AI, Ophthalmologist Friend or Foe? A Human Couple Takes Sides on the Question | AAO 2024

The topic of artificial intelligence (AI) has started more than a few arguments when it comes to how it might best fit into healthcare, and the 2024 annual meeting of the American Academy of Ophthalmology (AAO) held in Chicago was no exception.

Cecilia S. Lee, M.D., M.S.

During a session on opportunities versus threats, Cecilia S. Lee, M.D., M.S., took the opportunity stance when it came to AI while her husband, Aaron Y. Lee, M.D., M.S.C.I., took the opposing “it’s a threat” position. The Lees are on the faculty at the University of Washington in Seattle and lead a lab that researcher the uses of big data and AI in healthcare.

Cecilia presented first, mentioning the fun of having a "marital argument" on stage in front of an AAO audience. She ended her talk with examples of training AI that included photos of the couple’s daughter, which Aaron admitted it was hard to argue against, but not hard enough that he would fold. The dual presentations were labeled as from the Lee family, offering a very human element to an otherwise dry topic.

To that end, Cecilia highlighted the ways AI could help her provide more individual-focused care by taking the focus off the drudgery of completing fields to satisfy the demands of the of the electronic medical record.

“If we could spend more time face-to-face — listening to our patients instead of focusing on charting — we could provide the real care they deserve," she said. “While I’m having a conversation, it could provide real-time literature reviews relevant to our discussion.”

Her presentation highlighted three areas of opportunity for AI: clinical care, access to care and new research. In the future, Cecilia envisioned a world where AI could complete pre-charting work, fill out notes and provide visit summaries. Additionally, there could be less time spent outside appointments filling out paperwork, prior authorizations and other documentation.

With AI-driven treatment algorithms in place, there would be a better chance that high-quality care could reach all areas of the world, she added, and in the U.S., this range would include remote places and low-income areas. Cecilia also mentioned the possibility of better referral network if the algorithms were applied in primary care, shortening the path to need ophthalmic care.

For clinical research, AI has the potential to shift the paradigm, as it is applied across various settings. There have been examples already shown for AI to expedite novel molecule discovery, with the potential to further expand this. Additionally, different applications of the technology could lead to accelerated clinical trials and more personalized medicine options.

“AI can digest vast amounts of multimodal data and transform how we approach patient care," said Cecilia. “With the combination of the data we gather, we can move towards meaningful personalized medicine.”

She compared AI to having a child: At first, parenthood can be scary and then a far less frightening, joyful reality takes shape. To win the audience over, she showed adorable photos of her and Aaron’s daughter.

"The reality is that AI represents a transformative force," she said. "It is adaptable and trainable; the more data we provide, the better it becomes. Instead of resisting, we should embrace that AI is here to stay. With rapid advancements, we might even find enjoyment in this journey."

Aaron Y. Lee, M.D., M.S.C.I.

Aaron began his rebuttal with areas of agreement with his wife and added some details to how eye exams could be more automated. He also demonstrated a tactic for eye tracking and other methods that AI could mimic, many of which were procedures carried out in a typical eye exam. Following this demonstrated, Aaron showed an example of ChatGPT easily passing ophthalmology exam questions, even complicated ones. “The future looks promising, and we are in agreement that our goal is to provide better care for our patients,” said Aaron.

But he also cautioned against blindly following what AI offers up and said human judgment remains invaluable.

After showing the growing use cases for AI, Aaron then presented a somewhat dystopian picture of the future: Patients would enter the office on an assembly line, be seen by the imaging robots and an AI model would analyze their results. They would then either go home or be sent for a robot surgeon to correct the issue.

Following this example, he further emphasized the threat to the profession of ophthalmology by highlighting quotes from AI thought leaders.

“If we have all that information, it’s not too far-fetched to believe that we can have all the physical exam elements available for automation,” said Aaron. “What exactly does the ophthalmologist add? If you ask scientists, they might say nothing.”

To reset the mood in the room, Aaron switched back to talking up the benefits of AI and that current uses of it still have depend on having human discernment in the loop. He mentioned the cases when people driven into disaster depending on GPS-generated directions.

To avoid the threat of AI ophthalmology would need to take caution and noted that care is better when clinicians integrate with technology and overrule when necessary.

To keep AI from being threat, Aaron said the field of ophthalmology should take a cautious approach. Healthcare is better when clinicians integrate with technology and overrule it when necessary, he said.