In pursuit of our virtual twin: can maths recreate humans?

 9 mins | By Atreyi Chakrabarty
 | Biotech | AI | Sep 29th 2020

You and I have a body in space and time, which interacts with its environment and other bodies within it. The body itself is made up of interacting organ systems, made of different types of cells, various proteins and a single set of DNA. Now, imagine your body with its exact organ systems, cells and DNA, existing not in space and time, but in 1’s and 0’s, on a silicon chip. This body, defined completely by a set of mathematical equations, is what you would call a virtual twin, and research under way in Oxford could indicate how far away we are from creating such a digital entity.

go deeper

For decades scientists have dreamed about defining biological phenomena quantitatively, to capture the essence of what gives rise to life in equations. In fact, physicists had begun to grasp the complexities of the universe centuries ago with mathematical axioms and laws. Biological systems must satisfy the known laws of physics too. So, by applying the laws of physics, integrating statistical randomness and making reasonable assumptions, one might start to build models of biological systems, such as the human body.

This notion was launched in the form of the Physiome project, founded by Peter Hunter, director of the Auckland Bioengineering Institute, with the mission to find our mathematical twin. Yet this quest dates back to 1960, when the first mathematical model of the heart beat was created by Denis Noble, then a humble doctoral student, now Emeritus Professor of Cardiovascular Physiology in Oxford and author of the book, Music of Life. Just a few years before Noble’s model, Alan Hodgkin and Andrew Huxley had found the equations to define the electrical activity in a nerve cell.

“I had to read Hodgkin and Huxley’s papers as an undergraduate, and I found them exceedingly difficult,” Noble confesses with a modest smile, “but I was also highly impressed – could you really mathematise physiology?”

The power of a mathematical description of our body lies in precision and prediction. Just imagine being able to solve the set of equations to find out how your body might respond to certain drugs, surgeries, or changes to your lifestyle. Such a system would be the Swiss Army knife of personalised medicine.

Denis Noble, Emeritus Professor of Cardiovascular Physiology, Oxford. Pic supplied

Marco Viceconti, Professor of Computational Biomechanics in the University of Bologna, envisions a time when medicine will no longer rely on a devide et impera (divide and conquer) reductionist paradigm, which sees each constituent part of the body as separate. The four biggest killers in Europe are chronic conditions: cardiovascular diseases, cancer, diabetes and respiratory diseases. They have nothing in common, except that they are multidimensional diseases. Covid-19 is turning out to be a multi-faceted beast too, affecting the respiratory, cardiovascular and nervous systems, uniquely in each individual. By harnessing individual data about the physiological make-up of patients’ bodies and predictive simulations, a virtual twin could speed up treatment processes and recovery times. In a 2012 TED Talk, Viceconti set out his ambition for a ‘Virtual Physiological Human’ which would allow us to look at all these complex interactions in order to plan therapeutic interventions, proclaiming, “We cannot ditch complexity anymore. We have to face it.”


Video:The vision of VPH. Marco Viceconti was one of the founding members of the VPH initiative, for which Blanca Rodriguez has been involved in designing cardiac models.”

By definition, a model is a simplified representation of reality, the knowledge of which stems from experimental experience. It encapsulates the essence of a process, but may not be robust to changes in parameters. Our body, on the other hand, is incredibly resilient due to its built-in complexity and ability to compensate for loss of function. The first model of the heart beat was made of only four components. “It was so simple, that if your heart was behaving like my model back in 1960, you wouldn’t be here talking to me at all! That model in 1960 worked quite nicely. But it was extremely fragile,” Noble explains, “so this is why the complexity is there – it’s beautiful – nature has known how to build in resilience, great insensitivity actually, even to the genetics of what is happening.”

Do we, then, need to rebuild the human body in its entirety, all the way up from atoms and molecules to organs, to fully reproduce its complex processes? “Exactly so, which means it’s impossible,” says Noble, laughing at the scale of the challenge presented. Maybe we don’t need an exact atom-to-atom replica to build a virtual twin, we just need a reliable simulation of the processes that occur in our body. He believes following in the footsteps of physicists will lead to success:

“If they want to investigate the properties of a steam engine, they don’t bother about all the atomic elements in the system, they bother about the Laws of Thermodynamics … how gases and fluids and so on are constrained by the system they exist in. And that is exactly what we have to do in reconstructing the mathematical behaviour of the body as a whole.”

The best tool we have in our arsenal to deal with complexity is the computer. The more elements we can account for within a model, and the more real-life data we have, the more refined it will become. Now, with improving computational processing power, and machine learning to make sense of big data, our models are able to become more and more refined. Just like weather prediction systems have become more accurate over the past decades, so can physiological prediction systems.

Already, a computational cardiac model developed by Oxford computer scientist, Blanca Rodriguez, and her team is able to accurately predict drug outcomes on patients’ hearts almost 90% of the time. This is in stark contrast to the 75% accuracy rate using rabbit hearts for drug trials. Rodriguez is paving the way for in silico drug trials to become the new norm in medicine. “In engineering or other areas such as aeronautics or development of cars for example, it’s very normal to use modelling and simulation, it has been established for many many years. Nobody would [build] a plane without first simulating the ability of that plane to climb … In the medical therapy side of things, we are way behind,” she said at the 18th Annual Lecture hosted by the Fund for Replacement of Animals in Medical Experiments (FRAME). While Rodriguez’ work holds hope for reduced biological experimentation, this cannot be fully abandoned, since any development of modelling needs to be in an organic iterative loop with an understanding of real-life physiology. Blanca’s cardiac model is currently proving useful for pharmaceutical companies such as Amgen, Merck & Co. and UCB Biopharma to test the safety and efficacy of drugs on  simulated human hearts.

Electrical simulation of the heart from cellular units up to ECG level, integrated with anatomical model of the heart in the torso. Pic: Ana Minchole and Blanca Rodriguez

While the cardiac physiome has made impressive progress, it is still in isolation from other systems. As a neuroscientist, I was interested to know where we are with integrating the brain within the physiome. Alain Goriely, an Oxford professor of mathematical modelling, thinks we are a long way away: “The brain is of course a formidable challenge for any mathematician – in terms of modelling there are many questions at all levels … Right now we are trying to integrate various aspects of the brain – the lymphatic system, with cerebral flow, together with the tissue itself. And that’s already a big part. So I think we are nowhere close to integrating multiple organs, [but] eventually you need the brain in the picture,” he says. Goriely and his team are modelling the processes involved in dementia, a neurodegenerative disease, the economic impact of which is more than cancer and heart disease combined.

Visual representation of a brain model where the nodes are “regions of interest” and the colours represent the weight of the connection between nodes (number of axons/length). Pic: Alain Goriely

Now, you may be wondering what happens when scientists do finally build your virtual twin, brain and all. Will it be able to laugh at your jokes, and feel your pain? Will it think like you? Will it live on indefinitely, as you? “Well I think the brain physiome would just be how the brain works you know, the plumbing and all,” Goriely says, “but the Human Brain Project, that was their initial ambition – if we get all the brain connections and put them on a computer can we understand the mind. We’re pretty far from that … but I think mathematical understanding of consciousness or the mind will arise from a good understanding of the brain connectome.” He believes we can learn a great deal about the mind by looking at dementia, which results in significant brain atrophy, a literal disintegration of neuronal matter: “If you start with the mind, and see the evolution as it breaks down, maybe you can see how the pieces fit together before. You can ask the question, when does the mind or the consciousness or the presence of the person disappear during the disease – it kind of evaporates – what is the transition? When is a person not there anymore?”

The Physiome project and the Virtual Physiological Human initiative do not have explicit intentions to simulate the human mind too, but to really have holistic healthcare, we cannot separate the mind from the body. “Where Western science went wrong, it hived the mind off as a separate thing from the body,“ Noble warns us, referring to the 17th century father of modern philosophy, René Descartes. “You see, there are processes in relation to our health that clearly depend on the body as a whole, it’s interaction with the rest of the environment … and the mind, in particular.” This principle is often a common thread through Eastern and other ancient medical practices.

Given the steady evolution of science and medicine towards a more integrated outlook, as well as the exponential rate of developments in technology, perhaps one day we may see the digital reincarnation of humans. The implications of this are astonishing. Nick Bostrom, director of Oxford’s Future of Humanity Institute and author of the book Superintelligence, urges us to think far ahead and is devoted to ensuring that we are prepared to respond to intelligent entities before we create them. But do we have the right to control them? The HBO sci-fi series Westworld, while wildly entertaining, poses the poignant ethical and philosophical dilemmas of creating conscious androids, in which the fictional creator, Dr Robert Ford highlights the precarious double-edged sword, “You can’t play God without being acquainted with the Devil.”

We are curious creatures. Understanding the nature of our existence, our body and mind is one of the biggest endeavours of mankind, one which will require a concerted effort from all areas of science, mathematics and technology. Applying our knowledge to improve our lives and those of our loved ones is an even greater challenge. Technological advancements are transforming medical research and healthcare at an astounding rate. Our virtual twin may one day help to further reduce human suffering. But I also hope that we continue to apply the common sense and judgement that are ingrained in our heuristics, and treat our twin with the dignity, respect and wisdom that we crave.

Modelling the heart

The heart is made up of electrically active muscle cells, called cardiomyocytes. Each cardiomyocyte contains a multitude of protein channels, which allow the flow of charged molecules called ions. The flux of ions produces currents, which gives rise to the contraction of the myocyte, resulting in the heart beat. These channels have certain dynamic properties that define the kinetics of channel opening and closing. By defining these dynamic properties of channels, the activity within a myocyte can be simulated on a computer. Every person’s heart is different – it has a different shape, different size, different position within the chest, and it may even have different profiles of channels. So modellers like Blanca Rodriguez use experimental data from real patients’ heart cells, combined with MRI images of their chest and heart, to overlap the electrical activity with an anatomical and mechanical model of the heart’s movement. Scientists can then run simulations of hundreds of thousands of drug compounds and their interactions with the heart’s ion channels, to visualise the outcomes in the heart’s rhythms. The momentum is now there to integrate this into broader healthcare systems, so having your heart on a computer  is just a matter of time.



About the Author

Atreyi Chakrabarty

Atreyi is a PhD student at the University of Oxford, studying neuroscience. When not in the lab, she likes to write about science, and is a former editor-in-chief of The Oxford Scientist magazine.

Subscribe to our newsletter