Using machine learning for early detection of eye diseases: this is what DeepMind, a division of Google, and London’s Moorfields Eye Hospital are planning. The two partners are working on a five-year research project which aims to use algorithms to speed up the process for detecting eye diseases via scans carried out at Moorfields Eye Hospital. Ars Technica looked at the project in more detail.
Data science is no longer fantasy but reality. Back in December 2015, we talked about an algorithm that can predict how sugar levels vary from one person to another after a meal.
Now, eye diseases could benefit from machine learning, technology that creates predictive models from datasets.
Two million people are already suffering from sight loss in the UK, of whom around 360,000 are registered as blind or partially-sighted. Google predicts that the number of people suffering from sight loss could double by 2050. Better detection and earlier treatment could therefore make a huge difference.
Through this research project, DeepMind and Moorfields Eye Hospital hope that common causes of sight loss, such as diabetic retinopathy and age-related macular degeneration, can be spotted earlier and thus treated more effectively to prevent blindness. Google claims that up to 98 percent of sight loss resulting from diabetes can be prevented by early detection and treatment.
The project focuses on two types of scans: traditional fundus images, which are basically photographs of the retina at the back of the eye, and optical coherence tomography (OCT) scans, which are a cross-section of the retina. As Pearse Keane, consultant ophthalmologist at Moorfields, explains:
“If you have an OCT scan done, a machine learning algorithm will be able to tell if it’s something urgent, versus something that’s not so urgent.”
The previous collaboration between DeepMind and the NHS (National Health Service) was somewhat controversial: in April 2016 New Scientist revealed that DeepMind had access to the health records of 1.6 million patients in the UK – without their knowledge. This time, however, Google assures that all the data used is anonymous and will remain the property of the hospital:
“It’s not possible to identify any individual patients from the scans. They’re also historic scans, meaning that while the results of our research may be used to improve future care, they won’t affect the care any patient receives today.”