ARVO | Artificial Intelligence (AI)

Artificial Intelligence (AI)

All-Access Saturday — May 2 | Colorado Convention Center | Denver

Artificial Intelligence (AI) - Data science, real-world implementation, setting up an AI lab and AI-aided research

(all times are local MT)

This full-day course provides a comprehensive, practice-oriented introduction to artificial intelligence (AI) in ophthalmology, progressing from foundational concepts to real-world implementation. Designed for learners at varying levels of experience, the program begins with an accessible overview of deep learning principles, terminology, and evaluation metrics, enabling participants to confidently interpret and discuss AI technologies. It then advances to hands-on training in building and evaluating deep learning models for classification and segmentation tasks commonly used in vision research.

Lead organizer— Ching-Yu Cheng, MD, MPH, PhD

8 - 9am — Session 1: Fundamentals of deep learning model training — [BREAK @ 9am]

Session organizer— Aaron Y. Lee, MD, MSCI

Target audience—This is a beginner-level session. No prior experience with AI, programming, or tools is needed.

This session is for those who want to learn about artificial intelligence (AI) but are not interested in learning how to create it. The session will introduce the basics of AI in plain language. Didactic lectures will explain AI, its fundamental concepts, and related terminology. The last hour of the session will be devoted to an open Q&A with the audience. Come prepared to get the answers you want and need. Already have a question in mind? Want to ask a question anonymously? Send it to the organizers now via our question collection form:

Participants will be able to:

8am — Welcome and introductions of the workshop

Ching-Yu Cheng, MD, MPH, PhD, FARVO

Professor

Center for Innovation and Precision Eye Health, Yong Loo Lin School of Medicine, National University of Singapore

Singapore

Professor

Department of Ophthalmology and Visual Sciences,

Washington University School of Medicine in St. Louis

St. Louis, Mo.

8:05am — How to train your own model?

Assistant Professor

Medical University of Vienna

Vienna, Austria

8:25am — Metrics of AI Success

Professor

Department of Ophthalmology

University of Colorado

Denver, Colo.

8:45am — How to Set up an AI Lab?

8:55am — Q&A and Discussion

BREAK: 9 - 9:15am

9:15am - Noon — Session 2 (Hands-on lab): Training a deep learning model — [LUNCH @ Noon]

Session organizer/presenter — Aaron Y. Lee, MD, MSCI

Target audience—This is an advanced-level session. Attendees should be seeking to create AI models themselves and be familiar with basic programming skills.

This session will demonstrate how to train deep learning models for two broad tasks: classification and segmentation that is often used in research. We will cover training classification models using pretrained models, foundation models, metrics to appropriately compare results, and post-hoc visualization methods. We will also cover taking a dataset that was annotated using ImageJ or FIJI and then converting that data into training a deep learning model that can be used to automate the segmentation task.

Equipment requirements:

Attendees will leave this session with the ability to:

9:15am — Welcome and instructions

Professor

Department of Ophthalmology and Visual Sciences,

Washington University School of Medicine in St. Louis

St. Louis, Mo.

9:30am — Exercise 1: Training for classification

Medical Student

Johns Hopkins University School of Medicine

Baltimore, Md.

10:30am — Break

10:45am — Exercise 2: Training for segmentation

Medical Student

Johns Hopkins University School of Medicine

Baltimore, Md.

11:45am — Wrap-up and Q&A

LUNCH: Noon – 1pm

1 - 2:50pm — Session 3 (Hands-on lab): Use of large language models and AI agents for statistical analysis — [BREAK @ 2:50pm]

Session organizers — Quan Dong Nguyen, MD, MSc; the Society for Artificial Intelligence in Vision and Ophthalmology (SAIVO)

Target audience — This is an intermediate level session. Attendees should be familiar with the basics of AI and be seeking to use existing tools in their clinics or research.

This session introduces large language models and novel concept of AI agents, reviews their strengths and limitations, and provides hands-on instruction for how to use them and employ AI agents in your research projects.

Equipment requirements:

Attendees will leave this session with the ability to:

1pm — Welcome and introduction

Byers Eye Institute, Stanford University School of Medicine

Stanford, Calif.

Assistant Professor of Ophthalmology

Byers Eye Institute, Stanford University School of Medicine

USA

1:05pm — Exercise 1: LLMs for literature review & knowledge synthesis

Senior Investigator

National Library of Medicine (NLM)

National Institutes of Health (NIH)

USA

Assistant Professor of Biomedical Informatics and Data Science

Yale University

USA

1:30pm — Exercise 2: Using ChatGPT for data extraction from screenshots and basic statistical analysis (regression, plotting)

Clinical Instructor

Byers Eye Institute, Stanford University School of Medicine

Stanford, Calif.

Research Scholar

Byers Eye Institute, Stanford University School of Medicine

Stanford, Calif.

1:55pm — Exercise 3: AI agents for statistical analysis in research and clinical studies

Associate Professor

Department of Biomedical, Surgical and Dental Sciences, University of Milan

Milan, Italy

Assistant Professor

Department of Ophthalmology and Visual Sciences, University of Michigan Medical School

Ann Arbor, Mich.

2:20pm — Exercise 4: Advanced data analysis using LLMs (correlated data, genomic and proteomic data management)

Associate Professor

Lee Kong Chian School of Medicine, Nanyang

Singapore

Solutions Architect, Data & AI Foundation, Scania AB

Sweden

2:45pm — Wrap-up and Q&A

Assistant Professor

Department of Ophthalmology and Visual Sciences, University of Wisconsin–Madison

Madison, Wisc.

Exercise Facilitators

BREAK: 2:50 – 3pm

3 - 5pm — Session 4: Co-design of innovations in AI and digital eye health – Hands-on implementation science lab

Session organizer—Lisa Zhuoting Zhu, MD, PhD

Target audience — This is a beginner session suitable for healthcare professionals, researchers, or technology implementers who understand basic AI concepts and seek practical strategies to co-design strategies to support implementation of AI in clinical settings with stakeholders.

This hands-on session introduces practical frameworks and toolkits for co-designing AI solutions with end-users and stakeholders, including clinicians, patients, administrators, and IT. Participants will learn about processes to identify stakeholder needs and barriers to AI adoption, translate workflow pain points into co-designed strategies to support AI implementation and prototype a clinical AI implementation solution conscious of real-world constraints.

Equipment requirements:

Attendees will leave this session with the ability to:

3pm — Welcome and introduction

Ophthalmic Epidemiology Unit, Centre for Eye Research Australia

East Melbourne, Australia

Professor

The School of Optometry & Vision Science, University of New South Wales

Kensington, Australia

3:05pm — Consolidated framework for implementation research to implement AI for DR screening

Assistant Professor

Department of Ophthalmology and Visual Sciences, University of Wisconsin–Madison

Madison, Wisc.

3:15pm — Clinical trial design for screening diabetic macular edema using AI: A case study

Professor

Department of Ophthalmology and Visual Sciences, Faculty of Medicine, The Chinese University of Hong Kong

Hong Kong, China

3:25pm — Co-design for AI in healthcare—Principles, pitfalls, stakeholder roles & toolkits

3:40pm — Exercise 1(Hands-on, Groups) : Gathering perspectives (stakeholder touch points & barriers)

Wenyi Hu, MD, PhD

Jessie Huang-Lung, PhD

4:15pm — Exercise 2(Hands-on, Groups) : Prototype an AI implementation plan (integration, training, workflow impact, evaluation metrics)

Wenyi Hu, MD, PhD

Jessie Huang-Lung, PhD

4:50pm — Wrap-up: Key takeaways, resources and next steps

* Speakers and topics are subject to change.