Geeking Out about Data science in medicine with Roche’s Angeli Moeller
Harry Glorikian: Angeli Moeller is one of those people who seems like a born leader for the emerging field of data-driven healthcare and drug discovery. She’s a molecular biologist, a neuroscientist, a systems biologist, and a data scientist all rolled into one.
Which makes her a perfect example of the kind of multidisciplinary executive I’ve been saying for years that we’re going to need in this new digital health ecosystem defined by big data, AI, and machine learning.
She’s a founding member of the Alliance for Artificial Intelligence in Healthcare. She does extensive work for the nonprofit rare disease advocacy group Rare-X. She spent almost five years managing global data assets and IT partnerships at Bayer. And at the beginning of this year she became the head of international pharma informatics for Roche, the world’s largest drug company.
Angeli and I seem to be on exactly the same wavelength about how Roche and other life sciences companies should be taking advantage of the huge leaps in computing power now available to them to advance the field of data science in medicine. And every time we see each other we end up geeking out about the enormous potential of today’s advances in artificial intelligence and machine learning to unlock revolutionary new treatments and even a new standard of care in our medical systems.
Our last conversation happened back in February, and this time you get a chance to listen in.
Harry Glorikian: Angeli, welcome to the show.
Angeli Moeller: Hi Harry. So I’m really glad to be here today with you.
Harry Glorikian: Yeah. And you’re in this like really cool room. I almost wish I could like be there. It’s like, so it looks very social. I mean, I forget what social is like at this time in life.
Angeli Moeller: Well, that’s it. So I mean, I’m sitting in Berlin, Germany and but I can’t go out into the city at the moment. So I try to bring some of that feeling here in my virtual environment and for anyone online, who knows that studio. So this is great. And the studio. I’m having some fun on the weekends, making virtual backgrounds since during the lockdown. There’s not much else to do.
Harry Glorikian: God. Now that you mentioned it, I sort of missed some great places I’ve been in Berlin, so, Oh my God. Well, there’s this one place I remember I walked down the street. It was a beautiful sunny day it’s it was near the, like a river and. Just the greatest beers. We’re all like, Oh my. Yeah. So let’s not go there, but we’re supposed to be talking about data science in medicine and how it’s changing healthcare, not beers. But tell me, tell me about, you’ve got this new position at Roche. You’re head of pharma informatics international. What does that—has that role ever existed? What do you do and what does that mean?
Angeli Moeller: Yeah. Thank you so much for asking, Harry and, and the funny thing is you and I were talking just before I took the role, then just as I took the role, and now I’ve been in the role for eight weeks and we get to, to now have some more coherent answers and what that looks like.
So the role did exist. It’s pharma international informatics. So within Roche Pharma, we have an informatics teams. So, so the digital arm of all of the affiliates outside of the United States. So that’s our definition of what is the international group, it’s everything apart from Genentech, essentially.
So it’s all of our affiliates outside of the US, and the informatics team within that. We’re really then the digital arm of that organization. But one thing that has been evolving at Roche over the past a year and a half or two to three years is very much that it’s about, in each of these countries, in each of these incredibly diverse markets, how were we then engaging with the external ecosystem? How were we supporting patients through their journey and different parts of their journey, as they perhaps go from not having a diagnosis to having a diagnosis and then different disease states? So the work is just so diverse. I think that’s the thing that strikes me most about the work.
Harry Glorikian: So, so how has, how do I say this? How are you thinking about it from, do you have everything you need? Do you see what I’m saying? It sounds like there’s quite a bit there and I’m trying to figure out, like, it sounds like there’s new tools that you may need to create.
Angeli Moeller: Yeah, that’s a that’s an interesting question, Harry.
So I’ll admit I don’t tend to focus too much on the tool question. I think that there are new challenges that are, or perhaps challenges that have always been there, but now we’re looking at more closely in each of these countries and in each of these patient journeys and for each of these health ecosystems and I think that there are tools that already exist, but perhaps haven’t been applied to these challenges before, or haven’t been applied in a specific combination before.
And I think, for instance, what do you do in Alzheimer’s disease in these different countries as a patient is before diagnosis, as they go through diagnosis. What is the interaction, not just for that individual, as the first symptoms appear, as they’re told the first diagnosis, emotionally and physically, what they have to endure during that period, but what’s happening for their spouse? What’s happening for the family around them. What is useful information for their health care provider? Who’s having that discussion with them? And this just looks incredibly different in many different parts of the world. So I think—
Harry Glorikian: It sounds different, period. You just, it sort of describes an entire healthcare experience as opposed to one tiny sliver of a healthcare experience.
Angeli Moeller: And I think here, first of all, it is a very broad area, so that’s accurate. And of course, we do have our focus points at Roche. So there are certain therapeutic areas that we’re focusing on. I think that what we are really learning and the real strength of the international organization within Roche, but the approach that Roche takes is, is very much about what can we reuse? What can be applied and scaled across multiple countries, multiple disease areas so that we can as efficiently as possible, have the most meaningful impact on each of these healthcare ecosystems.
And I know that’s broad. So let me make it a little bit more tangible. If we’re looking for instance in the neurological space, we know that in that space changes in sleep patterns can be really critical. And we know that’s true also in rare disease. And we know that that’s true in several therapeutic areas and we know it can be an indication of quality of life.
What can we do and how can we make measurements on sleep patterns. There you go. For anyone who’s listening to this on the podcast, he’s holding up his Apple Watch. What can we do to really then, to understand how is that quality of life progressing? And, and I think there are these reusable components that really apply in multiple countries in multiple therapeutic areas.
Harry Glorikian: So, for those that don’t know, Roche is the world’s largest pharma company and, and, everybody also may not know that you guys have a fairly large US subsidiary and in Genentech. Can you guys give the listeners a sketch of the drug areas where Roche is most active.
Angeli Moeller: Mmm hmm. So in oncology, in neurological disorders, and in rare disorders, so these are all three largest therapeutic areas. And and, I think, this Harry that I also work for the charity Rare-X and and then that, that was one of the things that made it quite a good match with what we do in in the, in the rare space for, for me at Roche.
So it meant that I had some familiarity with Rare coming in. I think you also know that my PhD was in oncology and my post-doctoral studies were neurodegeneration. So I manage to have in my past done a bit of rare, neuro and oncology, but I think as well as you come in/ And that’s just the pharma division, the diversity, that was so rich, and that was so much to do.
I mean, looking at Roche total, beyond Roche Pharma we also have a diagnostics division which is suddenly had a significant impact around the world this year. And that’s the other key division within Roche.
Harry Glorikian: Yeah, that’s the one division I knew well from, being in, I started out my career in immunohistochemistry a long time ago. We did, I had, I had a consulting firm that really it would either would take a hard look at Roche or would help Roche think about what they were going to do. So which one of those areas that you talked about do you think is going to be most transformed by the impact of the combination of AI and all of it’s tools that fall under it and the data that you’re able to accumulate. And I know that we’re doing better on data accumulation in certain geographies, rather than other geographies, but hopefully it’s a heterogeneous population and we might be able to start to see patterns that we can utilize in more in certain places than others.
Angeli Moeller: Yeah, thank you for the question, Harry. And I have to admit, it’s a question I get asked, quite consistently, where is artificial intelligence, where is data science in medicine, where is data generation going to have the biggest impact. And I was reflecting that I think my answer changes. So, if somebody was to put together all the recordings and interviews I’ve done in the past year, I think my answer might evolve.
Because, I’ll be honest with you. I’m sure you have the same, Harry, because I’ve listened to your podcast. It’s moving so quickly, places where we didn’t expect to see impact, suddenly an amazing new discovery is made in, in machine learning or deep learning or in another space in data science in medicine that says something we thought we couldn’t do now we can suddenly do.
I think for me, what’s going to be most impactful for healthcare systems, for global economies, for patients, and for doctors is really what can we do to the patient that will also be meaningful pre-diagnosis. So that’s a question that we’re asking ourselves.
Harry Glorikian: Ok
Angeli Moeller:What can we do that will really be a signal we can detect in very complex data sets. I mean, think of the watch you have, I have the same watch. We have other similar wearable devices which are penetrating the market, and we also have electronic medical record data becoming increasingly accessible. What can we do at either very early stages of diagnosis at very mild disease states and pre-diagnosis states to prevent that continuation of disease severity, to prevent the point where somebody needs an acute intervention or a very serious therapeutic response. So we’re really discussing and looking at how do we make the healthcare system sustainable globally, by really understanding how we can use personalized medicine data science in medicine and detections and patterns of the data to really have those early interventions.
And the second shift that I’m seeing is also on real focus on quality of life and all of the, the wearable data and the high-density data science in medicine that we have now which is collected outside of a clinical setting is giving us a much better view on what does quality of life really look like. So how is that patient or that individual really doing outside of the clinic?
Harry Glorikian: Yeah. And it’s interesting, like, when I think about all these different parts that are being impacted by data science in medicine, historically, we’ve looked at them as different silos. So we treat them as they are like, I’m in this room today and I’m in that room tomorrow and I’m in that room tomorrow. And I, I’m not sure if everybody understands that the data actually coalesces the room. In other words, if I start to see somebody that’s in early diagnosis, being diagnosed early, first of all, what data analytics platform do I put there to capture that? But that information might indicate that either a therapeutic regime or a clinical trial might be relevant.
So that bleeds with what you’re doing and might also influence drug discovery. Right. And then the outcomes, data of how we manage you. I mean, I think of it like Google maps. If there’s not a constant communication between all the points that are moving around live on the map, the map can’t rejigger itself to tell the next person, how long is it going to take you to get from point A to point B? I mean, the map is a little simpler than what I’m saying, but it’s having all the data sort of swirling in an ecosystem that can share and be accessible at one time. Maybe I’ll see that in my lifetime. I’m not sure.
What is the role of data science in medicine?
Data science in medicine plays a significant role, specifically in areas such as precision medicine, drug discovery and development, clinical trial design, and healthcare analytics. Using data science techniques, such as machine learning, data mining, and statistical analysis, medical researchers can gain insights from large amounts of medical data to improve patient outcomes, develop new treatments, and optimize clinical decision-making. Additionally, data science in medicine can also be used to analyze electronic medical records, monitor public health trends, and identify risk factors for disease.
Angeli Moeller: I think here, so back to the question you asked before, which is, where might we see earliest impact, I think the rare disease space is moving very quickly, because of need, because of really critical need. And also because the patients themselves and their caregivers,if its parents, as these diseases often affect young children, are really actively promoting data sharing, actively asking for data sharing and putting out that data science in medicine into platforms where it can be accessed and shared because they want to drive forward that conversations.
And in some of the Rare-X discussions we have, so we have discussions with different partners who want to work with us in the charity Rare-X on how to study and analyze that patient data for the sake of data science in medicine. So we go out to ultra-rare diseases and also to some larger disease communities. They’ve normally been creating their own registry. Looking after the data themselves. We host the data. We connect the data. The data stays patient owned, continuously throughout, but we give them the tools to make it really easy for them to withdraw or give consent at any time during that journey and also forever so that, they can change their mind and say, no, I want to withdraw consent or yes, I’m happy to continue to have this data science in medicine available.
So we try to put all the power back in the patient’s hand. And here, it’s really then—we have debates, I’ve been sitting sometimes in an hour-long debate. What’s most valuable. Is it the electronic medical record? Is it the patient-reported outcome for the natural history study? Is it the genomic test thing that we can offer on top in our collaboration with the Broad Institute? Is it the wearable data that we can collect and able to detect if a patient’s going to have a seizure or not? And I see you smiling. Cause I think you agree with me. It’s all of it. All of it goes into data science in medicine.
Harry Glorikian: I was thinking of the multiple choice in where the bottom one, it says all of, all of the above.
Angeli Moeller: Exactly, exactly. And I think that’s I understand why this question keeps coming up because people have to know where to invest and where to invest at which time. And as somebody who spent a lot of time in academia before, I think that’s why there’s a lot of value in defining which scientific questions you want to ask and then prioritizing which data you’re going to clean and collect based on that. But at the same time as this field emerges, we’re seeing there are more questions we can answer that perhaps we didn’t even have in our minds when we started to collect that data for data science in medicine.
Harry Glorikian: And if, it’s interesting, right? So I was talking to Joel Dudley at Tempus, right. And one of the things he said is, well, you get a sample. We do everything we can do on that sample. Even if it’s not going to be used right this minute, we know that that is going to play in the symphony that’s being created like, maybe French horns. If we had French horns, we could like, this would sound better. Right. And everything that I’m finding in the data space is, yeah, we do our analytics for data science in medicine. We look for a signal, but if we could add another component to it, it gets, this gets better over time. This we’re adding to it. It’s like trying to determine whether based on one marker, which is impossible, you have to look at, hundreds of things that are happening to be able to, and they don’t even do it well, even with the with the hundreds, but let’s hope we do better, better in healthcare.
But let’s jump, jump back for a minute because when you were explaining your, your background I mean you’re a unique duck in a sense, right.
Angeli Moeller: Duck?
Harry Glorikian: And like, I mean, it’s not many people that have all these different areas. Plus the data science in medicine in one individual. Like, I think we need to rewrite the curriculum because we need more like you, if we’re going to make the advances that we’re having. But, How did you master all these new fields? How did, how did it, did they just fall into place? Did, how did you come about going in that direction? Because I’m hoping some young people might listen to this and think, Hmm. I get this question all the time: Harry, what should I study? I’m like, Hmm, not sure. You should definitely understand computer science and you should understand finance. And then you should learn how to learn, is my last one, because it’s ever-changing. But how would you, if you were giving somebody advice, the younger, maybe you what, what would you say?
Angeli Moeller: I mean, I mean, very similar. How are you making me smile low? Cause you were calling me a duck today. And I know when you and I were at JP Morgan earlier this year at the virtual JP Morgan, you referred to this profile as mutts.
Harry Glorikian: Well, it’s true, right? I mean, actually it’s funny because I gave my kids this book called Range. Because I was trying to explain to them, being super deep in one area. Like that’s great if that thing lasts forever, but having range allows you to think about a broader area that, and what we’re finding is that area that we didn’t think it mattered, actually matters. And if I was only deep in this one area, I’m not going to see the, how it’s all coming together.
Angeli Moeller: Yeah. And I would definitely support that and agree to that. But also, and I do go out to universities, a lot to either do mentoring with students or do careers talks and I think, I always say, you’ve got to love what you do. You spend so much time at work and it can be such a big and rewarding part of your life that normally the first thing I say is, do something you absolutely love. And then I say for me, that happened to be this, and that’s why I do this. That’s why I’m working on data science in medicine.
The other reflection on. So, yes, I have switched fields quite a few times and that can be a pro and a con, but in the end, it’s my personality. Right. And I think also people know this when they hire me. People know this when they work with me. So I know that when I started off choosing what I wanted to study, I guess the one red thread is it’s always been healthcare. This was the first step for me to get into data science in medicine.
So when I was at school, I was thinking Médecins sans frontières, and I was always thinking, what can I do in healthcare, in that space? And at the age of 16, I was volunteering in an emergency room back in England, just thinking, okay, I’ve really got to do something that’s, that’s helping patients and that’s really focused on that. And then and then at the time, being in Newcastle in the North of England, Dolly the sheep was happening. And it was, it was a time, and that wasn’t happening far away. That was just in Dundee. So a couple of hours drive and. And so I, I just thought, yes, this is something I want to be part of, I don’t know what it is, but I want to be part of it. So it just began with a commitment to genetics and biochemistry. And then when I turned up at university on my first day, they said, we’ve transformed that into a molecular biology degree. It’s the first time we’ve had that degree and you’re our first student, and three of you. And then after that, as I, went into my PhD it was again, okay we’re going to work in single chain, antibody engineering and nobody thinks that these re-engineered antibodies will ever make it into humans. We’re going to start, but we’re a long way from that. So that’s how I got closer and closer to working around data science in medicine.
Harry Glorikian: But that’s the funnest part of this stuff, right? Like, I don’t think I’ve ever done. I, and sometimes I think I’m crazy is I don’t think I’ve ever done the same thing twice. It’s always some left turn, right turn, building on what was there. But every once in a while I’m like, you’re a nut case, you’re just making your life so much harder. There’s something new to learn every time, but I can’t help myself.
Angeli Moeller: Yeah. And I know Harry from talking to you before that you have this personality as well, but I have to say I have full respect and would also encourage somebody who has a consistent passion that they want to stay with and something that is in the same field. And they say that this is what I want to do forever. And I have a lot of friends who pursued those sorts of career paths and find them very enriching. So I would always say, do something you love. I mean, for me then as I moved into neurobiology and as we started to have IPS cells and be able to differentiate them into neurons, I remember that moment in the lab and how exciting that was. I was always truly amazed at everything concerning data science in medicine.
I think I just feel so honored because if I look at single-chain antibodies, if I look at IPS cell differentiation, if I now look at, whereas we start to, to really capitalize on this machine learning revolution, I feel like, for my generation, I’ve been just really lucky to be at the right place at the right time and have those opportunities to be part of what to me are some of the major milestones of my generation. And maybe that’s not very scientific.
Harry Glorikian: No, I mean, I, I totally agree. I mean, when I try to explain to people, I’m like, well, this is happening. And then we went from this to this, and it went in this period of time. The problem is, is most people don’t understand like what the timescale was and what the timescale is, and what I can see the timescale will be. Right? And the impact is. And the science is just like, we can do what, like we did, what? I get super excited and everybody around me in my house goes. Again, with this data science in medicine discussion, like, can you, can we talk about something else? But…
Angeli Moeller: I’m having those same discussions at the dinner table, but I think the one thing, and maybe building on the theme, things you and I have in common, is that is now the excitement of connecting that to something that’s sustainable. And that sustainable also means that it’s working economically and financially and that it becomes sustainable, and that’s the journey that I would say I’m now very much thinking about sustainability. And, and how do we make sure that this innovation to patients really becomes sustaining, moves past academia and moves past the lab and moves past the computer and the algorithm, and really becomes something sustainable in terms of delivery to patients for data science in medicine.
Harry Glorikian: All right. Let’s, let’s dig into that a little bit because I mean, I think you’re, you’re part of this group called the Alliance for Artificial Intelligence in Healthcare and, and you’re the treasurer. So I don’t know how you have time for all this stuff. Like, wait, let me get the spreadsheet. Let’s see how much money we spent. No, no, we gotta make that system work and it’s gotta work for those patients in wherever. I don’t know how you have time for all this. But what’s the, what’s the, what’s the origin story of the Alliance and, why are you so passionate about it?
Angeli Moeller: Yeah. And I certainly am passionate about it. So we, when I was in my former role at Bayer we started on on our digital transformation journey there. And I was invited to lead the workstream on artificial intelligence, so, for all of the, the pharma division across the pharma division, and he I, I started to just think, okay, who’s moving this area who I might know already. Who’s doing what in this area? Who could I talk to? And Naheed [Kurja], was somebody who’s the CEO of Cyclica at the time. And he and I had recently had lunch. He’d just been in Berlin and we just had lunch. And I saw online that he was attached to a post about the Alliance for Artificial Intelligence in Healthcare, and they were just brainstorming what that could be. And so I pinged him and I said, what is this? Can I get involved? What’s going on here? What are you trying to do? And that was three months before the JP Morgan in 2019, which is where we launched. So we had a three-month, very intensive period in deciding who we were and what we were.
But the frustration and the opportunity that led to the AAIH was there were quite a lot of extremely technical CEOs in the health tech space who were meeting constantly, who were all at the same conferences, who were always all together, and who were seeing a lot of confusion around the topic of artificial intelligence in healthcare and data science in medicine. And we’re feeling that they were constantly having the same conversations, that they were constantly trying to push past the hype cycle into something more tangible.
And they were also seeing a lot of companies coming up with the .ai and, and not much else behind them. And so they were seeing this trend, they wanted to shift the conversation into something more concrete and more grounded in good engineering and good science practices for data science in medicine. So this was their driver. And my driver was, I want to learn from you guys. I want to learn from you guys. And I think that you’re wonderful. And then in that spirit, and as we come together now, where we’re about 45 companies, that was the: this is bigger than any one of us. This is more important for the movement of our industry than anyone company than any one individual. And we really believe so much in that. We believe that patients are missing out on innovations they could use today because there’s so much hype and so much confusion around artificial intelligence and data science in medicine. And it’s taking away opportunities from patients because they’re not getting access to things that could help them just because of this fog of confusion. And we felt so strongly about that, that we put it above our company roles and we decided that we would together found this organization. And it’s been an honor to be…
Harry Glorikian: So let’s talk about the, let’s talk about the impact of this. Cause I always think to myself, all right. So two years ago when I said AI, people were like, huh? Right. Now, when I say AI, it seems like if every CEO is not talking about or implementing something in AI, they’re behind and, we’re starting to get to the point is if you haven’t already put something into place, you are going to be so far behind. So the curve of, it forever for some things to come up that curve. And now this one seems like it’s crashing in on itself. From a timescale perspective. I mean, where do you guys see the organization having its largest impact, so that you can do what you want to do, which is get these products or services or both into either an organization or that’s going to have an actual impact on a patient, right? Assuming that the organizations you’re trying to get it into or already trying to do this themselves. Right? So what’s the overarching—how are you going to do that for data science in medicine?
Angeli Moeller: Yeah. Thank you so much. And, and it’s actually very timely you asked, Harry, because we just had our strategy workshop last week as well to have a refreshing and a good look and have an intense discussion amongst our board of directors about how we see our strategy.
So one piece which won’t be surprising to you is the data sharing aspect of data science in medicine. So data sharing and to talk to regulatory bodies about how we can incentivize data sharing in specific areas. So you may know that Roche, our clinical trials in the COVID space, that we shared the data from those trials. And we do see that that’s happened in other specific areas. So we want to work together with regulatory bodies to really incentivize data sharing, particularly where we see an acute need, and also sharing of models. This will truly help the growth of data science in medicine. And models that can be used to bring new solutions to patients. So to have more open sharing of algorithms that are developed, but in a sustainable way. So in a way that still allows the innovation to be rewarded for the individual companies and the individual data scientists who are doing that innovation. So we’re looking at how to make that sharing possible, but also sustainable.
The second part that we work on, which is a lot of our, a lot of our man-hours, let’s put it that way, is working with policy makers, business leaders, healthcare professionals on myth-busting. So we just spend a lot of time doing educational sessions, on preparing webinars, on running conferences, on going, even doing smaller sessions to really answer questions. So we’re so lucky that this is a growing expert community. And and that also our founders are fairly strict in, in the technical excellence and in what good engineering and good data science in medicine looks like. And that means that we just want to go out there and be a resource, to also take away fear, to take away misconceptions around artificial intelligence, to maybe move away from the HAL [in 2001: A] Space Odyssey to, this is just something in your smartphone and you can rely on it. So that’s a really big part of, of how we spend our time.
And then I think the third part, which we’re really looking at very tangibly and which may be a new thing for us this year as an organization that’s now getting a little bit older and a bit more established and a bit bigger, is we want to run some joint projects together. So we’re looking at which joint projects. Basically, we just looked at each other around the table and we said, we have some very smart people here who I’m privileged to look after the finances for, but we sit around the table. And I imagine if we all took a very important challenge, a very important healthcare challenge, and we all worked on it and we took all of our great data scientists who do great work for data science in medicine, and all of our great biologists, chemists, cloud engineers, such a mix of diverse talents. And, and then we just really worked on a very important challenge together. So that’s the angle that we’re really looking at this year.
Harry Glorikian: Well, being on the investor side, that sounds like a roll-up, like an incredible company. But it also sounds like you guys are developing or want to develop something like GitHub where there’s a repository of algorithms already available to people that they can use. The data sharing, everybody’s not super good at that. COVID was an incredible exception. I don’t think I’ve ever seen data sharing like that before. But I’m not sure that how much it’s going to continue when the world is not being threatened. That’s definitely going to be a possible challenge for data science in medicine.
Angeli Moeller: And I think, I think on some topics you do have to already now be starting to make sure we’ve got things in place to keep momentum. So Paul Howard from Amicus, he and I did an AAIH panel yesterday evening. And this was one of the topics we intensively discussed. I think here, with the library of models. So that all great things on GitHub, I don’t want to make a new GitHub. I mean, that’s all great. It’s more about working on what is validation look like and what does good look like and how do we have a repository of validated models that are of a standard that would make a regulatory organization happy? And how do we build up that library? So that’s really where we’re shaping the conversation. I think, I think for pure academic brilliance, there’s already repositories out there. There’s already great libraries out of that.
Harry Glorikian: I was going to say to you, I think, I think, well, I’m going to add one more thing in your career. You’re going to need to write a book based on this, all your experiences,
Angeli Moeller: All right. So what would we call the title?
Harry Glorikian: No, we’ll come up with something. Listen, I’m working on number three right now, so don’t worry, it’s totally doable. Let’s jump to, to, Rare-X. So you’re Roche, AAIH, Rare-X. I thought I was doing a lot. You’ve got me beat hands down. And you have a life. I mean, let’s, let’s add that to the table, but what drew you to that organization?
Angeli Moeller: And you can see today, and Nicole Boice is going to be so proud of me ‘cause I know we’re being videoed, but I’m also wearing my Rare-X sweater. But what drew me to it? Well, I was visiting Anthony Philippakis at the Broad Institute. And and he and his organization, are part of the AAIH, and we were talking about informed consent, which is one of my pet passions. How do you make informed content manageable and work for all parties involved? And he, and and Morry Ruffin who also helped found the AAIH. They said, you’ve got to meet Nicole Boice. And I hear this story. So often people are taken aside somewhere and I told you’ve got to meet Nicole Boice. And Harry, if anyone ever says this to you, the answer is yes, please. Nicole, is the founder of Global Genes. And now sits on the board for that organization. And and she’s also our founder and CEO at Rare-X and she’s worked in the patient advocacy space for rare diseases for most of her career. And for me, Nicole is a moral compass or she’s increasingly become that because every time we’re talking about what we could do together, and as we talk about sustainability, and there’s often a time when you can look at short term gating and also short term revenue, and Nicole is the person who in every conversation brings it back to, what does this mean for the patient? What does this mean for that parent, for that child, for this rare disease community, for data science in medicine? And I value what Rare-X has brought to my life and to my career so much for that for just being within network of people who ask themselves that every day and for that to be trained and to become such a big part of my life.
The solution. I mean, the reason I started to get into it is that I really liked the technology as well. Right. As much as, as much as the commitment to the goal. So it was a, it was a few things I cared deeply about coming together. The platform is based on Tara Bio. You might’ve seen recently that Anthony’s group got an additional investment from Verily and Microsoft into their Tara Bio platform. And the way that they’ve set it up with the different modules means that we can go out to these rare patients. We can help them host that data for the sake of data science in medicine, but they always see their own data. So they can always see how they are doing in comparison to an aggregate of other individuals with similar phenotypes, with similar genotypes, with similar clinical progression. That’s where data science in medicine plays a big role.
And for a lot of the ultra-rare disease patients, they find out or rather their parents find out they have a mutation. They try to find out what does that mean for life expectancy? What does that mean for breathing problems? Sleep problems? All of the symptoms look so unique that they’re seeing in their child. But then they can also map to other phenotypes and other families who are similar phenotypes and who are also then seeing what treatments are effective. And they can see that in an anonymized way and it can start to give clues to them and their health care providers for this very ultra, even N of 1 diseases that are struggling so hard to find what is the right treatment path for me?
Harry Glorikian: Well, I’m going to, yeah. I mean, I’ve spoken quite a few times to Robert Green, who’s done BabySeq. Right. And I’m actually, I’m going to catch up with him next week. But you can see as you’re looking at this, first of all. And Sharon Terry, right? And you start to understand that the power of the technology is the N of 1 is no longer the N of 1. Right? It’s, you may be geographically the N of 1, but in an aggregate, you can get more of them in one place. And as soon as you can see more than one, it’s better for us to try to figure out what’s going on. I remember when one of the people that Applied Biosystems sequenced their own kids and found what was wrong and was able to give them an over-the-counter drug that made a huge difference in the person’s life. Right. That was sort of the first shot across the bow. Now I think it’s taken the rest of the world forever to catch up with what I think was almost, 15 years ago. Right. I mean, it’s a long time. And the system you’re describing of being able to look at myself along other patients, shouldn’t that be standard of care, like for everybody, everybody who has cancer or anything else.
Angeli Moeller: And then Harry, I mean, you really getting onto the vision of the future and where all these things fit together for me, because, with the AAIH, then, you can start to have the conversations with patients and healthcare providers and policymakers to create this shared vision to talk about the practicalities of, if I imagine my parents having this information on, on their health care journey, and then they need to understand that it needs to be in a digestible way, but their healthcare professional has to be open to talking to them about it. Even though they don’t have an acute disease, they have to be open to saying, let’s look at this data together. Let’s think about what it can mean about you as an individual, not you as an average person who has, in the case of my father Type 2 diabetes, but you as an individual and how to your phenotypes, genotypes and clinical progression look on an individual level when we look at your data compared to the aggregate.
And I think with Rare-X, one is before I’d worked with Nicole, I hadn’t worked in the rare disease space before, so there’s so much about that space that I had to learn, but it’s also, the patients are so engaged because the need is so acute and they really understand the value of that data for data science in medicine, and the value of having as many researchers as possible looking at that data and being able to integrate that data for meta-analysis. And they’re so engaged on that journey with us that I think it’s an opportunity to showcase what that could look like in a faster way, in a more tangible way.
Harry Glorikian: Well, it’s interesting, right? Because you are talking about children and patients parents aren’t. Yeah. They will do things for their children that they might never do for themselves. So that’s that’s always a driver but, we talk about rare disease and I, I know that we’re talking about, like, I think we’re, we should start calling it the ultra-rare disease, because if we look at breast cancer or neurological diseases, every one of them is going to go down a different branch and there’s going to be subsets. And they’re all going to be a rare disease at some point. I just, I can’t imagine that we’re not going to get better and better at targeting something. And then maybe thinking of combos because it’s different pathways we need to hit at the same time. And I don’t know how we’re going to do that without some level of artificial intelligence and, the entire toolbox that comes underneath that, that can help identify that.
Angeli Moeller: I mean, I completely completely agree. And I think it’s just about, the problems are there already. I mean, now I know your background’s in immunohistochemistry Harry, I can, I can get into it, but I come from proteomics. And if we look at…
Harry Glorikian: Way more complicated.
Angeli Moeller: I mean, you’ve got, I’ve got all the information at a genetic level. Then I’ve got all the transcriptome information, then I’ve got all the proteome information. Then I’ve got every single post-translational modification on top of that proteome, I’ve lost your whole audience now, it’s hard. That’s why I went to use these tools in the first place, because, at that level, every cell is extremely unique. Never mind every human individual.
Harry Glorikian: Yeah, no. I remember when we were at Applied Biosystems. Okay. We’re going to do the genome. Oka, I was like, all right, chemistry, we can do that. Like, that’s not a problem. Right. And then somebody said to me, and our next thing is, we’re going to do the proteome. And like, we’re going to do what, like. I went to my wife. I’m like, I think I should sell some stock because this is going to be really hard. Right. And look at how long it’s taken us to just start to scratch the surface of that whole, methylation and this and that, and trying to bring all that information together. It’s not trivial.
I still think there’s low-hanging fruit just on the genomic side. I mean, let alone everything else. I do believe like one of the next big areas is going to be spatial genomics, like basically immunohistochemistry, but looking at it from a, which cells are lighting up and how much gene expression and what’s going on in that space and being look at it relative to other cells. I mean, for me, that’s just molecular immunohistochemistry.
Angeli Moeller: Yeah. And, and I think, I think that, the actual tools that you need, and here I’m talking about laboratory equipment, the actual quick pace as well, at the same time, the, the algorithms used by system biologist developing. And I think when you add onto that, okay, that complexity at the cell level, the complexity you have now at a pharmacometrics level with all of the different organs in the body, talking to each other and what that looks like, and then you bring it up to the complexity of our population level. And now you asked me, I think it was the second question you asked me is, how did I come into the informatics space? And I would say here. I’m not that old. Right? So I’m 37 and I would, I always tell people at the time I went through university, there was no molecular biologist in the lab who wasn’t also doing bioinformatics.
And I see the same now for people doing marketing degrees today, or people doing other traditional degrees in many different areas. I think informatics has now become part of every profession. And I think that, you can do a marketing degree, but it’s going to have digital marketing, and it’s just going to be inherent and you can do an MBA, but it’s going to have a lot of big modules which are going to be focused on informatics. And you can become a biologist, but you’re going to do a lot of big modules on informatics. And that’s just the nature of where we are today.
Harry Glorikian: Well, just to put it into a timescale, and I would say I’m a, I’m a little bit older than you. Not that much, but just a little bit older than you. And I would say that, when we were doing the genome. We’re like, we need this bioinformatics, right? It’s like, what the hell is that? Well, get the comp sci guy and get the biologist, put them in a room and have them figure it out. And they could barely talk to each other. Right. And that wasn’t that long ago, relatively speaking. So it’s interesting. I always wonder, like the university curriculums are not, I don’t think they’re keeping up with the pace of what needs to happen for us to keep this momentum going? Because like you said, everybody has .ai. Well, I start digging under the covers and I’m like, you don’t have what it takes to don’t what you’re saying you’re going to do. You don’t have the people, we haven’t graduated enough of them. So to keep that momentum going, I think we really do need mutts to come out of the woodwork because otherwise I don’t think we’re going to achieve that next level of, of growth. I mean, we keep taking physicists and putting them into this area because they’re so good at the math. I think we need a physicist crossed with a biologist, not just one or the other, because they don’t, sometimes they don’t think about the problem, the way that they should. I’m bringing my biases into this, but….
Angeli Moeller: I love the diversity we have in the Alliance for Artificial Intelligence in Healthcare. So, there are astrophysicists there. There are people with MBAs. There are chemists, molecular biologists like me, pure computer scientists. I think that often the slight differences in the way we approach a problem, how — we also have public affairs specialists and lawyers — and the slightly different way we approach a problem often is what helps us find the solution to that problem in the end.
I think, though, the one thing I definitely don’t underestimate is the value of hard skills. So I think I been, I’d love to hear your thoughts even on this Harry, but as digital, as we move away from the roles that we’ve just learned and know and accept, I’m a molecular biologist, I’m a bioinformatician, I’m a chem-informatician. These are things that mean something very tangible to me. There are now a lot of job titles with digital in them. I’m a, I’m a digital lead. I’m a digital officer. I’m a digital transformation officer. And I think, I think there, in trying to pick a way in the same way we have with the .ai. What does that mean? And what are you going to do? That’s it, that’s an interesting question for the industry, right now.
Harry Glorikian: No, and there’s very few people that I’ve spoken to where I’m like, Oh, this person gets it. Like this person really understands it. Right. And they understand it at a level where I’m, I’m struggling to just keep up with where they are. And that is the, there not tens of thousands of those. Right. There’s, there’s few of those. And I’m not sure talking about just the machine learning of the AI. That’s just, okay, great. I can go to Silicon Valley. I can find a 23 year-old that can, right, probably run circles around me in that sense. But that understands how some of these pieces are going to come together, how they need to think about the math differently than just taking what was and slapping it on there. I mean, in some cases, and I’m talking to somebody about this now is, some of the math we’re using is just old, and it was we’re using it because that was as good as they could do at that time. Well, we have this thing called a computer now. Like we should be able to, like, improve that math to a certain degree to actually come up with a new mathematical pathway to this problem. And I’m reading a whole bunch of papers right now so that I can continue my debate with the individual, but this whole field is changing so rapidly that every week, I’m having a conversation that I’m seeing it move forward. The problem is, is I don’t think the existing status quo can keep up with the how quickly it’s moving.
Angeli Moeller: And Harry, sometimes I’ve felt like that. And and again, if you put all my interviews side by side, I think you can see the days when I felt like that. But I’ll be honest, I was having this discussion with some colleagues just this week, and I think the emotional intelligence that most senior leaders have, will get them through whatever comes, whatever digital, whatever machine learning, whatever informatics brings. Because, you, and I know you’re not going to have a team of 50 fresh machine learning graduates producing something immediately fantastic, patient-centric and commercialization, and that we’re going to need seasoned leaders with good business acumen still playing a key role and critical role there and the skills that are taking them through every other twist and turn of business life are still going to take them through this next digital transformation and also mean that they can really unlock the power of what machine learning and other new technologies can bring in the same way they did when single-chain antibodies came out and when pluripotent stem cells came out and they used the same emotional intelligence.
Harry Glorikian: I agree. And I disagree. Right. I agree because I totally understand the historical line that you’re drawing on the biological technologies. I think that leaders need to be really looking at what tech is doing, how quickly tech is advancing, what are the, the arcane things that are going on there that they, they is not even in their view on a daily basis, and then be able to superimpose some of those what’s going on there into our world to actually see how this is going to happen. Or what’s going to change because I do think that there are things that are happening there that people in our world don’t fully understand the impact of, which I think is, is the coolest stuff that’s going on.
Angeli Moeller: How many of our colleagues audit the new Nvidia graphics card and understood what that could mean for healthcare?
Harry Glorikian: Well, it’s also just trying to, I mean, I remember the impact when we were at Applied Biosystems and, Intel released a new chip, and all of a sudden we could do 72 hour unattended sequencing. We had nothing to do with that. We just took the chip, plugged it in and off we went. Those changes are happening….I’m having trouble keeping up with some, I don’t know if you saw. Samsung is releasing a new memory chip where it’ll have AI machine learning capabilities on the memory chip. So if you start to rethink the architecture of the computing platform and then superimpose that on what we’re doing, there are big changes that are coming, that if you talk to people in our field, they’re completely unaware of the, how quickly it’s coming. And so as a leader of an organization, you need to preplan for some of that, right? Otherwise, you can’t absorb it. So that would be my 2 cents.
Angeli Moeller: We’re really getting into it Harry and we’re probably going to have to do another recording sometime. But I mean, I think it’s about the mindset and, and sometimes, I’m like, it’s about the hard skills. Because you can’t get away from the hard skills. There are hard skills you needed in your organization. So let’s say that’s a given at the leadership level, it’s been, the mindset becomes even more critical because it’s about, I think with software it just moves so much faster than drug development, clinical development than pharmaceutical development, it just moves so, so, so much faster. And even what we would call a traditional IT project where you choose the solution, you roll out a solution, it’s there, you don’t retire a solution as you roll out the next solution, anyone in the software industry, nobody thinks like that anymore. Maybe they did 20 years ago. And it’s about, how do we prepare for the fact that everything is replaced the second that goes live and how do you prepare everyone to be comfortable with that. That I’m going to have DevOps, I’m going to have my data science in medicine platform it’s going to go and have a new release every two weeks based on what there I immediately get from the end-users. I’m not going to ask the end-users what they think, but I’m going to have different metrics that the software immediately picks up to see how they like it, how they use it. I’m going to throw out features to this group and do AB testing. I think.
Harry Glorikian: Yes, but this is why I think we need to have skunkworks areas that can move this thing forward, but also a leader that can understand the implications of if that’s skunkworks is successful what is the implication on the organization? And that’s hard in a big organization, right?
Angeli Moeller: Yeah. I mean, absolutely. I can, I can see challenges, but I’ve got to say, I feel, I feel really optimistic. But people understand, what it can mean to have an agile transformation at an enterprise level, and also about what sort of mindsets are going to keep them safe during this journey.
Harry Glorikian: Yes. I think it, again, leadership sets the tone, right? As an investor, I’m looking for the Series A guy, right, that’s going to revolutionize something. Right. And it’s 13 people, right? Or 15 people. It’s, it’s not, 5,000 people, right? Hopefully, maybe the organization will grow to that much. Although I think, I don’t know if you ever need that many people anymore to change the world. It’s I think it’s a smaller group.
But look, it was great to talk to you. I wish we were actually sitting at that bar. Right behind you and able to relax. Cause I have not left this room, I don’t think since last March. But it was great to catch up with you. And I look forward to continuing the conversation.
Angeli Moeller: Absolutely. Harry, and it was always a pleasure to speak to you. Thank you so much again and have a great rest of the day.
Harry Glorikian: Thanks.
Harry Glorikian: That’s it for this week’s show. We’ve made more than 50 episodes of MoneyBall Medicine, and you can find all of them at glorikian.com under the tab “Podcast.” You can follow me on Twitter at hglorikian. If you like the show, please do us a favor and leave a rating and review at Apple Podcasts. Thanks, and we’ll be back soon with our next interview.
FAQs about Data Science in Medicine
I often like to include some answers to frequently asked questions for a specific topic I’m covering in an episode. I find this helps my listeners and readers get a better understanding of it. For this episode, here are some FAQs about data science in medicine.
What is medical data science?
Medical data science is a field that combines the skills and techniques of data science with the knowledge and expertise of the healthcare industry. It involves the use of data science methods to extract insights and knowledge from large and complex datasets in healthcare, such as electronic medical records, genomics data, clinical trial data, and other types of medical data.
Data science in medicine can encompass a wide range of activities, including:
- Predictive modeling and machine learning to identify patients at risk for certain diseases or to predict treatment outcomes
- Natural language processing to extract information from electronic medical records
- Data visualization to help physicians and researchers understand complex medical data
- Data cleaning and pre-processing to prepare data for analysis
- Statistics and data mining to identify patterns and trends in medical data
The goal of data science in medicine is to turn raw medical data into actionable insights and to support the development of new treatments and therapies.
What are examples of data science for medicine?
Data science in medicine is a field that involves the use of data science techniques to extract insights and knowledge from medical data to improve patient outcomes, support medical research, and optimize clinical decision-making.
Examples of the applications of data science in medicine include:
- Personalized medicine using genomics data to predict individual patient responses to treatment
- Clinical trial design using data science in medicine to optimize patient recruitment and improve trial efficiency
- Healthcare analytics using data science to monitor public health trends, identify risk factors for disease, and evaluate the effectiveness of healthcare treatments and policies
- Medical imaging analysis using data science to automatically identify patterns and features in medical images that could be helpful in diagnosis and treatment.