Intended for healthcare professionals

Feature Artificial Intelligence

The computer will assess you now

BMJ 2016; 355 doi: (Published 24 October 2016) Cite this as: BMJ 2016;355:i5680
  1. Stephen Armstrong, freelance journalist, London, UK
  1. stephen.armstrong{at}

Stephen Armstrong reports on work to develop machines that can diagnose conditions faster and more accurately than people

Can a game of space invaders defeat the largest cause of sight loss in the developed world? Research at the heart of a controversial deal between Google’s artificial intelligence (AI) company DeepMind and Moorfields Eye Hospital aims to find out.

DeepMind, a UK based company founded in 2010, recently signed a series of deals with NHS hospitals to develop diagnostic tools for a series of common conditions, including age related macular degeneration with Moorfields,1 head and neck cancer with University College London Hospitals (UCLH),2 and acute kidney injury with the Royal Free.3

“I approached DeepMind’s founder Mustafa Suleyman after reading an interview with him in Wired,4” says Pearse Keane, a consultant ophthalmologist at Moorfields Eye Hospital, who specialises in conditions of the retina, including age related macular degeneration.

Over 600 000 people in the UK have age related macular degeneration, and 200 people are diagnosed with the condition every day.5 Over the past 10 years, optical coherence tomography (OCT) has revolutionised diagnosis.6 The technique uses light waves to take cross sectional pictures of the retina. Several high street opticians already offer OCT scans in the UK, but the problem, according to Keane, “is that they don’t always have the necessary training to read the scans.” He says that high street opticians are referring large numbers of people with false positive results for urgent NHS assessment. “We’re struggling to deal with the workload, and it means patients who do have macular degeneration may not be seen in time while those who don’t spend time worrying.”

DeepMind’s machine learning seemed to offer a solution—in February 2015 the company published a paper in Nature showing the results of an …

View Full Text

Log in

Log in through your institution


* For online subscription