Digital Medicine - 6 steps for a better future

What are unmet needs in ophthalmic care that AI can address? How can research in AI be democratised? What are common myths about AI? These were some of the questions asked in the Digital Medicine session on Saturday, 17 September, in Milan.

Farhad Hafezi MD, PhD, FARVO, Consultant Ophthalmic Surgeon in Dietikon, Switzerland, suggested the development of Smart Mobile Affordable Reliable Technology (SMART) for use in low- and middle-income countries (LMICs) will change care for patients. He used keratoconus as an example of where the technology could be applied.

“Now that we have a treatment modality to stop the disease— first comes detection,” Prof Hafezi said. Prof Hafezi investigated his home country, high-income Switzerland, asking how many ophthalmologists had access to the most basic placido type topographer. He found only 60% of ophthalmologists had direct access to a topographer for their patients. And in other countries, one can assume that access is likely only worse. This lack of access risks later diagnosis of keratoconus, a potentially blinding corneal ectasia that early screening can detect and allow for treatment with cross-linking.

Keratoconus screening can be carried out without a topographer using a smartphone-based keratographer (SBK), which is currently at prototype stage. The device, featuring a lens and a forehead mount attached to a smartphone, will be more affordable than available topographers, with an estimated future cost of $1,500. It can be operated freehand or mounted on a slit lamp and gives a readout similar to a conventional topography machine.

“It’s a screening tool, not for making decisions about surgery,” Prof Hafezi said.

WHAT ARE COMMON MYTHS IN AI?

Sunny Virmani, Google Product Manager at Mountain View, United States, explored important misconceptions in health applications of AI. One is that more data is all you need for a better model. Not so straightforward.

“You need not just quantity, but quality,” Virmani said. “Another myth: an accurate model is all you need for a useful product. A product must be useable in real-world settings.”

Did you know what can be revealed by a photo of your eye? Not a fundus photo, an external photograph? Virmani presented exciting data from an article published by Babenko (et al.) in a 2022 issue of Nature Biomedical Engineering.

“External eye photos for detecting disease,” were investigated in the study. It included a large sample: a training set of more than 140,000 patients with diabetes and a validation sample of more than 40,000 patients with diabetes. A fundus camera took the external photographs to train a deep-learning model that went on to accurately predict diabetic retinopathy, diabetic macular oedema, and poor glucose control. The prediction had a better performance than logistic regression models using demographic and medical history data. It is unknown whether other cameras will replicate the results, Virmani said.

If ophthalmologists want to research in the AI space, will they need to learn how to code? Short answer, no. Pearse Keane, Consultant Ophthalmologist in London, United Kingdom, showed alternative, democratic approaches.

“Clinicians will play an important role in the next phase,” Keane said. He believes upcoming developments in AI will be led by those who have the best ideas for clinical applications. And there is a need to make ophthalmology services as throughput and lean as possible. “Nearly 10% of all clinical appointments in the NHS are for eyes. We have standing room only a lot of the time in our clinics,” Keane said. “AI can play at least some role in mitigating these challenges.

“What I’m excited about, though, is code-free machine learning,” Keane continued. There are cloud-based platforms that allow clinicians to upload photos, which can train an AI model. He suggests using these platforms to get to a proof-of-concept stage and that a subset of models may progress to clinical application. While in the past, there was concern among some clinicians about AI replacing doctors, currently, AI is becoming more and more accessible to clinicians to advance their own research ideas and potential clinical applications.

Tags: 40th Congress of the ESCRS