Global Ophthalmology, Practice Development, Cataract, Artificial Intelligence
Advancing Vision Research
National Eye Institute initiatives encourage AI-based innovations and data interoperability.


Cheryl Guttman Krader
Published: Monday, March 3, 2025
“ I am excited about all of these projects and being able to work with everyone in our community to make the scientific advances that will help us take better care of patients. “
Established in 1968 as a branch of the National Institutes of Health in the United States, the National Eye Institute (NEI) manages national efforts in vision science, directing and funding vision research around the country with a budget approaching $900 million in 2024.
Speaking to retina specialists at AAO 2024, NEI director Michael F Chiang MD highlighted the agency’s support for two initiatives designed to leverage the power of big data—the development of artificial intelligence (AI)-based imaging applications and improving clinical data interoperability.
“In 2019, Dr Eric Topol tweeted that ophthalmology was the field driving AI innovation in medicine,” Dr Chiang said. “Yet on the current list of FDA-approved AI-based diagnostics available for different specialties, ophthalmology ranks only sixth.”
This observation begs the question of how to get more AI-based innovation into real-world practice. Building larger data sets is one component, and the NIH Common Fund’s Bridge2AI offers an opportunity to stimulate the effort.
The goal of Bridge2AI is to generate diverse flagship data sets across multiple disciplines that can be used for knowledge discovery across domains of medicine. Its $130 million of funding spans four years, and among the recipients is a team undertaking four data generation projects for diabetes insights (aireadi.org).
“This is really exciting and a huge opportunity for the ophthalmology community,” Dr Chiang said.
The NEI is also working to stimulate oculomics research through a three-year initiative supported by the NIH Common Fund Venture Initiative Program, which aims to develop novel ocular imaging methods coupled with AI models to diagnose and predict systemic disease. The funds—totalling up to $14.4 million—have been awarded to three recently announced groups.
The awardees are:
1) Amani Fawzi MD (Northwestern University) and Stephen Burns PhD (Indiana University), who are developing cellular level vascular oculomics technology to monitor systemic vascular health in real time;
2) Vivek Jay Srinivasan PhD and Laura Balcer MD (both of New York University), who are developing next-generation OCT to detect biomarkers for Alzheimer disease and Parkinson disease; and
3) Jianhua Wang MD and Liang Liang PhD (both of University of Miami) and Yuhua Zheng MD (Doheny Eye Institute), whose focus is creating novel imaging techniques to measure blood flow in retinal capillaries to detect cerebral small vessel disease and dementia.
Clinical data interoperability
Dr Chiang explained the NEI focus on clinical data interoperability recognises the importance of data sharing and collaboration in improving clinical care in the twenty-first century and the existing challenges from very inconsistent data formatting in different electronic health record systems.
“Even two systems from the same vendor may not be able to naturally share data with each other,” he said. “Standardisation of data and integration into a multimodal data ecosystem would have extremely powerful ramifications for the vision community.”
Toward this end, the NEI is collaborating with the American Academy of Ophthalmology to build a common data model for ophthalmology that will allow for the creation of standard cohorts and federated data analyses. Data sharing through federated data networks has emerged as an alternative to data centralisation, the method offers several advantages.
“Data centralisation—in which all sites send data to a single repository—has benefits in that it provides access at a single point,” Dr Chiang noted. “It is simple and intuitive. However, sending data from site to site involves extensive legal, privacy, and security agreements that can take years to set up. In addition, transmission of the large image files can be impractical logistically.”
In a federated data network, the data is retained and analysed at the site where it is collected, and only the analysis is shared. This approach eliminates the need for data use agreements but does require data harmonisation.
As proof of the feasibility and power of the federated data network approach, Dr Chiang described a retrospective cohort study led by Dr Cindy Cai that aimed to characterise the incidence of kidney disease associated with intravitreal anti-VEGF exposure and compare the risk between the various biologics.1 Using the Observational Health Data Sciences and Informatics (OHDSI) network, data were collected from 12 international databases, encompassing 485 million patients. Within just nine weeks, the researchers built a cohort of 6.1 million patients with blinding diseases and identified approximately 240,000 patients meeting the criteria for inclusion in the analyses.
To stimulate development in this area, the NEI created the OHDSI Initiative for Eye Care and Ocular Imaging Challenge. This competition seeks proposals for research questions that can be answered using clinical and/or image data in a federated vision network. It is awarding four $250,000 prizes. More information is available on the challenge website.
“I am excited about all of these projects and being able to work with everyone in our community to make the scientific advances that will help us take better care of patients,” Dr Chiang concluded.
Dr Chiang spoke at AAO 2024 in Chicago, US.
Michael F Chiang MD is Director of the National Eye Institute, Bethesda, Maryland, US. michael.chiang@nei.gov
1. Cai CX, Nishimura A, Bowring MG, et al. “Similar risk of kidney failure among patients with blinding diseases who receive ranibizumab, aflibercept, and bevacizumab: An observational health data sciences and informatics network study,” Ophthalmol Retina, 2024; 8(8): 733–743.
Tags: data, imaging systems, streamline workflow, digital systems, image guidance, digital environment, artificial intelligence, AI, AI applications, deep learning, data sharing, data sets, database, Michael F Chiang, OHDSI, NEI, ocular imaging, early diagnosis, Bridge2AI, research, data interoperability, AI models, flagship data sets, data centralisation
Latest Articles
Committing to a More Equitable Ophthalmology
Charity Committee enhances ESCRS’s reputation as a socially responsible society.
Virtual Community Unites Ukrainian Eye Specialists
Microlearning platform proves vital to saving eyes and lives in areas of conflict.
ESCRS Survey Shifts Industry Dial
Annual survey reveals trends and issues regarding the digital OR.
Always Measure the Epithelium!
Epithelium thickness mapping key to better refractive outcomes.
Advancing Vision Research
National Eye Institute initiatives encourage AI-based innovations and data interoperability.
Best Candidates for Phakic EDOF IOLs?
Patient profile suggests limits for phakic EDOF approach.
Knowing Iris Repair: Iris Repair with Prosthetic Devices
New devices on the market expand available options when iris tissue is insufficient.
Targeting the Kallikrein-Kinin System to Treat DME
Novel oral plasma kallikrein inhibitor shows promise in early clinical trial.
Faricimab VOYAGER Study
Real-world experience with faricimab for treatment-naïve nAMD and DME is underway.