Sometimes medical breakthroughs come from odd places. For 35 years a French software company called Dassault Systèmes has been making 3-D “digital twins” of real objects like bridges, buildings, and automobiles to test, virtually, how they will behave in the real world. Using such a digital twin, engineers can see how a bridge flexes during heat waves, how buildings sway during earthquakes, and how cars behave during head-on collisions. That allows them to identify—and perhaps fix—weaknesses, ideally before shoddy bridges, buildings, and cars have a chance to hurt anyone.
But Dassault wants to take its concept of digital twins into medicine and to simulate the human body. But the foundation of that future likely has to be laid organ by organ, with realistic and customizable models of each jigsawed together. So far, the company has just one organ—the heart—on which they can do in silico analysis, or research by computer modeling. Dassault calls this organ “The Living Heart,” even though the point is precisely that it’s not living.
Simulated organs could change how medicine works, making it hyperpersonal and less invasive. A medical scan—of a broken elbow or a mild concussion—could become a 3-D model that doctors could use to diagnose and test treatments. When all medical records become electronic, information skimmed from them—assuming they are stored in some universal format, a task in itself—could begin to construct a digital picture of each person’s inner workings. “Over time, [my medical] record could be a digital twin of me,” says Steve Levine, the chief strategy officer for Dassault’s Simulia division. In its most evolved form, this shift would allow doctors to replace generic assumptions with individual ones and then digitally probe individuals.
But that will require coordination across care facilities, software companies, and governmental organizations, which may be reluctant to standardize and share their proprietary data. And, of course, software engineers must build more organs, based on high-resolution scans and probes of IRL organs. After all, says Alexandra Golby, the director of image-guided neurosurgery at Brigham and Women’s Hospital, “Any 3-D model is only as good as the data itself.”
Dassault is hardly alone in pushing this agenda forward. Collaborations like the In Silico Oncology Group and the Virtual Physiological Human Portal, among others, also aim to create and integrate parts of your body inside a computer.*
For Dassault, though, it all begins with the heart, and the project’s challenges map out some of the obstacles in the road to true “digital twins.” For Levine, this project is personal. Inside his daughter’s heart, the left chamber is where the right one should be, and the right one is where the left should be. Because of that reversal, Jesse received her first pacemaker at age 2. But the wire leads kept breaking, and by age 20, she was on her fourth device. Her doctors said that to understand her rare condition and come up with better treatments, they needed more data. And that data would be difficult to retrieve. Given all of the metal in her chest, they couldn’t give her an MRI. So Levine decided to create data about his daughter—digitally. And so in 2013 he and Dassault—which had recently branched into life sciences—decided to apply Simulia’s technology to the heart.
We know the basics about the heart—for instance, the average ticker beats 100,000 times every day, pumping the equivalent of 7,000 liters of blood through. But the small-scale details have remained mysterious—as we all know, the heart is complicated. Ventricles are one-tenth the width of the atria; the atria can also be tangled like clothes just taken out of the washing machine, and their stiffness changes rapidly during a beating cycle. Electricity and blood propagate through different parts of the heart differently.
Others have tried to make these sorts of models in the past, but they fell short. Some simulated the left ventricle, fewer simulated both, and even fewer took all atria and ventricles into account. Some accounted for electricity, others blood flow. None could handle all of the physics and geographies. They were oversimplified and unrealistic: spherical cows, as physicists would say. On top of that, their interfaces were kludgy—hand-built, as software often is for academic research. Their creators could use them but had not made them user-friendly (or didn’t have the expertise to do so) because they didn’t always plan to share: The data undergirding them was often proprietary, and different groups held their own poorly made models close.
So while most of the information needed to make a heart model existed—somewhere—it was scattered across labs, split between people with different needs and desires. When Levine called for researchers and organizations to join their data together, he promised not to share the individual pieces of data itself with anyone else—just the final, put-together puzzle. Tell Dassault what you know about the heart, he said, and Dassault will weave it into an easy-to-use, customizable model: the Squarespace of simulated hearts.
Dassault commercially released the Living Heart—the first realistic model that accounts for electricity, mechanics, and blood flow—in May 2015. The software can turn a 2-D scan from an individual human into a personalized full-dimensional model of his or her heart. The user can manipulate it—stick in pacemakers, reverse its chambers, cut any cross section, and run hypotheticals. Just like meteorologists combine radar and satellite data to make models—predicting how much sun you’ll see next Wednesday and where the strongest winds will be inside a superstorm—medical simulators see both inside and into the future of virtual organs, without needing to put on scrubs.
It took the group two years to weave the Living Heart together. Today, it is made up of 208,561 tiny, digital tetrahedrons, each of which has its own electrical and muscular properties. Those properties can be adjusted to account for individual patients, conditions, and devices. So far, experiments with this virtual heart mimic the behavior of the real one.
But certainty that a simulation actually behaves like a real organ—sure enough that you’d bet someone’s life on it—is serious business. To ensure safety and learn how best to integrate digital medicine into physical practice, not just for the Living Heart but for the simulations that come after, the Food and Drug Administration has entered into a five-year partnership with Dassault.
But not everyone believes simulations show the way to the future. First, clinicians and researchers have to be willing to use digital twins, and then they have to learn how to do so correctly. “Most doctors are receptive, but without training, they will not go for it,” says Victor Chang of Leeds Beckett University, who contributed to the book Advancing Medical Practice Through Technology. “E-learning for staff training is very important to ensure doctors are well-equipped.”
Programs like that aren’t widespread or formalized, although software simulations are regularly used to train apprentice doctors and to improve the view during invasive procedures. “There’s a fair amount of work going on in terms of augmented reality during surgery,” says Golby. But she cautions that not all organs are as simulate-able as the one behind your ribcage. “The brain is many orders of magnitude more complex than the heart,” she says. “I’m not sure that I would see a comprehensive brain model emerging any time soon.”
Other critics remain skeptical that any virtual body part will ever be good enough to trust. “There are basically two camps of experts on this general topic,” says Kirby Vosburgh of Harvard University and the Center for Integration of Medicine and Innovative Technology. One camp says that big digital data will take over medical knowledge, and digital models will, in turn, take over the everyday practice of health care. The second camp says that digital knowledge can’t expand to represent the variety of conditions and qualities of every individual. “You treat individuals,” says Vosburgh, “and can’t do that well by making each person ‘average.’ ”
In camp two, Jeffrey Olgin of the University of California, San Francisco, for example, fears human bodies are too complicated to be confined to a computer. In an interview with KQED, he said, “Unfortunately, the human body doesn’t always follow the rules of physics.”
But, replies physics, that can’t actually be true. Everything has to follow the rules. And when the human body doesn’t do what we expect, that simply means we don’t know its details well enough to understand the physics. It does not mean the details are unknowable or that the body is breaking universal laws. It rains when the meteorologist says it won’t because the meteorologist didn’t map enough particles between the atmosphere’s edge and the ground, not because storm fronts behave miraculously.
The birth of all these digital twins won’t be an easy labor, if it’s possible at all. But if they’re out in the world, they could change it and us. We would have the power to understand ourselves as well as we understand the stress and strain of a bridge—what happens when we work right, what happens when we go wrong, and how to turn those wrongs into rights.
This article is part of Future Tense, a collaboration among Arizona State University, New America, and Slate. Future Tense explores the ways emerging technologies affect society, policy, and culture. To read more, follow us on Twitter and sign up for our weekly newsletter.
*Correction, Feb. 12, 2016: This article originally misstated that Harvard University’s Wyss Institute’s “organs-on-chips” team works on digital simulations. They are microchips lined by human cells, which the institute plans to link together to mimic whole-body physiology.