AI tool successfuly used to warn of heart attacks

It’s hard to read, listen to, or watch much nowadays without some mention of AI – artificial intelligence. Often that mention will not be positive with concerns about accuracy, plagiarism and job losses. But experts from the University of Oxford have developed an AI tool that does a better job of predicting heart attacks than doctors.

More accurate prediction means better preventative measures, and lives saved.

How can AI tools predict a heart attack?

The key to answering that question lies in CT scans, and the ability of the human eye to detect warning signs – or more accurately in this case, the human eye’s inability to detect those signs.

Many patients who present with chest pains will have a CT scan. The aim is to find evidence of narrowing or blockages in coronary arteries. A majority of those patients – about 75 per cent – historically have scans that reveal no such evidence. The patients are given the ‘all clear’ and sent on their way, suitably reassured.

However, a significant proportion will go on to have heart attacks regardless. In those cases the doctors haven’t failed to be careful enough, they simply didn’t have the ability to see other tell-tale signs within the scan. But the latest AI tools do.

A newly developed AI tool can detect changes in the fat around inflamed arteries, something that’s not possible with the naked eye. It can also provide information on narrowing and other clinical risk factors. That extra information, basically invisible to doctors examining a CT scan, helps predict the risk of heart attacks or strokes.

Is this technology available now?

Many developments in medicine require years of trials before being approved to ensure there are no unintended adverse consequences. The good news for the new AI tool, is that, to use a marketing cliché, ‘the future is now’.

Thanks for this goes to the good people of the University of Oxford’s Radcliffe Department of Medicine. In the new study, funded by the British Heart Foundation, the Oxford team, led by Professor Charalambos Antoniades, analysed data from over 40,000 people undergoing routine cardiac CT scans at eight UK hospitals.

Many of the 40,000 were accurately predicted to have a risk of a serious cardiac event by traditional analysis. That analysis of scans revealed significant coronary artery narrowing. But more than twice as many whose scan showed no significant narrowing still went on to experience heart attacks or even cardiac death.

The new AI tool was trained to look at not just narrowing but also blood vessel inflammation. It was further trained to look at changes in fat around those inflamed blood vessels. Using the combined data, the AI tool – in layman’s terms – hit the jackpot.

Next steps

Testing on 3393 patients over 7.7 years revealed the AI tool could independently and accurately predict risk of cardiac events. This successful real-world trial has now been extended to a pilot in five hospitals. Researchers hope it will soon be introduced more widely.

Said Prof. Antoniades: “We demonstrated that providing an accurate picture of risk to clinicians can …  potentially improve the course of treatment for many heart patients.”

Prof. Antoniades hopes the AI tool will soon be implemented across Britain’s National Health Service. This, he says, will help “prevent thousands of avoidable deaths from heart attacks every year in the UK.”

With luck, Australians who are unknowingly at risk will soon have access to the new AI tool. And this might just boost the reputation of artificial intelligence’s potential for the good of humanity, rather than the alternative.

Have you been a sceptic when it comes to usefulness of AI tools? Does this new study change your attitude? Let us know via the comments section below.

Also read: What your heart rate reveals about you