Prof James Wolffsohn, professor of optometry at Aston University, said: “AI analysis of retinal images has already shown the ability to extract far more than clinicians ever thought possible, such as age, sex and blood pressure.
“Therefore it is not surprising that it could be used for early diagnosis of glaucoma.
“In terms of the quoted study, true artificially intelligent algorithms need thousands of images to learn, and it is not clear whether those not officially referred were screened to see how many cases of glaucoma were missed.
“The ground truth will have been an ophthalmologist determining whether the referrals had glaucoma in their opinion, which is variable and doesn’t necessarily reflect the long term prognosis of those patients.
“Hence this research is important, but we have a long way to go.”
Contact details are available to account holders.
Create a free account to get in touch with this expert.