our co convener Switzerland. Thank you very much and we hope to see you again this week.