Neural networks developed by researchers at Google’s UK AI division, DeepMind, are as capable as world-class clinicians at detecting more than 50 eye diseases, according to a landmark study.
A paper published in Nature Medicine on Monday reveals that the AI system, built in collaboration with Moorfields Eye Hospital and UCL, can make the correct referral decision 94 per cent of the time, matching the clinicians’ score.
During the study, the system’s two neural networks were trained on thousands of “historic de-personalised eye scans” in an attempt to teach them to identify the characteristics of a range of eye diseases.
The research marks the first time DeepMind’s AI has been applied to healthcare data and lends weight to the idea that medics and intelligent machines will one day work side-by-side.
The researchers hope that the system could soon compensate for the fact that the number of eye scans is growing at a faster rate than experts are able to interpret them. It could pave the way for earlier diagnoses and treatments.
“The AI technology we’re developing is designed to prioritise patients who need to be seen and treated urgently by a doctor or eye care professional,” said Dr Pearse Keane, a consultant ophthalmologist at Moorfields.
“If we can diagnose and treat eye conditions early, it gives us the best chance of saving people’s sight. With further research it could lead to greater consistency and quality of care for patients with eye problems in the future.”
As well as offering referral recommendations, the system also provides an illustrated explanation of how it reached its conclusion and a percentage to demonstrate the level of confidence.
Researchers are now preparing to test the technology in clinical trials. The goal is to roll out the technology across Moorfields’ 30 hospitals for free for at least five years.
“We set up DeepMind Health because we believe artificial intelligence can help solve some of society’s biggest health challenges, like avoidable sight loss, which affects millions of people across the globe,” said Mustafa Suleyman, DeepMind’s co-founder and head of applied AI at DeepMind Health.
“These incredibly exciting results take us one step closer to that goal and could, in time, transform the diagnosis, treatment and management of patients with sight threatening eye conditions, not just at Moorfields, but around the world.”
The extent to which private companies should be able to access NHS datasets is a matter of ongoing political debate. Writing for NS Tech in May, the MP Darren Jones said the government should “expect to share in any profits made off the back of [a private company] using an NHS data-derived algorithms anywhere in the world”.
Martin Cordiner, head of research at the College of Optometrists, welcomed the paper, noting that eye-care professionals are excited about the potential that AI has to assist them. He added: “Further support in prioritising the referral of the patients with the greatest need is in the interests of both practitioners and patients, and we look forward to the results of clinical trials of this technology.”