Applications of Quantum Computing in Biomedical Research

Podcast Transcript

Introduction: Neuro Pathways, a Cleveland Clinic podcast exploring the latest research discoveries and clinical advances in the fields of neurology, neurosurgery, neuro rehab, and psychiatry.

Glen Stevens, DO, PhD:

In March of 2023, Cleveland Clinic and IBM announced the deployment of IBM Quantum System One, the first onsite private sector IBM-managed quantum computer in the United States and the first quantum computer in the world to be uniquely dedicated to biomedical research. Installed on Cleveland Clinic's main campus, the quantum computer is part of a 10-year partnership between Cleveland Clinic and IBM called the Discovery Accelerator that is focused on advancing the pace of discovery and healthcare and life sciences by using high performance computing. In this episode of Neuropathways, we're discussing how quantum computing could be harnessed to hasten biomedical innovations. I'm your host, Glen Stevens, neurologist/neuro oncologist in Cleveland Clinic's Neurological Institute. Joining me for today's conversation is Dr. Lara Jehi. Dr. Jehi is a neurologist in Cleveland Clinic's Epilepsy Center. Cleveland Clinic's Chief Research Information Officer and the Discovery Accelerator Executive Program leader for Cleveland Clinic. Lara, welcome to Neuro Pathways.

Lara Jehi, MD:

Thank you, Dr. Stevens. Happy to be here.

Glen Stevens, DO, PhD:

We've known each other for many, many years, but I'm sure there are individuals out there who don't know you. Tell us a little bit about yourself and how you came to the Cleveland Clinic many years ago.

Lara Jehi, MD:

I did come to the Cleveland Clinic many years ago all the way from Lebanon where I did my medical school at the American University of Beirut. I specialize here in neurology at the clinic and then epilepsy, and I've been on staff since 2006. My journey has evolved to build a research program around predictive modeling and analytics and in the context of epilepsy surgery, but then grew from there to the point where I was fortunate to serve in this role of chief research information for the organization where my job is to link technology with research for investigators across the health system.

Glen Stevens, DO, PhD:

Well, I've always known you as a great epileptologist and I'm so excited that you're in this role and I'm sure you do a great job. Really looking forward to the conversation today and what happens in the future. I'm an old guy. I grew up in a small town in Canada, I hate to tell you, but I remember using an Abacus and then in university in the seventies, I remember the card readers. We used to do punch cards and everybody would carry around a shoebox and in order to do a very simple math formula, you would need a size 12 shoebox with all the punch cards in it to do it, and inevitably you would trip, you would fall, the card box would spill ...

Lara Jehi, MD:

Yeah, have to start again.

Glen Stevens, DO, PhD:

... And it was a disaster and everything was easy, right? It was binary, and I'm sure you're going to tell us a difference. It was zeros and ones, it was on, it was off. It sounds like we've moved ahead quite a bit. In 2021, Cleveland Clinic and I B M entered into a 10-year partnership to establish the Discovery Accelerator. Share with us some of the background on what this partnership entails.

Lara Jehi, MD:

It's a futuristic partnership plan, and for us, it came at the time when Cleveland Clinic had to think way ahead. If you recall, 2021 was the year we celebrated our centennial as an organization. We took stock of what we accomplished in the previous 100 years, and we thought of what do we want our next 100 years to look like? What kind of impact we want to leave in the world when our grandkids look back 100 years from now and think about what we've accomplished. We were in that long term, very long term mindset when we decided that we have to think very creatively, innovatively, and take risk. The pace of biomedical discovery now, we have a huge growth, as you know, in the amounts of data that can generated, whether it's imaging, genetics, electronic health record information. We're bombarded with data and the scope of that is growing so much and we are limited by how effective the computational tools that we have now are able to deliver.

We are also limited by expertise, workforce, people who know how to use and apply all of this computational technology. We went into this partnership with IBM Research to address all of those challenges. It is organized around three pillars, high performance computing, artificial intelligence, and quantum computing with research teams that are organized between the two institutions, Cleveland Clinic and IBM partnering together on projects that use all of that technology. The undercurrent beneath all of this is education. We're developing courses and AI courses in quantum computing, working with the local schools all the way from high schools locally in the area and through the web, up to graduate students in the neighboring universities. What we are looking for is to build an ecosystem of biomedical research across all industries that are sharing with us this excitement, this interest in looking at what the future of biomedical research can bring.

Glen Stevens, DO, PhD:

Well, it sounds like a very, very exciting time to be in healthcare. Now I'm going to let you show your computer expertise, talk to us specifically about quantum computing in a way that I could understand and give us a brief overview of what it is and how it really differs. I think that's really the secret sauce, right? It's not just a really fast computer. It's really completely separate. Take us through that process.

Lara Jehi, MD:

Yeah, you're exactly right, Glen. It's definitely not just it. It's the difference between say when the light bulb was first invented versus the way people were doing lighting up to that point with candles, right? You couldn't have enough candles to give you what a light bulb could provide. That is that transition between classical computing, the candles that we have right now, the way the regular computers work, including supercomputers, including the computers that we apply AI on, those are the candles. They're limited with what they can store only zeros or ones. Each bit is either a zero or a one, black or white, on or off, versus what the light bulb electricity transition, what quantum computing can bring, it's a completely different way of doing computing. The zeros and ones is what we call deterministic computing. There is no choice about it. Quantum computing does what we call probabilistic computing.

You can think of the unit of computation in a quantum computer as a sphere rather than a box that's either black or white. Think of how the earth is, and you have the north pole and the south pole. The north would be, say the zero, the south pole is the one, but the units of information can exist anywhere along that sphere and wherever it is that's captured as a probability. It's an infinite amount of data that you can store in each unit of compute, which is what we have with those qubits. To translate this to English, so you have a sense of the scale. If you want to store the data that you get from sequencing, the DNA in one individual, the human genome, you need 1.5 gigabytes, giga, so the 10 to the nine bytes of regular bits in a classical computer. The equivalent of that in a quantum computer is 34 qubits. 34 will do the job of 1.5 gigabytes. If you want to store the human genome of every single living human being on the face of the earth, you need 68 qubits. Just by doubling the number of qubits from 34 to 68, you have that exponential growth and the amount of data that can be stored because that's the power of exponential computing that you get with the quantum. The computer that we have here in Cleveland Clinic is 127 qubits.

Glen Stevens, DO, PhD:

Well, I was doing a little bit of reading. I love the word qubit. I've learned some new words as we've come through, so maybe I'll have to get a tattoo of that as I move along. Quantum computing can be harnessed for incredibly complex projects, yet is still in its very nascent phase. We're very early on in this. Tell us a little bit about how its current capabilities are being utilized specifically at the Cleveland Clinic.

Lara Jehi, MD:

Sure, so we have several projects going on now using the quantum computer. In general, you can think of it as serving three different types of scientific questions. There is the quantum simulation, there is the quantum machine learning applications, and then there is quantum optimization. Quantum simulation, this is the easiest way to communicate that is when you have problems where you need to go from a formula on paper to a structure that you can look at in space. This is fundamental for drug discovery. You start with the chemical formulation of what a compound could look like, but before you can get it to a drug that is usable, you have to either make it and then test it, or you have to simulate what that formula would look like in space and then simulate the experiments that tell you how that drug would fit with receptors in the body, et cetera.

That simulation requires many computational paths that have to go in parallel to be complete, and traditional computers get stuck with that. They're limited with how much they can simulate. Quantum computers do this more efficiently. We have projects that are going on now, one that was just recently funded by the Welcome LEAP organization to look at how we can simulate protein folding using the quantum computer. That pretty much is along the drug discovery path. Quantum machine learning is a field where we are taking all the models that get stuck with their predictive accuracy despite applying the best AI technology that's out there. Then quantum helps with improving the accuracy or reducing the number of features, meaning in imaging for example. There are so many different aspects of that MRI that you can include in your model to come up with a prediction. The more features you include, the more complex the computation is going to be. Quantum can help with reducing the number of features, so you don't have to test like a thousand different things to come up with a result. You can test much less. Quantum optimization is helpful in clinical trials. You design those trials better, make them more efficient so that they can finish faster.

Glen Stevens, DO, PhD:

Well, it's so exciting. They say, I think on average that from a discovery to implementation of a drug is about 17 to 20 years.

Lara Jehi, MD:

Correct.

Glen Stevens, DO, PhD:

With this type of system, do you want to lay a guess on what happens to that timeline?

Lara Jehi, MD:

Well, our goal is to cut it down by 10 x, so that's what we're working towards. That's the vision. That's why it's a 10-year program and we're not going to get there tomorrow, but unless we start now, we're never going to get there. We really want to change the way medicine is done.

Glen Stevens, DO, PhD:

Mm-hmm. Let's focus on the neurologic space since we're both in that realm. How do you see quantum computing being applied in a clinical or a surgical setting?

Lara Jehi, MD:

I think it's a bit soon to think about it directly in a clinical context. Right now, it's still a research tool. We have to define its applications through research first and then we can transition it to a clinical application. Bigger pictures, I cannot give you specific examples there, but the bigger picture vision from it is that the quantum computer is not going to replace traditional computers. In a way, we can think about it like how say now you've scaled up your desktop, your robots in the OR, whatever you use so that you can integrate in it, whatever AI can bring to accomplish the clinical goal that you want to do. The vision from Quantum is that it will be able to seamlessly link to the computers, the tools that we have now in clinical care that use this, use computation in general so that it's seamless and it works on the back end and whichever computations can happen with classical can go that route. Whichever ones need the quantum will go there. I suspect it's going to be the same as how now computers are just, if you can't even imagine your job without them, sometimes we want to, but we can't. The quantum computers will be the same. They'll just be in the background of everything that we do.

Glen Stevens, DO, PhD:

Well, it's interesting you mentioned that sort of the three pillars that AI is part of it as well, which I guess is part of the quantum machine learning, or am I correct about that?

Lara Jehi, MD:

Yes, yes, yes.

Glen Stevens, DO, PhD:

I was down in Gamma Knife the other day doing a case and one of the radiation oncologists was down there and he was planning a case for a patient that had GI cancer and they have to draw out the colon and they have to physically sit there and draw the pictures and try and separate it from the normal, and they put it through an AI software that would then interpolate what was small bowel, large bowel, where the intersection was, what was normal, what wasn't, and he said it cut off about two hours of the time that he would take in order to draw. Certainly, I can see these applications from learning what's real and what's not real and really cutting down the time that physicians would be taking to do these things. I'm sure in epilepsy as well, right? It will help you differentiate cortical migration problems, those types of things from normal tissue. I mean, it just seems an explosion of possibilities.

Lara Jehi, MD:

Absolutely, Glen, one of the projects that we have now ongoing with the Quantum is looking at its applications in radiation therapy for breast cancer, where the team is looking to see whether quantum computing can help with designing the radiation treatment, the path better than what's currently available to avoid toxicity and make the treatment more targeted. This is a project that's going on right now. Another project that we have going on right now with MRI data is looking at knee MRIs where one of our researchers is wanting to predict the recovery course after knee replacement using MRI signals and the ratio of fat and muscle in there, and she's stuck with AI algorithm that's there. It requires 20 features. She's looking to see whether quantum can help.

When I meant I don't see it now in the clinical space, it's because as you know, usually you need to go through a lot demonstrating that something first works in the research space and then go through the regulatory approvals that are needed, which have not been defined yet for a quantum, but from a trajectory perspective and applications perspective of what type of work is going on now with quantum and research and how it can translate directly to patient care, there is a lot of promise, whether it is with better design of therapy, like you mentioned with the colon example, in the radiation therapy example that I mentioned, or interpretation of imaging, early detection of disease, biomarker discovery. I mean, it's limitless.

Glen Stevens, DO, PhD:

The first thing I think of with our conversation is if I had a great idea, how can I move forward with it using the quantum computing? Do you have a process that people want to do a project with you? Let's say I have a great neurologic oncologic project. I might think it's great. You may not, but what's the process?

Lara Jehi, MD:

Well, Glen, I'm sure it's great. You're full of great ideas. We have processes now in place both for internal, so researchers or clinicians who are in Cleveland Clinic, so that's an internal process that we have, and we have another process for external collaborators, startups, companies, academia from anywhere in the world we're collaborating with. Internally right now, we have to get our pipeline of projects developed. What we've done is we've had a competitive process, so we put out a request for proposal every quarter, and we get applications from around the clinic. We have a scientific review committee that looks at them, and we support five of those every quarter. What we've done to help the ideas mature to have successful applications is we've been organizing workshops and seminars and open houses where the data scientists just are there, make themselves available, and the clinicians and the researchers with the ideas come and can meet with them and just start the conversation.

The process of developing research, as you know, it's never a one, I have a great idea, then I keep it to myself. You have to go through the exercise of being challenged and betting it and building it to the point of a proposal. We offer that, and then in parallel, there's the competition. We built people so they're ready and then the proposals come in, and then for external collaborators, we have a communication that can happen through the Center for Computational Life Sciences that I lead. That's the administrative entry point if you want, for all computational work that people can reach out to and start the conversation.

Glen Stevens, DO, PhD:

It's really like learning a new language, isn't it? I mean, there's so much education that needs to go on. If I'm sitting out here listening to this and I'm thinking, boy, I don't understand enough about it. I want to learn more. I want to come to an educational workshop. Is there a web address that they could go to? Is there an easy place they can go to find this information? If you have that, do you want to share that?

Lara Jehi, MD:

Sure, yeah, I'll be happy to share it. We have a lot of information that we put on the center website, the Center for Computational Life Sciences. If you want to learn more, you can visit clevelandclinic.org/discoveryaccelerator. All of the seminars are recorded so people can listen to them whenever they want. There is articles that we put together, so we are working very closely with our education institute, our leaders for research education, to develop all of these courses. We've developed a curriculum, a course that's now integrated in the medical school curriculum and data science right before our medical students have their research here. They are sort of going to be thinking about this exactly like you said, Glen. I mean, it's an exercise and education, and unless you learn to think differently and you open up to what this can offer, you would be missing out.

Glen Stevens, DO, PhD:

The partnerships for 10 years, best case scenario, what happens at the 10-year mark?

Lara Jehi, MD:

Well, best case scenario is don't get me started. We would have every medical student who is so familiar with this and that they would look back and really not see anything surprising in what we developed. The best case scenario for me would be 10 years from now, somebody would look back at the way we are thinking about medicine and doing research now, the same way you started talking about the punch cards. We have to make progress where we revolutionize the way research is done, but then for people, then this is just the way it's done. This is the new normal. We should get from question to answer much faster than we are right now. Learning, we have to learn fast, fail fast. We have to discover new treatments at the pace that is much faster than that 17, 20 years that we started with. Data science education would be an expectation in learning how to do medicine. I mean, I just imagine a very different world from where we are right now.

Glen Stevens, DO, PhD:

Yeah, it's shocking, isn't it? I was with my seven year old grandchild recently, and I had an issue with my phone. I couldn't figure something out and she goes, "Let me see that," papa," and she just, within seconds, figured out and has never owned one and just sort of sees her brother play with one. It's just, as you say, it becomes the normal.

Lara Jehi, MD:

Yeah.

Glen Stevens, DO, PhD:

They don't have to think about it and they go, "Well, how did they ever do that before with a regular computer as it goes through?" Final takeaways for our audience, anything specific or things that we haven't covered that you feel are important to discuss?

Lara Jehi, MD:

I mean, the most important message from me is that I want to encourage people to be adventurous, to be risk takers, to not worry about making mistakes and failing. Research is just an exercise and failure. We have many ideas that we try and we have to try many times before we get it right. The future is very bright. All of these technologies can really change the way we do things. It's an invitation for us to go deep dive into all of this discovery exercise and see where it takes us.

Glen Stevens, DO, PhD:

Yeah, I like that because I hear a lot of people talk about AI and they sound a bit scared of it, and I think we're just scared of things that we don't know about. I think it behooves us all to educate ourselves and figure out how it's going to work within our life because it's going to happen as we move forward. Well, listen, Lara, thanks for joining me for this discussion. This is incredibly fascinating. Maybe I'm going to have to look into taking one of these educational courses, but I'm really looking forward to see what happens in the future and what's next. I'm just so excited for you, and I've loved watching you grow in your career, and I'm so proud of you.

Lara Jehi, MD:

Well, thank you so much, Glen. It means the world to me. Thank you.

Conclusion: This concludes this episode of Neuro Pathways. You can find additional podcast episodes on our website, clevelandclinic.org/neuropodcast, or subscribe to the podcast on iTunes, Google Play, Spotify, or wherever you get your podcasts. And don't forget, you can access real-time updates from experts in Cleveland Clinic's Neurological Institute on our Consult QD website. That's consultqd.clevelandclinic.org/neuro, or follow us on Twitter @CleClinicMD, all one word. And thank you for listening.