By Patrick Landon Ferree, PhD Candidate in Cell Biology, Duke University, CIFS Health Junior Associate
When we think of digital twins, we tend to imagine a twin to a human being, like the one in the picture above. However, the concept of the digital twin is a little broader. A digital twin is a simulation that incorporates the minute details about any physical object. Imagine a ball. A digital twin for this ball (the one you’re imagining) will be a computer model that incorporates all of the irregularities that distinguish it from other balls. Those irregularities will include information about the texture on the surface, and where the texture is wearing. Quantitative information about the durability of the bladder, sturdiness of the windings, and health of the carcass. But also, information about the air pressure and temperature, and deformations in the shape of the ball as it is dribbled across the field. The goal of the digital twin is to be hyper-personal and hyper-specific to its underlying object.
Formula 1 engineers build digital twins for each of their individual racecars. Hundreds of sensors throughout the cars relay information back to headquarters for performance analysis in real time. They can then visualize exactly what went wrong around a corner, or how their tires hold up under certain weather conditions.
Digital twins are popping up everywhere. With rapid improvements to computation and simulation capabilities, it makes sense to study and test out the limits of an object in silico (on a computer) before making any major manipulations to the physical object as it is in the world. F1 engineers tweak certain parameters in their digital twin and study how the engine holds up. If the digital engine fails, they know not to make that particular change to the real-life car.
Besides simulation capabilities, over the course of the last decade, deployment of digital twin capabilities has accelerated due to increase in sources of data (with real time monitoring), interoperability (as integration of digital tech has increased), visualization (including advanced data visualisation), instrumentation (IoT sensors etc), and rise of platforms (hence, the availability of powerful computers).
Benefits to healthcare
It is not terribly difficult to imagine the benefits that digital twin technology will bring to healthcare. A short perspective piece in Science predicted back in 2013 that the future of biology would be dry (Service 2013). That biologists and biomedical scientists would work comfortably from their laptops studying complex computer models of cells, tissues, organs, and organ systems. Today it is mostly recognized that there must be a synergy between wet and dry science. The wet scientists collect the data that the dry scientists build the models.
But there are also already reports of digital twins in the health science literature. For instance, scientists have simulated in silico certain individuals’ hearts and used these models, along with artificially intelligent algorithms, to make predictions about the risk of arrhythmia (Maleckar et al. 2021). The goal of this type of research is to prevent unnecessarily invasive surgeries prior to having knowledge about the problem. Just like with the racecar, it makes sense to test out the options on a computer simulation first.
Genomic and Mobile Twins
The genome is another area of interest for digital twin research. As the genomes of more and more humans are sequenced, it has become clear that they harbor quite a bit of natural variation. The future of personalized medicine promises to take this variation into consideration for treatment plans. Perhaps it will become common for electronic health systems to house each person’s digital genomic twin, which will feed helpfully into the decision making.
But the more information that a doctor has about a patient’s unique biology, the better they can diagnose an illness or prescribe a medication. Many of us already wear health sensors on our bodies. Our watches measure our heart rates and even the levels of oxygen in our blood. This type of real-time mobile health data can also be incorporated into our digital health twin.
The consolidation of health data into personalized digital twins may radically improve the process of drug discovery and testing. Reports show that 38-75% of patients do not respond to drug treatments for many of the common diseases (Björnsson et al. 2020). Hence, there is a movement to use digital twins to screen thousands of drugs to find the one that works.
Digital Twins in Metaverse
Regardless of what Metaverse,this mysterious place, turns out to be, there is a lot of interest in providing healthcare to patients remotely, and this has led some to speculate that the Metaverse may be the right place to do that. Throughout the COVID pandemic it became common to visit digitally with healthcare providers.
Rapid improvements in virtual reality technology suggest that perhaps visits to the doctor will take place behind VR goggles. And more than that, there is interest in surgeons performing operations remotely -an activity that is already underway but will be radically improved with VR technology. A similar story can be told about the training of doctors. No one wants to be the first one to receive a novel surgical operation, or even be a surgeon’s first living patient. These are problems that the Metaverse and its digital twins may solve.
Semi Digital Twins First?
There are, of course, some obstacles to this digital healthcare utopia. The digital twin is intended to be an exact replica of an object. But biology is extremely complicated and difficult to replicate exactly. Whereas with the F1 racecars we literally know all the details, this is not true of biomedicine. We are still working out the molecular basis of many, if not most, diseases, and unknowns are difficult to simulate. So, it seems that most computational models of biological phenomena are currently, at best, semi-digital twins.
Furthermore, at an individual level, not everything is easily quantifiable into a set of parameters (Popa et al. 2021). At the end of the day, there will be health-related variables that do not enter into even the best digital twin-like models. Not that this is such a bad thing. Some scientists have even argued that true digital twins would be an overkill, and require unreasonable levels of computational power to generate even just one human-like digital copy.
One last challenge to the future of digital twins in healthcare is (no surprise here) the whole collection of ethical concerns that it raises, not least amongst them, the acquisition and proper handling of huge amounts of personalized data. As with any other up-and-coming Silicon Valley Super Technology there are concerns about who gets to keep the infinitely detailed models of humans. The goal, after all, ought to be the improvement of health outcomes and the flattening of socio-economic disparities, and not the invention of new ways to market hyper-personalized unnecessary products.
Edited by Leo Petersen-Khmelnitski, LinkedIn