Call 08000 277 323 any day, any time

AI diagnosis – a new age for medicine, or a potential liability nightmare?

In an exciting development for both technology and medicine, an Oxford hospital has developed an artificial intelligence (or AI) that can look at scans and diagnose heart disease or cancer. The hope is that the technology can save the NHS millions of pounds, at a time when it is under significant pressure to reduce costs.

However, it does present some difficult questions when it comes to negligent treatment.

How does the new AI work?

Currently, doctors review echocardiograms to identify any problems a patient may have. However, even the best of them get things wrong a fifth of the time, and patients go on to have a heart attack or an unnecessary operation.

The AI on the other hand – known as Ultromics – uses something called machine learning to understand patterns that show heart disease.

What this means is that it reviews a database made up of thousands of echocardiogram images, extracting various bits of data to develop an understanding of the patterns that show different diseases. The team at Ultromics in Oxford state that this has improved the accuracy of tests by up to 90%.

Are there any drawbacks to this new AI?

No new technology is perfect, and this AI is no exception.

Ultromics is not designed to be a magic bullet for cardiographic diagnosis but rather, as Prof. Paul Leeson puts it, “now there is a possibility that [heart specialists] may be able to do it better”.

Whilst Prof. Leeson’s words are confident, the work Ultromic has done is not yet peer-reviewed, so their theory has yet to be confirmed by other experts in their field.

Finally, from the point of view of a clinical negligence specialist like myself, this AI poses a somewhat difficult question about liability.

Who’s at fault if things go wrong?

Whilst the reduced rate of negligence is certainly welcomed, the errors in diagnosis will not be totally eliminated with this new technology.  And when errors happen, it’s not yet clear who will be at fault should complications arise.

For example, should someone be given the ‘all clear’ but go on to have a heart attack, should the AI result have been double-checked by a specialist? Is this possible? If not, and this isn’t reasonable to ask of clinicians, who is responsible for the technology when it makes an incorrect diagnosis? Is it the hospital that uses the technology, or the manufacturers?

Hopefully, all of these questions will be answered prior to the introduction of Ultromics in the summer. In the meantime, we should all be excited by the possibilities this new technology offers to both clinicians and patients – could it signal a new age for medicine?

 

Want to know more?

Share this

Leave a Comment

Your email address will not be published. Required fields are marked *

Explore our site