Introduction

About our AI and Technology Innovations Task Group

Our AI and Technology Innovations Task Group was established in October 2024 with a remit of exploring the issues, implications, obstacles, and pitfalls of innovative technology adoption in optometric practice.

“Understanding product limitations and trusting your own judgement based on that knowledge will be essential. We hope that this resource will help you in your journey.”

Following a series of discussions and lectures from experts in the field, the Task Group have developed this resource providing members with useful information on AI, designed to offer members easy read guidance. We hope that it will help ensure that implementation of these exciting technologies is in the best interest of optometry practitioners and their patients.

With an abundance of materials and resources on AI already available this resource explores only the primary considerations for optometrists. We hope it will make an in-road into helping you become more “AI-Ready”.

In its simplest form, AI combines computer science and robust datasets to enable swift and efficient problem-solving. More advanced forms feature reasoning and decision-making based on data inputs and complex layers of parameters. The widely prevalent AI sub-fields of machine learning and deep learning are frequently found under the broad umbrella of AI. These sub-fields are usually assigned specific tasks and do not “think” on their own but can hone their abilities to achieve the task as they continue to process data in real time.

Indeed, the technological innovations under the banner of artificial intelligence already permeate our daily lives – both personal and professional. They can be found in a range of everyday applications, including virtual assistants, search engines, navigation software, and facial recognition systems to name but a few. From the popularity of “frontier AI” products using Large Language Models (LLMs) such as ChatGPT, generative image and music software, and the development of “AI Agents”, its wider influence on culture seems to be unstoppable.

AI is already being used in clinical settings for eye care. For example, ocular coherence tomography (OCT) is benefiting from tools that aid image acquisition and noise reduction^{i ii}, OCT angiographyⁱⁱⁱ, and also in retinal layer segmentation^iv,v. In recent years, there have been software developments that can detect the optic disc and macula in fundus images^vi, and further developed to provide clinical decision support or even to make diagnoses autonomously.^vii,viii Not all of these products are commercially available or have regulatory approval.

Research suggests that optometrists are generally very positive about embracing the new technologies that can help them perform their daily clinical and business tasks though erring on the side of caution when it comes to trusting the results, particularly where clinical referrals are involved.^{ix x xi}The concept of trust in AI is crucial and a recurring theme in the attitudes of practitioners and patients and is also a core underpinning of the attempts to legislate.

The Royal College of Radiologists (RCR) have conducted research into public perceptions of AI in healthcare more generally^xii. The findings spark positive reinforcement that human oversight and interaction when delivering healthcare will remain relevant for years to come. However, they also highlight the need for professionals to be ahead of the curve with the technology. Patients trust clinicians to guide them and their views on using AI in health settings. If clinicians trust and understand the tech, the evidence suggests that patients will follow.

This very much underlines the importance of the optometrist-patient relationship and emphasises trust as an overarching message to the optometry profession. While other terminology shifts and weaves alongside the political will, the term ‘Trust’ remains steadfast in regulation and governance documents.^xiii If trust in the tools is there from both clinician and patient, it will lead to more widespread adoption, more net benefit and better patient outcomes.

Adopting and trusting AI-assisted tools should not be an administrative burden fraught with audit requirements, regulatory pitfalls and confusing guidance. The core message is that the majority of AI technologies likely to be in regular use in the short to medium term will not make autonomous clinical decisions. They will expertly synthesise data and offer advice to practitioners based on pre-determined parameters such as guidance or standards. Understanding product limitations and trusting your own judgement based on that knowledge will be essential. We hope that this resource will help you in your journey.

The AOP AI and IT Innovations Task Group

References

ⁱ A guide to optometrists for appraising and using artificial intelligence in clinical practice

ⁱⁱ A guide to optometrists for appraising and using artificial intelligence in clinical practice

ⁱⁱⁱ A guide to optometrists for appraising and using artificial intelligence in clinical practice

^iv A guide to optometrists for appraising and using artificial intelligence in clinical practice

^v A guide to optometrists for appraising and using artificial intelligence in clinical practice

^vi A guide to optometrists for appraising and using artificial intelligence in clinical practice

^vii A guide to optometrists for appraising and using artificial intelligence in clinical practice

^viii A guide to optometrists for appraising and using artificial intelligence in clinical practice

^ix Glasgow Caledonian University

^x Attitudes of optometrists towards artificial intelligence for the diagnosis of retinal disease: A cross‐sectional mail‐out survey