When an artificial intelligence (AI) beat the world champion at Go, it made international headlines. AlphaGo’s victory – masterminded by scientists at Google’s secretive subsidiary DeepMind – was seen as a major leap forward; Go should have proved too complicated for a six-month old program to learn.
Yesterday, DeepMind announced its involvement in a very different project. Its partnership with Moorfields Eye Hospital might not be as flashy as its previous projects, but it could be far more immediately useful. A DeepMind system will analyse a million digital eye scans, learning the patterns that indicate early signs of blindness. The system will then create an algorithm that can detect those warning signs in previously unseen scans.
DeepMind is looking to spot two conditions in particular, wet age-related macular degeneration and diabetic retinopathy. The latter of those is the fastest-growing cause of blindness in the world.
“There’s so much at stake, particularly with diabetic retinopathy,” DeepMind co-founder Mustafa Suleyman told The Guardian. “If you have diabetes you’re 25 times more likely to go blind. If we can detect this, and get in there as early as possible, then 98% of the most severe visual loss might be prevented.”
Moorfields and Google: an unlikely pair
The collaboration between DeepMind and the Moorfields NHS Trust came about after Pearse Keane, a consultant ophthalmologist at the hospital, got in touch with DeepMind through the company’s website. Like many others, Keane had been impressed by DeepMind’s achievements in training an artificial intelligence agent to solve Atari games.
Having read up on machine learning, Keane realised how well it could be applied to his area of expertise:
“I had the brainwave that deep learning could be really good at looking at the images of the eye. Optical Coherence Tomography is my area, and we have the largest depository of OCT images in the world. Within a couple of days I got in touch with Mustafa, and he replied.”
OCT certainly seems a good candidate for deep learning, which has already been mooted as a useful tool in treating depression and diagnosing cancer. NHS higher-ups have already implemented several robotics solutions at a number of trusts, but AI has been implemented much more slowly. Despite Jeremy Hunt’s promise in February that computers would be diagnosing patients at home within two years, the NHS’s infrastructure remains incredibly outdated.
In that context, Pearse Keane’s outreach is the more praiseworthy. Keane took it upon himself to start a research project that could massively cut the amount of time doctors spend diagnosing these conditions. That would free his fellow ophthalmologists up to do more valuable work, making Moorfields more efficient.
What are Google’s plans for the NHS?
This isn’t the first time DeepMind has collaborated with an NHS trust. The company partnered with the Royal Free Hospital in North London to develop an early warning system for patients at risk of developing kidney problems. That agreement sparked debate over whether or not Google had permission to access such a huge cache of personal data – but the Moorfields collaboration uses anonymised data, so is much less likely to cause trouble.
Though the scans are anonymous, they’re paired with information on eye condition and disease management. Alongside the scans, that data provides a fuller picture of patients’ conditions.
Using that data, DeepMind’s agent could make the diagnostic process many times quicker, handling data at a much greater rate than a human would be able to:
“It takes me my whole life experience to follow one patient’s history” said Prof. Peng Tee Khaw, head of Moorfields’ ophthalmology research centre. “And yet patients rely on my experience to predict their future. If we could use machine assisted deep learning, we could be so much better at doing this, because then I could have the experience of 10,000 lifetimes.”
Google is known to want more NHS data – and despite patient disgruntlement, I suspect they’ll get it. Obviously this is a divisive issue, but I tend to think it’s unlikely that an NHS doctor – or a Google engineer, for that matter – is laughing at your bunions. Google will acquire much more of our data – and our NHS will be better for it.
Category: Artificial Intelligence