 It's a great privilege to now introduce the next panel. This is a panel that's now happened year on year for the last five years, reflecting the trends in our industry around data, digital, and technology. It's an incredible group of panelists that I'll let our wonderful moderator, Najat Khan, introduce. Just briefly, Najat Khan is a figurehead and a leader in the field. She's the chief data and science officer at Janssen. And she also leads the strategy and operations group. And one of the things that I've noticed about Najat as I've been on the circuit with her in various places is how much energy she brings and how patient focused she is. So Najat, with that said, I will hand it over to you. Thank you so much for the very kind introduction. Thank you, Karin, for having us all here. As Angie was saying, we've had this panel for a few years now, and it's really a testament to the fact that we are moving from ideas and concepts of what we can do with data science to real impact that's happening. Now, the advent of General Data AI has certainly helped propel a lot of what's been happening as dinner table conversations, which is fantastic. But we're going to focus on health care today. And my esteemed panelists here have actually looked at it from many different angles, from industry to academia, and then also from all the way from discovery, the inception of understanding what's driving a disease, which we frankly have a lot more to do and work on. The work is definitely not done to clinical trial execution and then ensuring that there's access to these really important medicines. And why is that important? Because we serve our patients and every single day there are patients that are waiting. And if we can improve the probability of success that something becomes a medicine and make it better and faster, I think we have all served our core purpose. With that, what I will do is just set the context up. My philosophy would be a lot of the times people have asked, is it hyper-hope? And it's neither. It's just real. And to be competitive, and this is what we need to do. The question is, how do we do it really well? How do we collaborate even better to ensure that we have impact at scale? So with that, I'm going to start with my first question. And it's for each of the panelists here. A couple of minutes to answer. First, please share sort of an example where you have solved a significant problem or are solving a significant problem that would help make a medicine, transformational medicine for patients. So it can be a discovery, development, you pick. And then another example that we're not working on yet today, because it's not prime time, but you anticipate will be prime time in the next few years. I'm going to start with Alice. Hi, thank you. So I'm Alice, I'm the CEO and co-founder of Virginomics. We're using AI to develop new drugs for these kind of huge unmet needs, like ALS Parkinson's disease. And in my view, one of the reasons I started the company was because I think one of the biggest limitations right now, especially in applications of AI is really the input data. We've seen huge advances in AI over the last 5, 10 years. But the fact of the matter is that the cost of R&D still continues to increase. And so what's the gap here? And I think what I see is that it's really doable to generate large quantities of data from cell models. But ultimately, cell models aren't predictive of success in humans. And so I believe that in biology, the biggest problem is really a missing data problem. So what we've been doing is instead of starting with cell models, we actually go directly to humans. We have a massive collection of human brains, human tissue from patients that have directly passed away from disease. And this has allowed us to quickly identify new mechanisms in ALS and also a new target in ALS that we discovered entirely from a platform and developed a clinical trials in just four years. So that's actually a new treatment for ALS that's currently in clinical trials. It's in a phase one trial based entirely on human data and based using our AI algorithms. And I think that's just such a promising example of how we can address this fundamental challenge of using AI, not just to test more drugs faster, but really to fundamentally think about how can we increase clinical success? Because even an incremental change there would have massive implications on drug development writ large. And I think our fundamental belief is really the sophistication of your algorithms and AI on its own is insufficient if you don't have the right data to train and learn from. So that's kind of our hope at verge is that by really building the right datasets that are truly predictive, we can get to better outcomes not just faster, not just more, but higher quality outcomes. That's such a great point, Alice. I mean, it starts with understanding what's driving the disease. So your point around novel insights is going to change the way is so, so important. All right, great. We'll go to Anne next. Thank you, Najat. And I'm Anne Hetherington. I'm head of the Data Sciences Institute and within our own data, Takeda. And it's an honor to be here with my fellow panelists. So I guess the example I'd like to give is going back to patient centricity, which is there are many diseases where it's actually really hard to really understand the disease and monitor the disease and the patient. If you think about the brain and the intestine, it's quite hard to get inside. And I know a couple of people in this panel are actually helping us in this space as well. And so with Takeda, we're developing digital biomarkers and tools to help us get inside these organs. So for instance, for Celiac disease, we are using an old technology, a Pil-Cam. But what we've done is we use that to photograph and create video of thousands of thousands of frames along the length of the intestine and then use AI algorithms to read those frames, understand the extent of the impact of Celiac disease along the whole length of the intestine. And therefore we can monitor the impact of any subsequent treatment on those patients and really help us really connect the pathology of the intestine to the Celiac symptoms that patients generally have to talk about. The other organ we're getting inside is the brain and particularly in sleep disorders where we're using novel AI ML techniques to monitor EEG outputs to really help look at sleep staging and diagnosis of narcolepsy and thinking about wakefulness, quality of wakefulness during the day and then sleep staging at night. And we've a group of data scientists and statisticians that are developing multiple different algorithms here to allow us to understand how disorders like narcolepsy really impact patients. And so a bit like Alice described, changes in how we get at medicines, we want to really understand our patients and really develop drugs that impact their disease through a deep understanding of their disease. Thank you, Anne. It's about how you actually connected the pieces around understanding what's driving the disease, but how do we measure it to determine are the outcomes getting better or even stratifying patients bringing precision medicine to reality for all diseases, right? So now I'm going to go to Dina because I know Dina does some great cutting edge work in terms of not just digital endpoints, but how to bring, you know, vision medicine and the patient centricity with devices and solutions that are much, much easier for patients and allow decentralized trials. So Dina, over to you. Yeah, thanks, Nisha. So Dina Kachabi, a professor at MIT and also the president and CEO of Emerald Innovations. So Nisha, at the end of the morning, actually I asked Joaquin about basically why clinical trials still like 90% of the drugs don't make it. And I think that is very, very important question that we should ask. And this is where I think AI and data science, digital technology can help a lot. So I really very much like what both Ann and Alice were saying. Particularly Alice mentioned something about going to the human, understanding diseases and target in the humans. And then you take this, these new drugs to the clinical trial. And now what we need is particularly in diseases that are complex, like in neurology, like ALS, like Parkinson's, Alzheimer's, we need to be able to decide very quickly that we are on the right path or not, that we are seeing and creating through digital biomarker. I think here play very, very important role. Can we get very early on, very quickly, insights whether these drugs are able to provide statistical significance between the arm of the drug and the alternative arms. And this is actually where we have proven recently to publication in Nature Medicine and Science Translational Medicine. The example that we use is Parkinson's, but I think what we did extends beyond Parkinson's to other diseases like Alzheimer's, ALS, and others. But basically that we can use our data to very quickly reduce the sample size by an order of magnitude to reduce the time of the trial and quickly be able to give statistical significance on differentiating the difference arm of in Parkinson's clinical trials. So I think we need to move to the human as soon as possible. And this is what we are seeing increasingly. And we need to be able to provide statistical significance on optimizing the portfolio even before thinking about like, okay, taking it to the FDA because today only 10% goes to the FDA. Yeah. And that's such an important point because if you can make that go-no-go decision with the right methodology earlier on, you really change the trajectory in terms of the probability of success. So Dina, that's a small sample size. Exactly. Smaller, more smarter, efficient trials. All right. Next, I'm gonna go to Junaid. You know, Junaid tell us, I mean, you've been hearing the various examples, but would love to hear your perspective from the Microsoft tech side and also being a physician yourself, a little bit about the examples that you're seeing that are making an impact. Thanks, Nijia. Always difficult to follow such clever people who've come up with all the right answers before you. So I'm Junaid Bajo, I'm a physician by training background and the chief medical scientist at Microsoft Research. So as I thought about your question Nijia, I think it's interesting how we think about both the hope and the hype and how do we make it real? There's been so many potential options for what we could do, so many papers that have been written, but I think it still requires us collectively to take responsibility to how we think about real world impact as a consequence of some of these tools, technologies. And there's huge opportunity across the entire farmer value chain, right? From discovery to development all the way to commercialization and what we need to exist in the real world. From a data perspective, I think we recognize the unlocked potential that we have across all types of data. So what we have in electronic medical records, what we have in social determinants of health are genomic profiles, our individual behaviors and others, which is beginning to allow us to unlock novel digital biomarkers in addition to the biological markers that we already consider and have access to today. And the things that excite me on the discovery side is when you look at ecosystem players and others who are doing some really novel work in optimizing small molecule drug candidates today and bringing that actually into the fore. So you look at the work that generate biomedicines are doing or you look at the work that excite are doing and you think what needs to be true and how do we accelerate some of that work and bring that to life. And they work through a partnership mechanism with other partners of course as well. And the work that we're doing at Microsoft is truly to think about what do we do and what must we do to empower the chemists and organizations, the trialists and organizations and colleagues across the farmer value chain to unlock the potential that they have with their data to ultimately deliver value to the beneficiaries, the patients, public and society for who need faster, cheaper, more specific medicines in the future. Great, Junaid, I mean, end to end, how do you enable to do this better? Now, moving on to Shastri. Shastri, tell us a little bit about your perspective. The end-to-end platform, the ability to do all of these great examples that Dina and Alice and Junaid mentioned at scale in a reliable, transparent and understandable way is super important. So we'd love to hear your perspectives on that. Yeah, absolutely. If I can also answer the question about where we've seen the successes and where we see the opportunities, if we've been talking about data and digital and analytics for more than a decade in life sciences in a pretty serious way. And I think the biggest areas that we've now recorded successes across the industry are in the use of real world data for regulatory approvals as well as in decentralized clinical trials and then in the area of clinical operations around identifying new sites and investigators to be able to put the sites in. And then the biggest opportunity is something that we've talked about in this panel as well as the previous panels is this concept that I call around digital biology, which is how do you collect all of this high resolution multimodal data around the patient to be able to drive new insights be it around disease progression and sub-characterization of disease or simply identifying drugs that won't work in the patient so that you have a better chance of success around putting the patient on the right drug. Now to your question, Najat, about platforms and the platform story, we are still on that path a decade in around building these big robust scalable platforms that are reliable. And the analogy I draw is around electricity. When electricity first came in as a utility, it took about 20 years for electricity to be truly integrated into the operations of the industry. And because everything had to change, the way the plants were laid out, the way that workers were trained around the use of electricity as well as drawing this boundary between do you have a power plant in-house or do you actually go by the utility from someone outside? And I think we're under a similar journey around AI about how do we start to build these platforms at scale and how do we partner? What standard should we be adopting as well as how do we ensure that we have robust and reliable data that we're able to use for training our algorithms because that's really what drives the downstream use. And it's also iterative. All of us as an industry are trying to build the plane while we're flying it. So it's not like the platform is built and then we do a big bang launch and then everyone starts to use it. We're driving this continuous innovation where we continue to tweak what the platforms look like and then people in the big farm organizations as well as biotechs are guiding us around here are these cases to prioritize and here is what we wanted to in-house which is what we want to do outside. So I think we're in the second decade of our journey around the broad based adoption of these technologies and the drug development process. I believe the processes will fundamentally change at the end of this decade and it's gonna be a lot of co-creation and partnership as we continue to go along this journey of building these platforms at scale. Super helpful Shashri. And just reflecting before going to the next question on the answers here, I mean, you all have shared examples starting from Target ID to denied dimension sort of the de novo design that's happening small molecule and biologics to digital endpoints for stratification monitoring outcomes Shashri to the clinical trials piece and RWE in terms of evidence generation. So there's really value across the board platforms are not built perfectly yet but we are getting there and it's totally gonna take collaboration. No one person is gonna be able to do it on their own. I guess the question for me is, when you think about unlocking that value end to end there are still some with any new innovation or any new approach challenges that exist whether it's the data quality, the data integration whether it's the deployment of some of these new approaches because the ecosystem has to also mature while you're trying to do it you have to do it all at the same time and then there's the good old cultural change. So I'd love to just get perspectives from a couple of you of what do you think will it take to really unlock this value because if not, we're not doing the right thing for our patients. So I'll start with Anne first just your perspective from what you've, you're seeing at Takeda but many industries before that many pharma companies before that and then I'll also go to Dina from the academic perspective and being on the scientific advisory boards and just be a rounded awesome person in the space that's innovating and pioneering your perspective as well what is it gonna take? So starting with Anne first. So I think it can take many things and I actually considered which answer I give for this. Like many people here I could give multiple answers but I think the answer I'm gonna give is within a company and I'll speak, I could talk externally but for this answer I'll talk internally within a company we need to really take our responsibility for data very, very seriously and we need to take control of our data in terms of its governance, its quality, its storage and its accessibility and so within Takeda we have done two things actually that are really leaning into that notion. The first one is, is that we've brought our technology and data partners into the business of R&D and we have brought the disciplines that really think about how we use data in clinical trials like our statisticians our clinical pharmacologists to sit beside our technologists our data scientists our data architects or data engineers and so by bringing the technologists and the business experts together we're really galvanizing everybody to really think about the business of R&D and how do we develop drugs with a greater insights to data to get them to patients faster. So that's one big area that we have done. The second big change we've made is that we're halfway through a build phase for our clinical trials where we are really taking control of our clinical trial data and we're really taking a quality first automation a first mindset for this. So we're thinking about how do we ingest our data no matter where it comes from? You know, it could be digital tools as I mentioned before it could be ePros it could be those captured at sites. How do we ingest that data? How do we derive its quality? How do we make sure by the end of the trial that we're ready just to push a button at our database lock because our data is high quality? And by doing this in a very repeated way we're just where our goal is to have high quality efficiency in our system and then through good governance enable appropriate access to that data for both primary and secondary use. So they're two things that we're doing internally to really move us along this path within R&D. Excellent. Andina? Yeah. So I would say, okay, so there are many challenges of course but I wanna pick one challenge because I think it's probably the most fundamental challenge is the culture, I think. And you said that I'm in academia and not just in academia I'm a computer scientist also. So take it from someone who's outside your industry and looking at it but for the last six years I've been interacting I mean I'm on this tab with Janssen and work with many of you guys. Looking at it from the outside from a computer scientist perspective and I really love the industry. I can see, like I work with the scientists I work with the R&D people, crisis aside and I said different question but the industry has a level of the scientists the chemists, the biologists the people who develop drugs have immense, immense ability to innovate on the drug design, on the chemistry, on the biology but they are really conservative when it comes to digital when it comes to the thing that they don't really know and therefore they don't trust. And as a result like naturally and like everything that we are not so comfortable was because we don't understand it that well there is a huge gap there that they are way more willing to accept things like genetic drugs or gene therapy, cell therapy stuff like that because it's closer to that understanding when it comes to digital AI it's this black box and there is the cultural gap I think is one of the biggest challenges and pharma is by nature a conservative industry particularly when it comes I think to digital is even a bigger conservatism that you can see in the history like even the electronic PROs we still don't see them everywhere it's just like sometimes we have clinical trials with paperwork and you ask why like why we are still in 2023 have that. So I think there is a need to all understand and all accept a bit that you have to take some challenges of course you have to be very, very careful when it comes to patients and people's lives but you have to be more open to digital and AI and digital health and look at it and work with those concepts and adopt them in your operation much more. Yeah, I mean both of your answers super insightful and I mean if you don't have the data platforms hard to do at skill and Dean of the cultural aspect it's a really good insight because on one hand I would just say that pharma companies it's a highly risky business 10% success rate, I mean these are risk takers I mean I would say but it's a new discipline I'm very innovative, it's amazing I really like working with pharma because actually the level of innovation and the mentality I work with scientists in the pharma industry. Yeah, so I mean but I was gonna say that I do think representing pharma today like it is a very high risk is high tolerance for risk and innovation but it's a new discipline that's merging and coming together and the sooner it can be adopted I couldn't agree with you more I think the better off we will all be you know Alice I would love your perspectives on this too you know just from a company from a biotech tech bio I don't know what's the perfect term these days but really looking at how you figure out the right targets drivers of disease in a different way and also the culture and mindset must be different so how are you pulling that together both the science medicine and the data science aspect? Yeah thanks I think it's a really interesting time and you know on the topic of tech versus biotech startup versus pharma I think that you know we're entering this new era I'm a believer we're entering a new tech driven era we're just increasingly technology is advancing at just an unprecedented rate the cycles are getting compressed and when this happens you have to really think about what is the right culture to leverage all of these advantages I like I read a quote once from the founder of Genentech that said you know I always maintain the best attribute we had was naivete and I think in times when the playbook has yet to be written and where the technological challenges can be daunting oftentimes innovation may be even more important than experience right and naivete can actually be an asset so I think we're starting to see actually this new breed of CEOs that don't look like what other CEOs have looked like in the past where you know being very nimble and very quick about how you make decisions you know in the field that we work in AI and genomics you know if it takes you months to make a decision the landscape looks entirely different we've seen that with you know chat GPT and generative AI in months and I think it's important to think about how are we evolving our culture I've heard we have had a very different culture than I would say the vast majority of pharma companies and biotechs we have a huge emphasis for example on emotional vulnerability where we actually we start meeting by naming our emotions we meditate for 10 minutes every day and I think that this reflects a larger shift in the generations now that are entering the workforce because I think that there's a shift from a leadership style of pounding your fists on the table saying we're crushing it every day to one where people really want us the authenticity they want to be brought along they don't want the leader that has it all figured out and I think that you know our workforce being very heavy on engineering and AI is tends to be a younger kind of population and that's you know been really key for us to really attract and retain these folks especially when we live in a neighborhood when Google and Facebook are you know constantly approaching our employees and offering them packages of four times what they're making now? Yeah no I mean I think it's a couldn't agree more it's a new new breed of talent that you know I've used the word bilingual before but understand both disciplines I think it's a balance you know but I'm biased but you know we have to get on board and really figure out how to do it effectively thank you for sharing Alan's you know one thing is you know we hear about the data the platform the cultural change and the change in the type of leadership as you mentioned Alice right like all of that is changing but it's in the midst of that I think one thing that concerns people quite a bit is around you know how are we using AI for good you know how are we using it data science in a responsible way in an ethical way in a transparent way because for centuries any new innovation can do a lot of good and the reverse can also happen so the guardrails are really important and the understanding of it and how you apply it is important so with that I'd love to pose the question starting with Junai at first just you know tell us a little bit about what does responsible AI mean to you and then you know what are the aspects to really think about around the data the algorithms, development, deployment and so forth Thank you for the opportunity so the exam question that I focus on at Microsoft is how might we transform the practice of medicine with trusted reliable human-centered AI so it's not just about AI but it's all the precursor words I said that are really really important around the trusted nature of what we do the reliability of things it has to work but within that there's an ethical equitable and responsible component to it and so there's a conversation and an ethos that many communities in the AI space have been thinking about for a long time we have a set of principles and standards within Microsoft that we call the responsible AI principles that we try and adhere to and we think about both the development of tools and products in addition to then the deployment of these tools and products and so we think about the importance of fairness of reliability and safety of privacy and security of inclusiveness of transparency and of accountability and there are checks and balances and a governance process that we have to go through in throughout the course of the research development cycle through to product development cycle and then to the deployment cycle and then what's probably required on the receiving end if we've built what we have how does that look within some of our partner organizations how equipped do they feel around the checks and balances that we have put into the checks and balances that they'll probably have to put in on their side and I think it's an uneven taking thought at the moment where the tech companies have their principles and approach but actually I'm not sure how many farmer companies have a set of responsible AI governance processes in their organizations today that they're willing to adhere, adopt or combine with and then you extend it out and then the farmer companies are probably thinking so what are the regulators thinking about this space and how do they feel about responsible AI and if we were to submit something to a regulator of your choice even if we've adopted these, is this good enough and I sometimes wonder whether everybody's waiting for somebody else to do something first somebody's willing to go to the FDA first or the MHRA first or EMEA first and waiting for them to give permission on the approach but everybody's learning this at pace and scale so we've spoken about generative AI I'm sure if we had this conversation last year we wouldn't even mention generative AI in this dialogue today but now it's in the hands of everybody 100 million people adopting it within six weeks it's democratized access to these tools, techniques and solutions and it will be affecting all aspects of our life as a general purpose technology what do we need to do to ensure the safe deployment of these from healthcare and life science purposes within the organizations that you represent and how we do it from a technology side and work hand in hand to do it in a safe reliable manner I think is the challenge before us collectively The challenge and a huge opportunity so I'm just gonna ask one question lightning round I'm gonna start with Shastri 30 seconds who's gonna be, we have folks from the tech background or tech companies focusing on the space pharma big and small and we also of course have tech bio the new age or biotech tech bio companies we're gonna be the winners in the next five years who's actually gonna make a significant dent in what we consider as valuable impact using data science and making medicines Shastri first but 30 seconds All of them the space is just so big I don't think there's gonna be any single winner that takes all we're gonna see tremendous progress across all of these players and hopefully as a community A, we learn from each other to be able to move the ball forward a lot faster and share best practices and then secondly to build on what Jeanette said I really hope that as an industry we come out with good AI practices just as we have good manufacturing practices good clinical practices we need to come out with good AI practices that allow us to be able to promote and use ethical use of AI because the training data sets actually amplify the disparities in healthcare and the lack of diversity of clinical trials and we just can't keep amplifying the problem we need to figure out how we break it Love it Anne? So in my mind if the winner isn't the patient we can all pack up our bags and go home Seriously That's the best So either we directly get good drugs to patients faster by speeding along the development and discovery cycle or we indirectly make it easier to take away the operational components and things but no matter what way we slice it if the patient's not winning we're not doing our job Alice? I totally agree but I also love being controversial on panels because I think it makes it more interesting I will say that for any fast-moving technology obviously from a biased perspective that ultimately I think that smaller companies are more well-equipped to just move more quickly and capitalize initially on those technologies you know many pharma companies are so talented but I think just structurally there are a lot of challenges that misalign incentives both on the financial side the bottom line and what's really needed to really see a highly risky technology and innovation through so I think if it follows history I think my prediction is that the biggest initial advances will be made on the startup side when there's clear kind of de-risking then pharma companies will come in and collaborate more broadly but I think a lot of the initial innovation will be made from startups as a startup founder myself Great Dina? So I don't think that there is a sector that's going to be a winner or a loser I think actually all of them will have winners and we will have losers and what I think is going to be very important is the role of leaders because there are going to be many big decisions that have high stakes and people are going to make different decisions companies are going to make the right decisions are going to be the winners and companies who are going to be taken by the inertia they might find themselves irrelevant And Junaid, before we go to Q&A bring us home I should have gone before Dina I think it is leadership but it's leadership with a long term view so not just about short termism I think it's sustained leadership and leadership that is not just willing to make the investment but go on the change curve and change and transform business processes within their organizations to enable and unlock the value of these technologies moving forward I couldn't agree more I think the leadership aspect and leaning in day in and day out when it gets hard and actually ensuring we keep the talented folks inside and outside they are empowered so they can actually have an equal share of voice and make a difference I think that's going to be what helps us all Thank you to everyone Now we have a few minutes left and we have a couple of Q&A questions I'll start with Chris Benko Chris, take it away Nice to see you all Thanks, Najaf We have really great representation on this panel as we often do We've got some really well capitalized technology companies pharma companies and other early stage companies represented Right now the capital markets are really tough and I know that we're seeing a lot of earlier stage companies with technology capabilities not have the capital required for their companies to be going concerns or to generate all the clinical evidence they might need and my observation in this field in the last 10 years is that very often people have two out of the three capabilities but you need them all You might have technology and capital but you don't have the clinical development capability or you might have the clinical capabilities like pharma you don't understand the technology So my question back to you all is are resources really allocated in the right place right now across the industry or if you had the opportunity to squeeze the balloon if you will and shift some resources from one place to the other where might you do that in order to make sure that promising innovations actually make their way all the way through to patients and their use at scale I can start with again I want to put in controversial views out there So I would say one area where I think resources are not as democratized as they could be really is in venture capital right now I think a lot of the classic formation of biotech companies is concentrated within very few VCs they go out to academic groups they decide what is promising and then they start these companies from within the VCs oftentimes perverse incentives to flip them and take them public right and I think that's what's led to a lot of the challenges in the bubble that we see right now I think that the field is changing I think we're moving more and more towards like a tech-like model where there's more organic company formation that are coming directly from the scientists the postdocs, the grad schools, out of grad school that are started because there's someone that's passionate about the problem on the idea and I think that overall democratization of company formation dollars into the hands of many rather than the hands of few will lead to better innovation because that will lead to supply demand like what are the ideas that are actually needed by the field rather than just what people think can be profitable in the next five years If I can jump in Chris to answer your question I think just given the resource crunch there is gonna be a greater need for what I'd call translational tech organizations which is how do you work backward from the patient and take all of this amazing innovation that's happening in biology and technology and turn that into treatments that get to patients faster I think we have a lot of innovation that's happening in pockets nobody is really taking a step back and saying how do you string all of this together to be able to fundamentally change the patient outcomes? Sorry, Anne, you were saying something. Yeah, no, actually a similar but different answer actually I think tech companies are really good at the ideas and pitching the ideas and starting there the piece that's missing for me is the scalability for clinical trials and for testing So that piece, if the tech companies were able to have more resources and more understanding of that scalability piece that's so important for us and Pharma and the rigor piece, so scalability and rigor I think that if we could bolster that in the tech side I think that would be really helpful. Thank you, Anne. All right, I'm gonna go. Thank you, Chris, for the question. It's a great question. Deepa, are you on? Deepa, can you hear us? All right, thankfully, I know Deepa's questions. I'm gonna try my best to ask the question. So one of the things that Deepa had mentioned is we talk a lot about the various different data sets and I think Shostri you and others mentioned a little bit around EHR clinical omics data and so forth. Connectivity of the various data sets is really important to be able to do this better at scale. And of course, there's a lot of questions around using it in a way that doesn't add to any bias. So just a couple of answers from the panelists here in terms of how do you see that evolution happening from a multimodal data perspective to answer some of these really, really important questions. I'll throw out two controversial perspectives. The first one is the reason we've hidden behind large data sets of large N numbers is because we want to make up for errors in data. If we had really well curated data for small subpopulations, I think we should be able to get to really good insights a lot faster. So I would like to see a shift away from this big rush we have to get the largest number of data sets to actually get a really clean meticulously curated data sets that are multimodal as well as high resolution that allow us to get insights faster. And then secondly, I think back to my concept of good AI practices, how do you actually make sure that these declining data sets are truly representative of the world so that these algorithms that come out are explainable and actually come out with something that we can rely on is gonna be a big thing that we have to solve for. And you know, there's a part of it that came up and maybe I'll just ask Dina and Alice to comment around perspective data too, right? In terms of generating data that's actually really comprehensive, doesn't have the missingness and is curated the way you need for the question you need to answer because a lot of real world data doesn't have that today and then having really good methods to solve for it. So maybe Dina, do you wanna comment on that and then Alice too? Sure, yeah, so I do agree and actually I work as AI model, that's my bread and butter and I do agree that actually it's very important to like a smaller data set that is right data set is much better in achieving the goal than a bigger data set that is noisy and has missing and wrong information. I particularly think that one of the data that we are missing much of the data that we have is Snapchat like medical record or genetic data. And one of the things that I think it's really important is to be able to have dynamic data that's tracking disease and symptoms and response to medications. Yeah, I think a big learning we've had at Verge is that single data types on their own are not particularly helpful but really looking at where signal converges across multiple data types is where you get the smoking gun of disease pathogenesis, large-scale genetic data sets being a great example of this, I got one in thinking we would find lots of new therapeutic targets, instead we found hundreds of variants each with a small risk of disease and now the missing question is how do these different genetic hits actually tie together? I think the nice thing about prospective data is that it gives you an opportunity to iterate using the models that you've developed. So there's always gonna be learnings on where the gaps are in the data and if you don't have an ability to actually address those gaps quickly, then that can be a big disadvantage versus prospective, let's say you, we discover that certain disease stage or longitudinal data is really important that allows you to alter how you're collecting the data to fill the gap quickly. So with that, great answers from everyone, just wanna say a big thank you to all the panelists. If I reflect back on each year we've had this panel and we go back even two, three years ago, there's been such a progression to see from impact use cases and just a level of sophistication around what works versus not and the collaboration that's increasingly happening to figure it out. So much more to do again for the patients that we all serve. Thank you so much everyone for your time and perspectives today. Back to Andy and Karun. Could we have the pulse light please? Data science panel really terrific and really appreciated the focus on culture as much as the focus on technical. So for audience participation, the question that follows this panel is the following, where are digital technologies likely to make their greatest impact on drug development in the near term? A, digital endpoints, B, digital therapeutics, C, patient recruitment, D, trial decentralization and E, real world evidence collection and interpretation. So I don't think there's a right answer there but we'll see where the audience feels the friends will take us in the near term. So again, please contribute to the polls and once more, thanks to the panel. Could you bring up the results please? Surprise actually, a clear winner actually in terms of the audience. So almost half of you feel that the greatest applications are gonna be in real world evidence generation and interpretation. So really, really terrific. An area that's taking off. And of course as we heard earlier from our FDA commissioner, Rob Califf, an area that's primed for excellence and really to contribute. So terrific, thank you very much.