The Future of Digital Twin Hearts

DR. JAZMÍN AGUADO-SIERRA

When I first did the simulation, you could see a very good rotation of the heart during contraction. Where you could see the actual blood going out and expelled in a very efficient manner.

ROMA AGRAWAL

The idea that we might be able to make digital copies of our organs may feel impossible. Like something out of a science fiction novel. But for Dr. Jazmin Aguado-Sierra, who is responsible for the world's first heart simulation, this kind of radical innovation is part of her extensive research as a bioengineer at the Barcelona Supercomputing Center.

DR. JAZMÍN AGUADO-SIERRA

The ethical question that you’re saying, like, who is going to have their own virtual heart.

ROMA AGRAWAL

Unveiled last year in the Engineers Gallery at The Science Museum in London. She used scans of her own heart to capture the complex interactions of blood flow, muscle contraction and electrical impulses. All through the power of a supercomputer. It's a vital opportunity to understand this very complex organ, from testing treatments before giving them to patients, to the development of digital twins. Is this transformative innovation the future of predictive medicine? Joining Jazmin and I to unpack these questions is Roger Highfield, Science Director at the Science Museum Group, fellow of the Academy of Medical Sciences, and the co-author of ‘Virtual You’.

ROGER HIGHFIELD

Why stop at one digital twin of yourself, when you can actually have armies of digital twins of yourself.

ROMA AGRAWAL

I'm Roma Agrawal, and you're listening to Create The Future from The Queen Elizabeth Prize for Engineering. So thank you so much both for joining me. I've got Roger Highfield and Jazmin Aguado-Sierra with me and we're going to be talking about what I think is one of the most fascinating organs in the human body. So actually in my book, I have a chapter on the pump and I start off with this idea that the heart is this incredible pump, that engineers are really struggling to recreate. And if you try and think about making a pump that doesn't damage the blood that can change its capacity depending on if you're running or sitting or when you're stressed, it's almost impossible to recreate. So, Jazmin, can you tell me how you're approaching this idea of recreating the heart? 

DR. JAZMÍN AGUADO-SIERRA

So how we're trying to do it, it's trying to simplify a little bit on the function in the terms that we do not need to go and model every single cell. But we use an idea that the tissue is a continuum, and then we apply different mathematical models to try to solve the problem. Because when you go deeper and deeper, the problem becomes even more complex. In a way we try to simplify it to the bare minimal physical terms. In terms of mathematics, physics, to try to describe all of it from the cardiac mechanics up to the blood flow and all of the interactions in-between. The more sophisticated you go, the more computing time you require. So, one of the very interesting things right now is, how deep do we need to go to try to simulate the heart in a good way? It's a really complex thing to answer because it really depends on what question you're trying to ask to your model. But, you know, we're trying to find this balance in which we can add as much complexity as possible, but still make it usable enough so that we can obtain interesting answers from it.

ROMA AGRAWAL

It reminds me a little bit of like, if I'm trying to design a steel column as a structural engineer, I don't really need to understand how each atom is arranged and what the crystals exactly look like, in order to understand the wider functioning of it. So it's basically trying to find that particular level of detail, to create in this virtual environment. But Roger, if I could ask you maybe to define what is a virtual organ?

ROGER HIGHFIELD

I mean, if you look at Jazmin's heart, which actually I'm pleased to say, is beating in The Science Museum. Not a life sized version, a person sized version of a heart is beating away. It looks like, a kind of Hollywood recreation of a beating heart. But in fact, this is, a virtual model is a mathematical model brought to life, in Jazmin's case by the Barcelona Supercomputing Center. And actually, it doesn't just look like her heart, it actually behaves like it as well, which is really important. People talk about digital twins as well. This is a phrase introduced by NASA years ago and actually it’s very common in engineering, where you basically do simulations in a computer, whether it's the way that winds go over an aerodynamic surface, like a car or a plane, or you can do a digital twin of a factory and so on. But actually, the precedent of digital twins goes back a long way. In fact, NASA likes to say that if you look at the, really one of the worst space disasters in history, although it had a happy ending I hasten to add, Apollo 13. The spacecraft blew up most of the way to the moon. They lost a lot of power, and they were in a desperate state where the astronauts, it was a battle to get them back safely. And of course, there’s that wonderful movie, the Ron Howard movie, that shows that they made it. But one of the secrets of NASA's success was, they used simulators on the ground to try out strategies to save power and save the spacecraft. So, they had a choice, they could have had the astronauts playing around with the spacecraft in real time and risk their lives. Or they could do experiments on the simulations on ground and that's the whole philosophy embedded in engineering. If you'd asked a biologist ten years ago, “can we do this in biology?” They would have just kind of looked very sceptical and said, “Biology is way too complicated to make a digital twin of a heart”. But in fact they're wildly out of date, you know, the creating digital hearts goes back to work by Denis Noble in the sixties in the University of Oxford and what Jazmin has done, using an off the shelf model, called Alya Red, which is so famous it's even in a Dan Brown novel. She's customised it with data about her own heart, to behave like her own heart. Well true Digital twins should talk to her heart, so really strictly, it's a simulation of a heart. So the twin gets data from the heart and you know, there's a bit of a dialogue between them in real terms, but these are really useful, already being used and accepted in drug trials. You know, why do an experiment on an animal that isn't really like you? Why do an experiment on you? Use your digital twin instead. So, Jazmin, I hope you've been doing lots of experiments on your heart.

DR. JAZMÍN AGUADO-SIERRA

I mean, I have extremely interesting projects coming on. I am currently pregnant and I have been doing this MRI imaging of my cardiovascular system, of my heart. And the aim now is to try to extend those simulations and look at all of the cardiovascular changes happening during pregnancy.

ROMA AGRAWAL

Can we take it a couple of steps back and maybe just talk about the purpose of the virtual twin or the virtual organ? So, Jazmin, if you can maybe start by telling us what is the purpose of you recreating your heart?

DR. JAZMÍN AGUADO-SIERRA

One of the main purposes is we're trying to figure out how to use some of our tools towards personalised medicine. This is potentially a big, big step and this is something that we need to figure out what data is important. How can we use these models to potentially predict future events, future disease, or potentially to use them in conjunction with other models to recreate the behaviour of a population. So, the main aim is to try to, at this point, is to try to understand how can I use my model to make it, you know, more predictive in the future? What is the data that are required to personalise it in the best manner? And to make it useful?

ROGER HIGHFIELD

It really helps to look at conventional medicine. And conventional medicine is a bit like driving a car down the road, while looking in the rear-view mirror the whole time. You know, modern medicine is guided by trials done in the past on people who are genetically a bit like you, who might be in slightly similar circumstances, but might not. You come up with average, one size fits all medicine and the whole point of digital twin and virtual organ research is to make medicine truly personalised and truly predictive, because it's neither at the moment. You know, we've all had an experience with our doctor where you say there's something wrong with me and they say, “Hey look, try this drug and if that doesn't work, I'll give you a slightly different one”. So they’re kind of experimenting on you. Wouldn't it be great if instead they can take a digital twin of you and say, “Actually, Roger, this is the drug that you should be taking” and cut out all that mucking around. You know, that's the dream of this research.

ROMA AGRAWAL

Okay. So we understand the purpose of why we're trying to do this. But if we kind of strip back what you're doing, you're putting a huge amount of data into a computer, essentially. To then, run that data and learn from it and I watched a video, Jazmin, of this huge supercomputer in the facility. Can you describe that to me a little bit? The size, the scale, the terabytes, the you know, what's going on with the supercomputer?

DR. JAZMÍN AGUADO-SIERRA

Definitely supercomputers are our main tool to create some of these models because, as we mentioned before, they can be as complex and you know simplified, to have a full powered simplified, you would still need a supercomputer. But in any case, the magnitude of the problem becomes very, very big, very, very quickly. To be able to capture all of the physics within the model in an accurate way. You need to force time or look at different time steps, and we need to cut that into small time steps, where all of the physics are moving and interconnecting and interplaying. So we were looking at how much of the human time would it require for us to do that. If we, you know, sat with almost pen, paper and start doing those, all of those equations. And it's really impossible to calculate, we were counting billions of years.

ROGER HIGHFIELD

I've got the figure, 57 billion years you figured it out to be.

DR. JAZMÍN AGUADO-SIERRA

Yes. It was, it was a challenge to try to calculate it that way and it would come to that. So it is really, really necessary to use our supercomputers because we can definitely create terabytes of data.

ROGER HIGHFIELD

Just on the supercomputer front, I mean, Jazmin uses MareNostrum, which is probably the most beautiful data centre on the planet. It's in a deconsecrated church in Barcelona. But in fact, the biggest machines on the planet are called exascale machines. Which is like a million, million, million floating point operations a second. In fact, the co-author of my book Virtual You, Peter Coveney, he's using the most powerful supercomputer on the planet, called Frontier in the States. And these machines, you know, they are kind of hitting the limits of what you can do. They’re using the amount of energy equivalent to a town, like 20 to 60 megawatts. In the case of Aurora, which is another exascale machine. I talked to Rick Stevens, who's putting it together. They've got cooling pipes which are three feet across, pumping tons of water through the hot heart of the machine, and Peter's using it. He's actually got data on a South Korean lady who gave her body to medicine. Which has been digitised and Peter simulating the flow of blood through her whole circulatory system, which is many, many kilometres, right the way down to the microscopic level. And of course, the next step would be to couple virtual circulation to something like Jazmin's heart. You might think, well, crikey, this is too rarefied. We can't use digital twins in real life. But actually, Jazmin’s kind of hinted at the answer to that. It's a bit like imagine navigating across London. You know, I could do it by giving you like, a high risk satellite map of London. Or I could go another step and say, hey, I'm gonna to do a LIDAR scan of London. So you've got a whole laser scan which gives you elevations and positions of buildings and everything. Or I could give you a tube map and if the aim of the exercise is to get around London, the tube map is really good and really useful. So, the hope is that, you know, we're talking about the really bleeding edge digital twin simulations. But in a doctor surgery, you’d use a stripped down version, a bit like the tube map, which captures just enough detail of what you're trying to do to give you a bit of predictive reality.

ROMA AGRAWAL

Give me a couple of examples of how this is being applied today.

ROGER HIGHFIELD

Well we’ve already mentioned, you know, digital twins of hearts. There’s also a lot of digital twin and virtual simulation research, in terms of drug trials. Research by Blanca Rodriguez in Oxford has shown that you get better results than if you do animal testing, using a human digital twin. At the cellular level, there’s been wonderful simulations of virtual cells by Markus Covert in Stanford University. My co-author’s doing virtual circulation, so you can track the way the blood clots move around the body and differences in blood flow. We’ve had virtual breaths, so we can actually see if you breathe in an inhaler, and depending on your lung anatomy, where the drug gets deposited in the lungs. There’s a team under Peter Hunter, in New Zealand, that’s doing all sorts of organ systems. You know, from the microbiome to other things. You could even think of the computer models used to predict the pandemic as, kind of variants of digital twins, where you’re looking at whole population behaviour. So, there’s actually huge amount of research right the way from the molecular up to the whole body level. And even very tangible things, like you can simulate a sneeze and see how far particles bearing Covid virus can travel. So, you name it, someone’s trying to simulate it in a computer.

ROMA AGRAWAL

So one of the, the big questions I have is, you know, this all sounds so exciting and brilliant, but I'm interested now in the ethical and social side of it. Like who are the people that this technology will actually benefit? Is it just gonna be rich people who can pay to get themselves replicated? Or, you know, what, what is that going to look like? What are some of the questions that scientists and engineers like yourself are asking about the ethics of this field?

DR. JAZMÍN AGUADO-SIERRA

There's several, several things about this. One thing is always access to clinical data, because we want to try to replicate the function of a human. And to do that many times we require to access some of that clinical data. Now these get sorted, you know, proposing your clinical protocol to a hospital and from there, there's this ethical committee that can allow you to, you know, get that data from the patients. Now, there is also the ethical question that you're saying, like, who is going to have their own virtual heart? I think this is something that is coming more towards the future. The one thing we're trying to focus now, is to represent humans as closely as possible and with as much diversity as there is as possible. So, for example, women are a minimal percentage of subjects within clinical trials, particularly cardiac safety, drug testing. And this, of course, has many reasons, one is just they need to be tested in normal humans, and males tend to have a lot less effect from these drugs. If you want to try and get your drug out in the market, well, you need to prove that is safe. But generally the populations are very, very reduced. Currently, what we're doing is trying to extend that into not only normal populations because having nausea is not just a woman thing or a pregnant person thing. It can be any elder person, any person with other comorbidities. And what we're trying to do is actually to include that diversity. So in the end, what we're trying to do is actually to better the current practices, for everyone and this is a very, very important part from where we can use some of these digital twins to actually better all of this.

ROMA AGRAWAL

Roger, talk me through some of the ethical challenges that you're dealing with or that people should be considering with this technology.

ROGER HIGHFIELD

We've already thought about a lot of potential ethical issues. So let's say, for example, a close relative has got Huntington's Chorea, which is a terrible disease. Do you want to know whether you're carrying that gene or not? It's so terrible, the disease that a lot of people say, I just don't want to know. So that's one issue. Let's say you're making thousands of digital twins, experimenting with different circumstances, and if they keep on giving you bad news, do you want to know that news? Well, you know, I would rather have the choice than not have the choice at all. There are issues about data and I think that we're in, again, a similar position we are with, say, the banking sector. We know that there are risks of abuse in the banking sector, but we also know that there's huge convenience. And I think that, what people will be willing to, have a digital twin of themselves, if they're convinced, and I'm certainly convinced that there'd be medical benefits for them. But it might have some curious novel things to think about. So, for example, what do we mean by healthy? So, you might think that I'm looking gloriously healthy from what you can see of me at the moment, but actually my digital twin could be saying that, you know, hey Roger, that three bottle of claret, an evening habit is gonna take ten years of your life. So what does it mean to be healthy? If you look healthy, but your twin is saying that you're knocking years of your lifestyle. And of course, humans get very anthropomorphic about things. And I could see that if someone has a very elaborate digital twin of themselves and they do die, would the family feel that they had to keep the twin going in some sense. There might be some novel issues to come up there.

ROMA AGRAWAL

One of the things that strikes me is this idea of the California tech bro trying to bio hack and increase their lifespans as much as possible. Will having a digital twin be the luxury of the rich?

ROGER HIGHFIELD

There's always a risk with technology, particularly cutting edge versions of digital twins, there’s no doubt you’d need enormous resources to do it at the moment. But I do think that in years to come we might see simplified digital twins that are much more accessible to doctors that can help deal with very real problems. I mean, just to give you one example, if you've got a confused parent who's a hundred and five or something, and they're on ten different drugs, and they've got high blood pressure and low this and all the rest of it, you know, you get elderly patients with lots of different morbidities. There's lots going on. They might be pre-diabetic as well. Trying to figure out whether it's the side effects of drugs or two drugs interacting or something to do with the blood pressure and the diabetes is very hard at the moment. And my hope is that digital twin models will give us a much better sense of what's going on in these kind of cases and not too distant future. You know, they're already being used for high end applications, you know, like developing heart medicines and heart implants. But I do see the technology getting, like any technology is expensive to start off with, but once it becomes useful and you get those simplified models, you know a bit like the tube map that I mentioned earlier, I think they will become democratised.

ROMA AGRAWAL

And so on that note, Jazmin, could you tell me a little bit about what the future of personalised medicine could look like? And let's, let's be as speculative and go as far forward in future as you would like.

DR. JAZMÍN AGUADO-SIERRA

Well, what we're aiming for and this is, of course, the future, but I see it more closely. Is that there are certain parts or certain organs or certain functions where we will be able to produce a personalised version of them to try to provide you with the best therapy possible for yourself. And this is not only regarding the heart. This is also regarding, you know, many other different systems. We are working on some women's health, for example, modelling the uterus, modelling the overall function. And we are trying actually to provide and simplify some of these models to make them work for you and for your own therapy. So I do see this working more effectively for specific purposes where we haven't really focused on in the past. This would be a big breakthrough coming within the next few years.

ROGER HIGHFIELD

Well, I suppose looking into my crystal ball, why stop at one digital twin of yourself when you can actually have armies of digital twins of yourself and you can be doing all sorts of things to them, you can put them on vegan regimes, you can up their exercise, you can down their exercise, you can take that micro dose of aspirin each day. You can let yourself repent and have a bottle of claret every night, and you can actually explore your possible futures with a digital twin and figure out, you know, it's not really, I know Californian billionaires have got a fixation about living longer, but I’d hope that in the short term, these armies of digital twins will tell you how to live better for longer, if you see what I mean. Which is much more important and they'll do it for you, rather than the current status quo, which is really one size fits all medicine that you hope will apply to you. So I’d hope that it would go beyond, you know, straightforward treating disease to actually preventing disease and looking at lifestyle choices and then being even more speculative. There's always lots of talk about biohacking and could an implant make you live longer or better. Could you augment your memory with an implant and so on and so forth. Well, again, I'd rather not have a surgeon experiment on me with that, but you could actually use your digital twin to see whether that would really be worth taking all the risks of that sort of enormous intervention. So, I think there's lots of exciting possibilities there. But to be honest, my modest ambition for digital twins is just to make medicine truly personalised and truly predictive, which is not where it is at the moment.

ROMA AGRAWAL

You've been listening to Create The Future, a podcast from The Queen Elizabeth Prize for Engineering and Peanut and Crumb. This episode was presented by me, Roma Agrawal, and produced by Tess Davidson. If you enjoyed this and you happen to be in London, make sure to check out Jazmin's digital heart at The Science Museum. You can find it in the Engineers gallery and you might even run into my face. To find out more, follow QEPrize on Twitter, Instagram, and Facebook. Thank you for listening and see you next time.