Liesbet Geris (45) has just had breakfast in her hotel in the centre of Amsterdam. Her parents and son, who are accompanying her, are going for a walk in the city, while the Flemish professor (at the University of Liège and KU Leuven) takes time for an interview. "If we want to preserve the healthcare system we currently have, we need more personalised medical advice," says Geris. "Digital twins can play an invaluable role in this."

Geris is the coordinator of a European project that kicks off the development of digital twins for the entire human body. From patients to hospitals, from companies to universities: Geris has brought everyone together over the past few years around the virtual digital twin. Due to this innovative work, she was awarded an honorary doctorate from the University of Amsterdam two weeks ago.

The idea is to use real-time patient data as input for computer models that give a signal when something might go wrong. Moreover, doctors can test in a digital version of the patient what effect their actions will have in advance. For each condition or body system, you can design a specialised twin, and these specialised twins will continuously "talk" to each other.

How would that look in practice?
Geris: "Let's take an imaginary patient in ten, fifteen years: Rosa, a 55-year-old woman with a family history of type 2 diabetes. She leads an active lifestyle and proudly tells her GP that she has a smartwatch and exercises a lot.
The GP suggests that she shares the data from her smartwatch with her personal Virtual Human Twin Platform. This is a digital environment where all her health data is stored in a secure vault, and she always has access to it.
"In the platform, the GP looks at which digital twin programs are specifically relevant. Because of Rosa’s family history, for example, he recommends downloading the diabetes program, and the bone program is also indicated for her due to her age and possible risk of osteoporosis. Rosa also decides to share her glucose data and information about her diet.
"After a year, Rosa starts experiencing muscle cramps, causing her to move much less than usual. As a result, the bone program automatically generates a warning to both the GP and Rosa.
"Rosa undergoes a physical examination at the GP's office, and they discuss the digital twin data in more detail. The diabetes twin, based on recent glucose readings and eating patterns from the past months, has detected early signs of diabetes. Since this could also increase the risk of osteoporosis, the diabetes twin has alerted the bone twin. Together with the twins, the GP provides Rosa with personalized advice."

But maybe not everyone wants to know that such a risk exists?
"Yes, but with this, you can effectively prevent something. As a society, we must act because the healthcare system we currently have is neither affordable nor sustainable. So, if we want to preserve our healthcare, we need to be able to intervene earlier. And it's not like in the movie Minority Report, where actions are taken before something happens. This is about processes that have already started, and where you can still intervene now."

Do doctors already use digital twins in healthcare?
"The beginning is there," says Geris. She opens her laptop and shows a picture of a heart. "Look, the left chamber of the heart has an appendix, the left atrial appendage. Sometimes blood enters there and clots, leading to blood clots. If those clots travel to the brain, it can cause a stroke. That's why doctors need to place an occluder, a kind of stopper in the left atrial appendage. That procedure is called left atrial appendage closure. But the exact shape of the stopper needed looks different in each case. On images, it’s not always easy to see what shape that stopper should have. But if we create a virtual 3D model of the left atrial appendage in advance, we can fit the stopper in and then simulate the blood flow in that chamber to see if it works properly.
"We also use this for stents and 3D-printed bone implants. Companies offer it as a service. Doctors send images through a secure link. Those companies create a 3D model, analyse it, and provide options to the doctors. We see that doctors feel more confident during surgeries, as if they're supported by a colleague. It’s also a cost-saving measure, because often multiple stents or stops are used before finding the perfect fit. By using these types of models, researchers have shown a reduction in the number of stents used."

This is a computer model of something physical: a stopper or a stent. But to what extent will digital twins in the future be more virtual?
"When we talk about digital twins, it always involves both a physical and a virtual component. In industry, the link between the physical and virtual is often in real-time. For example, sensor readings from an aeroplane provide updates that are immediately entered into a virtual model, and that model can also send data back right away.
"But in medicine, this is rarely the case because you're dealing with people. You can create computer models of human cells, tissues, organs, or systems to learn more about them. And with that, we can get a broader, virtual picture of an individual's overall health. So, it goes beyond just the geometric shape of a specific part of the body. In that case, you should think more about Rosa, who receives advice based on combined models."

Is there still a role for artificial intelligence to make digital twins better?
"Partially. If we don't have much data but we know how something works, we use the physical mechanisms we know. Take a CT scan, where signals are converted into a black-and-white value. That’s based on pure physics. But when we compare data from a patient's smartwatch with a large amount of data from others to predict for that specific patient, we use AI. In practice, you often see combinations of models with AI and models without."

Are digital twins a form of prevention?
"Yes, but with 'regular' prevention, we are often too early in the process. People are not sick yet, and then suddenly the government steps in and says: you need to go for a walk, you need to eat so many grams of vegetables every day. These are general pieces of advice that people often shrug off. With the digital twin, people receive advice and warnings based on their data, indicating that something might go wrong. People will respond differently to that.
"At the same time, researchers can use the data, with the patient's consent, for research where there is a lack of data, such as research into rare diseases and conditions. My twin data can also help others in that way."

How will all those digital twins collaborate effectively in the future?
"We need to build a public infrastructure. A place where researchers can communicate with each other, where we can link models and exchange information between them. A platform where, as a researcher, you look at a specific model and, based on your choices, are also recommended other models, just like with Spotify or Netflix. This is important because, without large-scale collaboration, we won’t solve the big problems in healthcare. With funding from the European Commission, we are now building that platform."

CV Engineer

Liesbet Geris (45) was born in Hasselt, Belgium. Trained as a biomedical engineer, she became a professor at the Universities of Liège and Leuven in 2008. Since 2022, she has been the coordinator of Edith, a European multi-million project that provides a platform for building the virtual human twin.

Author: Sanne Bloemink 

Published in De Groene Amsterdammer on December 14, 2024. Translated by the Institute for Advanced Study.