Digital twins for health care and research

Bhavya Krishnan and Dr Elizabeth Redrup Hill discuss the applications of digital twins in the healthcare system, why there is growing interest, and some questions that they would like to see be explored in order to make digital twins an effective tool for healthcare and medicine.

Welcome back to Making science work for health, the PHG Foundation podcast that explains the most promising developments in science and their implications for healthcare.
 
In each episode, host Ofori Canacoo discusses with a PHG Foundation policy analyst, the underpinning science, the ambitions for improving population health and the impact it could have on patients, on society and on the people delivering your healthcare.
 
If you would like to find out more about what was discussed in this episode, you can find additional information on our website, phgfoundation.org.

Regarding digital twins, you can read our explainer about this engineering concept.

If you have any questions about the topic then you can email us at intelligence@phgfoundation.org. 

Transcript

Ofori: Welcome to 'Making science work for health', the PHG Foundation's podcast exploring developments in genomics and related emerging health technologies. The progress being made by teams of scientists and researchers around the world is gaining more interest and attention. Many of the latest advances feature genomics and omics related technologies.

The field in which the PHG Foundation has more than 25 years of experience, helping policymakers get to grips with practical, on the ground delivery. 'Making science work for health' aims to look behind the hype and explain what new science means for patients, health professionals, and members of society.

My name is Ofori Canacoo, part of the communications team at the PHG Foundation, and host of 'Making science work for health'. For this episode, we're talking about digital twins. A not so new concept in other industries, but certainly novel in the realms of healthcare. This is a virtual model of a particular process or system that could, amongst other things, help with certain forms of decision making.

Bhavya Krishnan, Policy Analyst in Biomedical Science, and Dr. Elizabeth Redrup Hill, Senior Policy Analyst in Law and Regulation, join us to discuss the applications of digital twins in the healthcare system, why there is growing interest, and some questions that they are keen to be explored in order to make digital twins an effective tool for healthcare and medicine.

Hello to you both. 

Bhavya: Hi.

Elizabeth: Hi. 

Ofori: Thank you very much for joining me today on today's podcast. , first of all, would you like to tell us just a little bit about yourselves? Elizabeth, would you like to go first? 

Elizabeth: Hi, so I'm Elizabeth. I'm a Senior Policy Analyst in Law and Regulation at the PHG Foundation. My areas of interest include all things digital health related, such as, medical devices and AI, data protection, computer modeling and simulation, and large scale data spaces for health.

Ofori: And Bhavya? 

Bhavya: Hi, I'm Bhavya, a Policy Analyst in Biomedical Science. My areas of interest include application of omics technologies, global health systems, AI and digital innovation in medical practice, and addressing health inequalities. 

Ofori: Great, thank you very much. So, we are talking about digital twins today.

So I think my first question for you both is going to be, what are digital twins? 

Elizabeth: So this is a great question. It's certainly a concept that's been around in other industries for a long time, but it's quite new to the health space. And essentially, a digital twin is a virtual representation of a biological system or process.

So it could be a single cell within the body, or it could be a whole organ. And possibly in the future, we might even be able to use these to create a digital twin of a whole human body. I think what's important is to remember that the aim of creating these virtual representations isn't to map every single cell and to match that to a patient's particular composition.

It's about just getting down the main, a map of those systems and how they interrelate and to work from there. 

Bhavya: So for example, within healthcare, there are examples where teams have used computer modeling and simulation to replicate the heart function of heart murmur patients to see how they can better treat these patients.

Or there have been the use of twins within the pharma industry to kind of replicate the whole drug discovery process to predict drug reactions within patients and outside the system to understand how the mechanisms work. And there was a particularly useful example where they tried using digital twins of patients who had previously had C-sections to kind of see how the C-section scars would impact future pregnancies.

But twins have also been used, sorry, digital twins have also been used to replicate hospital departments to improve health system efficiency and the operations within particular departments or an entire hospital. For example, orthopedics or radiology, where they can track real time how the resources are being utilized and kind of understand the maintenance of the equipment that's being used.

Elizabeth: Yeah, I'd actually like to just go back to that part about the C-section scars, because Bhavya and I were at a conference in November 2023 at the Royal Society of Medicine, where we saw some fantastic research, and it's fantastic for many reasons, obviously one of them being it was about female health, which is really under, under researched, as we all know, but it was really incredible listening to the science behind it, so they were mapping the geometry of C-section scars and how, depending on how the scar, or how the...

Bhavya: Surgery was done. 

Elizabeth: Yeah, how the surgery was done, like if it was done vertically or horizontally, and then even just, you know, how the tissue separates differently. You'll have a slightly different geometry, and depending on that, if you then had a subsequent pregnancy, you could be at risk of rupture more than say someone else's scar just because of that unique geometry and it's just really fascinating.

I just thought it was something worth highlighting. 

Bhavya: Yeah, and especially this kind of research kind of shows why digital twins in health are being discussed with so much interest currently, because especially for maternal health, it's such an under researched area within healthcare, but a really important one.

And like the C-section example it's to predict pre-eclampsia, which is a condition that affects many women and a lot of women die at childbirth due to pre-eclampsia. And we don't really understand a lot of it because it's so difficult to conduct trials in pregnant women. This particular research which showed that just by mapping out how the scars would predict the outcomes, it's really useful and kind of gives an example of how this technology could be used.

Elizabeth: It's pushing new boundaries and it's enabling us to do things that we couldn't do before. You know, there's obvious ethical and legal reasons why you wouldn't want to test certain drugs or devices on pregnant women or, children, et cetera. These kind of protected groups and this technology enables us to do that so we can still serve those communities and those groups.

It's a really exciting area of research to be looking at. 

Ofori: So do either of you have any other real world examples of places where digital twins are being used? 

Bhavya: Mm hmm. So there are several examples where they are being used already. Most of them are pilots to see how well they work. For example, the Living Heart Project is an initiative that uses digital twin technology to kind of replicate the functioning of a human heart. And this tool is intended to help developers understand how new innovative treatments for like heart conditions, like cardiovascular conditions can be developed. Or this has also helped surgeons kind of practice really risky operations.

And practice and try different minimally invasive techniques to see which one works best to kind of minimize the risk of doing it on an actual patient so they can practice multiple times before they actually perform the surgery. In terms of hospital administration, there have been a couple of examples at Guy's and St. Thomas and the Leeds hospital, the Leeds teaching hospital trust, where they've used digital twins to kind of map out patient flow and resources being used within the hospital to make... to optimize efficiency and kind of plan rotas and plan the maintenance of medical imaging equipment. But these are just examples of how twins are being... how digital twins are being used in healthcare delivery.

However, they have also been used in the research context, which Liz might be able to help you with. 

Elizabeth: So in the research context, they're essentially a form of computer modelled and simulated trial, which sometimes are known as in silico trials. But what the real benefit and value that they therefore offer is that these are, you know, these kinds of forms of digital twins are particularly valuable in patient cohorts that for possibly ethical or legal reasons, you can't test in them, or perhaps the cohort is just so small that we don't have enough data to, for regulatory approvals of say like a drug or a medical device.

So rare patients, pregnant women, children, all of these underrepresented populations in healthcare can now be serviced in a way that they, they haven't been before. I also think from, like, theoretically speaking, these trials are... well, they have a real benefit in the fact that they can be undertaken an endless amount of times, and you can technically run a whole new trial by just tweaking bits of code.

Obviously, I say theoretically, because from a regulatory and ethical perspective, depending on how much you changed it from the original spec, you might need to then get new ethics approval, for example, because it's changed too much and, you know. But, the value that they're bringing, therefore, is that they are perhaps less resource intensive than you say, running a clinical trial and then you realize that you want to tweak one kind of factor to see how that might impact something else. And then you have to do a whole new ethics approval, a whole new trial, get more participants in again. So they have some real value.

There's another form of digital twin, which I think is really important to talk about, particularly in the health space, which is digital twins that are designed to feedback in information into a system to improve its functioning. So these could be an important component of like a learning health system, for example, which is where the boundaries of care and research are going to be blurred for their mutual benefit.

 A real world example of that might be a virtual ward. So the digital twin is a twin mirroring the real ward and it's monitoring patient intake and outgoing patients to see, for example, maybe how many beds are free. And based on that, it can then feed back information to say, "Hey, we now have five new beds available on X ward".

So it can really help the learning, the health system best utilize its resources, which is so critical at the moment, given the policy outputs that are coming out about how under resourced the NHS is and how it's struggling particularly with things like bed spaces. 

There is just one kind of caveat I would say. Which is that whilst it's important to note that digital twins offer this great... all these great benefits and values that we've been discussing, their purpose isn't to entirely replace human testing. So the point is to refine and reduce human testing as much as possible. But I'd say the consensus, generally, is that they won't entirely replace human testing because there's certain... there's certain factors, there's certain things you just... you just can't mimic and you can't account for.

Bhavya: Yeah. So it's, like Liz said, it's... digital twins are not meant to be replacement, but assistance for clinical trials and healthcare research. And the simple reason is that the human body is too complex and cannot be replicated in every single aspect. And accurately replicating it requires extensive validation, which no matter how hard we try is really hard to do because even in a human body, we don't understand how everything works yet. And there's always going to be unknown factors that we can't account for, like how the environment influences certain drug reactions or certain disease development. And when we don't understand this in a biological system, it's very hard to replicate this into a twin, which would be able to predict it accurately.

So yeah, like Liz explained, this was, digital twins are just meant to be assistants for making human trials or making animal trials more efficent and reducing how invasive these trials are but may never be a replacement because there will be factors that you need to understand in a biological system. 

Ofori: What are the future facing considerations for digital twins?

Bhavya: So a major limitation for their accuracy is the capacity of current computing models and their ability to handle large amounts of data. So creating accurate twin models would require processing large data sets. And it requires immense supercomputing power. And there are limits to how many variables the supercomputers currently can handle.

Increasing the number of variables doesn't necessarily mean you increase the accuracy of a model. There comes a certain threshold, usually around like 80 to 90%, after which the model cannot accurately process the data that's being inputted. So there are limits to how much you can feed into the model to get details.

And the amount of data required varies depending on how complex the biological system you're trying to model actually is. Like if it's a single cell or if it's just one organ, the amount of data required is much lesser than say, trying to map out an entire human body. So in the future, as supercomputers kind of improve in their ability to process data, there might come a time where they are able to create digital twins, which resembles a biological system more closely and give us more accurate insights. But at present, that accuracy is quite limited. And sustainability is also a key concern because supercomputers and large server farms that are required to process the amount of data that goes into them, occupy large swathes of land and maintaining these farms and like the amount of resources and energy that they consume is extremely high.

So sourcing them over a long period of time is definitely going to be a concern. 

Elizabeth: I think another major area is the... that there needs to be... well, there's calls for greater certainty within the regulatory space. So, for example, I was earlier talking about in silico trials, and a major uncertainty that the industry face is they are uncertain about whether... also researchers... they don't know whether regulators will accept the evidence that these trials produce, or these digital twins produce, as being as good as real world evidence, if not better.

Also, there's uncertainty about how medical practitioners or the general public may view that evidence and whether they might, whether they accept it or not. We know that there is some cross government work underway to hash out what uncertainty remains among these groups and also how regulators will view this evidence and we also know that there's initiatives to help develop much needed guidance on good simulation practices and principles for model assurance, but further clarity is still needed. And another area where regulatory certainty has recently been... regulatory uncertainty has recently been flagged is in the... off the back of the MHRA's Point of Care Regulatory Framework, which was under consultation in early 2023, or at least the results of which came out in early '23.

And this framework could be relevant to digital twins because, certain digital twins that have a feedback loop in their system with the aim of informing better care decisions could be brought under it. For example, a digital twin that monitors for, say, the production of CAR T-cells and predicts the optimum time to harvest them could be relevant to this framework.

The benefit of such technology is that the patients can receive earlier treatment by not relying on the standardized harvest times because we know that people's cells reproduce at different rates. at different speeds. Because of that, by sticking to this kind of standardized time, we could actually be preventing patients whose cells replicate faster to get their treatment faster, and as such, not actually avoiding harms that we could.

So this digital twin, by being able to predict when the... the optimum time of harvest is around, could kind of avoid those harms. But, as I said this kind of future facing need is really based on regulatory expectations and clarity around that. So, what do the regulators accept as good as evidence?

And, I think that's very much needed for the industry and research and development to have a confidence to continue to develop these technologies. And not only that, for us to actually realize the full potential that digital twins can offer healthcare and research, both in the near and future term. 

Ofori: What question on digital twins would you like to see answered the most?

Elizabeth: Well, for me, it's the regulatory uncertainty. It'd be really great to see that clarified as soon as possible. So, you know, are regulators going to accept in silico trial evidence and digital... digital twins that are a form of in silico trial. Will they accept that evidence as, as good as, if not better than the current real world evidence?

How accepting are they going to be of digital twins that have almost this AI component in the sense that they're in a feedback loop and constantly learning and okay it might be easier to do that in say putting that into a virtual ward kind of situation as opposed to trying a drug or changing someone's insulin levels from back home.

Yeah, I think that uncertainty needs to be answered and the sooner the better if we can. Because of the benefits that these technologies really offer patients research and the healthcare system in the UK in general. 

Bhavya: So in line with my interest, I really want to know how, like, they intend to use genetic and genomic data within developing digital twins and see how that plays at all because there's a lot of talk around it, but not a lot of in depth information. So I'd really like to know more about that and also from like the adoption point of view.

I want to know what the public clinicians and like the health administration feels about digital twin, do they actually understand what it is and what benefit it can have to their everyday life and also see what they think, like the impact of it. It could be whether it'll make their life harder, whether it'll be useful, where they think it'll be useful.

And yeah, just understand how it fits into the existing healthcare system workflows that are in place because they're not going to change the entire workflow just to incorporate digital twins, but it also shouldn't be that they just kind of pigeonhole it into a system which doesn't work. 

Ofori: Following on from that then, do you have any expectations on what the answer to your question would be?

Elizabeth: It depends. It depends on the digital twin itself and the context and the kind of drug or device that they're trying to get approved. But generally speaking, I think they are very open to these discussions and we should be able to clarify them quite quickly. I mean, look at the work that they're doing on the point of care regulatory framework.

Yeah, so fingers crossed. I think quite soon we will have an answer that is quite positive but measured. 

Ofori: And Bhavya? 

Bhavya: I feel like most people are open to having discussions about this topic, simply because no one understands it completely yet. And it is something that's quite new within healthcare.

There's a lot of work surrounding it, so people are interested in understanding what it actually is before they make a judgment on it. Like Liz said, people are... practitioners and clinicians are quite cautious about it. They aren't entirely sure what the actual benefit is and whether it might just increase their work with very little benefit or whether it has a huge benefit with like very little work.

So yeah, I guess it depends on whether there's a successful pilot, large enough pilot, say a national level one, which can be used as an example to kind of say, look, this is how it could help you. And I feel like if that comes across a lot of people would be more open to it. But yeah, at the same time, there is a lot, a lot of risk involved.

And that kind of makes everyone quite hesitant about it. So I guess it depends on the twin and how much the technology advances and how the regulators respond to kind of convince the general public. 

Ofori: Okay, well I think that's a great place to end this episode. Bhavya, Elizabeth, thank you very much for for joining us today and we hope to have you on again soon.

Elizabeth: Thank you. 

Bhavya: Thank you.

Ofori: And that brings us to the end of the episode. If you liked it, please leave us a rating and review, and make sure to subscribe. If you would like to find out more about what was discussed in this episode, there are useful links included in the podcast description. You can also find additional information on our website, phgfoundation.org. And if you have any further questions about the topic, then you can email us at Intelligence@phgfoundation.org. Thank you for listening and we look forward to bringing you a new topic in the next episode.