In this episode of Interventional Mindset, Dr. Joaquin O. De Rojas, a specialist in cornea, cataract, and refractive surgery, discusses the impact of artificial intelligence (AI) in ophthalmology, including its current applications and future potential.
Dr. De Rojas is the Medical Director of the US Eye platform in Florida and is based in Sarasota. Besides his clinical duties, he also engages in software development, focusing on AI tools to create software that supports both ophthalmology and the operational and business aspects of ophthalmology practices. Click here to demo three of these AI tools.
The role of AI in ophthalmology
Dr. De Rojas views AI as an upcoming "orchestration layer for ophthalmology and healthcare” that will help improve clinical decision-making, reduce or automate low-value tasks, and empower clinicians to take control of their workflows.
He believes that while AI will not replace ophthalmologists, it will augment clinical care and automate certain tasks, thereby extending the reach of doctors who learn to delegate and leverage AI effectively.
This shift enables physicians to devote more time to clinical judgment, surgery, and building trust with patients, while spending less time on charting, repetitive tasks, and administrative duties.
Types of AI used in ophthalmology practices
Three types of AI can have an impact on ophthalmology:
- Predictive AI or Machine Learning: Used to predict diagnoses by analyzing images or patterns, such as in studies using retinal optical coherence tomography (OCT).
- Generative AI: Large language models “LLMs,” such as ChatGPT, Claude, Gemini, or Grok, that comprehend meaning and can summarize notes, letters, and instructions, assist with research and learning, and communicate verbally.
- Agentic AI: LLMs that can reason and “do things.” They can be connected to systems such as electronic medical records (EMR), imaging, scheduling, patient outreach, and email to get work done.
3 pillars of AI's impact on ophthalmology
Dr. De Rojas outlined three pillars for how AI will change the job of the ophthalmologist:
- Clinical Diagnosis and Prediction: Focuses on better risk stratification, screening, diagnosis, and clinical prediction. Examples include:
- Retinopathy Screening: Autonomous AI diagnostic systems assist with retinopathy screening.
- Glaucoma: Algorithms can analyze fundus images and OCTs to help estimate progression risk.
- Keratoconus: Machine learning is used in conjunction with tomography and biomechanics to improve early detection of subtle forms of the disease and to estimate the risk of progression.
- IOL Formulas: Formulas such as the Kane, Hill RBF, Ladas, Pearl DGS, and ZEISS AI formula in Veracity are at least partially or fully AI-assisted. Dr. De Rojas personally has used AI to create a nomogram for LRIs/arcuate incisions, achieving over 90% accuracy within 0.5D of astigmatism.
- Automating and Augmenting Workflows: Utilizes generative and agentic AI to handle tasks, serving as an orchestrator for various administrative and clinical responsibilities.
- AI Scribes: AI can serve as a scribe during a conversation with a patient, create a structured ophthalmology note, and reduce physician burnout and improve well-being.
- Voice AI: Used for patient communication, scheduling, and call centers. Dr. De Rojas' practice is initiating an outbound calling system for their ASC that makes routine calls to patients with pre-surgery instructions and can connect them to a human if needed.
- Prior Authorizations and Coding Support: Companies are developing solutions to reduce the administrative burden of prior authorizations. For coding, AI can review the entire patient note to audit the coding and ensure compliance with ICD-10, CPT, and E&M codes.
- Developing a Builder Mindset: AI tools enable non-coders to create custom software, dashboards, and tools using plain English. Dr. De Rojas has used this technology to develop a personal assistant called Iris, which can respond to emails, generate reports, plan vacations, and assemble a morning briefing from listserv emails (Keranet, RSA, WCRSVS) by organizing them into a database. This system is controlled through voice commands to the AI assistant.
Important caveat on using AI
All AI workflows that involve patient information must be HIPAA-compliant, with business associate agreements (BAAs) and audit trails in place. Dr. De Rojas mentions he has used Google Vertex, which offers access to various LLMs in a fully HIPAA-compliant way.
Dr. De Rojas urges ophthalmologists to stay current with and utilize AI, or risk falling behind. He concludes that AI will not replace ophthalmologists, but it will replace certain tasks, transforming a doctor into a "10x ophthalmologist."
For more information about Dr. De Rojas’ AI projects, visit: https://derojas.ai.
