Imagine having a digital twin that gets ill, and can be experimented on to identify the best possible treatment, without you having to go near a pill or a surgeon’s knife. Scientists believe that within five to 10 years, “in silico” trials – in which hundreds of virtual organs are used to assess the safety and efficacy of drugs – could become routine, while patient-specific organ models could be used to personalise treatment and avoid medical complications.
Digital twins are computational models of physical objects or processes, updated using data from their real-world counterparts. Within medicine, this means combining vast amounts of data about the workings of genes, proteins, cells and whole-body systems with patients’ personal data to create virtual models of their organs – and eventually, potentially their entire body.
“If you practise medicine today, a lot of it isn’t very scientific,” said Prof Peter Coveney, the director of the Centre for Computational Science at University College London and co-author of Virtual You. “Often, it is equivalent to driving a car and working out where to go next by looking in the rear-view mirror: you try to figure out how to treat the patient in front of you based on people you’ve seen in the past who had similar conditions.
“What a digital twin is doing is using your data inside a model that represents how your physiology and pathology is working. It is not making decisions about you based on a population that might be completely unrepresentative. It is genuinely personalised.”
The current state of the art model can be found in cardiology. Already, companies are using patient-specific heart models to help design medical devices, while the Barcelona-based start-up ELEM BioTech is offering companies the ability to test drugs and devices on simulated models of human hearts.
“We have already run a number of virtual human trials on several compounds and are about to enter into a new phase, with our product being ready and deployed in the cloud for external access by pharmaceutical clients,” said the ELEM co-founder and chief executive, Chris Morton.
Speaking at the Digital Twins conference at the Royal Society of Medicine in London on Friday, Dr Caroline Roney, of Queen Mary University of London, described efforts to develop personalised heart models that would help surgeons plan surgery for patients with irregular and chaotic heartbeats (atrial fibrillation).
“Often surgeons will use an approach that works on average, but making patient-specific predictions and ensuing that they predict longer-term outcomes is really challenging,” Roney said. “I think there are many applications in cardiovascular disease where we will see this sort of approach coming through, such as deciding what type of valve to use, or where to insert it during heart valve replacement.”
Cancer patients are also expected to benefit. Artificial intelligence experts at the drug company GSK are working with cancer researchers at King’s College London to build digital replicas of patients’ tumours by using images and genetic and molecular data, as well as growing patients’ cancer cells in 3D and testing how they respond to drugs.
By applying machine learning to this data, scientists can predict how individual patients are likely to respond to different drugs, combinations of drugs, and dosing regimens.
“You can’t do this repetitively with the real patient with multiple drugs and drug combinations, because every time you try a new treatment it is a clinical trial,” said Prof Tony Ng, of King’s.
“We are trying to find a solution while the patient is still alive, so if they come back with a recurrence [of their cancer] we will know how to treat them, or which clinical trial to put them on.”
Proof of concept trials are expected to begin next year.
Researchers are even developing digital twins for pregnancy, which could help develop drugs for conditions such as placental insufficiency or pre-eclampsia, and a better understanding of the physiological processes underpinning pregnancy and labour.
“You can’t do experiments on pregnant women in many cases, and there are also not good animal models for human pregnancy,” said Prof Michelle Oyen, the director of the Center for Women’s Health Engineering at Washington University in St Louis.
Oyen is building placenta models from ultrasound scans taken during pregnancy and high-resolution images after birth in women with healthy and complicated pregnancies, and training an algorithm to recognise and construct a digital replica of the various tissues.
“Our goal is to try to figure out things that we could measure on a live person to predict who is likely to have problems with placental function during pregnancy, and intervene to prevent things like stillbirth,” Oyen said.
Her collaborator, Prof Kristin Myers, of Columbia University in New York, is constructing models of the cervix, uterus and the membranes that surround the foetus. Their long-term goal is to combine them all into a single model of an individual that could predict how the pregnancy might play out.
Myers said: “My hope is that we could take a simplistic ultrasonic scan of maternal anatomy and be able to assess how this uterus is going to grow and stretch, and better time when labour is going to happen.” It might even predict a long or complicated labour, and help women make a more informed decision about whether to have a caesarean section, she said.
Other researchers are building a digital twins of hospitals to try to improve the efficiency with which individual patients move through the healthcare system.
“By tracking digital signatures that are made every time anything happens with a patient – from when an X-ray is ordered, performed and reported, to when that patient is booked for an outpatient appointment and attends it – we can build a highly detailed, real-time picture of how patients with similar conditions move through the system,” said Dr Jacob Koris, a trauma and orthopaedic surgeon and digital lead at Getting It Right First Time, a national programme designed to improve the treatment and care of patients.
“Doing so could identify areas we need to improve, but also good practice that improves patient care, which we can use to redesign how we look after patients.”
