Easing the strain on doctors with A-eye

AI can help diagnose sight-threatening eye conditions for early treatment
16 August 2018

AI is helping diagnose eye diseases to help improve the chances of patients saving their eyesight. Source: Shutterstock

When simple eye diseases cause patients to turn blind, it can have irreversible consequences.

According to data from the World Health Organisation, there are 285 million people globally who are visually impaired. Eighty percent of these cases could’ve been prevented if a diagnosis was received in time.

However, ophthalmology is heavily dependent on retinal scans, takes quite a lot of time to diagnose, and therefore, fails to provide a prognosis on time.

DeepMind is helping solve this issue with the use of AI. Working with Moorfield Eye Hospital in London, they have trained algorithms to detect over 50 eye conditions that can cause visual impairment. This cuts down the time doctors spend on diagnosing, allowing them to focus more on providing 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 in a statement.

“There is a risk that this may cause delays in the diagnosis and treatment of sight-threatening diseases, which can be devastating for patients,” he explained.

In fact, a clinician’s time is often squeezed, making it difficult for them to perform a detailed diagnosis for a patient.

A study by BMJ recorded that across the world, consultations generally lasts less than 5 minutes. In the UK, patients generally get 10-minute slots, which is hardly enough for treating complex cases.

DeepMind’s algorithm is able to recommend an appropriate referral decision and prioritize patients according to the urgency of treatment based on diagnosis. An earlier detection and treatment can give patients a better chance of saving their eyesight.

However, the diagnosis made by the AI system isn’t meant to be taken as an absolute. Rather, it serves as a guideline, whereby medical professionals can further assess if the diagnosis is accurate, before going ahead with treatment.

Keane added that further research can help improve the consistency and quality of care for patients with eye problems in the future.

While DeepMind’s algorithms were trained with one particular type of eye scanner, it is compatible with any model. This means that the AI can be used regardless of hardware, even when equipment may be replaced or updated in the future.

As the world population grows older, cases of visual impairment will also be more prevalent. Estimates by the International Agency for the Prevention of Blindness (IAPB) suggested that by 2050, the aging population can contribute a threefold increase in people suffering from blindness.

Although AI isn’t able to make 100 percent accurate diagnosis as of yet, the ability to speed up processes for doctors will help ease the strain on the healthcare services. This also ensures patients get the best care they need.