Roche Holding AG (ROG) Earnings Call Transcript & Summary

[Operator Instructions] At this time, it's my pleasure to introduce you to Karl Mahler, Head of Investor Relations and Group Planning. Karl, the stage is yours.

Yes. Thanks a lot, Henrik, for the kind introduction. So welcome to our seventh Science Day this year. The focus is on digitization. And according to my [indiscernible], this is the only Digital Day in Pharma of its kind. So due to the heterogeneous and complex regulatory environment in pharma and health care systems [indiscernible] is we had a comparably low digital maturity compared to other industries, but things are really moving, particularly here at Roche. And we will show you how digitization creates value for Roche along the value chain using very specific 27 use cases. This is the agenda. So we will start off with Alan Hippe. Alan is the host of our event today, is the CFO and CIO of Roche in that capacity, of course, instrumental for the success of digitization in Roche. We have 2 people joining us from the REDs from Early Research. Mark joining us from San Francisco. Christian is with us here in Basel, showing the impact of digitalization on research. Then we have Jacqueline Law, thank you that you could join us from Flatiron. You're also based in the Bay Area. Soon, she will talk about real world data and how that can be used in research. Then we have Steve Guise with us, Global Head of the Pharma Informatics. He will focus more on the later part of the value chain, so development, manufacturing, commercialization. And then Moritz Hartmann, I think you just arrived in Santa Clara, so thanks that you found time to be with us. From diagnostics, he will exemplify how digitalization will have an impact, in particular on point of care and digital pathology. So also very specific, his case. As Henrik mentioned, we have a questionnaire for you, would be pleased if you could fill this questionnaire for us that we can improve for next time. If the Q&A doesn't work for you, we have the Zoom, you can also drop me an e-mail under [email protected]. And I just look at the participants, so we expect about a bit more over 500 people joining this event. With this one, Alan, over to you.

Yes. Thanks, Karl. Thanks for the nice introduction, and thanks to the team for setting up the event. And certainly, thanks to the panelists. I think we have really a distinguished panel here, great that we brought such a fantastic group together. So I'm sure you will enjoy the presentations and it will give you a lot of insights about the use cases that Karl mentioned, and that means this is not really a very broad digital presentation. We will really dig into things and you will see tangible examples of what we're doing here. And then you can judge yourself whether you like them or not. Yes, welcome. I hope everybody is safe and healthy. For some of you, it's the second time in 2 days that you see me. Well, everything got a little bit condensed after the Novartis transaction, which we're certainly excited about. As said, today, it's about how we create or how we work with digital in Roche and how do we make our digital investments tangible, so we'll explain that. And as Karl mentioned, I think we will go through the value chain, and we will show you the use cases, 27. So you will see 27 examples how we approach digital here in Roche. When you look at my agenda, 3 points I would like to touch upon not too long, one about the industry trends. I think Karl has stolen my thunder a little bit at the beginning because you will see that the pharma industry is a little bit behind, and I will explain that. it's quite interesting. It doesn't apply to diagnostics, I guess. Diagnostics is perhaps a little bit more upfront. Then I will talk about Roche itself and have 2 of the 27 use cases. And then I look into the future, which is nothing else, and outlining the opportunities that we have along the value chain. Good. With that, let's go to my first content slide. And before I begin to it, I think we all know why digitalization is really something which is driven. I think, well, we have an aging population. We have pressure on medical costs, now we have the pandemic on top. And I think it's pretty clear that digital and the digital mindset is helpful, really, to move things forward and to create efficiencies, productivity, but also access for patients in such an environment. And when we agree to that, I think you see really on that slide on the left-hand side, 5 trends, which are pretty clear and pretty obvious when it comes to the health care industries. And the first one is that companies go beyond the pill. And what that means is there is definitely, yes, the try and the wish to really connect directly with the patient, so approach the patient directly. Second point is increasing the consumerized health care. That means you combine the delivery with a digital solution. Number three, patient outcomes are easier to measure, which we think is quite an advantage. And you also see there is something with patient -- patient data behind and how you manage it. Number four, the non-healthcare players entering the market, creating ecosystems around patients. I think you've all seen that. To which degree that will happen is a huge debate, but that we create ecosystems altogether. I think that makes a hell of a sense, and we know that we have to strive for that. And last but not least, patients are starting to managing their own health. So more extensively, if you like. So really, digital services empowered to take decisions and even patients take decisions here. On the right-hand side, you see really these impacts and they go through these -- through the layers, which are outlined here. I don't want to go through them. And then, really, through the whole care pathway. Good. Having said all of this, so evidently a lot is happening in pharma, and I would argue as well in diagnostics. The reality is, and you see that on the next slide, is that pharma is behind when it comes to digital maturity. And we have really asked ourselves where is pharma standing. We got some data from BCG, and you see it outlined here on that slide. These are basically self-assessments from companies. And you see really when you look at Asia, that in Asia, well, evidently, pharma is quite well positioned when it comes to digital maturity. And you see, really, progress. The PH 20 to PH 21, I think that's the progress Pharma made from 1 year to another. You see Europe, I think, quite weak. Pharma has made a significant step here, as you can see, but still really pretty behind. And I would argue the same applies to the U.S. So the question is a little bit. I think evidently, I think, that is moving. There's a strive in that direction, so why is that? And I would argue that's pretty intuitive why that is. Well, I think we are in a pretty heterogeneous, complex regulatory environment, which is hard to work against. And we have a rather slow customer adoption of digital tools. And we have, and we all know that and went through this, we have a fragmented health care systems technology landscape, I would argue. So I think these are just a few obstacles. The other one is really dealing with the data and bringing the data together and all the questions behind them, which are justified questions, but also that we have to solve and work with. So the first conclusion is there's a lot of will, there's a lot of drive. But evidently, this development will take time. What you see is when you look at it in digital health, a lot of investment, and increasing investment coming into the field. So I think the hope is justified that there will be significant progress in the next years. But as said, looks like this is more a long-term game than, really, a short-term game. Same is when you look at partners that you can team up with. And these partners are well funded, I can argue, and they really go across the value chain. So evidently, and that's really then the conclusion for Roche, when you look at us, I think what are we doing? I think we invest ourselves and we invest quite heavily and you will spend in digital is roughly $3 billion-plus per year. We have more than 300 key initiatives ongoing, so that number has not changed from last year. And the other piece is, really, we have a lot of partnerships we go for. So I think we really benefit from the environment. We try to make use of it, and the result is that we have more than 30 software products already in the market. On the right-hand side of that slide, you see really the partner, but also the acquisitions that we have done in the field, so we are heavily engaged. When you really look at the digitalization along the value chain, and it's the topic of today, we want to present you 27 cases, use cases. And you see in which field they are on that slide. And I have the pleasure to talk about ASPIRE, which is basically an ERP program plus, I would say. And the other piece is myBuy, which is a global procurement tool. But I can say, even with pride and conviction, I think you will see much more exciting stuff to come in the other areas. Good. With that, let's focus a little bit on Roche and our philosophy is over there when it comes to digital. The first one is, really, the vision from data to insight and how do we want to do that. And I would argue, a key topic of a digital backbone is that you really empower domains to manage their own data. So really liberalizing, democratizing the data. That's something which is always on my mind when I think about these backbones that we're building here. And no matter in which area, could be finance, could be in genomics, could be in a lot of other fields. But tapping into the data, using the data for certain use cases and making that access not so difficult, easy, I think is something which hopefully will deliver the breakthrough we're all hoping for. And what you see is that Roche is operating with a federated Roche-wide data and insights ecosystem. And then on one hand, on the left-hand side, we would like to converge within individual demands. And the other piece is we want to distribute across the enterprise. So very clearly, what we're aiming for, not so simple to do. And here is a quick example of how we integrate certain use cases, and I would argue it's pretty intuitive. So we go and that's the bottom of the slide from a functional data environment, then build really data products, which means we expose commonly used data as data products. And then really go for the cross-functional integrated data, which then is used in certain use cases. So that's the basic principle we're aiming for, certainly very hard to realize. Here is an example for finance, where we have decided that we go from a really fragmented infrastructure that we're having to a seamless data infrastructure. That is not applicable everywhere, admittedly. But here, in that case, I think it is doable, is nevertheless a major endeavor and a major effort, I can say. But if you achieve that, I think, as said, I think we can free up our data, we can democratize it. And certainly, the simplicity, the time to value and the scale, I think, is justifying the effort as well as that you have security and privacy by design and you don't have to reinvent the wheel all the time. When you really dig a little bit more -- a little bit deeper into it, I think really creating the digital backbone in finance as said. A lot of people ask me, how can you work with a network organization in finance where everything is about process, stability control? Well, I think you deliver a major element for agility. And agility, in my mind, is really speed, flexibility and stability. You deliver a lot of stability with a digital backbone. It's exactly what we would like to do and what that slide illustrates. So let me very quickly summarize about the digital and finance asset, ASPIRE. There is one enterprise-wide program for ERP, and I would call it ERP-plus because we have other tools involved even beyond ERP. And you see really a key element here is one process house across Roche, so we are really harmonizing and simplifying processes all across, which I think is a major opportunity and a major endeavor moving forward. Good. I think that we have partners in place here to do all of this is key, but this slide shows very clearly that Roche is shaping the industry standards, and that's what we want to do with that program. And that also facilitates it because there's even co-investment involved here, which certainly, I think, is attractive because, while very clearly, don't want to spend all of our money in digital backbones. I think where we want to spend the money is really at the very end in R&D and innovation, not really to go the next step. Good. A very quick comment on myBuy in finance. This is really about the procurement, predominantly integrate procurement where we drive solutions forward. And that's very helpful because that's also a digital backbone where you would say, why do you need different systems in the company? We don't need that, and we go now here for a harmonizing and simplifying approach. Same applies when it comes to cloud. Roche, in itself, we have around 6,000 applications in the company. I would argue over 25% of these applications are already in the cloud and cloud native, so I think that's quite -- quite a portion already. But certainly, we would like to drive that further. We're not dogmatic here, it's not like to bring everything in the cloud. We will end up, I think, with a hybrid multi-cloud environment. And certainly, we still will have data on-premise, especially when it comes to more historical data. But very clearly, the cloud is a key tool for us moving forward and creating these data pots and data lakes that we would like to use to create more insights in the future. Good. With that, I mentioned future already, a look into the future, and that's my last slide. It's really about digitalization along the value chain and what we can get from it. And I don't want to go now through each and every benefit that we can create. And you see really, on the bottom, efficiency gains all along the way. I think that's perhaps the common denominator here. But we're very clearly new insights in biology, new insights in human disease, new drug targets is a key topic. And then you can really go through this and see the benefits that we can create with that approach. So you see, really, we are we're high on digital. We appreciate the investment. We will invest further in the field, there is no doubt about this. And hopefully, we will be one company and one element, which helps that the perception is not -- that the digital maturity of Pharma is still relatively low. I hope it will help to drive the whole field to get better, and at the very end, to score a little bit better. So more patients will benefit at reduced cost to societies, that's what we have on mind. And with that in mind, I would like to pass it over to Mark. Mark, welcome to the panel.

Thanks very much. Pleasure to be here. If you want to move to the next slide. My name is Mark McCarthy. I'm calling in from California. I head Genentech Research and Development, or gRED, efforts to use human genetics to understand the biology of health and disease and to drive innovation in the discovery and prosecution of therapeutic targets and biomarkers. I'm also part of some broader efforts within gRED that are building capabilities in the generation and analysis of large-scale, high-resolution experimental and clinical data. And really, this is all to enable a digital transformation of many of our research efforts. Next slide, please. Now at Roche, our goal, as you heard, is to deliver more medicines at lower cost to society, a goal that cannot be met by incremental scaling of existing strategies. The approach, set out by Aviv Regev, who leads gRED, identifies 4 levers that we can exploit to accelerate our drug discovery efforts. And digital transformation underpins 3 of these, the investment in human biology, the generation of high-resolution data at scale and the use of advanced computational methods. And it also actually contributes to the fourth, helping us exploit a growing diversity of therapeutic modalities. And our goal, of course, is to harness the power of these data-driven approaches to highlight potential therapeutic targets and then to deploy the phenomenal capabilities we have to turn those aspirational targets into safe and effective medicines. Next slide, please. Now these efforts require us to generate meaningful data at scale, but equally to ensure that the data are made available within the company and of and beyond according to the principles of fair that the data are findable, accessible, interoperable and reusable. The data we need comes from many domains, clinical data from our trials or real-world data, genetic and genomic data, imaging and patient-generated digital data, for example. And together, these can provide a holistic and systematic perspective on the biology of health and disease. We need to scale these data across many individuals so that we can take account of the range of genetics, social and environmental diversity and develop more personalized approaches. And we need to extend the data collection over time to track the changes that reflect disease progression and therapeutic response, for example. Next slide. I'll focus mostly on human genetic big data. That's my area of expertise, but it was also amongst the first areas to bring big data and digitalization to biology. So human genetics studies the consequences of natural experiments of gene and pathway perturbation in free-living humans. And those mirror the perturbations we seek to enable, of course, with our medicines. Unlike many observational studies, human genetics can highlight fundamental processes involved in disease through direct causal inference. And that's because when we find a genetic association, we know that the genome and genetic variation is driving the phenotype and not the reverse. Indeed, there is increasing evidence now that therapeutic targets, for which there is human genetic support, are, when compared to those without the support, twice as likely to result in successful drugs. Next slide. To date, we've only scratched the surface of the genetic variation that is the substrate of human genetics. In fact, the mutations involving each and every one of the 3 billion bases in the human genome, provided they're compatible with life, are segregating in dozens of people across the globe. And as we expand our genetic studies beyond the historical focus on individuals of European descent into Africa, South America, into indigenous populations worldwide, we're building more complete inventories of genetic variation and more granular links between that variation and the clinical phenotypes we care about. To do this, we work to build phenotypic precision, sample size and diversity across multiple data sets. Sequence data from trial participants provides valuable phenotypic data, but often in small numbers. And contrast data from massive biobanks and increasingly direct from health care systems brings scale in connecting genetics and genomics to ever richer clinical and phenotypic data. Next slide. I'll highlight 2 examples of genetic discoveries. The first concerns data gathered from 2 large biobanks. UK Biobank includes clinical genetic and other data from 0.5 million U.K. participants, and FinnGen is a parallel effort in the isolate population of Finland. And genome-wide association studies in these biobanks have been carried out for many phenotypes, but here we're focusing on glaucoma and intraocular pressure, and these generated very clear hits for coding variants in the ANGPTL7 gene. Though different variants drove the effects in U.K. and Finland, they consistently demonstrated that loss of ANGPTL7 function led to protection from glaucoma, suggesting that inhibition might offer therapeutic possibilities. But would such a treatment be safe? These biobanks provide detailed information across thousands of clinical endpoints, so it's pretty easy to ask whether those glaucoma-associated variants have any other effects that could point to other potential treatment indications, of course, or it could hint at toxicities. And here, on the panel on the right, across several thousand phenotypes in UK Biobank, the only hits we saw were with glaucoma. So much more work needs to be done on such a target. These data provide compelling starting point evidence of both efficacy and safety in humans. Next slide, please. A second example relates to the checkpoint inhibitor, Tecentriq, which is licensed for treatment of multiple tumor types. Tecentriq releases the brakes from the immune system, encouraging host immune defense against the tumor. This can, however, set off autoimmune reactions, of course, skin, [indiscernible] and endocrine side effects. Interestingly, those who develop those adverse effects are actually more likely to mount better responses to their tumor. And the likely explanation is that individuals differ in their constitutive immune set points. In some, the immune system is already pretty revved up and releasing the brakes results in a florid immune response against both for tumor and against the non-tumor cells. In others, the immune system's quiescence and releasing the brakes has much less impact. And we were able to confirm this idea using genetic and clinical data gathered from participants in our Tecentriq trials. Participants for the genetic profile that put them at higher risk of autoimmune disease outside cancer were more likely to develop autoimmune adverse events when treated with Tecentriq, but they were also more likely to benefit in terms of cancer survival. And this maybe has some implications on how we might think about using these medicines. In addition, we could use the genetic risk scores to tease out the genes that were driving these effects, highlighting some novel processes that point to some promising therapeutic candidates. Next slide, please. So next slide. So far, I've focused on human genetics. But the most exciting current developments are driven by single cell technologies, bringing a far greater resolution than previously possible in an understanding of the genomic and molecular content of individual cells and how those changing development and aging and how they change in response to environment and disease. And this is just one example from a recent preprint describing a single-cell atlas constructed from human lung tissues, gathered from those who unfortunately succumbed to COVID-19 infection and controls. And this sort of data set provides us foundational information that can aid understanding of the biological impact, in this case, of SARS-CoV-2 infection. Combined with human genetics, data such as these transform our picture of disease. Next slide. It's one thing generating lots of data, it's another maximizing their use. Biomedical data sets have been getting larger and more diverse, but they've often been sparse and disorganized. But recent computational advances, particularly in the AI and ML field have brought us to an inflection point where we can use that scale and diversity to advantage. Now that advantage plays out in different ways. AI, ML methods are adept at reducing complex multimodal data to lower dimensional representations. Those representations can be used for prediction, just as algorithms predict our individual movie preferences. Based on the patterns of other's behaviors, we can find molecules that have cellular signatures that mirror some desired clinical outcome. These models can be extended to provide causal inference, highlighting the processes where the therapeutic impact of intervention may be greatest. Next slide. I could show many different examples of AI approaches, but focus here on early efforts in AI-driven drug design. Now one reason why AI-driven drug discoveries had limited traction so far is that AI models do not generalize well. They often do a good job of predicting compounds very similar to the ones we already know, but don't do a good job of predicting properties in novel compounds, which is the ones we really need. So leveraging recent progress in the field of deep learning, colleagues in gRED built GNE prop, which is a model that can make the leap, namely, it can generalize well on novel compounds. And this is an example applying it. The antibiotic discovery focused on the need for new molecules that are active against gram-negative bacteria. Next slide. They leverage our internal existing high throughput screening data, built a model that precisely characterizes activity and, in particular, captures activity cliffs on molecules that are very dissimilar to those in the training set. Those model predictions were then tested on a catalog of 2 billion synthesis on-demand compounds, and several hundreds of those with predicted activity have now been ordered for internal experimental screens. Final slide for me. I have given you a taster of how the generation of digital data at scale, particularly when gathered in humans, and using high-resolution assets can accelerate our understanding of the biology of health and disease and highlight ways in which we can manipulate that biology for therapeutic benefit. And with that, I'll hand over to Christian.

Thank you very much, Mark. My name is Christian Gossens. I'm heading, here in Basel, the digital biomarker area out of the pRED organization. And it's my pleasure today to talk you a little bit through the new digital normal in our organization, in particular for the patients related to our work, the clinicians and the researchers. So imagine an R&D ecosystem in which all participants are increasingly interconnected. And so our strategy here helps us to evolve on their journey towards making our work more efficient, but also gain complete new insights and drive new digital translational science. So you have, on the left-hand side, the ecosystem of the researchers and scientists that going to be using increasingly the power of predictive algorithms for drug efficacy and safety investigations. We're going to make more and more use of modeling and simulation and enabling self-service data analysis. We need to get much more interaction with the patients in our clinical trials and also gain a much deeper understanding of their individual disease biology. And on the right-hand side, you see what I just mentioned, is the operationalization that digital can really drive going forward. Now the very first massive investment that we are doing is the One-D program that was launched 3 years ago and investing really in the sustainable increase of pRED's productivity. We are aiming for roughly 20% until 2025 here, mainly investing in 2 things that also Mark mentioned: automation, clarification, digital pathology, getting better in touch with our patients in the clinical trial setups with clinical platforms, using artificial intelligence across the entire value chain, leveraging modeling and simulation, and also exploring completely new emerging technology like quantum computing, which I'm going to talk to you about on this slide briefly. So like with every emerging technology, it is very important that you investigate this very properly and find a good link between that emerging technology and your respective use cases. So we have pulled a task force together to do this and also to help these emerging providers of this new emerging technology to connect to our needs in the space of chemistry simulations, for example, for affinity calculations, how the docking of the molecules happen to a certain target, more complex protein folding activities, and last but not least, imaging classification has a huge computational power need and we investigated where we are there. If you want to find out more, there are public preprints available on the links underneath. Summary maybe is, in a nutshell, very promising, but very early days. The -- if you move a little bit further down the value chain, let's come to the preclinical safety. So in particular, everything that is related to animals. Our colleagues concentrated in a brand-new state-of-the-art facility that also won an award last year. Internally, it's highly automated. It's really impressive. It looks like a factory, very kind of close to sterile environment where all the trolleys are moved automatically, where robots are shifting the cages, bottles and feeding of the animals happens automatically. And on the right-hand side, you see very nicely how also then the lab of the future is actually already relative now where you have this smart lab environment where the building IT system is merged directly with the research IT needs. And so all the experimental conditions that our colleagues are creating and having during the setups that they're doing there, they can be replayed and reinvestigated afterwards. So if any parameter was missed, maybe during the setup of the experiment, you can be certain that it was measured with any of the sensors built in there. Going further down into clinical trials. Recently, we had a very interesting need for Neovascular AMD. They are -- the state-of-the-art or the gold standard is a pretty invasive, slow and costly procedure. And here, if we move towards what is called SD-OCT scans, which is a noninvasive scanning of the eye, which is still stationary, so the patient has to go somewhere. But you can capture them, these pictures, and these high-resolution pictures of the eye can be investigated in a 3-dimensional way. And deep learning allows us there to detect pathologies, a segment here, the setup, and then this -- the learnings that we then have and these features that we identify can then be used in a second step with machine learning to be classified. In this case here, for a more classical or cold type of a certain setup. That is obviously very powerful for patient stratification and might be used now in all future clinical trials of this indication. Now if you think about bringing sensors even outside of clinical trial sites, this is where the rest of the talk are going to be focusing on, and that mainly means like we equip our patients with sensors, with devices that they can take home when they leave the clinical trial site. And we are focusing still very much on the neuroscience space there. You see the areas here, disease areas, from Parkinson's, Huntington's, SMA, multiple sclerosis up to schizophrenia. And you see there, there's a lot of synergies that we can gain here because very similar assessments, for example, in the cognition space can then be reused in other indications or disease areas. And so we learn a lot building this metric, this portfolio that can be used to be more sensitive, precise and objective in our data collection compared to, in particular, classical physician rating assessments. This is also, as we can see, a great release of the burden to the patients, and the uptake is very well. Now this has been really, really fantastic during the COVID pandemic so far. What you see here is essentially that patients were facing a reduction of their life space of, really, yes, let's say, 50% at least on the left-hand side, right? So you can see everything that's more on the left-hand side of the chart is before the lockdown, then the greenish bars are the activity just before -- when we learned about COVID just a couple of days before, essentially, then lockdowns were executed. And you see that on the right-hand side, the bluish bars, there was a substantial reduction of life space in patients. Well, that obviously prevented them partially from even going to the clinical trial sites. The good news is, on the right-hand side, you see that patients just kept doing their active tests as they were asked from home. And that obviously, in these cases, was a fantastic influx of information for our researchers going forward. Now how do such assessments look like? You see on the life -- left-hand side here on this chart, in orange, some of these active tests. And so one that is then depicted also on the right is around patient holding a smartphone in his or her stretched arm. And we're picking up the acceleration from the arm, depicted there on the y-axis, essentially, and then plotted versus the x-axis. The physician rating assessment from 0, meaning like no tremor, to 4, very high tremor. Well, 2 key observations from that chart is look at the healthy controls measured by the sensors, essentially being much, much lower compared to the non-tremor observation by the physician in the clinic. So there's already obviously a difference. And the second thing is that you see among the patient group that is classified as having no tremor when they saw the physician, you clearly see there are numerous patients, which clearly showed tremor at least some point in time. Now very impactful or very impactful -- was this methodology also, in particular, in a situation to judge and make decisions on the future of drug program, in this case here for prasinezumab, where the primary endpoint, as can be seen on the left chart, was kind of inconclusive and didn't show essentially much effectiveness at all. A subscore , an exploratory subscore around [ bradykinesia ] showed, however, that there might be a signal of efficacy. Well, the digital score collected from the devices I just mentioned to you then really helped clearly distinguishing between the placebo and pooled patient arms. And the good news is even now, 2 years after starting the trial, that trend continues, as you can see on the very right chart. So this then encouraged our senior management to continue investing into the program and starting a Phase IIb to collect additional data. In particular, in this time, including digital assessments. Now having internal discussions is great, but we need to convince the community, and we do this by working in a very collaborative setup. Building the consensus, engaging the scientists across the organizations to finally work with the regulators to accept and shape the setups that we need in order for them to approve in a reliable, predictable way the setup of our -- the efficacy of our experimental medicine programs. Last slide, the multiple sclerosis setup, which is a nice collaboration also with our diagnostics division, which acts as a legal manufacturer. And so the assessments that we have been using and developing in the clinical trial setup are now being exported, if you want so, for out of clinical trial usage in the clinics by physicians who can prescribe this to patients. And so you have the ability to track disease progression and have a much more educated dialogue between physicians and their patients. With this, I want to hand over to Jackie.

Thank you, Christian. Hello, everyone. I'm Jacqueline Law, the Vice President of the Corporate Strategy at Flatiron Health. Pleasure to be here. Next slide. Next slide. So for those of you who don't know us, Flatiron Health is a health tech company dedicated to helping cancer centers strive and deliver better care for patients. So through our clinical and data science expertise, we translate patient experiences into real-world evidence to improve treatment, inform policy and advance research. Flatiron's north star is to transform the way cancer is treated and researched around the globe by developing disruptive and durable technology and scientific solutions. Flatiron is an independent affiliate of the Roche Group. The company was founded in 2012, acquired by Roche in 2018 and remains a separate legal entity. This autonomy underpins the Flatiron's work with many life science companies, community center practices, academic medical centers, regulatory bodies and efficacy groups. Next slide. So what are our priorities? So we continue to unlock more real-world data and real evidence in use cases to help answer a wide range of research questions, expanding beyond the U.S., linking EHL data and other modalities including genomics, claims and imaging. We continue to advance novel methodologies and engage with stakeholders like health authorities to accelerate the use of real-world evidence in decision-making. We are increasing connecting research and insights with everyday care. And we continue to invest in supporting cancer centers in the transformation towards a value-based future. I'll touch on all these topics in the coming slides. Next slide. So at the core, Flatiron is a technology company. We provide technology services that help community oncology centers deliver better patient experiences, operate healthier practices and facilitates modern research. We have built the leading U.S. oncology real-world data source derived from the electronic health records of millions of patients across the United States. We curate, aggregate and [ de-identify ] clinical genomic and other data. The real-world evidence generated from the data helps researchers, drug developers and regulators to answer critical, medical and scientific questions. Our data come from [indiscernible] platform as well as other electronic health record systems. So after almost 3 million patient records available for research in our network, 80% come from community purchases and 20% come from academic cancer centers. Next slide. Our biopharma partners have used Flatiron data to address evidence, get and generate actionable insights across the product life cycle. We have seen use cases span across discovery and translational research. A lot of these cases in clinical development, applications for regulatory approval and market access, as well as post-approval and commercialization. This is nonnot exhaustive and growing list of use cases that can be supported by the Flatiron real-world data, and I'll share some examples of them today. Next slide. It's very important to shape the field in the real-world data and bring everyone along. We collaborate with many to advance real-world evidence for regulatory and HTA decision making. Our collaboration with the U.S. FDA began in 2016, and that continues to inform our approaches on data quality and analytical methodologies. In July 2020, we have announced a 3-year partnership with the U.K. NICE, which instantly helping NICE to expand the use of real-world data in service of better outcomes for U.K. cancer patients. And we continue to shape and inform the real-world evidence through many long-lasting partnerships. For example, with the Duke Margolis Center for Health Policy, we will [indiscernible] collaborative and with the Friends of Cancer Research. Next slide. Flatiron real-world data can support meaningful inside generation relevant to a global cancer population. Evidence generated from our data have informed decisions by regulators and HTA bodies around the world, and these intrusions have a direct impact on patient care. So I'd like to share 2 case studies now. The first is the use of Flatiron data to support a request from the European Pharmacovigilance Risk Assessment Committee for Kadcyla. In this case, Flatiron data was used to fill an important evidence gap in safety, enable a label change for Kadcyla in Europe. I'd like to highlight this retrospective real-world data study enabled a sponsor to submit data 5 years earlier than their prospective study would have allowed. And importantly, this real-world data study led to a label update that provides important information to physicians and patients when making treatment decisions. Next slide. The second case study was to use Flatiron real-world data to support a U.K. NICE decision on the coverage of Tecentriq in locally advanced or metastatic non-small cell lung cancer patients. In this case, the Flatiron data was used to provide an estimate on the long-term survival for patients treated by the standard of care. The Flatiron results were being considered because of the quality and the recency of the data. NICE has agreed with the sponsor's approach of using the Flatiron data and reversed the initial decision, and this positive coverage decision allows more patients to have access to an innovative medicine. Next slide. Historically, there's been a gap in bringing together large representative longitudinal clinical and genomic data. The ability to correlate genotypes and phenotypes is critical to advanced cancer research and care. The CGDB chemical genomics database links Flatiron's longitudinal EHR-sourced clinical data with Foundation Medicine's genomic testing data. We have seen CGDB help researchers to conduct outcome studies in biomarker-defined populations to discover [ rightful ] targets, understand mechanism of resistance and to more effectively design clinical programs and clinical trials. As of Q3 2021, the CGDB included data from about 87,000 patient records. Next slide. So building on the concept of CGDB, the prospective clinico-genomic study, PCG, is piloting a novel technology-enabled platform, collecting non-standard of care biomarkers at the point of care. This platform aims to reduce operational burden and streamline workflows at the point of care to support clinical research. The obstructive of the PCG was to evaluate serial ctDNA as a predictor for response. PCG was initiated in December 2019, and despite last year pandemic's impact on clinical trial execution, it was fully involved with 950 patients within 18 months. This study is a great proof point on the potential to transform how clinical trials are conducted, making research more feasible at the point of care and more inclusive. Next slide. Here are some recent high-impact use cases supported by the Flatiron real-world data and clinical genomics data. There is a wide variety of applications here, including using Flatiron data to create an external control arm to contextualize clinical trial results, to support race and ethnicity disparity research, to enable study of real-world effectiveness and to have clinical trial design using real-world data and AI. Next slide. Continuing on the theme of linking EHR data with different modalities to unlock more use cases, we recently financed a partnership with Komodo Health, joining Flatiron's EHR-derived data with deep clinical data and longitudinal follow-up and outcomes with Komodo's close claims data, which include medical and pharmacy encounters. The combined data set can enhance our understanding of the entire patient journey and provide even more insights into what happened to cancer patients outside of the oncology side of care, such as health care resource utilization, treatment persistence and adherence, adverse events, comorbidities, concomitant medications. Next slide. Clinical imaging, all right, is a very important modality used along the entire anticancer patient journey. So we see imaging being used from detection, diagnosis, staging to treatment response assessment. We have built a clinical scan database with a large-scale scan data collected at the point of care linked with longitudinal clinical and genomic data. We think that clinical scan data can support both real-world research questions, for example, to supplement real-world outcomes like real-world response and real-world progression as well as to support AI algorithm development. So we are very excited about the possibilities of generating insights from this data. Next slide. So back to the point of care, oncology practices and care teams face many challenges in their day-to-day practice. More therapies emerge every year and there's an increasing complexity in the reimbursement space. Almost 40% of U.S. community oncologists use our OncoEMR platform and they rely on Flatiron's technology and services to bring better outcomes to patients. So we continue to invest in supporting cancer centers in the transformation towards a value-based future. As an example, Flatiron Assist is a new tool that supports evidence-based care and streamlined clinical decision support. Next slide. We've pioneered in many ways to maximize the value and importance of real-world data routinely captured in the EHR. So looking forward, we are increasingly connecting research and insights with everyday care by building a novel platform to bridge real-world care with clinical research. With our point-of-care technology solutions, our deep expertise in oncology real-world data and our exciting steps to integrate even more sources of data, we believe Flatiron is well positioned to help transform cancer research and patient outcomes. Thank you. I'll hand it to Steve.

Thank you, Jacqueline. My name is Steve Guise, based here in Basel. I look after the Informatics team supporting the Pharma division. And I'm going to talk through a couple of use cases a little bit further along the value chain, starting in commercial. We'll spend a bit of time looking at our go-to-market strategy, talking about clinical development, particularly clinical operations and how we execute clinical trials, moving into new manufacturing paradigms with individualized treatments and spending some time talking about how we're supporting patients and physicians more effectively, and lastly, but not least, pandemic response from an IT perspective. So the first point is really around our commercial go-to-market model. And as Alan pointed out at the beginning of the session, our industry has lagged behind a little bit, other industries in different sectors. And in one area, how we engage our customers in the field has been a relatively traditional model for some time. A number of years ago, we decided to start to transform our go-to-market model and our commercial customer engagement model. And this is both an organizational transformation but also enabled through digital technology. As we build towards our 10-year ambitions, one of the ambitions is that we engage our customer ecosystems around the world increasingly as One Roche, bringing together pharma, diagnostics and our insights offerings into a more integrated solution. We've built here a road map that shows a time horizon of how we're going to get there. The first time horizon is to really focus on our key stakeholders today and to make sure that our customer experience is really leading from an industry perspective. The next time horizon is to really partner at the ecosystem level, bringing together our One Roche offerings to our customers in a more integrated way. Finally, being a more integrated influential player in the market. Underpinning that is a major investment in technology, which we call EpiCX, it's a program I talked about last year in a similar event. And we're significantly down the road now of this investment. And it's really an end-to-end ecosystem of digital solutions that will transform our customer engagement model, and the intention is to make our customer experience as meaningful as our science. This is a very data-driven approach, end-to-end driven by analytics, looking at how we engage our customers through digital channels. Of course, during the COVID period, this has been significantly accelerated because we've had little access to our customers in the real world. So the digital investments that we've made have really enabled us to build a great customer experience despite the challenges of the pandemic. So this is now rolled out in its entirety in a number of markets. The content hub, which really supplies all the information that we want to share with our customers, is rolled out already globally. On this slide, it details actually our Veeva Vault PromoMats solution, which is underpinning the content hub available to all countries around the world. We've been building up a library of assets that we can reuse along all of our different markets. And as you can see in the graphs here, the available material has grown significantly over the last year, and our ability to get content created and approved is reducing in time. A key metric that we're looking at is how much we reuse content. So there's a significant saving opportunity here. When we started this program, we were actually reusing a relatively small amount of content around the world, which meant we were duplicating and recreating work in many markets. This has already accelerated to 14% across the globe. But in the countries where we've really focused on leveraging our content services lab, we're already at 78% reuse. So these investments have enabled significant cost savings that we are then able to reinvest into R&D. In clinical operations, we know that our ability to find sites, to find investigators, to recruit patients, to get them enrolled and to complete the protocol is a real critical barrier to executing our pipeline at speed. This is a short example here, which shows how we've used a much more data-driven approach. This example comes from the U.S. where we're using data to find sites which are high performing, make sure that we can recruit our patients effectively. We can also discover new clinical sites that perhaps we've not used before and engage our clinical trials much more effectively. This will result in a significant reduction in time and significant reduction in costs. When we are launching patient solutions, medicines and the digital solutions that Christian mentioned, we also need to be able to help patients gain access to our medicines. So we're equipping our field-based patient roles with a solution that will help them manage and navigate the patient journey to help them with reimbursement challenges, gaining access to medicines and supporting them more effectively, including home delivery of some of our key medicines that require timely delivery of product. The next example focuses on manufacturing. As we are moving into novel modalities and new types of medicines, we are, of course, continuing with our current paradigm of manufacturing which is making large batteries of medicine at scale and distributing them through the supply chain. But we're also moving to much more highly individualized therapies like CAR T-cells, RNA-based vaccines, individualized cancer vaccines. So we are developing new platforms that underpin a supply chain that may start with the patient with a biopsy where we take a sample from the patient. We sequenced the cells, identify the right kinds of mutations, synthesize the medicine and ship that back to the patient. That is an end-to-end process which is very new for our industry, and we're building the technology that will enable that. That is true for one-to-one kind of patient paradigms, but also where we have much smaller batches of medicine that require late-stage customization and distribution. So underpinning that are our 2 key platforms that we are building. One, which is facing our physicians. So it enables them to register new patients to submit their samples through to our testing paradigm, watch the custody -- the chain of custody of the samples through to a treatment generation and shipping that back to the health care center for delivery to the patient. On our side, we have a similar solution that enables us to receive those samples, to do the required work to personalize the treatment and track the individual information of patients end-to-end in a safe and compliant way and deliver those medicines, again, backed by Korea, to our physicians. And this process, especially in the field of oncology, needs to happen in days, not weeks and months. The last topic that I wanted to talk about was really the response that our broad informatics organization has taken to the COVID pandemic, which is, of course, now 18, 20 months in. But a couple of graphs that we put on this slide here show that we have essentially overnight, and I think Alan talked about this last time, enabled the workforce of nearly 100,000 people to seamlessly work remotely from home. This continues, and we're now finding our new normal where we are having blended meetings like the one we have today, where we have in-person and remote collaboration. And we're moving more and more of our workload to the cloud or making it accessible remotely on any device from any location. And what I'm very proud to say is that the kind of productivity of our workforce has actually increased during this time, I'd say the amount of work that can get done has significantly improved. We've also significantly reduced our travel budget. I think the expectation for the future is that we'll have half the travel budget or less in the future, and we've been able to reinvest those monies elsewhere in the value chain to create additional value. The other element, of course, is the sustainability side. So the amount of CO2 reduction from less business travel is significant, and this is one of our commitments for the future that we will maintain. So with that said, I would like to hand over to Moritz, who will talk us through some diagnostics examples.

Thank you very much, Steve. And once again, good afternoon, good morning from California. My name is Moritz Hartmann, the Global Head of Roche Information Solutions. I would like to address 2 use cases with you, which showed how we're expanding beyond our core business by leveraging the strong footprint that we have in diagnostics, but also by delivering and leading towards some of the mega trends in health care that Alan pointed out at the beginning of this presentation. Next slide, please. The foundation of our offering are strong platforms that we have both in our laboratory customer setting, where we have an on-premise or hybrid platform of which you will hear more in the course of next year. And also a cloud platform called cobas Infinity Edge, that we're launching later this year and which I'll speak to later in this presentation. On the health care provider side, we have our NAVIFY platform that has been built around several use cases in oncology and that is now expanding beyond. But in this space, we are also collaborating with third-party platforms, allowing us to access customers that have already an infrastructure in place and still can benefit from applications that we're providing to them. Of course, these platforms are as well part of our product offering in the course of a Platform as a Service business model. On top of our platforms, we have front ends, and these front-end applications actually allow us to enable another business model, which is a marketplace. And again, I'll speak about how we're using the marketplace to build ecosystems in this presentation. And then last but not least, over this marketplace, we're then deploying Roche-built and Roche-owned applications and algorithms as well as third-party applications and that in a Software as a Service business model. Next slide, please. Looking now closer into the cobas infinity edge cloud solution that is particularly built for point-of-care offerings, the core part, which you see at the bottom of the slide, is really the connectivity to multiple point of care and home care devices. This really enables our customers that, as well, increasingly integrated health networks to connect all the different parts of their network from the brick-and-mortar hospital setting all the way through GP or pharmacy to a home setting to administer and connect and remain connected to all their deployed devices and to bring the data back into their systems. Notably, the devices that you see here on the bottom are exemplary and our Roche devices, as well as third-party devices, that can be connected, including sensors that patients carry themselves. I would like to, on the next slide, give the example of how this can actually make a difference to our customers with the example of the cobas pulse, let's say, hospital glucose meter that we're launching as well later this year together with the cobas infinity edge cloud platform. Again, you see the health network on the left side. And the cobas pulse meter is actually based on an Android technology, so it has the ability to host a number of applications on the meter itself and to display that on the screen. Now this allows the nurse, bedside in the hospital, to not just take hospital glucose measures, but to actually, through third-party applications, have a number of different jobs done right next to the patient. A few examples are clinical decision support. Since we're speaking of glucose monitoring, this could be a bolus device application. But it can also be vital signs being tested through the camera system such as the heart rate or digital biomarkers, which have already been explained by Christian earlier, they can be all generated next to the patient in the hospital. And on the next slide, we will see what then actually the nurse can do with this. Our solution connects in with what we call the Unite part. It actually connects it to the administrative system of point-of-care devices. This means the point-of-care coordinator has the ability to see where are the devices, who is using the devices, and are these users enables and qualified to use the devices, and have the devices been updated and have the necessary quality control measures in place. Very important as well is the Scribe functionality, which allows not just the glucose measurement, but as well all the other measurements that can be taken through the applications to be directly integrated in the customer's EMR system. The Smart functionality actually is enabling this marketplace that I mentioned, but also allows for remote software updates, which is a critical element and a key enabler for such ecosystems. This is coming still in December this year, and the future then goes in the area of analytics, where we can think about the number of public health or as well, hospital setting reportings and analytics that allow the hospitals and the integrated care networks to improve their efficiency and to provide a better and more integrated care to patients. If we move to the next slide, please. Another example of this is in the area of digital pathology. Here, once we have digitalized the slides that previously have been processed in a fairly analog and manual way, once they are digitalized, we again have 2 platforms. One, the uPATH platform is an on-premise deployment for customers that would like to have their software not stored in the cloud, but being actually installed on their premises. As well as the NAVIFY Digital Pathology cloud solution, again, allowing to leverage cloud technology for the deployment of the digital solutions to customers. On both systems, we are then again deploying applications. Roche applications, such as our breast cancer panel or our personalized health care algorithms, but also our third-party algorithms. And here, we recently announced the collaborations and partnerships with PathAI and Ibex. And of course, if we move to the next slide, these solutions are integrated in our NAVIFY oncology suite of products. Another example here is our digital pathology PD-L1 algorithm, which helps the pathologists to interpret the biomarker. The NAVIFY Mutation Profiler then provides to the oncologist therapy options as well as identifies clinical trials where patients can be enrolled on to benefit from new and innovative treatments. The NAVIFY Tumor Board integrates data and information from radiology, pathology and clinical data together, allowing that the Tumor Board can effectively, and on the basis of the best information available, take decisions for the patient therapy moving forward. And of course, as outlined by Jacqueline earlier in this presentation, we also have the pillars of Flatiron Health and Foundation Medicine, which are increasingly integrated in our overall oncology offerings. And with that, I'd like to hand back over to Karl and [ Basel ].

Yes. Many thanks, Moritz. Many thanks, all speakers. That was a real value run-through through our digital solutions, which is really state-of-the-art and leading in pharma health care.

I think we'll have the Q&A. The poll now open as well, so if you would like to participate, we would really appreciate it. It will help us to improve next time, and it's also a very nice tool to be integrated into the Zoom meeting so you can do it in parallel. With this one, I wanted to open the Q&A, so you can ask your questions via the chat or you can drop me an e-mail. So we even have time now for, let's say, 15 minutes for the Q&A. And [ Vimal ], you are the first one. I'll open your line now.

Very interesting presentation, clearly, lots of activity. So can I just first ask, and maybe it's a high-level question, but how much can Roche actually save from digitalization across the value chain? I appreciate it's quite a difficult question to answer, but particularly within R&D and SG&A, you've highlighted plenty of examples today. But if we were to take today's spend levels, are you able to give some form of quantification of where we could get to from a savings perspective? And to be clear, I'm not really asking about margins, as I know that these savings could be reinvested, but I'm just trying to think about the potential savings. So I appreciate it's quite hard to answer, but curious to hear your thoughts. And then my second question is on the $3 billion of digital investment per year. And how does Roche actually measure the return on that investment? Again, difficult to quantify, I'm sure, just given the digital applies across the entire value chain. But how do Roche ensure that the returns on these investments remain strong?

Yes, let me give it a try. And I think really, [ Vimal ], I think you made the right point really at the beginning, hard to measure. And I would argue, we are debating today use cases here because it's a really fragmented endeavor that we're going through. I think in some use cases, the return is quite obvious and measurable and quantifiable. In other areas, it's more directional. I would call here R&D as an example. So I think for us, we all know that really investing in the area directionally, I think, makes sense. Let me say where it is very well measurable, certainly, is in all the areas that we've talked about with ERP and the programs that we're having here, so I think finance is very clear. We have the transformation. We measure that, and that's one thing. I think I'm sure Steve will make a couple of points because when you look at EpiCX and what we're doing over there, I think here we have very clearly measurable returns. So I would argue that's the thing. And then certainly, the more we go into more the exploratory fields like R&D, et cetera, I think it gets a little bit more fuzzy. And here we know the direction is fine. But is it like that we can put a clear number on it, and that's not the case yet. You know that at Roche, I think we are used to take risks. That's what we're doing. I think that what this company stands for and we are built on innovation. And we think that digital is a major enabler for innovation in our company. I think you made the right point about the margins, and that would have been my, if you like, my answer right at the beginning. Certainly, what we're doing is we work on profitability. And what we would like to do is we want to defend our margins. I think we have comments on what's happening on the diagnostics side, and certainly, we have to be careful. COVID is a major boost for diagnostics. But even when COVID slows down, I think the expectation is that we move in the right direction, and digital will play here a major role. And the other piece is really on the pharma side. If we can enable the pipeline to be even more productive than it has been in the past, I think the return, if you like, would be quite measurable. So I think overall, I can say, well, you're right. The savings will be reinvested, so defending the margin is really what's on our mind. Some areas, very measurable in some areas, quite fuzzy. But Steve, you have a comment, I think.

Yes. I think one point that I would add, which I'm sure Alan and Bill have shared before is the significant reinvestment we've made from sales and marketing, G&A into R&D and the commercial go-to-market digitalization journey that I talked about earlier. There is a significant technology investment that we've been making over the last 2 years, but it's also enabled a huge transformation in our field operations. So we have moved more exclusively to digital engagements with our customers and reduce the number of infield roles that we have, and those kinds of savings have been reinvested elsewhere in the organization. So that's one clear example. And content creation in the commercial space, we were running at a kind of run rate of about CHF 800 million spend in the content creation space with a relatively low reuse rate. By driving up that reuse rate, that is immediately savings that can be reinvested elsewhere. So -- but 2 concrete examples.

Yes, very good. And just to underline that in the cost of sales in the year 2020 had savings of roughly CHF 1 billion and in M&D of roughly CHF 600 million in 2020, and it was surely enabled by all the digital approaches that we have taken.

Christian, you mentioned before a 20% improvement, which you aim for. If you can give a bit meat around the bone, what exactly you're aiming for?

Exactly. So you have seen that overall, this pure-digital strategy that we have been coming up with, right, is really endorsed with the ambition to get down in the next 5 years in terms of operational activities that we have there, increase the productivity. And here is a combination of operational activities, plus doing better decision-making. And in total, we really are challenged also by our management, very, kind of on each and every use case that we bring forward, right? It's like, okay, you want to invest in this, what do I get in 5 years in return? And I think that's really the right way to manage this investment. And obviously, now we are very ambitious there, but I also think that we're going to see that return.

Okay. Very good. There's a question for you, Alan, from [ Ulrich Mulner ]. He was wondering, the use cases show -- are very well related to the current business. So he was obviously happy with that, what we have seen. But do you see the digital solutions also in the future as a stand-alone business, IG or EG as digital therapeutics or integrated care models where the [ road trucks ] would only be part of the offering?

Yes. I think I'm clear yes to that. I think some of the endeavors that we're going through at the moment will take their, if you like, their stand-alone business case and have their stand-alone business cases. That's also why we've shown the use cases. I think, certainly, everything around the patient journey is something where we build business cases as well and where we will see how it plays out. I have to admit that some of these use cases are not so -- they are still fuzzy. And how we can monetize them is still a question mark. But it is very clear, when you think about oncology, I think oncology, I think today, you have roughly -- and please, panel, correct me when I'm wrong, but I think you have roughly 1,500 alternatives how you can treat a cancer patient. In 5 years from now, external opinions say it will be more than 30,000. So the question is, how can you navigate in the field? So this starts really with comprehensive diagnostics. And we see, especially when I look at FMI, we see a certain underutilization of the opportunities in the market, and we can even say that. We have a couple of drugs which are very focused on specific mutations. I would argue the demand for some of these drugs is not on a very high level. And given the prevalence, we would argue, there is perhaps more opportunity. And why is that? Because these patients are not found. When these patients have a specific mutation, it's not really that there is a clear pathway how to find these specific mutations. So evidently, not every patient is really taking the opportunities and getting the opportunities that, really, biology and drugs provide already today. So it starts with comprehensive diagnostics. It goes through insights. So really then saying, okay. And based on the data that we generate from the patient, we benchmark this, we look at clinical trials and all that stuff, creating insights and then, really, the oncologist with a lot of support besides on the specific treatment for the patient. And I think when you just look at oncology, I think it's very clear what's happening there. And I'm sure there will be opportunities to monetize it. Let me also say, if we succeed in that patient journey, and oncology is just one example, also, we expect that health care costs might go down in these specific cases because we still see that there is still, how should I say it, is the hit rate for the right treatment for patients perfect for the time being? I would argue that's not the case yet. I think there is room for optimization in a world where every patient gets in the first treatment, the right treatment. I think would mean a lot of savings in the health care system. So I think, really, that's an incentive for everybody to work on this. And I'm sure that this will be, how should I say, that you can monetize this approach as well. And I'm not -- I don't want to go now to multiple sclerosis, I don't want to go to ophthalmology. But I would argue, in these disease journeys, in these patients' journeys, they are basically the same patterns you can follow as you see them very tangibly in oncology already. It was a long answer, [ Ulrich ], sorry. I hope it came across.

That's perfectly fine. There is one coming in, [ Kenny Smith ]. It's about going across the audience here from your, not sure who is best -- Christian, maybe Mark, maybe Alan, you. It's from a clinical trial development point of view, what would Roche like to see CROs assisting in the digitization process? I mean it's a bit also in your field, Alan, where the -- with your initiatives for the [ procurement ], but also in the clinical trial process itself when it initiated. So what would you like to see from the CROs in the future where digitization is concerned?

Yes, I can certainly put a thought there. I mean we're partnering very closely with our CROs, and they play a key and important role in helping us with things like site monitoring, identifying sites to -- I talked about this in my example, the clinical trial optimization through data. This is something that we do both ourselves, but we also rely very closely on our CROs to do that more effectively. And of course, they are under increasing pressure with their margins. So I think the digitalization of clinical research on their side is also a huge advantage for them. So this is clear.

Yes. Thank you. So it's an integrated process. There is one coming in, maybe also going to you, Steve, on -- have you discussed the use of digital-based endpoints with the FDA? So this is also a [indiscernible], yes. Mark and Christian. It was [ Sarita Kapila ] was specifically wondering about Parkinson, but maybe not only about Parkinson disease, neurological diseases, so yes.

Maybe it's a question I can take. So the path forward regarding digital endpoints is not yet a straightforward one. Those tools were not existing until very recently. Hence, there are no final guidelines yet available. So the very important part is, and I hope that came through at least a little bit in my second last slide, that working with the community with all the key stakeholders in that ecosystem is very much key, engaging early on with the FDA. So we are having those discussions now for multiple years in Parkinson's, in Huntington's, in MS where we are active. Because without having that very closely tied in discussion with the authorities, the path forward won't be clear. And then we would take a huge risk kind of collecting data that in the end might be jeopardized by things that we hadn't considered in the onset. So building that mutual consensus that I mentioned is absolutely key there. The good thing is the regulators are now fully -- they've bought in, right? They have even themselves created teams that are dedicated to digital biomarkers and digital tools overall also for us MDs. And with those teams, we have now really these expert exchanges and can ultimately ride together the guidelines. Obviously, they're challenging us as much as they can, and we're showing them what we can provide. And this together, kind of teamwork, will end up, hopefully, in even primary digital endpoints.

Moritz, anything from your side? Yes, there is a question to you, Moritz. Maybe we can go beyond this one, the last speaker. This is you. Obviously, there was a question from [ Teresa Ramos ] on the integration of your systems. So you talked about digital pathology, NAVIFY, FMI, Flatiron. How connected are these systems at the moment? So can you really make the use cases out of the system build what you have?

Yes, definitely, and thank you for the question. Everything that is under the NAVIFY platform, so those products that carry as well the NAVIFY brand in front like the mutation profile, digital pathology and the tumor board, are basically integrated as well from a data perspective and can be deployed and are in some customer sites. I think the link really was Flatiron and FMI is something that we're currently working on and effectively speak to a number of customers, but we would not today be able to disclose a customer site where such an implementation is already completely existing. It is in part, and maybe jump in, you want to add to that. I think it is working in parts with the digital pathology and Flatiron Health.

Okay. Yes. Thank you. [ Charles Pitman ] is asking on the analysis of large data sets. I guess this is the one for you, Steve, to determine trends. What are the primary hurdles to democratizing and analyzing the data you are using? IG, are there any legal restrictions in collecting and holding even anonymized data? And do these vary from country to country?

Yes. I mean I can start with an answer to that, and then I would hand over to Jacqueline or Moritz to make a comment. I mean, the first point is that when we are enrolling patients in clinical trials, we try to seek broad consent on how we can use their data. Of course, there is an initial primary use of that data, which is associated with the trial. But we would like to seek a broad consent to be able to reuse those data on studies in the future, maybe even the ability to recontact patients if we have tumor samples that we want to interrogate. So consent management is something that's very key, and people can withdraw their consent at some point in the future, and we have to be able to trace that back and then remove their data from various data sets. So we are very mindful when we make data freely available within the company that we're respecting those kinds of privacy and consent issues. Maybe Moritz, or Jacqueline, you want to add anything?

Yes. Just a point. From a real-world data standpoint, for sure, privacy is top of our priority when we engage with our external stakeholders and partners. So for example, within the U.S., we follow HIPAA, and as we engage outside of U.S., we follow the local regulations as well. This is absolutely high priority.

[ Charles Pitman ] was also asking about the current trend. Is there any kind of an improvement you see maybe also, let's say, supported by COVID and vaccine development so that you actually have seen an improving trend currently?

I think actually, more and more patients are willing to participate in broader research and usage of their data for drug discovery purposes. I think now we're in the habit of asking for that more broadly. I think we see great enthusiasm. I see nodding of heads from my colleagues online.

Definitely. And Steve, to add to what you said is there's a number of regulatory initiatives in various countries going on as well if you think about the secondary use of data. So I think there's a positive development that we currently see here in thinking about how such models can be enabled in the future.

Yes. Thank you. We have 2 more minutes left. There is a question via the phone from [ Eric Le Berrigaud ] from Stifel. Eric, I opened your line.

Yes. Hello, do you hear me well?

Yes. Eric, we can hear you.

Okay. Very good. Well, thanks for the very informative presentation, and sorry, maybe, for the very basic questions. Maybe back on the first question about the $3 billion and the return on that. $3 billion is a significant amount, but at the end of the day, this is 7% of total operating costs. For the second year in a row, you're mentioning this $3 billion investment. How should we think about this number? What was the number maybe 5 years ago? How should we think about it going forward? Are you managing this kind of budget separately from the rest and back on the kind of return we may expect from that as you said, and we understand that there is a shift in probably cost from more traditional type of costs into digital and kind of saving going from one to the other. So how should we think about the net cost of those EUR 3 billion rather than the growth? And the second question, since you rightfully said at the beginning, it's maybe the single event of that kind across the industry. And so it's maybe difficult for us to see how Roche stands versus competition. Where would you say that all that you presented today is very much kind of an industry trend versus where you think Roche has a clear edge over competition?

Yes. Perhaps I start. And I think on the second one, I would really ask the panel to come in because I think very clearly, there are different positions that we're having. I would say, in general, we are at the forefront. I think we've started to invest heavily into kind of a data business, if you like. You see Flatiron participating. Foundation Medicine was another play we went into. And then really, Moritz mentioned it, I think with NAVIFY, there was a completely different approach that we have taken and was a very novel one for the industry as a whole. And I think what we're seeing is I think one of the major advantages that Roche has, we have a Diagnostics business as well as a Pharma business. And therefore, it's for us much easier to create this insights business, let me call it that way, that you would like to see in the future, and that goes multiple direction. It goes to improving patient care, but it also goes into a direction that we really improve R&D at the very end. So really, there are multiple purposes behind it. Having said this, I think that leads me directly to answering the -- your first question. Honestly, I think that's nothing we monitor centrally. This is really about the different use cases, the different business cases that we're following. And as I've said, some are very tangible. I would argue the Aspire project, which is quite significant, is really related to transformations that we do in the company. So I would argue we have a clear business case behind that, and that's what we follow up. In other areas, like on the R&D side, a much more difficult to come up, really, with a business case. Here, it's really about improving the situation, creating more insights, but it's very clear if we get to better hypotheses and better probabilities for success with certain compounds that we build and certain treatments that we build, I think that would be a major breakthrough. We all know that. So I think for me, also pretty tangible. What encouraged me also, and I think that should also be encouraging for you as investors, is the point that a lot of money is flowing into this, as I've shown before. I think it's not alone. We're not alone, and we're not the only ones investing into that area. You see a lot of money is coming in here. But let me emphasize once again what I've said at the beginning. I think really we are a trendsetter here in a lot of areas. I think we had the guts to take money in our hands to go for the investment and then go into these directions. And you will see some things will be enormous. That's what I guess, some things well will fail as some of our approaches that we are taking on the drug side, on the treatment side. So we will see how it goes, but I would say very clearly, we are at the forefront. But I invite the panel here to make comments.

Maybe one comment I would add, Alan, is of the $3 billion, around $2 billion of that are investments that are really driving value in our internal business today, both for Pharma and dia and also the burgeoning insights businesses that we are growing. Each one of those investments has, in its own right, a business case. Some of those business cases are saving other costs elsewhere in the business, some of them are building assets for the future in the R&D space. So we -- the aggregate number that we presented there is a real blend of investments. But many of them are competing for monies versus other investments within our operational business. So it is investment that's flowing towards digital in preference to other investments internally.

Many thanks. We are a bit over time. Before I hand over for the final comments to our host, Alan, allow me to thank [ Bruno ], that was the heavy lifting was on your side. So thanks a lot for that. [ Jared ], for you as well. I wanted to thank our back office, [ Melanie ] and [ Ifa ], who organized this call very well in different time zones at different locations. So thanks a lot for this one. Thanks to the speakers. And Alan, over to you.

Thanks, Karl. I think I just can add. I think, really, first, thanks to the panel. Really with Mark, Christian, Jacqueline, Steve and Moritz, it's been a pleasure here to be with you and to listen to your presentations here and excites me quite a bit what we have here in store and what we work on. And I'm very convinced that we can definitely improve patients' lives with all these approaches, which I think is an exciting purpose to have and an exciting target to have. Yes, let me also say, well, as I said before, I think I guess we are really well-positioned as Roche to play a major role in the different fields here. We are a company based on innovation. We want to improve patient care. We have very, very ambitious ambitions moving forward that we would like to realize. The most prominent one is perhaps the one on the Pharma side, but I can assure you that basically in every segment of this company, we have very ambitious approaches. And I would argue that digital is instrumental for all of them to be achieved and realized. And what I see is that there is a lot of support of the investor side, and let me thank you for that because it also shows the trust and the commitment from your side and also belief that we will -- not just investor money, but we also -- how should I say it, create good cases out of this for patients on one hand, but also for stakeholders in general. So let me thank you. Thanks for your attention. Thanks for your commitment and your trust and hope to see you soon. Thanks.