EURETINA 2024: Martin S. Zinkernagel, MD, PhD, on the future of real-world AI use

At this year's congress of the European Society of Retina Specialists (EURETINA), the Eye Care Network spoke with Martin S. Zinkernagel, MD, PhD. Prof Zinkernagel is head of the Department of Ophthalmology at the University of Bern in Switzerland, and General Secretary of EURETINA.

Our conversation focused on artificial intelligence (AI) in practice, drawing comparisons between real-world therapeutic evidence and data regarding AI. Prof Zinkernagel also spoke about making the most of EURETINA in the weeks and months following the conference. Watch the full video for his advice.

Editor's note: The below transcript has been lightly edited for clarity.

Martin S. Zinkernagel, MD, PhD: Hi, I'm Martin Zinkernagel. I'm the head and full professor of the Department of Ophthalmology at the University of Bern in Switzerland. I'm also General Secretary of the EURETINA. It was a great meeting, very well attended. We had excellent speakers. Really covered the broad range of applications of AI and retina in practice. I think this is becoming more and more important, because it's kind of moving out of, you know, scientific content to really moving into medical practice nowadays. So we're really...on the doorstep of AI being in our clinical day routine.

I mean, to be honest, [AI is] already here, you know. It's already in everyday life. It's already in some devices. So the biggest change probably will be that it will assist us. I mean, it won't replace our conversations with patients, but it will assist us in making the diagnosis more accurate, to make less mistakes, not to miss something in the diagnostics we do. As with drugs, you know. We're talking a lot about new drugs, but what we really want to see is real-world evidence, and this is currently missing in AI. So here, you know, it can do this and that, and it can do everything, but what it really can do in clinical practice, this data and this experience is currently not really here. I think this is something that we need to address in the next years.

With therapeutics, you can just have like, you know, you can use the endpoints we have from clinical trials, like visual function or OCT data. I think here, really, what do we want from AI? We want to be more efficient. We want to have, you know, better diagnostics. So probably, in the end, maybe patient-related outcome measures will play a role, where we can really ask patients, you know, "Is the quality of treatment better when using AI?" Also the efficiency in clinic. What about waiting times? You know, how efficient is your clinic?
Also an important question is whether patients really trust in this, whether they want it. Because, I mean, with drugs and therapeutics, you know, if you have a better drug, patients usually want it, and it's really driven by the unmet need; whereas with AI, it's a more difficult question to answer properly.

I think it's really important, what we all want to do, is we want to use the data or the things we learned at EURETINA to use it in our clinical practice. And I think when you go back to the office on Monday or Tuesday, and you see cases you're not sure how to handle, you can think back to these main sessions, or these instruction courses. We covered so many things, I'm sure there will have been something you've heard that you can apply to this patient.