The research, which appears in Nature Medicine, applied deep learning techniques to thousands of historical anonymised retinal scans. Following this training, the AI system was able to recommend the correct referral decision for over 50 eye diseases with 94 per cent accuracy, matching the performance of the top medical experts in the field. With nearly 300 million people around the world living with some form of sight loss, it is hoped the work could help doctors and eye professionals spot serious conditions earlier and prioritise patient treatment.
“The number of eye scans we’re performing is growing at a pace much faster than human experts are able to interpret them,” said Dr Pearse Keane, consultant ophthalmologist at Moorfields Eye Hospital NHS Foundation Trust and National Institute for Health Research clinician scientist at the UCL Institute of Ophthalmology. “There is a risk that this may cause delays in the diagnosis and treatment of sight-threatening diseases, which can be devastating for patients.
“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. 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.”
According to DeepMind Health, its neural networks are able to identify ten features of eye diseases from highly complex optical coherence tomography (OCT) scans. The AI then makes a referral recommendation based on the most urgent diagnosis, with the results checked against the recommendations of clinical professionals. To back up the referral, the system also provides the information that formed the basis for the decision, along with a corresponding percentage of confidence. This type of ‘explainable AI’ is key if humans are to act definitively on the recommendations of machines, in healthcare and beyond.
“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, 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.”
DeepMind’s access to NHS data came under scrutiny in 2016 and a recent report – commissioned by the company itself - warned that its existing relationship with the NHS could lead to it developing “excessive monopoly power” over UK healthcare data. The company’s business model at this point has yet to be defined, with DeepMind so far employing its technology for free in collaboration with NHS trusts and academic institutions. If clinical trials of its eye diagnosis system are successful and it achieves regulatory approval, Moorfields will be able to use it for free across all 30 of its UK hospitals and community clinics for an initial period of five years.