I have always made good use of my daily commute listening to BBC Radio 4. It remains my staple for news, comedy, intelligent debate, and insight into the world of the visually impaired through its excellent programme In Touch.
A couple of morning drive-time Today programme items caught my attention recently. Mid-August Pearse Keane of Moorfields and University College London, (the subject of In Focus, August 24) was grilled by John Humphries about his research into retinal imaging in collaboration with Google’s Deep Mind Health. Using retrospective OCT data Deep Mind has been trained to recognise 53 different retinal conditions with a startling 94% degree of accuracy.
Now you may be thinking that this is less than perfect, however, it is apparently slightly better than can be achieved by ophthalmologists with decades of training and experience. Deep Mind can also assess thousands of OCT scans in a few minutes. What other news reports didn’t point out was that it didn’t miss a single case of a sight threatening eye disease.
In the absence of perfection one would think it surely better to be safe than sorry. Certainly ‘better safe than sorry’ is a mantra of many optometrists, who believe they should refer any patient about whom they are unsure. These optometrists need to change; or be changed for artificial intelligence (AI); as they are causing patients to lose their sight.
As Dr Keane keenly pointed out on In Touch the same week, of a cohort of 7,000 patients referred to Moorfields last year with suspected AMD, only 800 had the disease. Eighty-eight percent of patients got the ‘sorry we wasted your time, but better safe than sorry eh?’ explanation, compared potentially to only 6% of Deep Mind referrals. It isn’t just that poor referrals waste precious NHS resources, the burden of seeing 6,200 people unnecessarily meant that many of the 800 people with AMD did not receive treatment within the critical two-week window and suffered irreversible sight loss prematurely as a result.
The second Today piece that caught my attention was an interview on the world of work with Shadow Chancellor John McDonnell. He elaborated the risks and opportunities posed by AI, and explained that while previous industrial revolutions had displaced manual workers, the current industrial revolution means AI will displace cognitive workers. Thinking jobs will go too.
On the flipside all new technologies generate new jobs as long as employers and educators can keep up with the need for skills training. Here for once optometry might just be ahead of the game with new courses in the pipeline at the University of Central Lancashire (Preston), Bedfordshire (Luton), Highlands and Islands (Inverness), West of England (Bristol), Teesside (Middlesbrough) and Huddersfield. So are these venerable institutions training the optometrists of tomorrow?
At first glance it seems not. The publicised courses seem little different to the existing provision. There is no talk of blended learning or apprenticeships, heralded as possibilities in the GOC Education Review, and where will demand come from? Adopting a system of optometry in England similar to either Scotland or Wales, or simply ensuring universal commissioning of enhanced services could increase demand massively, but more than a week after A-level results day every existing optometry course except Cardiff still had places on offer through clearing for 2018 entry.
Thinking of the 6,200 patients unnecessarily referred to Moorfields, this would not have happened if the correct optometric services were commissioned in the first place. If GOS was amended so that supplementary tests could be done as follow ups (as they are in Scotland), or if there was a universally commissioned equivalent of the Eye Health Examination Wales across England then the problem would be substantially solved.
The Optical Fees Negotiating Committee should look to reform GOS to the benefit of patients, by demonstrating the benefit of supplementary tests, howsoever commissioned, in improving the quality of referrals.
As well as NHS connectivity and universal e-referral it should also be pushing for the universal adoption of artificial intelligence including OCT in every contractor practice. Certainly referral quality would improve immeasurably and optometrists could sleep sound at night knowing they no longer need to play safe better than sorry.