Sight saved using artificial intelligence?

Sight loss around the world: Image: Deep Mind, Moorfields

Sight loss is a major global health problem. Many eye diseases can be treated effectively if they detected early, But the current imaging system needs highly trained experts. DeepMind, working with Moorfields Eye Hospital has developed a system using Artificial Intelligence to ‘read’ and interpret the results of optical coherence tomography (OCT) scans.

OCT produces a detailed map of the back of the eye, in 3D images. There’s some info about OCT here and a  normal OCT image below

Sight: A normal retina (back of the eye) in an OCT scan.Image: American Academy of Ophthalmology
A normal retina (back of the eye) in an OCT scan. Image: American Academy of Ophthalmology

Healthcare professionals at Moorfields have to go through over 1000 OCT scans a day at Moorfields. Which obviously takes lots of skilled time! The inevitable delays can cause sight loss.

This week, an Artificial Intelligence / OCT system was reported to be as good as, or even slightly better than, the best experts. There’s a less technical write-up of the study here.The application of Artificial Intelligence to Optical Coherence Tomography is surely very welcome if it can prioritise those who need specialist clinical input – and so speed up vision-saving interventions.  Happily, the initial results suggest this is the case.

Elsewhere in this week’s BMJ there is an article on ‘Patients roles and rights in research’ – making it clear that patient involvement is best practice. The BMJ say that they have extended their current reporting requirements on how patients and the public were involved in research they report.
I’d like to know how the Deep Mind / Moorfields team are going to involve patients in the work they still need to do for regulatory approval? As the BMJ Editorial suggests, it’s not just about token involvement for ‘virtue signalling’

Deep Mind are keen to sort out the ‘black box’ problem for clinicians understanding a new technology. What about the patients?

The AI/OCT package is intended to help those with, for example, macular degeneration and diabetic retinopathy.
How are they addressing the needs of those with multiple visual morbidity? These more informed patients may well have different communication as well as clinical needs.
Deep Mind is clearly leading on some truly innovative work.

Can they be as exciting in their patient / community involvement?

I’ve submitted a version of this as a BMJ Rapid Response. I’ve got some ideas about the answers – and the Responses to the Editorial make some more great points.


Related Post