Nov 21, 2024 — Michael Szkarlat, Senior Partner Development Director
Disclaimer: Medical information is not medical advice—read our disclaimer.
The September 2024 Eversight Academy webinar featured Travis Redd, MD, MPH, Assistant Professor of Ophthalmology at Oregon Health and Science University. As both a cornea specialist and someone whose research explores the application of emerging technologies to reduce the burden of corneal blindness domestically and abroad, Dr. Redd provided valuable insights into the current state and future potential of artificial intelligence in corneal care.
Understanding AI in medicine
Dr. Redd began by demystifying AI terminology, explaining the evolution from early rule-based systems to today's sophisticated deep learning approaches. He emphasized how deep learning's ability to process unstructured data, like medical images, makes it particularly valuable for ophthalmology applications.
Current applications in corneal care
Several promising areas of AI implementation were discussed, including:
- Infectious keratitis - AI models now performing on par with expert cornea specialists in distinguishing bacterial from fungal infections
- Keratoconus - Excellent performance in detecting manifest cases, with current limitations in subclinical detection
- Corneal transplantation - Successful detection of graft detachment and prediction of rebubbling needs
- Trachoma screening - Near-perfect performance in identifying inflammation, with potential for community-level screening programs
Implementation approaches
Dr. Redd outlined three main paradigms for AI implementation:
- Assistive - Supporting clinical decision-making with quantifiable measurements
- Augmentative - Working alongside clinicians to provide additional insights
- Autonomous - Independent screening and triage, similar to existing diabetic retinopathy systems
Barriers to clinical implementation
Key challenges include:
- The "black box" nature of AI decision-making
- Variable image quality in real-world settings
- Need for more diverse and representative datasets
- Unclear reimbursement mechanisms
- Limitations of supervised learning in medical contexts
Global health impact
During Q&A, Dr. Redd highlighted the potential transformative impact in lower and middle-income countries, particularly for:
- Infectious keratitis management
- Trachoma screening and treatment planning
- Community-level disease detection
Building trust
Dr. Redd emphasized several key factors for building trust in AI systems:
- Better communication of research results to non-technical audiences
- Testing with diverse, representative datasets
- Local validation of AI tools in target communities
- Transparent reporting of model performance
Future directions
To move AI from research to clinical practice, focus areas include:
- Development of more explainable AI models
- Creation of diverse data repositories
- Standardization of imaging data formats
- Implementation of federated learning approaches
- Greater involvement of clinicians in AI development
Dr. Redd concluded by encouraging interdisciplinary collaboration between clinicians, machine learning engineers, public health experts and patient advocates to bridge the gap between published research and practical implementation.
Thank you to our expert
Travis Redd, MD, MPH obtained a Bachelor of Science Degree in Anthropology from the University of Oregon before completing medical school, residency, and a Master's Degree in Public Health at Oregon Health & Science University and the Casey Eye Institute. He subsequently completed a fellowship in corneal and external disease at the Francis I Proctor Foundation & University of California San Francisco, where he developed particular expertise in the management of infectious and inflammatory diseases of the external eye.
Dr. Redd's current research interests include the application of emerging technologies to reduce the burden of corneal blindness domestically and abroad. He enjoys outdoor activities in the Pacific Northwest and spending time with his family.
Eversight's free webinars are a great way for you to connect, learn and train digitally with leading ophthalmologists and researchers from around the world. We invite you to RSVP for scheduled webinars and browse our recording library.
Have an idea for a future topic? Interested in receiving timely and relevant information from Eversight? We'd love to hear from you!