Human digital twins: medicine without the guesswork

Professor David McGiffin knows his trade. A top Australian heart-lung researcher and a retired head of cardiothoracic and transplant surgery at The Alfred hospital in Melbourne, McGiffin has seen huge advances in his field over a long and distinguished career.

But one thing that remains unsolved involves a life-or-death decision surgeons have to make around a particular cohort of patients with a rare, but nasty lung condition: chronic thromboembolic pulmonary hypertension.

“When patients have a blood clot go to the lung, in 96 percent of patients the clot is broken up and resorbed by the patient's own lytic mechanisms,” McGiffin says. “But with about 4 percent of patients the clot isn't broken down, and then it causes obstruction of the arteries to the lung. This is ultimately a fatal condition.”   

There is hope – in the form of an operation, but it’s a long and very complex procedure that’s incredibly hard on the patient’s body. If it works, it can save someone’s life; if it doesn’t, the stress of the operation on the patient, without the improvement in their symptoms, risks killing them. 

For most patients, the decision whether to operate or not is pretty straightforward, McGiffin says. Doctors can tell from scans and tests whether surgery will be helpful or not. But in some people it’s really hard to tell whether they have the form of the disease that will be cured by surgery, or the form that won’t.

“There's a group of patients where we don't have really objective evidence. There have been some complicated ways devised to try and separate those two causes of pulmonary hypertension, but none of these techniques have proved all that reliable. So we're often in a conundrum: is an operation going to be a good idea or not?

“What we'd love to be able to do is feed all the information we have on the patient – their demographics, their heart catheterisation, their imaging studies and all the other information we have, into some sort of model which predicts what the post-operative result will be.”

And that's where human digital twins come in – and the start of a research collaboration between McGiffin and his colleagues at The Alfred hospital, and researchers at the Auckland Bioengineering Institute at the University of Auckland.

Digital twins, computerised models of a structure or object, have been around for a while in other engineering fields. If you are planning a bridge, a wind farm or a new sort of plane, a digital twin allows you to test the design against a range of scenarios – severe weather, for example – and see how it performs.

Human digital twins are way more complex. The human body contains a vast number of systems, so trying to translate those into a single computer model is incredibly difficult.

But the idea is the same: build a digital model of an individual person using all the medical, genetic and demographic information you have about them, and you would be able to use it to see how that person’s body would react to a particular medication or surgical intervention. One day surgeons like McGiffin could know whether an operation for chronic thromboembolic pulmonary hypertension would kill or cure a particular patient.

Work on a human digital twin has been going on for decades, spearheaded by the Auckland Bioengineering Institute, where researchers have already produced digital twins of the heart, lungs and other organs and systems.

Professor Merryn Tawhai leads the ABI and is part of the team working on digital twins. She says a digital human could be particularly useful with outlier cases – the 20 percent of patients who have their doctors scratching their heads.

“These people consume a lot of resources and end up with poor quality of life. They are difficult to treat, and they need lots of follow up visits. We would be able to use the digital twin to test different options. It might be surgery, a change in drugs, or it could be adding in one drug as well as another drug.”

The pulmonary hypertension surgery project is still at the research stage - inputting past patient histories and actual outcomes into the model and seeing whether the digital twin would have come up with the right answer. But digital twins are already being used in other ways – to improve outcomes for people having hip replacements, for example. 

One of the problems for hip replacement surgeons is everyone has different sized and differently shaped bones, says Dr Duncan Bakke, a research bioengineer at Formus. They need to get exactly the right hip implant – and there are lots of permutations. 

The present system involves the surgical team looking at the scans and x-rays from the person going under the knife and using their experience to judge the size and position of the implant they will put in. However, a little under 10 percent of people who have a hip replacement need to go back for what’s known as a revision – or repeat surgery, often because the first surgery wasn’t exactly right, Bakke says.

Formus’ technology gives clinicians more information. “If we can go in and take a CT scan and make a 3d model of the bone, then we can figure out ahead of time which size of implant is going to be the right one, rather than having to try it out in surgery or with intuition.”

Digital health navigator

Another Auckland Bioengineering Institute project, backed by more than $4 million of Ministry of Business Innovation and Employment funding, involves developing individualised digital twins for people managing type 2 diabetes from home, and then linking those digital twins with an on-screen robot – or “digital health navigator”. 

Basically the digital twin will gather information from wearable devices being used by the person with diabetes – maybe it’s a smart watch or a glucose or blood pressure monitor. Then the digital navigator will translate that information into day-to-day health advice.

The human digital twin been a project 40 years in the making, but ABI director Merryn Tawhai has no doubt it is coming.

“I think back to my childhood and Star Trek. When somebody got ill there was that little handheld device that scanned over them, and they miraculously knew what was wrong with them and how to treat them. 

“Well, we're not quite there yet, but we're along that pathway.  Five, 10, 15 years’ time we would be looking to be able to scan part of you and get new information that tells us about your current condition, and that is linked to your digital twin. Then we will be able to more rapidly make a diagnosis and determine a treatment plan for you.”