 Welcome, everyone, here and everybody watching from home to the promise of precision medicine. We are going to explore today how individualized treatment will revolutionize healthcare as we know it. My name is Linda Henry. I'm the managing director of the Boston Globe Media Partners, which also built a media company called Stat News, which reports from the frontiers of life science. And Stat created the great graphics we will see today as they have been diving deeply into precision medicine. And as a little bit of housekeeping, this is a 60-minute session, which will start with a discussion amongst our fantastic panelists that we have today. And then we'll open up to questions from the audience. And I will be very precise and end this at 1.30, so for those of you who are interested in attending President Trump's speech, you will have half an hour still to get downstairs. And now it is my pleasure to introduce this world-class panel that we all get to learn with today. So Nancy Brown is the CEO of the American Heart Association. Jay Flatley is the executive chairman of Illumina. Dr. Scott Gottlieb is the 23rd United States Commissioner of the Food and Drug Administration. Dr. Vasan Narasimha is the CEO-designate at Novartis as of Monday. You will be CEO. And Dr. Tan Chorchan, the Minister of Health of Singapore. Welcome and thank you all for being with us here today. They all have incredible experience and credentials and expertise and I encourage you to look at their bios on top link to learn more about them. And so it is so interesting to have this conversation right now. About 42% of all drugs currently in development and 73% of oncology drugs are targeted therapies. Biopharma companies nearly doubled R&D investment in precision medicine in the past five years and that is projected to increase another 33% in the next five years. The next, the number of precision medicines in development will increase 69% in the next five years. So there is a lot going on right now. And I want to start by talking about the expansion of precision medicine. Vas, our ability to use your words, which I really loved, you talked about our ability to generate and interrogate and gain insights from data. It's rapidly increasing. Can you explain to us how this is accelerating precision medicine? Yeah, sure. And thanks for the opportunity to speak with all of you. You know, when you look at it, there's a couple of trends that are happening all around us. The computing power that we have at our disposal has increased dramatically such that we can actually take on so much data. The second is now we're able to tie together data sets, genomics, proteomics, phenotype, the biomarkers that actually enable us to generate deeper insights into human biology. It's another accelerator. And then the third is our analytical power has improved. Our ability to tie data sets together and then analyze them to get to the underlying determinants of disease. When I look at it, I think there's two big opportunities. One is can we mine our clinical data to find new drug targets that we wouldn't have identified before? We've seen that already in mining our large-scale clinical trials with proteomics. The second, and maybe the more tractable one, is how do we find through precision medicine high-responding patient populations? It's worked beautifully in cancer, you know, when we do have genetically derived tumors. It's worked now with gene therapies for eye disease. The question becomes beyond that, can we move into other areas where we can identify high-responding patient populations and target the medicines in clinical development through clinical trials so we get the patients who get the biggest benefit at the end when we actually launch these medicines? Wow, great. Nancy, there are some limits to this data that Vas is talking about in terms of the population that we're getting this data from. Yes, absolutely. I think Vas has rightly identified the power and the potential of using data in the intersection of high-compute capability with all of this data that exists in many different places. The concern that I would raise is that the individualized data is from very select populations. If we really are going to focus on high-responding patients, then we need to have data that's representative of all people. If you look at who participates now in clinical trials, who perhaps has participated in these long-standing population studies, we would find that not all populations are represented. In addition to thinking about how we interrogate the data that exists today and how we create data frameworks that can allow this data to come together and be analyzed among and across these Vas data sets, we must do more to recruit individuals who will be willing to share their data so that all populations are represented, because if not, the promise of precision medicine will not be realized. Chow Chan, you are dealing with this issue directly. Can you tell us about how you're helping with this? Yes, I think it's very important for us to ensure that the study populations that go into precision medicine are sufficiently diverse. In 2016, a survey of the genome-wide association studies showed that of 2,500 studies, only 19%, 19% of the participants were of non-European descent, and of these 19%, the majority were Asians. And this matters because in the underrepresented participants, ethnic groups in these study populations, genetic markers might be wrongly assigned to be disease-related just because they're underrepresented. And therefore, you run the risk that by creating more precision medicine for one group, you in fact create imprecision medicine for another group or groups. So a lot of work, I think, needs to be done for us to pool larger sets of more diverse patient representation so that we can be more confident of the validity of the predictive power of the biomarkers in different populations. So more diverse data will enable the precision medicine to be more precise to the individual? Across populations, ethnic groups. Jay, you've lived at the intersection of biology and technology. With all this data, and once we do get a better balance in terms of diversity, there's still a lot that we don't know. Can you talk to us about some of the limits of the data? Sure. So let me start by saying that the underlying tools that have kind of powered the discovery engine around genomics and precision medicine have increased in power dramatically over the last couple of decades, and have been used quite effectively by governments, research agencies, and commercial companies as well to create a vast knowledge about what's going on in the human genome and that of other plants and animals. But if you think about how much we actually know versus how much we think there is, that estimate's very hard to come up with, but people think it's still in the single digits somewhere. Some say low-single digits, some high-single digits. So we have a lot of work to do here, and the way that work needs to get done is through big science, and that big science, I think, is beginning to happen around the globe in the form of what we call population genomics programs. One great example of that is the genome-ingling program where they're sequencing 100,000 genomes through the clinical side of the NHS. All of us program in the U.S., there's a program in France, China obviously has a precision medicine program, many others around the world are beginning to come out of the ground, and so the combination of those projects, if we can figure out how data gets shared between them, which we've been talking a bit about already, will exponentially increase the power of those data sets, and kind of achieve what I think is the most fundamental objective, and that's to improve the clinical utility of the genome. Great. So, Scott, you've talked about how clinical trials are done, and how they have to be changed in order to catch up with the rapid advancements that we're having in precision medicine. Can you talk just about some of your initiatives? Yeah, I think I would start off by saying, first we have to decide what we mean by precision medicine, I think on the one hand we have the ability to change the underlying mechanism of disease, and that's one component of what I think about when I think of precision medicine, and when you think about changing the underlying mechanism of disease, you think about things like gene therapy, Cas9, CRISPR, and in those constructs, you're able to often observe the efficacy very early in small trials, sometimes these technologies are so efficacious that they're super efficacious, if you will. Some of the things that we've seen in early development, you're producing almost too much levels of certain proteins that you're trying to replace in some of the genetic diseases that are being targeted with these technologies, and that's been in some of the early literature. So the issues that the agency is going to confront are going to be less about determining efficacy, where there's a strong proof of principle and you can observe efficacy early, and more focused on long-term durability and safety and product issues, issues related to the off-target effects of these technologies and what the implications are longer term. And so I think that's where some of the focus on how we're going to try to steer the design of clinical trials in these realms is going to play out, and we're going to be putting out some guidance. We've talked about it early this year. I think we're really at an inflection point right now, where we're at a point in time where we're sort of defining the modern rules for how these technologies are going to be regulated, and we're going to be looking at accelerated approval end points for earlier approval on questions of efficacy with more vigorous long-term follow-up in some of these constructs, where we have authorities to do that under accelerated approval. And you're seeing a lot of engineering around the product specific issues, in part for commercial reasons, where you see companies changing how they develop some of these technologies because there might be stacked royalties and other things, patents that are blocking. So it makes the issue of looking at the product issues more complex. The other thing you think about when you think of precision medicine is instead of changing the underlying mechanism, manipulating the underlying mechanism of disease. And there you think about the traditional drug diagnostic combination and drugs that try to target a mechanism of disease in a pre-specified group. And I think there it's incumbent upon us to think of clinical trial designs to make it easier to pre-specify those groups and to target them where you can observe a strong efficacy signal earlier. So you think of things like tissue agnostic approvals and master protocols and drug diagnostic combinations trying to get the diagnostic approved alongside the drug. And we're putting out, we've put out a couple of guidance documents articulating those pathways as well. But I think they're distinct sets of challenges, if you will, and I think they need to be thought of differently. There was a precision medicine conversation here that Shoshan and I were at yesterday, I think. And part of the conversation where we're talking about approving the process as opposed to the specific therapy. So the process by which the patient is served versus the actual dosage amount, which or combination amount that had to be, which is the way things have done in the past. So instead of testing and approving each dosage amount, approving the process by which it is achieved. Well, I think we wouldn't necessarily regulate how things are applied in clinical practice. I think where you might see a construct where you're regulating a treatment system would be if you're trying to approve a drug and a diagnostic in combination. And it's been a little bit more challenging to define the pathway on how to get regulatory approval for the diagnostic at the same time that you get regulatory approval for the drug. In part because a lot of these diagnostic tests have been promulgated as laboratory develop tests. So they haven't been brought through a traditional regulatory process. I think that I've talked about the fact that I think it's incumbent upon us to think differently about how we regulate diagnostics. And I think it's time that the agency needs to work with Congress and stakeholders to develop very specific targeted legislation that would give us a unique set of authorities to regulate diagnostics properly. My view is that the old 510K PMA pathway, and it's a little bit jargony, but it's a technical audience, but the traditional pathway for approving medical devices doesn't really fit well with modern diagnostics. And we need very well-fashioned authorities when it comes to diagnostics. Vassa, what has your experience been with this? You know, when you first see your first question on the process and the product, you know, we had us recently licensed a cell therapy through the FDA, and this was a very precise therapy where we take a patient's cells, we reprogram the cells for B cell cancers, and then we infuse them back. And it's a completely different construct, right? It's sort of a living drug where the patient's own cells are part of the production process. That, as Scott points out, I mean, this is a completely different paradigm. And then I think on the diagnostic side, I think with the power of digital, and can you really do digital diagnosis when you think about the power of liquid biopsies and can we detect cancers earlier? I personally think that we're at the limits right now of what the human mind can ascertain in terms of biomarker signatures. We're going to have to have a machine looking across a panel of signatures and identify a signature that would then give us a patient subpopulation. Asthma is probably not asthma. Severe asthma is probably not seizure asthma. There's probably a whole range of subpopulations. And then we need, of course, the regulations that would allow us then to bring a drug with that signature or that diagnostic through clinical trials, which we'd still have to do prospectively to market. And the other way to interpret your question is exactly how you picked up first, which is, and that's why I did the bifurcation in my comments, which is, you know, on the biological side, the process is the production of the biologic in a lot of these cases where we have precision medicine like CAR-T, like gene therapy. And that's where a lot of the focus of the regulatory process is on the manufacture of these personalized treatments. Because that's where a lot of the complexity is and that's where a lot of the things can go wrong. You call this cell therapy, we call it cell-based gene therapy. Correct my language. Take a note. I'll defer to you how you describe your product. Nancy, I want to go back to data, because this is really what precision medicine relies on. And you've talked about how with larger data sets, we can improve care for data. We talked about the importance of diversity in our data set as well. And because the large data sets help physicians understand how targeted treatments have worked for genetically similar patients in the past and help doctors uncover treatments that they may not have thought of. But assembling the large data sets requires healthcare systems to share information. How do we solve the current barriers to data? Yeah, you know, I think one of the most important things to remember on data is what we're trying to do is create, in essence, the public trust, you know, that there is this promise of more personalized treatments for patients and what patients want is to live a healthful, you know, satisfied life. And so health systems feel that they are the fiduciary of patient data and they are. And we've got to find a way, you know, to inspire the patient voice in all of this because what gets lost, I think oftentimes, this is about process. And it's a very technical conversation, but at the end of the day, what we're trying to do is assimilate this mass amount of data on behalf of patients in the public. The other thing that we haven't touched on yet as it relates to data is what we believe is very important data that must accompany the data from health system, EHRs, from clinical trials and long-term population studies. And that's environmental data, exposure data, as well as data on how people are living their lives. And one way to push the system, we think, is to inspire people to be willing to donate their own data from their own wearable devices, from, you know, new sensing devices that are not invasive sensing devices. And if we could create this groundswell of people that say, hey, you know, I want something different for myself and my family than I have today and I'm willing to donate my data to make that happen. And by the way, that's my data in your health system and I would like that to be part of this as well. That's what we're trying to do at the American Heart Association. It's what Francis is trying to do at the NIH and I think the system has to be pushed by the people who are the ultimate beneficiary. You talked about the word trust and Shoshan, you have talked about the importance of trust as well. And this includes rules on who gets access to this data that Nancy was just talking about collecting. I think it's very important for us to have very early engagement to build a trust because I think the promise of precision medicine is when we can get more and more different types of data, clinical data, genetic data, data linked to research, environmental data, they are specific for an individual. And then in aggregate, we have a pool of data that allows us to see patterns in associations that allow us to create a more continually learning system. But that means that we have patient identifiable data across many, many different types of data which are specific to a patient. So I think people will be naturally and understandably concerned about privacy, about security of the data and also about who gets access to this data and whether there are protections against genetic discrimination, particularly in employment and insurance. And I think these are difficult issues but they need, I think, to be the subject of an early engagement with populations because this is, I think, the direction in which the entire field is going to move. Great. Jay, you had thoughts on this as well. On what Nancy was saying about getting more people to share their data, how do we drive the behavior of data sharing and how do we get past the challenges of safeguarding that data? Well, the challenges that we've seen here that people tend to hoard what data they have either for economic reasons, intellectual property reasons or geographic boundaries. Very often you'll find countries have rules about data not leaving the geographic boundary of that country. And so we've been involved in efforts to try to figure out ways to kind of break down those barriers. And there's a whole host of things that have to happen here, including things like standards around phenotypic data. You find some of that locally centralized health systems tend to do better than places like the United States in that regard. But then also the underlying technology about how you actually federate these databases and allow, if you have to keep data geographically separated, how you have high performance query systems that can interrogate across these databases and how do you exchange economic value? And so there's been some, and there's ongoing research on using technologies like blockchain to try to track royalties. And in fact, there was an announcement yesterday, I think that Amazon is beginning to work on this notion and there are other companies in the blockchain sphere beginning to do this as well. And I think that'll help break down some of the concerns about, do I get the economic value out of all the work and dollars that we invested to create my particular data set? There's also some standards being formed like GA4GH, which are beginning to create the kinds of standards we need globally to put these data sets together. Vas, in order to get beyond the treatment and actually get to the preventative care aspect of precision medicine, how do we establish a continually learning infrastructure with real-time knowledge? Yes, I think it's gonna take time to really get the signatures and data that we need to enable this. I mean, we have examples. For instance, many companies are now running trials in Alzheimer's disease where we look for a particular gene signature, the ApoE4 gene, and we know that patients who are homozygous have a much higher likelihood to get Alzheimer's. So you treat early and then you continuously watch. So that's a very simple example. It gets much tougher. We've been doing work, as many others have, looking at circulating tumor DNA and other signatures for early cancer. You can do the continuous monitoring, and that's instance, but how do you know if you're treating in an instance where you don't need to treat? Is there the positive and negative predictive value of such tests? Because we know the body is constantly in a homeostasis versus cancer. So I think there's gonna have to be a conversation about how much do we really wanna know about our own health and when do we wanna know it? Because it's only valuable to know it if we can do something about it and do something about it with a high degree of conviction and with a high degree of data. At least that's how we've been thinking about it. So this predictive modeling that we're getting to the presenting, preventing the diseases that we know we can help. That we know can help. There's I think a great example that I like in asthma where we know atopic March happens in these kids. We know if you have asthma very early on and you treat early, you could potentially prevent atopic March. So could we identify those kids who would benefit and then prevent the asthma from ever becoming severe? I mean, those are the kinds of use cases I think that would make sense. So Scott, one of the other big areas of precision medicine that we have to talk about is affordability. How can we bend the curve on the price of precision medicine knowing that we've had an incredible drop in the price of genome sequencing which is one of the first steps but what else can we do? I think when picking up on what Vas was saying, I think doing clinical trials for primary prevention are particularly hard and costly. You oftentimes need very large randomized trials and you're looking over long periods of time and I think if we wanna try to bring the cost down of primary prevention, we need to think of how we design those clinical trials better to use data to better pre-specify the population that might be at risk so maybe you don't have to look over as many patients and look out over as long of a period of time and also have better ways to predict how products could increase the background rate for certain risks and so for example, if you're gonna give a drug to someone over a long period of time to try to prevent the onset of Alzheimer's because they might be at risk for Alzheimer's disease, you wanna make sure that prolonged exposure to that product isn't gonna increase your background rate of cardiovascular risk but if you wanna discharge the risk of slightly increasing someone's propensity to have cardiovascular disease through the administration of a product that's gonna be administered over a long time, that's a very large clinical trial and it's a very large placebo-controlled trial and so we need better ways to predict those risks and to pre-specify the populations at risk so that we can bring the cost of development down. If I could just add one point, I mean it's part of the reason and many of the times we have an expert group in the room, it's part of the reason why companies move to the adjuvant setting in cancer because we know this is a patient who already had cancer, cancer was resected, now can you prevent the recurrence of the cancer? Getting into primary cancer prevention then the numbers get very daunting and so to Scott's point, I mean if we could really specify who would benefit in a much more specific way then I think the opportunity to do these studies increase. Yeah, most of the developments in secondary prevention for that reason because someone who already had an event, if you already had a heart attack, you're at higher risk for having a second heart attack, it's easy to demonstrate that you can reduce the secondary risk. There are constructs though where we have studied drugs in secondary prevention or in more at risk populations and then extrapolated into lower risk populations. The classic example being ACE inhibitors in heart failure where we started with class IV heart failure and marched our way down because we allowed for extrapolation from a regulatory standpoint so there are models for doing that and I think we need to perfect those models. Linda, I might add that this is my favorite chart. Great. That the price of sequencing has gone down by two to three orders of magnitude in the last 10 years depending upon when you start and end and that we've signaled that there's another order of magnitude decrease in the price of sequencing gonna happen over the next reasonable short number of years. Yeah. I think if you look at two big topics discussed in this forum on health, one is precision medicine, that is value-based healthcare and I think if we can find a greater convergence and overlap between the two, it could actually be very powerful. For example, within the portfolio of things that are being done, are there precision medicine applications that can help improve health while reducing costs? For example, by reducing the number of individual treatments, how many patients you need to treat, for all the patients that you treat, how many people actually get a benefit and how many people don't have any benefit and I was just reading a report that said that for the top 10 best-selling drugs, for every person that benefits from the drug, between three to 24 people, actually show no improvement. So I think there are many opportunities for us to reduce costs in other areas, so as to create more regulatory headroom for the expensive types of treatments that inevitably will come out, which will be targeted at very small numbers of patients. Yeah, the system will pay more for a certain benefit. A certain benefit, that's reasonable, yeah. One of the themes that I've heard through this, and it's also the theme of the entire Fourth Industrial Revolution that's been a theme of the forum, is the amount of collaboration that is going into driving precision medicine forward, the interactivity. Vas, do you wanna talk to us about, you're at the forefront of, you're incorporating AI into this and you're working with technology companies? Yeah, actually, before I'd wanna just make a plug, Francis Collins is in the room, and we are actually working in many coalitions as with public and private partnerships to try, and particularly in cancer care, to bring the data sets together, to get much smarter about how we can actually improve patient care, and that's just one example, but I think that's very positive for society. On artificial intelligence, I mean, clearly there's a lot of opportunity, I think particularly to organize the data and actually generate certain insights. The big challenge will be, can we use AI, like many of the partnerships we have and others do, to speed up drug discovery and speed up clinical development? I think it's still early days, I think we always have to remember with AI that when we do image analysis, there's a training data set, you can train the algorithm on, and then you can look at the images. In drug discovery, there's not a great training data set, you can't find drugs for some of these conditions. It will undoubtedly speed up our ability, and I like Gary Kasparov's concept, and when you take the chess master, when you take a smart person and put them with a smart machine, you can be a smart person or a smart machine, and so I think we're gonna have to get used to having a culture where you have smart people working with AI to power drug development. I might add on to that, the power of partnerships is so key, and we've really focused on that at the American Heart Association, we've created a data discovery platform that has 10 million data points, the American Heart Association in Amazon have created that marketplace, but we announced most recently on the issue of drug development a new partnership between the American Heart Association and the Lawrence Livermore National Laboratories, which has one of the world's fastest supercomputers, and the idea is can we use the fantastic scientists of the American Heart Association who are represented in academic medical centers throughout the world and the artificial intelligence capabilities to do more simulation and modeling of how drug molecules attach to the wrong protein to try to get this issue of side effects off the table, and we think we're in a proof of concept phase right now, but we're very excited about this use and application of artificial intelligence because it is the mind and the machine coming together in a way with this super computing capability that is soon to be 10 to the 18th power, which is unbelievable. Just to pick up on that point, it kind of reinforces the need to try to develop good natural history models of diseases, and it's amazing we don't have good natural history models. We still expose patients to placebo and hypertension trials to see what happens with untreated hypertension. We should know what happens. We know what happens with untreated hypertension. I mean, you still need a placebo arm because you're looking for safety, but we haven't been able to model that yet. I think that's a big opportunity for investment, and we've made some investments, particularly in rare diseases, but I think looking harder at that, that's gonna enable the opportunity for AI. I wanna open it up for questions in the audience because I know that there's a lot of really interesting perspectives that we have. I think that there's one over there. Hi. So for somebody starting a phase one two study today, it's gonna take quite a while to get to the NDA stage. What should they be thinking about given things are changing? Is that directed? I think that's it. You assume so. Well, first of all, I would challenge just the concept. I think you've seen some, I think as we move into a realm where the mechanistic understanding of how the drug's gonna behave in a certain disease is well understood, and as we see drug developers targeting more rare diseases and significant unmet medical needs, I think you've seen very efficient development programs because you can get proof of concept very early. And then it becomes a question of looking at long-term safety issues where you're not gonna be able to discharge risk pre-market, all the risk pre-market, and you have to have very efficient tools post-market to continue to look at safety issues. And we now have those tools thanks to new legislation. So I think I would think about how you could try to develop drugs under that construct, which is sort of the theme of this panel, trying to better target development programs so that you're gonna be delivering a more certain benefit so that regulators, not just the FDA, but the EMA as well, have more confidence on the clinical efficacy questions, and then it becomes a question of trying to look at long-term risk. Could I also chime in? I think the power of all these technologies is incredible, but it's also incredible to me when I look across biotechnology trials that often the basics aren't done, right? So my first advice would be to get the dose right, make sure you characterize the patient population, understand your target product profile, what you're trying to actually develop, and understand your endpoints. And of course you can use AI to inform better endpoints, but I think right now with the energy which is all positive, sometimes there is a miss on understanding those basics, and those basics ultimately are what Scott's teams first look at before they consider these higher-end technologies. I'll just add, I mean I've been on both sides of this now as some people know, working on the venture capital side before, and I've heard, I've seen companies be reluctant to come in and talk to the agency because they're, for fear that they're gonna get advice that that's gonna encumber them in some way in terms of how they wanna develop the product. And I think more often than not, particularly in very novel settings, it's beneficial to come in and talk to the agency because first of all you might get advice that's gonna help you avoid problems, but if you find out that the agency doesn't understand a certain disease well or isn't in sync with your understanding of the technology, it gives you an opportunity to educate them, and I've seen that be done very effectively as well. So I think that it's rarely, you know, a bad proposition to come in and talk to us even pretty clinically, and we see a lot of people doing that, especially on the CBER side of the house with some of the novel gene therapy and CAR-T and CAS-9. Question right here. Hi, I just wanted to tell you more about a project that we have in Israel. I work for Mr. Mars Khan, the founder of Amdox, and what he did in Israel, he founded the Mars Khan and Makabi Health Institute for Science, which is basically in Israel from the time you were born till you pass away, you are at the same health insurance company. So they have your daughter from the time you were born till you die. And usually the family stays in the same health insurance company. So what Mr. Khan did, he basically decided he started it as a philanthropic venture because he wanted to provide scientists with a daughter so they can work on the daughter that Makabi Health Insurance has. And just to help science Mars, I think it would be better if you explain it. Is there a question? I don't have a question, but we've actually got information from about two and a half million patients which we've had for 22 years. And this big data has given us a tremendous resource and we've provided the base for interrogating this database and we've actually come up with some very interesting findings for research. It's a very interesting and valuable tool for research. The advantage we have in Israel is that we can actually, we don't have the problem of using the information that we have. Great. Can we have a question, please? Right back there, thank you. Yeah, thank you. Great topic, Christian Medicine. I was wondering, social determinants are very important on the outcome of medicine. I've seen them integrated in the concept of precision medicine, the social determinants aspect and also precise prevention if the panel would comment on precise prevention in medicine. I'd be happy to jump in on the social determinants of health. I think it goes back to the very beginning at the end of all of this technical discussion. It's about real people who live in real communities that may or may not have access to a healthcare system or to healthcare workers. And we know and understand things like levels of education, access to healthy foods, access to medications, access to treatment are going to ultimately impact health outcomes in the world. And we can't forget, as we're thinking about high technology and wonderful ways to deliver more precise treatments to patients that all of these societal issues must still be addressed and I think this idea of understanding the impact of people who are more disadvantaged as it relates to access to things that will help them be healthier and what that does ultimately to their overall health and wellbeing is something that must be studied. It's something that must be focused on. And I think we're recognizing more and more. Food is an easy thing to talk about. If you look at the impact of people who over consume high fat, high sodium, high added sugar diets on overall health and wellbeing, when we talk about what we're able to study from a primary prevention point of view, we know very clearly skyrocketing obesity, type two diabetes and hypertension, which cause heart attacks and strokes. And so we understand this and we shouldn't forget that important part of this ultimate drive that we all have to deliver more precise treatments to the right person at the right time. I think part of the promise of potential is we're able to integrate social data, environment exposure data together with the clinical and genetic data. And if you're able to do that, then we have some additional ways in which we can help to identify people at particular risks of specific problems and conditions and presumably be able then to target the interventions. And in some cases interventions might not be drugs or the conventional treatments, but social interventions. Great, that's good, right here. Yeah, I'm an endocrinologist working in Delhi. So I have two questions. Hold on one second, we'll just give you the mic. One is that we develop these drugs in a very long process. They undergo first animal trials and patient trials and they're finally marketed. And how is that that after they're marketed, some drugs like for example, PPAR antagonists like Rosy Gritter's own, like for example some of the anti-obesity drugs, they suddenly then you discover after they're marketing that they have side effects they have to be shut down. Well, I can start and I guess we'll see if Scott wants to jump in. So you know, we of course are, when we do the clinical trials, we power them based on what we see in each stage of development. So in phase one we see signals, phase two we see signals. And then phase three, the studies are powered to detect of course efficacy and then safety and particular signals that we've seen in the development. Now what happens post-licensure is in many, in the case of diabetes trials, you're talking about exposing 15,000 to 30,000 patients. So you will have a certain ability to detect. But I can tell you, my background originally was a vaccine developer. You have to expose sometimes 70,000 children to actually find a very rare case of inus deception. So it's no different when we think about diabetes drugs. Once we expose millions of people, we will find signals. Now it's incumbent upon companies to collect that data, report it to regulators and keep updating the labels. And then sometimes a signal pops up after enough exposure, especially after longer term exposure in some of the cases you cite, that then has to change how the drug is used. So that's how this happens. But it's a, it should be great confidence because it's a very rigorous process. I mean, we process millions of safety cases every year. We have to report them within 14 days. And if it's a serious case within seven days to all regulators, in particular the FDA and EMA. I think over time, our expectation of safety around products has increased. And I think that that's very reasonable. Our expectation of safety from the cars we drive have increased. I think everyone would rather drive a used 2005 car than a brand new 1995 car because cars are much safer now. And so, we're looking to discharge societally, we're looking to discharge more remote risks in the clinical development program. That's why you see trying to discharge the risk of secondary cardiovascular side effects in diabetes trials now. I think you referenced that. And we've made a decision that's reasonable, even though it's gonna add to time and cost because our expectation of safety is greater in a world where there's a lot of already good therapies you wanna make sure the incremental technology is gonna deliver a benefit. I think what we haven't caught up, though, with our expectation of a higher sense of safety around new products is the tools to try to discharge some of these risks. And I think that's where better investments in regulatory science to try to look at discharging these risks through something other than just randomizing 70,000 patients would yield a lot of dividends. Yeah, especially the drug Ampaflos and CU did the drug trials because they were FDA mandatory. But they found out such a useful benefit, 30% of the population improving cardiovascular benefit. So that's a positive thing of... I will avoid being drug specific or product specific. We have another question over here. Francis Collins from NIH. Appreciate the shout out from Vass about the importance of partnerships because we are at a space, I think, where pre-competitive opportunities are really all around us. And I wanna ask you, Vass, in terms of what's happening now with the development of these large-scale cohort studies. Jay already mentioned the All of Us study which will launch this spring in the United States. We'll enroll a million participants, very diverse, with baseline information about their social circumstances. They'll answer lots of questions. We'll have blood samples and urine samples. We'll have genomic data. We'll have their electronic health records. And they're pre-consented for recontact for participation in clinical studies of all sorts by industry, by academia, maybe about prevention, maybe about devices, maybe about drugs. So from your perspective, representing a very successful pharmaceutical company, what would you wanna see as far as the sort of platform that would be developed by studies like that? And I might just say, because this is Davos and this shouldn't just be about the US. At last count, there are over 50 such cohort studies in the world that are enrolling at least 100,000 participants, all of whom are gonna gather in Durham, North Carolina in the next couple of months to try to figure out how to share data and to have data standards and make this available to be a real research engine. So give us a little idea about what would be most critical to include as far as the parameters of such studies. Yeah, so thanks for the question, Francis. I would think of three things just quickly off the top of my head. I think one, it's the power of actually understanding the natural history of many of these diseases. And I think that the truth is, well, biotechnologies come a long way. And the science of medicine, we have a lot to learn about many of these diseases. There's so many conditions where we treat them as a monolith, but there's probably a lot more complexity that's out there. I always like to think that there's millions of proteins that you could target in the body and we can target about 400. So there's a lot we haven't figured out yet. So I think one is that. I think the second is what I really hope that those platform studies bring is a connection between phenotype, genotype, biomarker over time. Because I think one of the things, big learnings from the large genomic studies is without phenotype and without some of the other parameters, it's actually very difficult to really interpret what's going on. So if we have strong and good data that we can interpret, that will surely fuel drug development. But the final element I'm most excited about honestly is the patient element. It's still very hard to enroll clinical studies. And suddenly with these platforms that you describe Verily's working on, you suddenly democratize this because suddenly you go directly to patients who are already engaged and interested, want to become part of clinical research and now we can reach them, which is a big advance from what we currently do, which is randomly selecting investigators and randomly hoping people might show up to those investigators with our inclusion criteria. Which if we could streamline that process, it would be big, I think. I just have a general sort of add onto that comment and that is that we think it's critically important that we deeply connect the clinical world to the research world and we create this virtuous learning system. And so many of the clinical results that happen today through the physician's offices don't ever close the loop and those data sets don't ever become subsequently available for research. And so I think some of these larger scale projects that are getting started are driven by the clinical needs and to the extent you can database that information and create the right privacy controls, you can then use that collective information for research purposes and complete the cycle. And we think that's really important going forward. Jo-Jean, did you have some thoughts on that? No, okay. Next question, right here. Thank you, Majid Jaafar, Crescent Petroleum and also the Lulu Foundation, which is a rare disease medical research foundation. Almost building on to Francis' question. So the technology is clearly there and the data is being gathered. But what about the administrative or structural hurdles to really getting forward and also the differences? Because in the US, you've got the challenge of sharing across hospitals and insurance companies. In Europe, you may have national health providers, but they're pretty slow to uptake and innovate and notwithstanding initiatives like Genomics England. So what would you see whoever wants to chime in as the key bottlenecks, the key challenges to overcome and some ideas to achieve that? Nancy, this is something that you've talked about. You've talked about, I love your example of, if there's an update on for Facebook, you get it sent to your phone and everybody updates it. But to change the procedures for physicians, it takes years to get it through. So this is sort of similar to that. Yeah, absolutely. I think this idea of creating a data sharing culture is really key. Francis mentioned one example of how bringing people together to inspire this movement will happen. You mentioned you were from a rare disease organization and I think no one better than rare disease organizations power patients who are not having access to what they need to come and march to getting what they need. And I really believe at the end of the day, it's not about the technology. It is about an ethical framework and a patient privacy and making sure that all of the types of things that can assure that individuals' data is protected is what we have to be able to illustrate and demonstrate and to some extent to fix. You look at what banks and credit card companies are able to do around privacy. There are privacy breaches that really worry people. And on the other hand, they've created this wonderful framework. Who would have thought you used the 1990 car? Like in 1990, would you imagine ever that you're gonna take a picture of a check you have to deposit and it's gonna show up in your bank account and you're gonna trust and everyone's gonna trust that that's gonna be actual money that you can then go spend? So my own thought is it's not about the technology. There's probably more technology allowing this kind of data sharing than is even needed at this point. It's not about a willingness of players to collaborate because I think there's this worldwide sense of collaboration. It's about the fiduciary responsibility that people feel and it is from my point of view gonna require a groundswell of the ultimate beneficiary patients in the public to say we wanna do this because it's going to benefit us in better outcomes and we're gonna have to test it and make sure the security's there. Can I maybe just chip in here and say that it's a complicated issue but in some cases the laws and the regulations get in the way because it's really a balance between the societal benefit versus the risks and many of our privacy laws were developed at the time I think when the benefits were not quite so apparent and so say in the case of Singapore some of these are being refined so that it would then create a more facilitatory environment for sharing of data for common good and I think this will also help to accelerate the data sharing type of culture within institutions. Can I just say one thing, Jay was gonna jump in. Yeah I was gonna say I think another element of this is the move toward consumer managed health and this is gonna be critically important. Many people wanna have control of their information themselves and be able to get it directly from their physician, look at that information that's beginning to happen but ever so slowly and our view of course is that consumers have the right to their data, even their genomic data even though people have tried to stand in the way of that and that over time that needs to be a forcing function to accelerate the changes in the health system and the medical treatment system in particular which tends to adopt change very very slowly. The other part of this equation is trying to get the information into forms where it's gonna influence prescribing decisions by providers and we've seen historically that labels are slow to update with new information. We're looking at ways to be more proactive with that the Cures Act gave us some authorities to do that to reduce the barriers to supplemental indications. We're looking to initiate proactive updates on old generic drug labels for instance, initiatives like that. I think where I'm a little bit more perplexed is that the provider community has in some cases been slow to update their own clinical guidelines apart from the regulatory process you see clinical guidelines that is sometimes out of date. Particularly we were looking at opioids now where we're talking to the provider communities about building into the label, creating clinical guidelines that would delineate how long the duration of use should be for opioids for different clinical indications. If we had those kinds of clinical guidelines developed by the provider community we can then incorporate that into drug labeling and the health systems and other providers can use that as a way to try to control dispensing. So it's one example of where clinical guidelines get married to the regulatory process but they don't exist and I'm a little perplexed why the provider community hasn't created those. But just to step back a little bit, we do have the ability, the legal authority if there are clinical guidelines developed by expert groups we have the ability to incorporate that into drug labeling and then it could become part of prescribing recommendations that we give. How was that? I was just on the guidelines as the writer and author of guidelines for cardiovascular and stroke care in America. Cardiovascular ones are very up to date. Thank you. I was just gonna say, I think that the other thing coming you can debate whether they're too fast or too slow because we recognize that when a guideline is changed it changes clinical practice in America. In November we issued new hypertension guidelines that redefined hypertension and a hundred million people woke up the next day as hypertensive and so you can't, just like you're not gonna approve a drug with haste we're gonna be very careful as we make sure we're looking at all the evidence. That being said, I think this thought about how precision medicine will really challenge if clinical trials change then guideline development has to change and we're going to need to be very quick in thinking about how we compare an old way of doing clinical trials to a new way of doing clinical trials as we're writing guidelines is one thing I would mention and then the second thing I would mention is guidelines can move quickly when new technologies are created. Earlier this week at our international stroke conference we issued new guidelines for the acute treatment of stroke because of the new clot retrievers that quickly went through clinical trial, quickly got FDA approval in the United States and the guidelines were changed in record times so that more patients could benefit and I think this is yet another piece of the continuum that has to change. So I'm gonna go to lightning round because we only have a few minutes left. I wanna hear what your biggest prediction is sort of your most wildest prediction for precision medicine is. I wanna start with you Shoshan. Well, so my biggest dream is that we'll be able to work with the patient community to aggregate all this data so that we'll be able to more precisely target people at risk with treatments and with preventative care. Yeah, I tend to think about this as a future state and it's always the one we envision of where we wanna get to and I think that ambition is that each of us have our genomic information, our germline data in the database. We have our microbiome profile, we didn't talk much about that today but it's an incredibly interesting emerging new area to study that we understand what's going on in our blood and that when drugs are prescribed you have all of this information. If it's cancer, you sequencing the tumor or you're doing a liquid biopsy to read out that tumor and then some deep learning algorithm is what can integrate all that information in ways that humans won't be able to any longer and essentially becomes an assist to the physician in doing the diagnosis. Not sure I can pick one, I'll give you two. One, I think AI is gonna power our drug development processes pretty soon and we're gonna be able to prospectively define high efficacy patient populations. And the second is that blockchain at some point will lead to interoperability of clinical data which will completely transform how we do clinical trials. I would just say building on that, the power of digital phenotyping and creating real value for all of this other data that really matters, environmental data, personally collected real-time health data that that will power with the things that Vast just mentioned really change the way patients are treated. I think we're at an inflection point right now when it comes to gene therapy. Similar to what we looked at maybe in the 1990s when you had early antibodies on the market but they were murion, they were derived from mice and they weren't very effective because of that. And then there was an inflection point where we gained the technology to make those antibodies humanized and then fully human. And all of a sudden antibody-based drugs went from being something that didn't work very well to being important therapeutic products. I think we're at an inflection point right now where gene therapy is gonna become a more common form of product development and the inflection point was the development of vectors that are reliable, that can reliably deliver the gene. And I liken the development of the current vectors to the development of the humanized antibodies. And MIT recently put out some data, I don't know if they're right or wrong, but they looked out on the next, I think about six or seven years in terms of how many approvals they'll be in gene therapy. I think they predicted 40 and they looked at it, they're predicting a run rate by 2020 similar to what we see in terms of the development of antibody-based drugs. Again, there's gonna be a lot of uncertainties here. I think the long-term issues are gonna be the ones that are most complex for the agency and for patients and for providers, but a lot of opportunity as well. So what I'm hearing is a future where there's a lot of the emerging technologies are merged into precision medicine and that we have much better care. Within that, we have enough people participating. We have one of the things that we didn't talk enough about was ensuring equal access to this sort of therapy and being able to really treat people with the care that they need and not based on who they are in a way that not only cures, but to be able to get to a place where we can do preventative modeling so that we can stop it before it affects them. Any last thoughts, lightning round? I think one area we didn't really talk about is the huge change management that would be necessary amongst the medical community because the way they're gonna practice will change. They would have to trust in AI decision support systems, the way they deliver drugs, treat patients, and so a lot of working with the medical community and in the medical schools to prepare the new workforce for the future I think would be essential. Yeah, we didn't talk a lot about diagnostics here, mostly therapeutics, but I think in the diagnostic space it's becoming deeply challenged for funding. There's very few new companies here and the reason is that there are significant challenges around reimbursement and regulations, speed of approval, and many of these small innovative companies just can't get through that process in a timely enough way to fund their way through and so we're not seeing these new companies start up and be successful and it's a big, big problem. I think we need more dynamic processes in the regulatory and reimbursement systems. I would say just from a patient perspective how will we make these precision medicines available to more and more patients around the world because I think in the long run the puritive power of these therapies or transformative powers is incredible but it's only gonna matter for global health if we have broad access. Broad access, great. And that requires really looking at this risk continuum and who's taking the risk. The company, the insurer, the healthcare system, the healthcare provider, the patient and how along the way do we shift that risk a little bit so everyone can benefit? Yeah, we just echo that. I think when you have the capability to alter the underlying trajectory of a disease you worry about a technology gap and I think that you look at oncology for example and the gap between what goes on in the community sometimes and what goes on in the academic centers has grown wider in certain clinical settings so you have to focus on that to make sure that you're providing equal access and equal opportunity patients. Thank you all. This is, I've learned so much from all of you today and the great questions from your audience. And 45 seconds to start.