Getting Eye Care Down to a Science

There is a very timely need in the retina subspecialty to transition to new, digitally-enabled models of care, and then implement them into real-life practice. To get there, we need to use the capabilities of digital health technologies, including telemedicine and AI decision-support systems – tools that can help professionals make diagnostic and management decisions for patients with common retinal conditions and at the same time improve the patient experience of care. Such tools require a number of validation processes, as well as evidence gathered through what is known as “implementation science,” so that we can better understand their place in clinical practice.

There are three main steps in the development and introduction of AI decision-support systems and other digital technologies. The first one is a proof-of-concept step: developing the algorithm, which can then be tested against professional experts on retrospectively collected data sets – this should show performance as good as that of retinal experts in making the correct diagnosis. The second step requires gathering evidence from prospective research in real-life settings – introducing the developed and tested algorithm into a hospital setting, as well as community optometry practices (high street opticians), and gathering evidence of how it performs in real-life environments and in the general population. The third step involves “implementation science” – exploring how patients and practitioners interact with new digital technologies, and how offering and receiving care changes as a result of the algorithm implementation. This final step can include any changes to the workflow, any enablers or barriers to successful adoption, perceptions of these technologies, and economic aspects (the impact on the healthcare system’s finances, and on various professions involved).

The FENETRE study is one of the implementation science projects that I’m leading. It is a multi-site, clinical trial, based in 16 hospital-based departments in England, and around 40 community optometry practices – and it’s just about to begin. FENETRE is funded by the National Institute for Health Research in the UK, and is looking at an alternative model of care for AMD patients, who could receive their care in community optician practices rather than in hospital settings. This model is facilitated by digital technologies, and looks at creating a link between community optometry and specialized hospital-based services used to deliver second opinions, as needed, to ensure safety, and to provide training and quality assurance for the community partners.

The exploratory part of this project will analyze all the data collected in the study to check that management decisions made in the community and hospital settings match the decisions recommended by an AI decision-support system, assessing its performance. We have also developed a way to evaluate the economic impact of this model – investigating how finances might be affected if decisions were made by AI instead of human experts. We hope that the project will help determine the best pathways for management of patients with AMD.

One interesting point that makes this study even more timely, is the fact that community optometry practices are increasingly equipped with advanced imaging technologies (such as OCT). If they are capable of undertaking more primary and high-volume routine care for patients with common retinal conditions, such as AMD and diabetic retinopathy– with easy access to secondary care as a safety net, it would provide a convenient alternative pathway to patients; these community settings are often easier to access and more patient-friendly. And it would release some of the significant burden of care and treatment  on the healthcare system, allowing better and faster access to hospital-based care for patients that require treatment.