View  Transcript

Today on The Fundamentals, our guest is Dr. Geoffrey Siwo, a research assistant professor in the University of Michigan Medical School’s Department of Internal Medicine and Research Associate at the Michigan Center for Global Health Equity. His research focuses on accelerated and equitable innovation, using emerging computational technologies such as artificial intelligence, programmable biotechnologies such as CRISPR systems, and frameworks for scientific discovery at a global scale such as open innovation challenges. He is a co-founder of Anza Biotechnologies, a startup focused on accelerating discovery and sustainable manufacturing of therapeutics. Previously, he was a research assistant professor at the University of Notre Dame, lead researcher at IBM Research Africa, and a co-founder of Helix Nanotechnologies, a DNA nanotechnology company. 

You can learn more about the work Dr. Siwo's lab is doing by visiting their website, and you can follow Dr. Siwo @gsiwo, the U-M Institute for Healthcare Policy and Innovation @UM_IHPI and the U-M Center for Global Health Equity @UM_cghe on Twitter.

Resources

• Find out more about the research stories mentioned at the top of the show at the links below:
  • Researchers overcome major barrier in artificial placenta research
    Innovative strategy allows artificial circuit to safely support premature sheep without anticoagulation
  • Artificial intelligence predicts genetics of cancerous brain tumors in under 90 seconds
    Researchers hope it will improve diagnosis and treatment, as well as clinical trial enrollment
• Learn more about other exciting research happening at the University of Michigan, by checking out Health Lab, Michigan Medicine's daily online publication featuring news and stories about the future of healthcare.

Transcript

Kelly Malcom:

Welcome to the Fundamentals, a podcast focused on the incredible research and researchers here at Michigan Medicine. I'm your host Kelly Malcom.

Jordan Goebig:

And I'm Jordan Goebig. This week we dive into a topic that I could not know less about. Artificial intelligence. Our guest has such a unique perspective on the topic, and I think he tempered a lot of unknown feelings that I had about it.

Kelly Malcom:

This is one of those topics with a lot of buzz currently, and people are definitely grappling with how to use data to improve our lives.

Jordan Goebig:

There are so many cool things happening in the eHealth space, but a recent Michigan medicine article about an artificial placenta stopped me in my Twitter scrolling. The development of this has already been happening by Michigan scientists, hoping to improve outcomes for premature newborns. The research team developed a special coating that prevents clotting without thinning the blood, and it's shown significant progress in the lab.

Kelly Malcom:

And a study that stuck out to me is also about AI. Our researchers developed a tool that can predict the genetics of a cancerous brain tumor in under 90 seconds, which can help doctors make decisions about treatment. Absolutely incredible.

Jordan Goebig:

Yeah, it really is. As usual, we'll provide links to the full articles and info about our featured guest in the show notes. Now let's get on to our guest.

Today's guest is Dr. Geoffrey Siwo, a research assistant professor in the University of Michigan Medical School’s Department of Internal Medicine and Research Associate at the Michigan Center for Global Health Equity. His research focuses on accelerated and equitable innovation, using emerging computational technologies such as artificial intelligence, programmable biotechnologies such as CRISPR systems, and frameworks for scientific discovery at a global scale such as open innovation challenges. He is a co-founder of Anza Biotechnologies, a startup focused on accelerating discovery and sustainable manufacturing of therapeutics. Previously, he was a research assistant professor at the University of Notre Dame, lead researcher at IBM Research Africa, and a co-founder of Helix Nanotechnologies, a DNA nanotechnology company. His work has been featured in several media outlets, including CNN, USA Today, Fast Company, Ozzy, and other media outlets.

We are so grateful to have you here today. Thank you, Dr. Siwo.

Dr. Geoffrey Siwo:

Thank you for having me.

Jordan Goebig:

Thank you so much for coming in today. We're going to jump into some of our questions and dive into the fundamentals. So first, how would you describe information technology?

Dr. Geoffrey Siwo:

So when we think about information technology, we know that at the center is the use of computers to store information and data. But in today's world, we think of information technology even more broadly, because what information means is changing. If you look at the future, we see that new technologies will emerge, such as quantum computers. So as we think about information we should think about not just today, but also the future that is to come.

Jordan Goebig:

Wonderful. You were very succinct at my big question. I appreciate that very much. You really distilled that fundamentally for us, so I appreciate that.

Kelly Malcom:

So I'm really interested in the intersection between data and health, and I think a lot of people are wondering about how their data will be used to improve their lives, and just improve our understanding of the world at large. Could you maybe get into how you see data science helping advance basic science and medicine overall?

Dr. Geoffrey Siwo:

I think data has always been a key component of science. If you look at the scientific method, which is centuries old, at the heart of the scientific method, is that we come up with hypothesis, we make observations, and those observations inform how our ideas and views of the world evolve. So data really is fundamental to all the sciences.

In the biological sciences or health sciences broadly, data has a critical role to play in discovery of fundamental biology. So whether it is to understand the process of disease, what are the underlying mechanisms of a disease, whether it's infectious disease or whether it's a chronic disease like cancer, we need data to advance our fundamental understanding of disease. Then data also is central, of course, to how we develop solutions. So solutions can be in terms of therapeutics. So in drug discovery, data can help us to come up with new drugs for very specific diseases. But of course data is the first type because we have to integrate data with other technologies. So for instance, we can use statistics, we can use machine learning or artificial intelligence and other kinds of approaches. And then of course, data can help us to design new vaccines rapidly. So the faster we can collect the data, the faster we can also come up with new therapeutics and vaccines.

We can also predict the risk of disease so that we can detect disease early. So I see it as a very important, all of these areas in fundamental biology.

Kelly Malcom:

Yeah, I think I saw a recent story about how artificial intelligence was able to predict a woman's breast cancer four years before it actually developed. So it's pretty amazing.

Jordan Goebig:

Yeah, that's incredible. So we are going to dive a little bit more into this topic soon, but I know that you have had a non-traditional path in your educational experiences, and I'm just really curious to hear a little bit more about you and where your passion for artificial intelligence comes from.

Dr. Geoffrey Siwo:

Yeah. So to look back where this passion comes from, when I was just in about eighth grade, my teacher was defining what cancer is as uncontrolled growth. And that surprised me because I lot of people talk negatively about cancer. So I thought, why should uncontrolled growth be dangerous? And I questioned that and I thought maybe my teacher is wrong. So that day, I still remember, I went on to try and read about cancer in the dictionary. Then I found the same definition. I went into a medical book that my dad had. I found the same definition again, and that really amazed me. And I quickly began to learn that to understand cancer, one needs to understand multiple areas of biology.

So we need to understand from cell division, which is at the center of how cancer emerges. We have to understand how single cells interact to form tissues and organs. We have to understand how growth and development happens, and we have to understand immune system. And today, one of the most promising avenues for dealing with cancer has to do with immunotherapies, which are based on the fact that the immune system can be very powerful in attacking cancer. So anyway, so that really drove my passion for biology. And because of that, my view of biology has always been like a journey. And I've found biology to be very complex. That when we think about disease, we need to think about any single disease in the context of multiple diseases, and in the context of the complexity of biology. And to do this, no single approach can work well. And I think this is where artificial intelligence comes in and where my interest developed.

And maybe just to add on that, because this is something that I'm really passionate about. When I was just in my second year of college undergraduate studies in Kenya, and I had this idea that within our human genome there are virus-like material that can interact with with HIV. And I wanted to test this idea. I didn't have a lab, and I talked to the professors I was working with and he said, "It's a brilliant idea, but we can't do it."

So I thought what if I went to a computer and test this idea? And I didn't even own a computer, but I went to a cyber cafe. We still have those in many places in Africa, where I could spend a dollar an hour to browse the internet, as we used to call it. But when I went there, I used to analyze DNA on the internet already. The human genome was available in 2003 on the web. There were tens of thousands of HIV sequences actually, already sequenced by the Los Alamos National Laboratory here in the US, and I use that data to test this idea. But this really taught me that when you take very complex problems in biology, convert them into data problems, then you can solve them at the speed of computation. And so I see artificial intelligence as one of the tools in this process of converting human problems into data or information problems, and that allows us to solve these problems at a very rapid scale, and that is where I see AI coming in.

Jordan Goebig:

I love that. I love your story of being so passionate about biology in eighth grade. I hope my future eighth grader has that sort of drive when they hear something in school, and that you weaved it all the way into artificial intelligence through college. And now you're doing such incredible work with those sciences here.

Dr. Geoffrey Siwo:

Well, thank you.

Jordan Goebig:

Yeah, yes. You've said some really amazing things from just reading, some background on you and some different stories, but hopefully you can succinctly talk about ... You've said some really cool things about the development of the human species and how important the advancement of technology has been in the human species advancement. So I'd just love if you could maybe speak on a little bit more of that and talk to us about how technology advancements helps the development of our species.

Dr. Geoffrey Siwo:

Yeah, so that's a great question. I've always wondered, today's scientists tend not to be philosophers. We know that in the history of science, some of the greatest scientists were also great philosophers. Like Aristotle, they were philosophers. So I also love philosophy and I always reflect upon the history of humans, what makes humans special. And at the heart of it relates to your first question, which was on information, what is information technology? But if we just think about what is information, I think if we go back into human history, one of the remarkable things that happens in our history is our ability to develop language among our ancestors.

When our ancestors lived in caves and they would draw art on rocks, the act on rocks allowed one human to pass an idea that they leave behind on rock. And that idea would be seen by other humans who were not physically interacting with this one single human. And I think this is beginning of technology, the ability of humans to communicate. So beginning with writing on the sand or painting on the rocks, and then quickly evolving into printing. And with printing, it means that humans were able to share messages across borders. So you could print a book. People in your own generation can read the book, but people also in the future generations could read that book.

And so this allows ideas to spread widely. But if you go ahead you see that with computers and now the web, what it means that humans can now learn not just from the present generation, not just from me and you in the same room, but we can learn from those who are thousands of miles away from us. We can learn from those who no longer live here with us. We can learn from all of that, from all of that knowledge. And so I think at the heart of that, beginning from the rock art, that's information. And now being able to take that information and look at it digitally means that it is even more powerful, because more people can have access to that information and they can also add new ideas onto that.

And I think there are two elements. Of course, there's the physical world, and there's the world of information, which includes ideas even before information technology. And so as our physical world and our information world get closer and closer, tens of thousands of years ago, humans were building tools that were simple, that were mechanical. But even building those tools required some knowledge and skills that humans had. And now with the digital technologies, we are beginning to bridge that gap of what we do in the physical world and what we do in the virtual world. And I don't want to go into the metaverse, but this can say that in a way, this is the transition. That if we bring those two closer together, then it means we'll be able to solve problems faster.

So if we go into the space of health, we can build a digital model of a patient. This is called a digital twin. And if the digital twin of the patient behaves like the real patient, then it means that we can do experiments in this digital twin experiments that we can do in the real patient. We can see how this digital patient or digital version of you would be, hey, if we gave you a certain medication, or if you change your diet in a certain way, or if you did some certain exercises, we could see how the digital version changes. And so I think this would be very powerful in future. And perhaps thinking of the metaverse is the beginning of that future.

Kelly Malcom:

It's exciting, and it's incredible if you think about the arc of human history and how far we've come with sharing information and knowledge with each other, and then digitizing almost all of human knowledge. Now we have things like Chat GPT, and I got an email about Google Bard, and it's a little bit off-putting. And I think a lot of other people might be scared about this metaverse or this virtual world that they don't really understand, or they might have these fears about computers outstripping humans ability to process data. Obviously computers can process data a lot faster than people can, so I think there is some trepidation about where we're moving into in the virtual space.

Can you maybe help put my mind at ease a little bit? Do you think there are any dangers to using data to create virtual versions of ourselves? Are there any caveats that you have for researchers who are entering this space, especially as it applies to human health or things that have life or death consequences?

Dr. Geoffrey Siwo:

Yeah, so this is a really important question. So I think there are of course [inaudible] that we have to face, because one of the things with Chat GPT, and broadly foundational models, which are AI models that are trained to be able to perform across a wide range of language based tasks. And as you know, one of the powerful things in our history has been language. So if we bring this ability into the technology we build, it could similarly have a huge transformation on our future, just as our own language has had a huge impact on us. So if the tools we build are developing language, then it means that it could have a huge impact. So I think because of that, we should approach this responsibly.

So I think we need to develop some norms of how to develop these technologies responsibly and ethically, and especially in the ways that are inclusive. Because one of the challenges we face with the technology we're building today, be it Chat GPT, or be it Google Bard, is that these technologies are being trained on data that is available on the web. And we know that the web is not an accurate representation of our world as we sit here today. There are millions of people around the world who have no access to the internet. Their stories are not available on the internet. Even for us who we have access to the internet, we only share certain stories. Certain of our stories are shared on the internet, be it personal stories, the Facebook posts we make, but also be it the research we publish. Scientists don't publish the experiments that fail. They only publish those that work. So I think we need to think responsibly and ask ourselves, the technology we are training on this data, what are the limitations? We have to understand the limitations in terms of the biases in the data used in training. And this can be biases in the demographics. So gender, race, we need to ensure that we have a balance of representation in the data.

So I think the bigger risk will come in, what are the biases in the data, and what are the biases in the people building the technology? So who are the data scientists? Who are the engineers? Do we have an inclusive environment for engineers that come from different backgrounds? Yes, I think it's critical that we try to innovate responsibly, and I think there's no one definition of what responsible is. So I think that is where we need to begin.

Jordan Goebig:

Yeah. I think this is a really good segue your answer into kind of my next question, which you just mentioned that when we have data, there's also these little demographic pinpoints about these people that we need to be considerate about. So I'd like to hear a little bit more about why it is important to humanize the data and the benefits of doing that.

Dr. Geoffrey Siwo:

Yes, it's really important that we humanize the data, be it by making the data or the solutions we develop to be more understandable to anyone. Because the solutions we're trying to develop are for humans. So we need to ensure that we engage other people when we're developing these solutions, be it an AI system that is trying to diagnose cancer. We need to ask ourselves what's more important? Do we need to be able to predict with a certain level of accuracy? Because when you say that somebody is at risk of cancer is, let's say 50%, is it 50% today? Is it tomorrow? Is it in 10 years? So we need to ask ourselves that.

And also we need to ensure that the framing of our problems to the AI systems is the same as what we really intend to solve. So generally when it comes to AI or robotics, there's always this question of is there an alignment of the goals of the machine and the goals of the human? So for instance, you can have a very good AI model that can differentiate between who has cancer versus who doesn't have cancer. But it doesn't mean that that AI model is actually picking cancer. It might just be picking something else in the data that is associated with the people who are cancer in the dataset. And so that can be a very big problem when you try to apply that AI in a new context.

So for instance, in academia, there has been cases where an AI system was trained to differentiate between patients with cancer versus those who don't have cancer. It performed very well. But in the end, the researchers found out that the patients who had cancer, they came from one hospital facility, and in that hospital facility, all the X-ray images had a certain mark in them. And so the AI learned the presence of the mark, and not necessarily how to detect cancer.

Kelly Malcom:

So really the output of the machine is only as good as the programmer. So you have to make sure that you're asking the right questions to get the right answer.

Dr. Geoffrey Siwo:

Yeah. And that can be really hard for computers.

Jordan Goebig:

Is that something in your research that you're looking at is finding those anomalies in the data? I'm just so curious about, until I started working with Kelly, and I've only been here a few months, honestly, as a normal person, I didn't realize there's data banks out there where people are collecting data and using it for research. I didn't realize all the different ways that research and data was being collected. So I guess maybe even stepping back, where do you get your data from? Is it specific projects you're working on, are there larger banks that you're pulling from? And what are you doing with the data? How are you analyzing it? Are you using it on cancer related projects, or other things? I know those are really big questions, I'm just super curious about how the data comes to you.

Dr. Geoffrey Siwo:

Yeah, so let me say first that my approach when it comes to machine learning or AI is not to be purely data driven. I know that there's a lot of excitement about big data, but I believe that we should combine data with knowledge, that for us to build AI systems that can be trusted, in many circumstances we have to combine data with knowledge. And by data with knowledge I mean that today when you build AI systems, you can feed the AI systems like pictures from the web, and then the AI system can learn how to generate fake faces or even fake movies. And I think that shows the promise of purely data driven approaches. But the weakness it also brings that is that in the virtual world, there are no physical constraints.

So for instance, in the virtual world you can create any image you want, even if that image may not make sense. But we know that in the physical world, there are constraints you know for instance that if something is hanging in the air, it'll fall down because of gravity. That's a physical constraint that you know. If you feed your AI data just from the web, it will not understand something like gravity, or something like light. And that can actually lead to catastrophe.

A good example is in self-driving cars. Because self-driving cars, many of them are trained on data that is gathered by a stream of video, or in some cases they use LIDAR, which is a laser based technology, to collect data on objects around the car. But it's not possible to collect data on all possible scenarios that can happen on the road. So we have seen Teslas being involved in some accidents when they're in the self-driving mode. In part because the AI system trained just on the static images or video stream failed to understand that, for instance, a police car or an emergency vehicle with the siren going on can still be static. It doesn't mean that the car is moving, but the Tesla AI system never saw a video of a police car that was moving with the siren on, and so you end up in this crash. Yeah. So I think fundamentally, I think we should combine AI or data driven systems with some knowledge driven system.

Jordan Goebig:

Really, I like that. Earlier you mentioned getting philosophical, and I think it's needed. It sounds like what you're doing makes it seem a lot less intimidating to me. Again, as somebody with no background in AI, you think of self-driving cars and robots taking over the world. And you see headlines that joke about that, but it's really nice to hear behind the scenes how thoughtful your lab is being about it's ... We're not just inputting data into a computer and letting it go off into the world. Part of what you're doing is bringing that human element, because without it, as you said, Kelly, it just isn't going to work and it's not going to solve problems for people and help human health.

Kelly Malcom:

It'll probably exacerbate some of the existing problems we have if we're not updating that knowledge. And that's why I don't think people will ever really be replaced.

Jordan Goebig:

Yeah, no, no, that's what I'm learning.

Kelly Malcom:

Because it's a biological world. We are just using this virtual digital tool to understand the biological world. So it's always important to have that element there. So I appreciate that, like Jordan said, that we have people who are invested in making sure all of this technology is grounded in knowledge.

So the medical school has a bold science research scout program, the idea of which is to identify researchers like yourself who are closer to what some of the new up and coming research areas are, and can identify people who are investigating really important projects and make a small investment in getting those off the ground. As a scout yourself, what types of projects do you think you'll be looking for, and what are you hoping to see from our researchers?

Dr. Geoffrey Siwo:

Yes. So I think the scout program is a really exciting program, modeled after a program by the Hypothesis Fund, which also funds bold ideas that would typically not get the conventional grants. And I think when I look for ideas, and I should say that I've already funded a project, but I look for ideas that hold promise in terms of changing the way we think about specific areas of science or biomedical research. And also I look for ideas that would not typically be funded under normal funding mechanisms. So these ideas tend to be ideas that are still very early stage, because normally when you apply for conventional grants, you need a lot of preliminary data. And for the scouts program, we don't need this ... Preliminary data is not a requirement. So that allows scientists to bring their wild ideas that they might have been sitting on, but they might have a good reason to think that these ideas hold some promise, either through the hypothesis they have, or some fundamental understanding of science they have. And I think this is ... It's really exciting program.

Jordan Goebig:

Since I didn't realize you have only been here a few months, I'd love to just hear how's it going in terms of getting set up as a researcher at the University of Michigan. What that experience has been like for you?

Dr. Geoffrey Siwo:

It's been amazing experience because Michigan is a highly collaborative environment. So one of the things I've really loved about Michigan is the community, and just the diversity of disciplines that are represented. I've been able to meet new collaborators, I've also been fortunate to join other groups on campus. So for instance, e-HAIL that is really trying to galvanize diverse disciplines across campus from engineering to the medical school to come and work on ways of leveraging AI for advancing human health. So it's been a really amazing experience.

Kelly Malcom:

Yeah, I think that's one of the differentiating features of the University of Michigan is the willingness for people to work together, collaborate, and tackle research questions from various angles, from various disciplines. So I think it's great that you're already engaged in that. And I wanted to bring it back to computational medicine a little bit, computational biology. How do we make it more accessible? Because it does sound like we need more diverse voices in this field to make sure that it is resulting in applications that can benefit everyone. So how do we make it more accessible to people in maybe under resourced areas, other countries, and just people from backgrounds that might not have known about the field, and might not be drawn to the field as it stands now?

Dr. Geoffrey Siwo:

Yes. So that is a really important question, and one that I really relate to, because I grew up in an under-resourced setting. And if you go back to the early 2000s, there were very few people with a computer that I knew in Kenya. Mobile phones were just beginning to come up. Now everyone has a mobile phone in Kenya. And so I really relate to this issue of under-representation.

And so I would say that the good thing with computational medicine, and more broadly computational technologies including AI, is that generally they're more scalable. So of course, at the heart of it, why it's scalable is that with the computers, for more than 50 years now we've been on the Moore’s curve. So the speed of computing has been accelerating exponentially, and the cost of computing has been dropping down. We've been able to fit more and more transistors, beginning from it used to be the vacuum tubes, then transistors, then now silicon chips, it's becoming smaller and smaller. And with that means, it means that the cost of computing has reduced dramatically.

And in parallel when it comes to biology, actually we've experienced even a large exponential growth. So if you take the case of the human genome sequencing, the first human genome was sequenced for more than a billion dollars, took a decade. Today you can sequence a human genome for less than $1000. Or something like that. So a million fold reduction in cost of human genome sequencing. And as a consequence of that, we have had also an exponential growth of biological data that is fully available today. As I mentioned before, even growing up in a very under-resourced setting, and not even owning a computer, because of this magnitude of growth in biological data, I was able to sit in a cyber cafe in Kenya, access this data on the web generated by laboratory in the US, and examine questions that nobody else perhaps in the world had examined before. So because of that, I think that the convergence of computing and biology is going to be a democratizing force of who does biomedical research.

So to do biomedical research, you used to need to have a lab coat, have a sophisticated lab, but with computational medicine, you can do biology without having a lab. You can do it on the computer. That's still important. Biology, you can do on the computer, eventually you have to translate it into the lab if you're developing a medication. But even that is undergoing an exponential wave. So if you take the case of DNA sequencing, you can analyze DNA on the web, you can order DNA sequence from the web, you can have it delivered at a lab that you can outsource. You can have somebody do for you those experiments without running a lab. And in the future, which is beginning today, you already have robotic laboratories already emerging where you can push this digital information into a robotic lab to go ahead and do the experiments for you. And I think because of all of this, we are going to have a wider participation. But for us to ensure that it happens in a more responsible way, we need to ensure that our research programs are more inclusive.

So we need to engage communities in underrepresented groups, and there's a lot of effort now both from the government agencies, as well as from private organizations to ensure that in terms of who gets funded, we can have a diverse community of grant recipients. We can also do targeted training where we go to underrepresented communities to bring them together and learn how to use data science in biology. So I think we have to be more inclusive from the research perspective.

Jordan Goebig:

I love ending on a happy thing in a way, because we talked about some of the scary things about AI, and then I couldn't help. I was I smiling at you while you were talking, and about how much technology has advanced things, and we were talking about our kids before we hit record. And I'm thinking about how if one of your kids follows in your footsteps, the places they can be when they're in eighth grade, when they get to that point because of technology, which is super cool to think about.

Dr. Geoffrey Siwo:

Yes. Yeah, it's super cool.

Jordan Goebig:

Yeah.

Kelly Malcom:

It's an exciting time just in history. And if my son decides he wants to do some DNA experiments from home, I will encourage that.

Jordan Goebig:

They probably send kits now for four year olds on how a DNA sequence on their own, which is amazing.

Dr. Geoffrey Siwo:

We have to do it responsibly. There are already community bio labs where even if you're not a scientist, you have this lab in the community, you can just go in there. It's like a [inaudible] space, but for biology. And anyone can walk in, learn how to extract DNA, learn how to sequence DNA, and of course the DNA sequencing machines themselves. There used to be huge machines, several rooms. Now you have a DNA machine that can sit on the desktop. And also now you're beginning to have DNA synthesizer that can sit on your desktop.

And so what that means is that you can do a discovery on a very small scale and you can manufacture things like RNA. So RNA is a good example because we have seen that in human history we have never had an RNA vaccine until just two years ago. With advancements we have now, there's going to be more democratization of RNA science, and that means that things like vaccines can be manufactured locally and faster than they have been during the COVID time.

Kelly Malcom:

That's great. Lots of hope for the future.

Jordan Goebig:

Yes. Okay. So to end, this is your rockstar moment. I'd love to hear if there's any shout-outs you want to give to projects or collaborators or any publications, this is your time to just give us a little bit of information about what you have going on. Or again, anybody you want to shout out.

Dr. Geoffrey Siwo:

So I've been very fortunate to work with amazing scientists. I'm really glad that about two years ago I came to know Dr. Akbar Walji, who is a clinician and also a data scientist. And I've been just so fortunate to work with him, because he's given me a different perspective coming from a clinical background of how basic science can have an impact on human medicine.

So as a biologist who is very passionate about fundamental biology, that is where my journey in biology began. I am really excited by one of the projects Akbar and I are collaborating on. Still very early stage, but exploring the idea that how can we make the advancements in genomic medicine benefit patients for whom genomic data is not available? Because one of the biggest promises of the sequencing of the human genome was that it'll transform human medicine. But we know that that benefit has not reached many patients around the world.

Today if you walk into a lab to get a test, they won't be sequencing your DNA. But you know what they can do in many cases is that they can run some routine laboratory tests. So looking for things like your white blood cell count, your hemoglobin levels and so forth, we think that we have a way in which we can link information coming from this very easily accessible data from routine laboratory tests that you can get in many clinics or hospitals around the world, that there are ways we can link that with the latest discoveries coming from the sequencing of hundreds of thousands of genomes across the world.

So we are fortunate, at Michigan here we have the Michigan Genomics Initiative. That is biobank of over 70,000 patients in Michigan who have had their DNA genotyped at several loci. And so how can you use this information and link it with the information that most people when they walk into like a clinic they can get their blood tested. And maybe if we combine those two pieces of information we can develop more accessible models of predicting cancer risk. So I'm really excited about that.

Jordan Goebig:

Yeah, I'm excited to check back in with you. It sounds like-

Kelly Malcom:

I think we'll need a part two with Dr. Siwo.

Jordan Goebig:

Yeah.

Kelly Malcom:

Well, we really appreciate you giving us this insight into the incredible field of computational biology and data science and where it might take us. So it's been super, super fascinating to hear from you. Thank you so much.

Dr. Geoffrey Siwo:

Well, thank you Jordan and Kelly. It's been a real pleasure.

Jordan Goebig:

Yeah, we really appreciate you spending your extra time with us today. We hear and know you're a very busy. And again, I just can't wait to check back in and see where your projects are.

Dr. Geoffrey Siwo:

Well, thank you. Anytime.

Kelly Malcom:

Thanks for listening. The Fundamentals is part of the Michigan Medicine Podcast network, and produced by the Michigan Medicine Department of Communication in partnership with the University of Michigan Medical School. Find us and subscribe wherever you get your podcasts.