Artificial intelligence (AI) is at the forefront of current eye health and treatment research. Session Two of the 2024 Focus on Eye Health Summit: Being Seen and Heard, in the form of a Round Table discussion, featured Dr. Sally Baxter as the moderator, along with panelists Dr. Michael Abramoff, Dr. Cecilia Lee, and Dr. Benjamin Xu, and highlighted the opportunities and considerations of using AI in eye care, with particular emphasis on the patient experience.*

Advances in AI for Eye Health

Dr. Baxter highlighted the exciting advancements in AI for eye care over the last decade. These advancements include AI’s ability to diagnose diseases from images and predict disease risks. 

“I think this is a very, very exciting time to be working in AI in eye care,” she said, pointing pointed out that AI has the potential to combine various data sources such as: 

• Clinical records, 
• Imaging data, and 
• Wearable devices to improve patient outcomes. 

The discussion centered on the practical applications of AI in eye care, and the benefits of involving various stakeholders like patients, clinicians, and policymakers. 

Challenges and Opportunities in AI Implementation

In addition to practical applications, the panelists also discussed the challenges of implementing AI in eye care, particularly in screening and diagnosing diseases like glaucoma and diabetic retinopathy, and underscored the need for patient-centered approaches in AI implementation.

“The importance of glaucoma screening can’t be understated… early detection is really key,” Xu emphasized. Unfortunately, he said, glaucoma is particularly challenging for AI because “there aren’t good standardized definitions for the disease, especially when it’s milder in its course.” 

This lack of clear definitions makes it difficult to develop AI models that can reliably detect the disease across diverse patient populations, he said. Different populations exhibit different signs of glaucoma, which can lead to AI models performing inconsistently. 

“We need a good data set to train with that is very representative of the population,” Lee said, explaining that many existing data sets are skewed toward certain demographics or collected from specific regions, which can introduce bias into AI models. 

Dr. Abramoff added that another critical challenge is ensuring that AI models are both ethically sound and clinically effective. He also highlighted the importance of patient consent and data privacy, particularly in autonomous AI systems. “Patients should always be informed that they are getting a diagnosis made by an AI,” he said.

Takeaways

The round table highlighted the significant potential of AI in transforming eye care, from early disease detection to improving patient education and clinical workflows. The panelists stressed the importance of involving patients and patient advocacy groups in the development and implementation of AI technologies to ensure these innovations lead to better health outcomes and equity in care.

When it comes down to it, Xu concluded, “It’s not just about developing the best performing algorithm, but producing a tool that improves the patient experience and outcomes in a way that can be adopted by providers and patients.”

*Prevent Blindness. (2024, July 12). The Many Eyes in Artificial Intelligence [Video file]. https://www.youtube.com/watch?v=K7NJJZ6SkqE&t=2s 

Responsum Health closely vets all sources to ensure that we always provide you with high-quality, reliable information. We do not, however, endorse or recommend any specific providers, treatments, or products, and the use of a given source does not imply an endorsement of any provider, treatment, medication, or procedure discussed within.