Intelligent solutions to support patient outcomes

There are so many potential applications for artificial intelligence (AI) in eye health, but two areas with immediate impact and appeal for eye care in the UK are already with us and growing. One is the use of AI in patient support – the simplest version is the AI chatbot that can provide an ‘always on’ solution to patient queries; and the other is the application of AI based on deep learning to medical imaging, including retinal imaging.

Detecting abnormal features in retinal vasculature, lesions, optic nerve head images and optical coherence tomography (OCT) images will standardise and improve care, whilst facilitating different models of care. Ocular imaging is key to audit, telemedicine and virtual support, all of which allow us to plan safe and effective care for patients without the most senior clinician personally examining every individual.

AI is already with us. Chatbots are in use across several medical areas, including ophthalmology, and several options already exist in glaucoma and diabetic eye care to help identify high probability of abnormality, but not all of them utilise deep learning. The most useful role, so far, appears to be screening and detection in pathology with characteristic changes in fundus appearance.

The number one question will always be: what does it do for the care of our patients?

Co-morbidities in an elderly cohort are fairly common – hypertension and cardiovascular disease to name just two, plus other eye pathology – so there will always be a screening element to the pre-assessment visit. The clinician has to establish suitability for surgery and if any co-pathology needs prior management. Any tool that helps us to identify retinal disease – particularly through hazy media – is most welcome.

At pre-op assessment we also grade lens opacities, but this is subject to clinician interpretation. If this could be automated to flag up posterior polar cataract, for example, this could be beneficial. Patients with posterior polar cataract are at higher risk of intra-operative complications so sure knowledge of this condition directly impacts surgical planning.


We are always looking for intelligent solutions that improve the care and outcomes for our patients, whether that be a specialised wheelchair that can be used with ophthalmic diagnostic equipment or a digital AI solution. Having access to technology that can easily identify abnormal retinal images and raise suspicion for clinicians before they examine the patient has the potential to increase the effectiveness of the consultation.

Digital imaging systems already facilitate shared care and virtual review, via the exchange of accurate information. If we can also rely on AI to remove false positives or false negatives from a screening or monitoring service, this is likely to improve collaborative working further, and reduce the frustrating number of unnecessary or inaccurate referrals into secondary care.

The number one question will always be: what does it do for the care of our patients? Every other consideration stems from that – disease detection rates, how does it compare with the clinician alone, do we detect disease earlier, does it reduce the need for visits to the clinic? Referral refinement, screening, monitoring stable eye conditions, and identifying progression are all resource-heavy areas for ophthalmology and optometry – AI technology can be part of the solution. I’m eager to see where it takes us.