 a little bit differently because this is the last panel. We're going to put the last polling question up now. So if I could put the last polling question up and then you'll have the 30 minutes during the panel to think about this. So we're hoping this time you get the right answer. So the question is, what is the biggest barrier to widespread use of real-world evidence in biopharma, lack of data standards, interoperability of data collection systems, data science talent in biopharma or concerns over data security? So this will be up for the next 30 minutes. So we can now come back. So it's a great privilege to actually have been displaced because typically I've sat on this R&D head panel now for many, many years. As many as I've been in R&D head, which I'm now in my seventh year and it's always been the Hallmark event, the last event of the USAIC summit, but no longer data and digital sciences have taken over and they're now the event that will take us home to the finish line. It's a great panel. I'll introduce the moderator and then the moderator, Sashtri Chilakuri will introduce the panel. So Sashtri is a rare intellect. He spent his early part of his career at McKinsey where he was helping companies figure out what their digital and data sciences strategy might be. And then he realized, hey, I've got a lot of really great ideas. So he started his own company, a company known as Acorn AI. Is that the name of the company? Yeah. And then you've been serially acquired, which is a great testament to how successful you are first to metadata and then to the salt systems. So Sashtri, I know you have a really exciting panel plan. So without further ado, I'll kick it over to you. Great. Thank you so much, Andy. Really excited to be here to talk about digital and data science. And we have an illustrious panel. The panel needs no introduction. We have Rob Califf, Daphne Kohler, Karen Najat, as well as Craig. So why don't I kick it off, Rob, with you? It's been an incredible year for many of us over the last 12 months where we've seen amazing innovation. It's a miracle that we have a vaccine that all of us are functioning as a society back again. And we've seen tremendous advances both in the use of digital and AI. However, it's also highlighted tremendous amounts of inequality as well as opportunity in terms of the broad healthcare and the opportunity in front of us. Why don't you frame it for us in terms of what do you see as the greatest areas of opportunity as well as the greatest areas where we've fallen short? Well, thanks, Senate. It's great to be on with a panel of real experts. I'm sort of a middleman of data science. I'm not a real data scientist myself, but I've always worked closely with data scientists whom I admire. And by the way, it was exciting to hear the heads of R&D all agree that their companies will just be data science organizations in the future. That is a real area of opportunity, I would say if the leaders of those groups say that. But I think while we're hopefully coming out of the pandemic now, and certainly many of us are feeling better about the direction of things, in many ways the pandemic just shown a very bright magnifying light on the problems that existed in our society with our healthcare systems and with the way that our pharmaceutical industry works and serves society. We're in the midst of an onslaught of common chronic disease at the same time we're having a pandemic. In fact, in the US, more people died of heart disease and cancer than of COVID-19 during the same timeframe last year. And this was on top of a steady decline in life expectancy in the United States over the last seven years, five out of the last seven years we've seen a decline in life expectancy. And this is not because of people like us on this call. We really live in the US at least in eight Americas and I think around the world there's a very similar set of issues going on where health and biology intersect with social determinants, which are dominant over biology in most aspects of health and longevity. And so I know the panelists will talk about drug discovery where in my view, it couldn't look better. The ability now to use digital technology and the amazing aspects of computing to look at systems biology and to refine targets and to find new treatments is truly profound. But we have other issues that we need to attend to. There's been a lot of discussion at this meeting about more efficient clinical trials but I also wanna say more efficient does not necessarily mean smaller. If the purpose of a clinical trial is to inform what should be done in clinical practice and maybe that we need larger clinical trials just done in the right places in the right way without wasting time and energy on, for example, manual human data collection that doesn't add anything to the assessment of benefit and risk. And of course, the biggest part of the use of pharmaceuticals is a post market phase which in many cases I learned as commissioner goes on for 50 or 60 years with new discoveries yet to be made. And so I think the intersection of digital information in the post market phase is going to be a huge area in both drugs and devices. And the one thought I have about the big opportunity I think we have which I hope Alphabet my current employer will be involved in is a creation of essentially a COVID tracker for all of diseases. I don't know about you guys but most of the last 15 months I've gotten up every morning and looked at the COVID tracker to see what's going on in the world, in the United States, in the state I happen to be in, in the county I'm in and even down to the local hospital with up to date information fed in from a variety of sources. And knowing now that just having a good drug is not enough we actually need to deliver the drug to the people who need it in a way they can afford it and in a way that's convenient. This ability of modern computing and digital technology to look at the intersection of health, biology, behavior, social determinants and even the economic wellbeing of local areas related to that I think it's gonna be the biggest area for us to look at. Right, thanks Rob. Why don't I turn it over now to Daphne and Karen. What you work on is extremely exciting and when I talk to both of you you talk about how this using deep learning and data science on high resolution data can allow us to address a lot of these challenges that Rob was posing. But at the same time there are challenges and what I'd love to get from you both in terms of what do you see as the greatest opportunities and challenges? And then the second piece is I use this example often of the pole vaulting curve and there are large periods of time where the pole vaulting curve didn't really move that much. And then there was suddenly a breakthrough be it in the use of the materials that they were using or a new athlete came along and then they saw a step change. So what do you see as those technologies that are right now poised for this breakthrough that'll allow us to be able to achieve this step change in productivity and innovation? Daphne maybe we start with you and then Karen maybe you can chime in. Great, thank you Shostri. So I think the exciting thing about the moment in time in which we find ourselves is that I think there's opportunities throughout the pharma R&D value chain. I sometimes hear people talking about machine learning and data science as a point technology kind of like X-ray crystallography perhaps that might improve a single step in the pharma R&D value chain. And what I think is really remarkable about this technology is that it can help in so many different ways all the way from a better characterization of human disease where I think the current characterization that we have is based on very coarse grained clinical phenomenology that really doesn't capture a lot of the underlying biological mechanisms. Oncology is one of the areas where we've moved away using our understanding of genetics so much finer grained, more mechanistic understanding of sub types of cancer that are driven by certain molecular mechanisms and that's created an incredible innovation in the kinds of treatments that we can provide to patients. We haven't done that across the board in multiple other diseases that are not oncology and I think that's a tremendous opportunity but that's just the beginning if you will of the pharmaceutical value chain. We've seen tremendous advances in accelerating and de-risking the design of novel therapeutic novel chemical matter. We've seen and also heard from Rob about better clinical trial design whether it's the selection of better clinical sites on the selection of patients that are more likely to respond or in better tracking of those patients so as to determine which of them are actually benefiting in a much more granular and quantitative ways then let's write some stuff down in a notebook that has characterized some of the earlier ways in which I think some clinical trials are still designed and then over time I think there's and we'll hear from some of the other panelists I know about opportunities in manufacturing and supply chains and even in sales all of which I think can be improved by this incredible technology that we have available to us. So I think those are some of the opportunities we can certainly go deeper on all of them but you also asked about the risks and I'm going to highlight two of those that I think are really important. The first and foremost I think is data both the quality of the data and the biases in the data that are often very subtle and easy to miss. In some ways modern day powerful machine learning is very much a two-edged sword in the sense that well while it's really incredibly good at picking up on subtle signal it's equally good at picking up on subtle artifact and that artifact can really make the models highly misleading in ways that are both biologically implausible and relevant or in some cases we are hearing more and more about are biased in the context of particular racial bias or gender bias and again it's really important to try and address that and the only way to really do that is by having higher quality data and this is really the epitome of garbage in garbage out. The other big risk I think in the context of delivering on the promise here is the lack of availability of people. There's already this huge sucking sound from so many industries tech but also more broadly in terms of qualified professionals in AI and machine learning it is becoming increasingly hard to hire people with that skill set and for us in the biopharma industry it's not just any machine learning person who will do we need someone who has the capability to speak across the chasm between the technology world and the life science world and not everyone is interested in putting in the effort that it would take to really understand enough about human biology or chemistry to have a meaningful dialogue with one's counterparts in the other discipline. So I think this is a place where the biopharma industry really needs to make a big effort not only to attract the right kind of people but also to create a culture that can make them be as productive and as impactful as they can be because a lot of times people with that skill set find themselves relegated to a little corner where they're like the tech corner and that doesn't let them really have the impact that they can and also doesn't make it a particularly attractive place for them. So those are to me the big opportunities which are many and also two of the biggest challenges. Karen, I can't hear you Karen, can the others hear her? Looks like I was still on the, sorry about that. Yeah. Can you hear me now? Yes, we can. Yeah, I mean, I think one of the things that we have to say that it's become really obvious during this pandemic is that really the amount of data that's available to us, the area of data that's available to us as Daphne and others said not just in drug discovery but throughout the entire value chain and post market is vast. The area is huge but the perimeter of our ability as individual scientists to analyze that data has really kind of gone beyond what any individual or any team can do. And I think the promise really of data science and that combined with machine learning is and I often rename AI which is currently going through some sort of height curve right now. It's really about augmented intelligence, accelerated insights that help us to make better decisions. So that's really the opportunity that I think is in front of us. I think what we've also spoken about today is a sort of impatience that we all now have especially again, seeing what we've been able to do as an industry over the last couple of years with COVID there's an impatience. Why can't we do more of this accelerated understanding, accelerated hypothesis testing around unmedical need? And so the area that I'm focused on on a daily basis is not technology for technology's sake but technology that allows us to ask more questions, to test more therapeutic hypotheses. As Andy said, I think really we need to be a little bit more unconstrained in taking on some of these technologies to empower us to speed up the process and to change what we have been doing for decades. So the opportunity again, I think is embracing these fundamentally validated approaches again because I think we live in an age where AI seems to be touching everything. What are those validated approaches that really allow us to accurately test hypotheses? And I think the combination of, for example, physics-based methods that are highly accurate combining those with machine learning gives us an opportunity to sample, in this case, chemical space. And I'll just say that when it comes to real world evidence, historically we've been using subsets of chemical space captured in a high throughput screen or in a fixed library. We now have the ability to interrogate in a real world sense all of chemical matter to find the molecule that has the best properties to go ahead and test that therapeutic hypothesis. That's very exciting. I think we're seeing more and more people embrace that. On the challenge side, I think we're seeing the use of these terms AI as a panacea that's sort of gonna solve for some very complex things that frankly we don't really understand. And I think Daphne touched on this, really getting to the fundamentals of understanding biology so that we can go ahead and raise a hypothesis and then go ahead and test it. I mean, those are, we're really at the very beginning of that. So I think that there are many challenges in sort of knowing where these technologies can be applied with a level of accuracy and a level of validation that allows us to move forward with confidence and balancing that with the unfortunate, I think excitement and sort of rush of investment into an area that really is not a panacea. So that's one of the challenges. I think the other challenge is that we don't have all of the substrate information. We don't really understand human biology. We need to carefully apply these technologies in a way that allows us to keep asking questions, keep figuring out if we're on the right path. And in the same way, we need to essentially not sort of be tunnel vision about one approach of doing things but trying a lot of these different approaches. We now have so many modalities to go after. Gene therapy, biologics, CRISPR, mRNA, small molecules, AI and machine learning can be applied across many of these but there are some places where it works and there are some places where fundamentally it does not work. I think we have to own those areas that where this really isn't making a huge impact and find out how to sort of reconfigure our energies that we're not wasting time and sort of own up to the fact that some of this is not really having a meaningful impact. Daphne touched on training. I do think we're in a period where we're redefining what is a team? A team I think traditionally was comprised of people with very similar training, perhaps a team of people who were interested in therapeutic hypotheses working quite separately from people who speak the language of data science and machine learning. I think we're at a point now where team is defined as truly multidisciplinary, touching on physics, math, computation, biological insight, therapeutic and clinical insight, bringing that team of people together at the very beginning of programs, I think has the potential to really unleash our understanding, the types of questions we're asking. And so I would say that one of the challenges we have is in reconfiguring what these teams look like in very traditional environments. I think it's much easier in smaller companies. I think we're getting to the place now and I'm seeing a lot of this obviously amongst my colleagues that we're redefining team in the moment now. Again, partly because of what's gone on with COVID, but also broadly across by a farmer embracing these different disciplines and bringing people together early on. Yeah. So Najat, the question for you is how do you take these bleeding edge capabilities that the panel has been talking about and wire it into a 130-year-old organization like Johnson & Johnson with 10,000 plus researchers in the R&D department? I would maybe frame the question a little bit. How do we transform the discovery and development of medicines by actually advancing the technologies versus accepting the technologies? Because I think the reality of it is anywhere in the ecosystem we're all still figuring out how to do this well. And I think Andy said that well earlier on, really around innovation and you have to accept some of the things that are going to fail and some things that will accept that will succeed. So how do you take those calculated risks? So in order to actually embed this, I would say there's two main components, right? One is really around, everybody likes to see that there are wins. If you have wins at the end of the day, people can fall behind that, come behind that. But what does a win, what does success actually look like in a large pharmaceutical company? It's all about the pipeline. At the end of the day, if you make a demonstrable impact on medicines and programs, that fundamentally is what will have impact on the portfolio. So I'll talk through a few of the examples there and then I'll get to another aspect, which is the how, which is the people, some really great topics that have been already raised. So as Rob was mentioning, one of the aspects is how do we find the right patients? How do we ensure that they're the right segments that we're focused on? So just a few examples, we're looking at, let's say in oncology, I think Daphne mentioned as we're looking at more precision medicine, genetic mutations, these patients by definition of precision medicine are rare. So we're leveraging imaging. How do we actually use imaging of different modalities, whether it's radiomics or pathology, histopath images, to actually be able to identify which patients have certain mutations and then not just publish a paper on it, but actually require it in trials and to the real world. That requires a complete reconfiguring to Karen's point of how we work. The second aspect's example for COVID-19, I know there was a lot of conversation about the good and the bad. And when we started on the journey with the COVID-19 vaccine program, data science and digital was central to what we did and it's kudos to the need of 15 for doing that, because it's tough to actually take recommendations on, I think Rob mentioned this, not making the trial smaller, but the making the trial better. And for us, better meant, where exactly do we go? I mean, the reality is it takes us time to set up sites. There's regulatory hurdles, but how do we use not just to Karen's point, traditional real world data, but also a lot of social data, right? Like people, like at the end of the different modalities of data sets to be able to predict social compliance and where you might see hotspots of COVID. And that's what we used to really be able to precisely go to the right locations that accelerated the timelines of the trial. But also another point, which is around diversity. We factored in, you know, over indexing on areas of high diversity, African-American, Latinx, but also socioeconomic status makes our trials better. We had a very diverse, more so than we've ever had trial. And the other thing I'll say, I think Rob also just pulling the thread there mentioned about long-term outcomes. How do we measure long-term outcomes of patient sub-population? We do that with registries and post-marketing commitments. What about the patient or participant sessions that we have in our trials? We actually, actually some of the work was done with Shostri State, worked on creating a concept called tokenization where you can actually connect real world data to clinical trial data. So actually being able to better understand the pre and post of a patient's longitude and long journey to then be able to follow through the outcomes of actually participants in our trial. And each of these examples, as I mentioned, they've had significant impact in terms of how we do the story and the development of our programs. And, you know, I have to say, it couldn't be done just by the data science team. And this is going back to the reconfiguration of the team. It had to be done in an integrated fashion, having the data scientists with an equal voice around a clinical development team, take the current workflows, but ensure that data science is not only a voice, but is around the table for the decision-making very early on. For COVID, we started from day one. And for a lot of our other programs, that is what we do systematically. So, Chastri, to your question, it's hard to have that sort of impact because you're in large companies and it's debt by pilots. You want to make sure it's systematic and you're changing everything from the SOPs to the team structure. And then from the top down, right? When we have development committee, we actually look at what's the data science strategy. So it comes from many different angles of how do we impact change. The one other thing I do want to say that is really important is a piece around talent and bilingual data science talents. The talent that understands data science and Karen beautifully mentioned a lot of the difference of disciplines, but then also understand clinical development, understand chemistry and biology. That overlap in this amount of time inherently is hard to find because it's in the space. So how do we develop our right teams? But the reason why, and I loved what Daphne said, the sounding sound of talent going away, the way we keep our talent in the bio-pharma space is by focusing on what is the question we're trying to answer. It's for a program, Parkinson's disease program, or lung cancer, really being able to resonate to the impact that you can have by embedding data scientists in the right way. And maybe just one last comment. I think in general, large companies struggle with the fact that data sets are fragmented across the board. It's not consolidated in one place. How are we actually using the data? Is there the right appropriate distribution in the data set to avoid bias that Daphne thinks is a huge risk that we always have to look for? How do we do that? One of the first things when I did this rule a little over a year ago was I ensured that across just in order to pre-clinical, clinical and wearable data sets are connected, there is one place where all the models are being developed. And you know, this is a conversation that Anne Heatherington and I have a lot in terms of how do we integrate all of these data sets? But that has been really, really important to be able to feel the work we're doing. So there's talent, there is data, and using data to stewardship in the right way. But then also partnerships. We've done a lot of different partnerships because at the end of the day, and this is how we started, technological advancements live outside and inside. And that's true for R&D broadly, not just data science. And we need to keep leveraging that, especially in a space like data science and digital, which is earlier evolving faster. So I think that's what I would say in terms of sort of what's been our journey. It's both having the impact and wins, but then also some of those in nature. Great, I'll keep us moving here. And Craig, you sit at the intersection of the two most important topics over the last 18 months. Therapeutic development and supply chain disruptions. So give us your perspective from your role in supply chain in terms of the role AI played in terms of keeping the service levels up as well as ensuring that we didn't face drug shortages as a society. Yeah, thanks Astri. And I'll try to be succinct in my remarks. I want to leverage something that Daphne mentioned earlier, which I think is very important for everybody to keep in mind. We can't glorify just the notions of artificial intelligence and machine learning by themselves. We have to recognize that these are additional methods that we have because of computing power capability now and algorithmic capability that need to be directed at specific real world problems, whether they're for access development, supply chain or whatever it might be. We certainly saw those problems in abundance during the course of last year. And I would say, as some of my other colleagues have pointed out, these are problems that actually occur chronically all of the time in supply chain and manufacturing in general and specifically last year. But we saw plenty of opportunities last year and plenty of cases where artificial intelligence was able to respond to dramatically different situations and demands quickly. There are many circumstances where the normal transportation methods that we would use downstream of manufacturing became unavailable to us. And we relied on pretty significantly complex and well-trained machine learning to redirect shipments where we needed to be much clearer about where product actually needed to get to and how it might get there. We also saw plenty of opportunity using the same set of tools to make sure that we were able to protect the product for patients as it extended around the world. This became very true for products that required cold chain shipment where you needed to be very, very clear that you could maintain the temperature conditions all the way to the destination, even when there were disruptions to the regular supply chains. We've also seen plenty of opportunities in the upstream of manufactured portion of the supply chain as well, where image recognition technologies have accelerated the ability to do biological assay analysis and also the ability to just improve productivity for manufacturing where you've got certain circumstances where the demands just far outstrip our normal capabilities. I think finally what I would say just in this space too is beyond artificial intelligence and machine learning, there are other technologies including technologies like blockchain and decentralized databases that have helped us also provide a greater degree of security and surety for product as it goes out into the market and has allowed us to make sure that the product that we promise is the product that is actually delivered. So there are many, many opportunities that we've come to be able to capitalize on as a result of newer and more available technologies now. And as some of my other panelists colleagues have said, those problems and those opportunities are only now more fully available to us than they have been previously. And I'm quite confident that we're going to see a new expansive use of these kinds of technologies to solve problems that otherwise haven't been readily solved for us. Great. Thank you. At this time, why don't I bring in Chris Banco and get the audience question in. Chris, are you on? I am. Saustra, can you hear me in signal? Yep, we can see you. Go ahead, Chris. Guys, this was a really, really great panel and I love the balance of experience that's represented here. Anjad, I thought you spoke really eloquently about something that's near and dear to my heart, which is the collaboration of data science and let's call it the more traditional engine of pharma and how important that is. And I know, for example, for you and Craig, you have senior executive positions in a big pharma where you're trying to drive that kind of collaboration. Daphne, I'm hopeful that as a CEO like me, you just tell people to do it and it happens because you got your own company. But I'm curious particularly to hear from Karen and Rob. You are experienced biomedical executives who work at what are predominantly tech companies. What is it like getting pharma to the table to be a partner with a more traditional technology and data player? And is that something that is happening seamlessly or there in particular frictions or lessons learned about how you blend the mindset of a company that's built on technology and data with a more traditional industry that's still figuring out how to utilize it? Yeah, Chris, great to see you. I can start. I would say that a little bit like I mentioned before, I think that when technology is the first and last word in the conversation, it's very hard to focus on what really biopharma needs out of these technologies. And so I'd say that there clearly is some friction, but I think there's also a lot of enthusiasm to sort of shift, if you like, the outcomes that we're all seeing. And I think Najad touched on this. It's really about the outcomes. How can these technologies accelerate what we're all doing? How can these technologies move the needle in terms of, for example, the quality of chemical matter that we're able to find with balanced properties? How can these technologies accelerate our insights? And I think when you focus on these outcome-based discussions, testing therapeutic hypotheses, coming up with molecules that everyone can get behind, I think that a lot of that friction goes away. I think what I've certainly seen since joining Schrodinger, where I'm now based, having come from a large farmer, is that there's an excitement about the idea that we can work together with these technologies in companies where, frankly, they have a lot of time to dedicate to sort of fixing the kinks, embracing the cloud, working through all of the sort of bathroom operational stuff that frankly I don't think is necessarily the bread and butter of a large farmer company. That's all being done sort of by the technology company. But I think where the conversation really takes off is around this idea of jointly reaching outcomes that I think we're all striving for, which is testing more of these therapeutic opportunities and accelerating our way to molecules that allow us to do that no matter what the modality. So I do think there are challenges still because at some level, and I think others have said it, at some level you're speaking very different languages, but there is a common language and that common language is about how do we get to these medicines faster? Well, Karen is an optimist and I tend to gravitate towards the problems. I'm actually an optimist, but I do gravitate towards the problems because I think they're more fun to try to solve. And I would just say there's no lack of enthusiasm at the higher levels because as evidenced by this meeting, everyone's excited about the potential of tech and biomedicine creating something that's new and different in as these different types of entities interface. And I'd say a lot of progress has been made sort of in the backroom areas as Karen referred to. And I'm pretty sure that the massive amount of funding going into early drug development is represented by people like Daphne in her company. It's gonna pay off in the long run. Right now though, I would say the culture of a tech company and the culture of a biomedical company, I don't know of anyone who's actually figured it out. And when it gets down to the actual working project, there are a lot of misfits. And I like when Karen said we're speaking different languages and I was taken aback at first and then I realized it was right. Amy Abernathy is just joining us that barely. And when I asked her what do you think is most important, she listed having a glossary as one of the most important issues in getting a tech company and a biomedical company to work together because of flat iron. She said she learned that people will use a term and they mean entirely different things in the two industries. And there's also the fact that in pure consumer tech, the project manager, the product manager and the lead engineer run the show. The executives like me, I mean, I came out of being FDA commissioner, I arrived at Verily, I might as well have been the food service provider because there was no real recognition that that had much importance in the tech world. And to a large extent, the executives are there to make sure that the food service is there and that the computers work and that sort of thing. Whereas the biomedical industry is much more hierarchical in terms of recognition of experience and perhaps the wisdom that comes from it. So we've got to work through these issues. I'm sure we'll get there. But if you look at the intersection of big tech and biomedicine right now, the turnover rate of the biomedical people is really high because the cultures are quite matching. And then I'll just mention in closing, one real big success, Najat referred to and we were partially involved in this massive effort to identify, in the setting of a pandemic, how do you find the right site at the right time where the pandemic is headed? And that wasn't all Google, but we participated along with the J&J data scientist and MIT data scientist. And I think it really works. So that was a good example. If I can just build on that really quickly, I think Rob is right. It's a journey and Karen, sort of beautifully, very different languages. But the minute, when we started, for instance, the example that Rob mentioned, the clinical team versus the operations team versus the data science team spoke completely different languages. But I think it is once you go through the storming and then hopefully there's a little bit of norming to see what are the outcomes. Because how do you put trust in something that has never been used before? How has this been found? These are all of the questions that come through. But once you actually get one of those examples, it really does create a level of respect for the different disciplines. And I think we're not fully there yet. But I can't think of anything other than some of the interesting examples that actually create that. And then being able, and then just one comment I wanna make, that a lot of the times in large pharma companies, when one area recognizes that something was done in a different way to make the trial better, there's a lot of pull from other aspects and other areas too. So this is true human behavior anywhere. And definitely good to leverage that in this state. Fantastic. I think we're on time here and we're encroaching on Andy's closing remarks clock. But thank you so much to the panel for your time and your insights. It was really excellent to talk to all of you. I look forward to meeting together in person next year. We're probably all at somewhere near the peak of the hype curve around AI and machine learning. So it'll be good to get together next year in person and compare notes, both in terms of the new successes as well as in terms of the next set of challenges that we've all uncovered. With that, why don't I turn it over to Andy for the closing session. Thank you very much, Sastry. And thanks to all of the panelists. We'll hope to see you back next year as we continue along this amazing journey of integration of tech and healthcare. So before I bring this 15th annual summit to a close, let's look at the polling results from the last question, please. We can get them up on screen or maybe not. All right. Well, I'll start with my profound closing comments and then at some point in the middle. They are there, Andy, it's there. Oh, I don't see them actually. Are they up on the screen, Karuna? I don't have time. Maybe you can bring the other screen. Oh, there we go. Oh, there we go. Good, good, good. Okay, so the question was, what is the biggest barrier to widespread use of real-world evidence in biopharma? And we had semi-convincing, though probably not statistically significant, playing through the data science panel answer, which was the interoperability of data collection systems. And I guess in a relatively close second was lack of data standards. So thank you all for participating in these polls. I think Karuna will be published in BioCentury if I'm correct. Yeah, they are going to write it and plus we are going to put on the website also. Good, good. So you'll have a chance to make an impact beyond the setting here.