A medically accurate digital double could revolutionise your life, f inds Linda Geddes
WHEN June slipped on an icy pavement and fractured her wrist, the doctors treating her thought little of it. She was a fit and healthy 65-year-old. Her wrist was set in a cast and she was sent home. But two years later, June tripped in the garden and couldn’t get up. She was still there the following morning, when her son found her and took her to hospital. June had osteoporosis. She had fractured a bone in her hip and needed major surgery to put in an artificial replacement. She later suffered two fractures to her vertebrae, deforming her spine and restricting her breathing. Confined to bed for six weeks, she developed a chest infection and died. Her case is not unusual. Osteoporosis, where bones become weak and fragile, is often missed first time round. But just four years from now, such an experience could be entirely different. After that initial fracture, June’s bone density would be measured and she’d be sent home wearing sensors that continuously recorded her activity levels, gait and posture. She’d be offered a CT scan to capture the exact structure of her bones and reveal areas of wear, tear or weakness. All of this data would then be plugged into a computer simulation that would create a virtual June, fast-forward her through the years, and predict her physical destiny Welcome to the Virtual Physiological Human, an initiative backed by the European Union that has attracted £200 million of investment and more than 2000 researchers. The next decade is already set to bring personalised drugs and diagnoses based on a growing mass of patient data covering everything from lifestyle to genetic profile. But the Virtual Physiological Human project has an even bigger goal in its sights. Ten years from now each of us could grow a kind of twin. Detailed simulations of the inner workings of our bodies, from the interactions of genes and proteins in individual cells up to organs and whole-body systems, will be combined with data about our medical histories and the kind of lives we lead. The result will be a digital body double that could be experimented on, testing outcomes for different drugs, surgical interventions and lifestyle choices. We would go through life with a virtual clone. “Finally, we will be able to say something about you,” says Marco Viceconti at the University of Sheffield, UK. “Not because you are the same age, or sex, or have the same genetic profile and disease as a thousand other people, but because you are you with your condition, your history, your bad habits.” With power handed to each of us, healthcare as we know it would be turned on its head. Armed with a personal health forecast, we can choose to take charge of our health rather than accept our fate. But a glimpse of the future brings a responsibility to act. How will we cope? Embracing complexity TOMMY PARKER The Virtual Physiological Human initiative sprang from a 2005 summit at which European researchers from many different areas of bioengineering, physiology and clinical medicine agreed that a common pattern was emerging. Trying to tackle diseases by breaking them down into smaller parts and then looking at each one individually was no longer working. “If you look at the big diseases that plague Western society – cardiovascular disease, cancer, diabetes, osteoporosis – the one thing they have in common is that their symptoms cannot be explained by looking at one part of the body,” says Viceconti. Take osteoporosis. We know that many fractures in elderly people are the result of reduced bone density, and that this is caused by a hormone-related shift in the activities of different bone cells. But fragile bones are only part of the story. Deteriorating eyesight or “ Finally, we can say something about you: your condition, your history, your bad habits” hearing, weaker muscles, a person’s activity levels and posture all increase the risk of falling and fracturing bones. A pure focus on genetics doesn’t tell you that much about your future risk of developing these complex conditions. “You can have a genomic blueprint, but it will not tell you precisely what’s going to come out at higher organisational levels,” says Peter Coveney at University College London. Doctors caring for patients with multiple conditions, such as diabetes and heart failure together, face a similar problem. “The links between different diseases aren’t always made,” says Liesbet Geris at the University of Liège in Belgium. “It’s often hard to envisage what the effect of medication for the heart problem will be on the diabetes, or how these two diseases will interact.” In short, we have reached a point where it is no longer possible to ignore complexity. We have to embrace it. That was the conclusion of those researchers who met back in 2005. And their solution? Build computer models that can combine the vast amounts of data about biological processes at the smallest scale – the workings of genes, proteins, and cells – with our growing understanding of whole-body systems. The models would simulate how the body’s microscopic machinery gives rise to everything else. This can then be integrated with a patient’s personal data, ultimately creating virtual models of us all. Already, researchers have produced digital models of major organs like the heart and liver, which can be used to design new drugs. They can even model the forces generated by blood flowing through the arteries in a person’s brain and predict whether an area of weakness known as an aneurysm is likely to rupture and cause a stroke. So far, most of these models aim to give a real-time window on activity in the present, rather than predict the future. But that is coming. Viceconti and his colleagues have developed an osteoporosis model designed to second guess an individual’s risk of sustaining a fracture, by combining existing patient data with CT scans and information from wearable sensors. The goal is that by 2018, people like June whose osteoporosis may be relatively mild, yet have a high risk of fracturing bones because they are more likely to fall over, could be better identified. “This would allow us to not only target therapies that would protect their skeleton, but design exercise interventions to address an individual’s risk of falling,” says Eugene McCloskey at the University of Sheffield, who is preparing clinical trials to test the bone fracture model. Other teams are also working on the musculoskeletal system. Geris, for example, is using computer simulations to try to speed-up the development of tissue implants to heal bone fractures. She has modelled the process of bone growth from the bottom up, starting with the interaction between different genes to identify the key proteins that influence cell growth. The simulations can then predict how adding different chemical factors might influence fracture healing at the level of the whole bone. “It’s all about trying to make better decisions than the ones we’ve made through trial and error in the past,” says Geris. The complete patient One big challenge facing researchers is figuring out how to hook all these systems together into a complete patient. For now, they remain a collection of disembodied parts. But we may not be far off. For example, Robert Hester at the University of Mississippi Medical Center in Jackson and his colleagues have developed a mathematical model of the whole body called HumMod, consisting of some 5000 variables that describe factors including measurements of blood flow, recordings of heart and brain activity, results of blood tests and scans of our bones, muscles and organs as well as features of the environment, such as air temperature and humidity. Tweaking the numbers lets you play out potential scenarios for particular patients. Maths is only part of the puzzle, though. Models like HumMod will then need to be stitched together with an ever-growing mountain of biochemical and physical simulations, personal histories of illness and injury, and descriptions of diet and lifestyle. The goal is for Virtual Physiological Human models to be able to integrate different types of patient information – MRI scans and genetic data, say – quickly and automatically, even in emergency scenarios (see diagram, left). But for a doctor or nurse to run such simulations on demand will require more computing power than most hospitals have at their disposal. Indeed, several researchers, such as Coveney, who models blood flow in the brain, currently run their simulations on a supercomputer. “If you ask me point blank who is using modelling and simulation to save people’s lives, there’s very little of it at the moment,” says Coveney. Yet the wheels are turning. The US Food and Drug Administration, for example, is already considering large-scale simulations for drug testing. The toughest challenges are unlikely to be technical, however. For starters, there could be resistance from the medical community who may feel their judgement is being taken away. The hope, though, is that doctors will be more confident that the treatments they offer will improve patients’ health. “At the moment, doctors often have just one chance to get it right,” says Norbert Graf, a medical director at Saarland University Hospital in Germany. “But if you have a virtual patient, you could select the treatment with the highest cure rate and the lowest toxic effect.” Doctors would still need to speak to the patient, make the diagnosis, and select and administer the treatment – but they’d have another tool at their disposal. “The difference is that there would now be a model selecting which is the best treatment for the individual patient, not a protocol that is the same for hundreds of patients,” says Graf. Having a virtual twin should make us pay more attention to our own well-being, putting responsibility for healthcare into our hands, say researchers behind the project. But when it comes to looking after ourselves, our track record is mixed. Not everyone agrees that this will motivate people to take charge of their health. Jane Wardle at University College London has investigated the impact that genetic screening for obesity or lung cancer has had on people’s motivation to eat healthily or give up smoking. “Giving people more detail about future risk has never produced the effects that people hoped,” Wardle says. Often people feel very motivated in the short term. “They say that ‘It’s a wake-up call’,” she says. “However, in the long run, it’s not actually achieving any behaviour change.” Even so, those working on virtual versions of us are optimistic. Evidence suggests that being confronted with a vision of your own future – rather than being told you have a 65 per cent chance of developing lung cancer, say – can have a big impact. “Visual imagery is more emotionally arousing than non-visual information,” says Hal Hershfield at New York University. Hershfield is interested in how thinking about our future selves affects our decision-making. In one recent experiment, he took photos of volunteers and used them to create aged avatars, complete with jowls, bags under the eyes and grey hair. The volunteers then took control of their avatar in a virtual world, where they were confronted by this aged image of themselves in a mirror. Decisions, decisions Later, they were given a choice of four ways to spend $1000: buy a gift for someone special, have a party, put the money in a current account, or invest in a retirement fund. Hershfield found that people who came face to face with an older doppelgänger put nearly twice as much into the retirement fund as those who saw an avatar who was the same age as themselves. A related study by Jesse Fox at Ohio State University in Columbus has shown that watching a personalised avatar lose weight by exercising can motivate people to go to the gym within the next 24 hours. At heart, we’re wishful thinkers. Asked to imagine how we would look and feel if we continue to lead unhealthy lives, it is easy to conjure up a better future than is realistic. Confronting the future seems to help. “Imagination can only go so far,” says Hershfield. But as Wardle points out, the challenge is translating short-term motivational boosts into long-term change. And herein lies another key difference between genetic screening and individualised medicine. As a lifelong companion, the virtual twin would allow us to track our health continuously. Many people already use health apps on their phones to track their diet, exercise and quality of sleep. Adding such data to your simulated self would refine the models and boost their predictive accuracy. “This isn’t just a gene saying you have a 90 per cent chance of getting cancer and there’s nothing you can do about it,” says Geris. “Your genetic profile won’t change. But if you change your pattern of behaviour and food intake, the model can update. If you let yourself go, your prognosis will change.” Interventions could also be tailored to specific people like June. By 2023, data from wearable sensors could constantly and automatically update your risk of fracture. New advice based on these updates could then be streamed back to a smartphone. Local weather conditions might be taken into account. If your digital double deemed you were at risk of falling, it could send you a reminder to put on the anti-slip shoes it had encouraged you to buy after noticing that your stability was getting worse. For some, this may be too much. Ultimately, the biggest roadblock may be our own unease. For the grand vision to work, all of our medical data will need to be put online, an idea that has already met resistance. We may also resent the intrusion of extra surveillance. And then there’s the very real possibility that a virtual twin may throw up insights that we are not prepared for, such as a better understanding of how and when we are likely to die. A diagnosis of terminal cancer, for example, leads some to vow to fight it and others to lose the will to go on. Deciding how much we each want to know will need to be decided on a case by case basis. It may be that our virtual twins will need to mirror more than just our biology. “There are different personality traits that are going to predict how people will react to bad news,” says Fox. “If you’re going to model the physical body, you really need to think about modelling a person’s psychological state as well.” ■
Aucun commentaire:
Enregistrer un commentaire