Doctor AI and your digital twin - the future of diagnosis

Could the key to your future health rest with a digital twin who acts as a crystal ball of future health that doctors can use for dry run procedures or to monitor reactions to treatment?

A scientist who champions the idea predicts the development will become commonplace but only if patients are confident that their data is as secure as a bank account.

Peter Coveney, a chemistry professor and computational scientist, says the rise of artificial intelligence means computers can accumulate knowledge like no doctor before them – and predict a patient’s health like a weather forecast.

The idea of giving you and your doctors a digital body double to examine has attracted interest from universities, major corporations such as Nokia and Siemens, and Saudi Arabia’s futuristic megacity Neom, among others.

But it gives rise to ethical questions such as: Who controls my data? Is my right to privacy being invaded? And would an AI medic lack the necessary human touch?

Prof Coveney, the director of University College London’s Centre for Computational Science and co-author of a book called Virtual You charting the path towards digital twins, believes there are “compelling benefits” that make this ethical minefield worth navigating.

“At a stroke you should be able to reduce, if not entirely eliminate, the use of animals, because these so-called animal models are never sufficiently faithful to humans anyway,” he told The National.

The idea, he says, is that people can deal with not just the short-term weather of ailments requiring a doctor’s visit but understand the climate of their long-term health.

“You can have access in these scenarios to your own data. Indeed you should have your own data, you should hold it yourself so you don’t depend on hospitals giving it to you,” he said.

“You can use that to help you to plan your own lifestyle, by making these longer-term predictions out of it, rather than just the short term where I need a fix because something’s gone wrong.”

Organ doubles

The immediate aim is not to replicate every one of the 30 trillion-odd human cells in your body. That would take too long, even with modern computing power. However, efforts are under way to give your vital organs a digital duplicate.

Siemens is looking at copying the heart, potentially allowing people to spot early signs of cardiovascular disease. It hopes to make this a reality within a decade. There is an EU-funded initiative to make a brain twin. US and UK researchers this month announced a $3.2 million project to digitally inspect your bladder.

These organ doppelgangers could then be programmed with your medical data – or even linked to your fitness watch – so that doctors are “treating a patient that’s in front of you, rather than a statistical inference of lots of others”, Prof Coveney said.

In Saudi Arabia the vision is of a "Dr Neom" system in which people can track their own health or consult a virtual doctor. It is hoped that every resident will benefit from the virtual database that tracks their blood type and unique genetic sequence.

Prof Coveney believes people should control the use of their data, and be able to opt in to clinical trials, to assuage fears of a Big Brother industry putting health records to unwanted use. He sees online banking security as a model.

“Most of us implicitly, if they don’t think about it too hard, are actually pretty happy with having their bank accounts accessible and managed on the internet,” he said.

“The technology exists to provide that level of security but we’re nowhere near realising it [in medicine], because it’s got all these complicated issues around ethics and who’s in charge of your data.

“But if we move to a position where everyone had their own data, and they controlled it in the same way as you have your own bank account and control it, you are then in a position to use that data in the ways you wish to for yourself and for the benefit of medical science and so on.”

AI as a doctor

If AI can crunch through all this data to make a model, can it also be your doctor? Can it take all this knowledge, maybe furrow its brow over a clipboard, and deliver a verdict on your health?

In principle, yes – but there are reasons to be cautious here too.

Software like ChatGPT may sound like a human being but it can state things as fact that are total nonsense – designers call this “hallucination" – and has no mechanism for demonstrating its workings, which is known as the “black box” problem.

The first is clearly a hazard for life-and-death decisions. The second could make doctors reluctant to trust it.

“It’s rather difficult for a regulatory authority to agree to allow a black-box solution to be used to save people’s lives if we don’t actually understand why it’s making the predictions it does,” Prof Coveney said.

That’s where people come in. Like a weather forecast, even the most sophisticated digital twin modelling would only ever be a statement of probability. It is likely you’re sick, based on the data, but every simulation will have a slightly different result. It will be up to a person to decide what to make of this.

“The sharp end is to find ways of making those predictions reliable such that we can have confidence in them, and a clinician can,” Prof Coveney said.

“We will be making informed decisions around probabilities as to what's likely to be most successful in a therapeutic context.”