“The most successful applications will be those that help clinicians make better-informed decisions, improve efficiency, and ultimately enhance patient outcomes while preserving the physician–patient relationship”
As artificial intelligence (AI), data science, and real-world evidence become increasingly important across ophthalmology, their potential is expanding beyond image analysis alone. From identifying clinical phenotypes in heterogeneous conditions such as dry eye disease to supporting risk stratification, patient education, and more personalized follow-up care, these approaches may help clinicians interpret complex data and improve decision-making in everyday practice.
We spoke with Dr Germán Mejía-Salgado (Universidad Autónoma de Bucaramanga, Bucaramanga, Colombia) about how AI and real-world clinical data are being explored in ophthalmology, where these tools could add the most value beyond diagnostic imaging, and what will be needed for AI to become genuinely useful for ophthalmologists and their patients.
Don’t miss out on hearing about our latest peer-reviewed articles, expert opinions, conference news, podcasts and more.
How are you using or exploring AI, data science, or real-world evidence in ophthalmology?
My work has primarily focused on applying data science and machine learning methods to clinical ophthalmology research, rather than developing image-based diagnostic algorithms alone.
One area of particular interest has been dry eye disease, a condition characterized by substantial heterogeneity and often a poor correlation between symptoms and clinical signs. Using machine learning techniques, we have explored whether clinical phenotypes can be identified from routine clinical data and whether these phenotypes are associated with biomarkers of ocular surface inflammation, such as matrix metalloproteinase-9 (MMP-9).
More recently, I have also been involved in projects evaluating whether AI can help classify complex ocular surface disease presentations, predict inflammatory activity, and identify patient subgroups who may benefit from more personalized management strategies.
Beyond AI, I am particularly interested in the use of real-world clinical data. Ophthalmology generates large amounts of structured information during routine care, and there is tremendous potential to leverage these datasets to better understand disease behavior, treatment responses, and long-term outcomes in everyday clinical practice.
Where do you think AI can add the most value beyond image analysis, for example in clinical decision-making, patient education, or follow-up care?
While image analysis has understandably received most of the attention, I believe some of the greatest opportunities for AI lie in integrating and interpreting complex clinical information.
Many ophthalmic diseases require clinicians to synthesize symptoms, examination findings, imaging results, laboratory data, and treatment history simultaneously. AI systems could assist by identifying patterns that are difficult to recognize through conventional analysis and by supporting risk stratification and clinical decision-making.
I also see substantial potential in patient education. Many ophthalmic conditions are chronic and require long-term adherence to treatment. AI-powered educational tools could help explain diagnoses, improve patient understanding, and provide personalized information in a way that is scalable and accessible.
Another promising area is follow-up care. AI could help identify patients at higher risk of disease progression, prioritize appointments, monitor treatment responses, and potentially reduce unnecessary visits while ensuring that high-risk patients receive timely attention.
What needs to happen for AI tools to become genuinely useful for ophthalmologists and their patients in everyday clinical practice?
For AI to become truly useful in clinical practice, it must move beyond demonstrating technical accuracy and show meaningful clinical utility.
First, AI systems need rigorous validation across diverse populations and healthcare settings. Models that perform well in a single institution may not generalize to other patient populations or clinical environments.
Second, integration into clinical workflows is essential. Clinicians are unlikely to adopt tools that increase complexity or administrative burden. Successful AI solutions must be intuitive, efficient, and capable of delivering actionable insights at the point of care.
Third, transparency and trust are critical. Ophthalmologists need to understand why an AI system is making a recommendation and what evidence supports it. Explainable and interpretable models will likely achieve greater acceptance than black-box systems.
Finally, AI should be viewed as a tool that augments, rather than replaces, clinical judgment. The most successful applications will be those that help clinicians make better-informed decisions, improve efficiency, and ultimately enhance patient outcomes while preserving the physician–patient relationship.
Cite: Where can AI add value beyond image analysis? touchOPHTHALMOLOGY. 17 June 2026.
Editor: Nicola Cartridge, Director of Content
Acknowledgement: Dr Mejía-Salgado has nothing to disclose in relation to this article. No fees or funding were associated with its publication.
