Reviewed by Konstantinos Balaskas, MD, FEBO, MRCophth
Ophthalmology, with its heavy reliance on imaging, is an innovator in the field of artificial intelligence (AI) in medicine.
While the opportunities for patients and healthcare professionals are great, barriers to full AI integration remain, including economic, ethical, and data privacy issues.
According to Konstantinos Balaskas, MD, FEBO, MRCOphth, retina expert at Moorfields Eye Hospital, London, UK, and director of the Moorfields Ophthalmic Reading Center and AI Analytics Hub, AI is a broad term.
Related: Cutting-edge neuro-ophthalmology: Combining artificial intelligence, eye tracking
“The type of AI that has generated a lot of enthusiasm in recent years is called ‘deep learning’,” he said. “This is a process by which software learns to perform certain tasks by processing large amounts of data.”
Deep learning is what made ophthalmology a pioneer in the field of implementing AI in medicine, as we increasingly depend on imaging tests to monitor our patients.
“Particularly in my sub-specialty of interest, medical retina, imaging tests such as optical coherence tomography (OCT) are performed very frequently and have provided the material to train, test, and then apply the systems to. AI decision support, ”noted Balaskas.
In the retina in particular, some of the most common causes of visual loss in the Western world, such as age-related macular degeneration (AMD) and diabetic retinopathy – require early detection, prompt initiation of treatment, and regular monitoring to preserve vision.
Balaskas said this is where AI decision support systems can help improve access to care and ensure optimal clinical outcomes for patients.
Balaskas cited the AI decision support system developed in collaboration between Moorfields Eye Hospital, where he is based, and Google DeepMind.
Related: Blockchain Technology Aims To Drive Big Data To ‘Train’ AI
“He is able to read OCT scans, interpret them, provide diagnosis and make management recommendations,” he said. “The other area where AI shows promise is the development of personalized treatment plans for patients by being able to predict their response to treatment and visual results over a period of time.”
When examining common conditions that threaten vision, such as AMD and diabetic retinopathy, Balaskas says AI’s decision support tools, once validated and once they’ve got the regulatory approval as medical devices, can help improve access to care.
“They can, for example, help community health practitioners to diagnose illnesses early,” he explained. “In the UK, where OCT scans are widely available in high street opticians’ offices, an AI tool would be particularly useful to help them interpret scans correctly and identify disease at an early stage.”
Likewise, in diabetic retinopathy, where patients require regular screening and follow-up, AI tools can dramatically increase the effectiveness of screening programs.
Balaskas pointed out that such applications already exist and can be particularly useful in diabetic retinopathy screening programs such as in underfunded health care settings.
“Other indications for the application of AI surveillance, such as AMD, are at an advanced stage of development but have not yet been implemented in real life,” he added.
Balaskas said there are problems with integrating AI into diagnostics and treatments for the retina.
Related: Integrating AI to Manage Diabetic Retinopathy in a Primary Care Setting
He noted that he has a personal academic interest in implementation science, which examines the gap between the development of a medical device such as an AI decision support tool and its deployment. in clinical practice.1
“The potential hurdles we need to overcome for the tool to be meaningfully deployed to improve outcomes for our patients go beyond testing and validation,” he said. “These include economic evaluations: how would such an automated decision support model affect the finances of a health system, so that it can offer value for money or achieve savings?
The next consideration relates to human factors, particularly how these AI-powered models of care are perceived and accepted by patients and practitioners.
What is the level of confidence in these technologies? What level of information and education for patients and the public is needed to build confidence in their use? Then there are considerations regarding the training and technical infrastructure to support these tools.
Balaskas notes that ethical and data privacy issues, as well as forensic considerations, are also important. Who is responsible for decisions made by an AI algorithm rather than a human? How do these tools affect the way healthcare professionals diagnose and treat disease?
Related: Deep Learning Algorithm Proven To Be Accurate For AMD Classification
“There is a phenomenon called automation bias, where practitioners are sometimes more likely to rely on the recommendation of the AI tool, perhaps even against their better judgment,” he said. .
Balaskas notes the problem of interpretability – that in many cases these AI tools are opaque in how they work.
“We do not fully understand how a specific recommendation is reached, whether it is a diagnosis or a management recommendation, and that the lack of transparency can exacerbate the medical, legal and ethical problems mentioned previously”, he stressed. “In summary, we have found that there are several hurdles to overcome before AI tools can be deployed in real life in a manner that is safe and that will improve clinical outcomes. “
Additionally, Balaskas said the lives of ophthalmologists may change in the future, and he has an optimistic view of AI in medical practice.
“Our field is becoming more and more complex and we have to deal with data from various sources when we assess our patients: data from many imaging modalities, genetic data and different types of omics, such as proteomics and emerging field of oculomics, where the characteristics of the eye exam can be indicative of systemic health issues, ”he said.
Related: Telemedicine Opens New Chapter in Eye Care
Balaskas also noted that data from home vision monitoring devices will become increasingly available.
However, Balaskas said it can be intimidating to make sense of all of this data in order to craft a personalized treatment plan for each patient.
“AI could become a very useful aid and, as described in the Topol Review on AI commissioned by Health Education England, provide time for patients and practitioners, giving them the opportunity to discuss and decide together on the treatment plan. optimal, informed by the processing of large complex data sources ”, he concluded.
Konstantinos Balaskas, MD, FEBO, MRCophth
e: [email protected]
Balaskas is interested in new ways of delivering eye care, including telemedicine, virtual clinics, remote monitoring, and AI decision support. He has no financial disclosure.
1. Campbell JP, Mathenge C, Cherwek H, et al; American Academy of Ophthalmology Artificial Intelligence Working Group. Artificial intelligence to reduce disparities in eye health: from concept to implementation. Transl Vis Sci Technol. 2021; 10 (3): 19. doi: 10.1167 / tvst.10.3.19