Leanne Labriola, DO, MBA, a retina specialist with a practice near Pittsburgh, diagnosed a woman who complained of ocular pain with multiple sclerosis. When a male patient came to her complaining of having trouble reading, she diagnosed him with Parkinson’s disease. And she diagnosed diabetes in several patients who complained of floaters in their vision.

Those cases led Labriola to pursue her own research into oculomics, the study of biomarkers in the eye that are signs of systemic disease. She has been working on a platform to test the tear fluid for protein biomarkers of conditions such as diabetes, cardiovascular disease, and kidney trouble.

“Oculomics has emerged with a very narrow definition that is most frequently associated with it, which is making systemic diagnosis based on visual changes in retina cells and nerve tissue,” Labriola said. “It’s fundamentally because there is a lot of money behind artificial intelligence and screening for disease with those cameras.”

However, she said the definition of oculomics was meant to be broader: “It was meant to be any way of making a systemic diagnosis and having ocular fluid testing be part of that is fundamentally part of oculomics but not as often mentioned.”

What Exactly Is Oculomics?

The concept of oculomics stems from two root words: ocular for the eye and omics which in biology means collective study of molecules and their functions.

The predominant research in the field for a decade has focused on retinal imaging, namely retinal photos, to diagnose and monitor systemic diseases such as diabetes. Oculomics has gained traction as research has focused on using retinal imaging hand-in-hand with AI.

“The current state of oculomics is truly transformative,” said Malvina Eydelman, MD, a former FDA official who oversaw ophthalmic devices and now chief executive of the Collaborative Community on Ophthalmic Innovation, a consortium of individuals involved in drug and device development. “The retina offers a uniquely accessible, noninvasive way to visualize the microvasculature and neural tissue, effectively serving as a window into systemic health.”

Advances in AI and imaging are accelerating potential applications of oculomics in neurodegenerative, cerebrovascular, cardiovascular, renal, and metabolic diseases, as well as biological aging, she said. “These innovations create opportunities for earlier detection, more timely intervention and ultimately prevention of downstream impairment,” Eydelman said.

The FDA had already approved imaging devices that can help detect and monitor diabetic eye disease, but oculomics takes imaging beyond that.

“The bottom line in oculomics is the idea of bridging toward personalized medicine, reducing costs, being able to actually focus on the eye, and perhaps lead to diagnosis and impact management of systemic diseases like cardiovascular disease,” said Alon Harris, MD, PhD, director of the artificial intelligence in ophthalmology program at Mount Sinai Hospital in New York City.

The veins and arteries in the back of the eye, which an optometrist or ophthalmologist can see when they perform an exam with a slit-lamp or a binocular indirect ophthalmoscope, are “part of the cardiovascular system,” Harris said. “The eye is definitely sensitive to any changes that we have in the cardiovascular system.”

Multiple researchers have focused on capturing images of the back of the eye, using either retinal photography or optical coherence tomography (OCT). An international team of researchers reported that three major advances have accelerated the field of oculomics: the widespread clinical use of noninvasive ophthalmic imaging, the availability of large studies to analyze links between ocular pathology and systemic disease, and advances with AI.

Harris was the lone ophthalmologist who participated in the Eye2Heart study, which evaluated a mathematical model to evaluate the interactions between the heart and retinal circulation. The model was able to reproduce cardiovascular parameters, such as stroke volume and cardiac output, and ocular hemodynamics, such as blood flow in the retina. The model also found a potential link between intraocular pressure and left ventricular compliance on both local ocular and systemic circulation.

Harris participated in another study, published this year, that used AI with retinal images to measure blood pressure. Cuff measurements “can be skewed,” he said. “This study basically found that if you use this AI model, it matched and even slightly surpassed the accuracy of traditional cuff measurements.” The study used doppler technology to measure blood flow in the eye while performing simultaneous echocardiogram, he said.

In China, researchers reported changes in the microvasculature of the retina on OCT can serve as markers of cardiovascular disease in people with diabetes, and a research team at Northwestern University in Chicago found patches of retinal damage caused by poor blood flow on OCT can indicate the extent of coronary artery calcium.

Other research has gone beyond cardiovascular disease. A group in Tokyo and London developed an AI model to estimate biological age and potentially disease risk from retinal photographs. At the University of Calgary in Calgary, Alberta, Canada, a team is using AI to analyze ocular biomarkers of neurologic disorders, a “fledgling discipline” they are calling “neuro-oculomics.” Harris said his research group is also evaluating the association between changes in blood vessels in the eye, such as capillary density and retinal perfusion, and other conditions, including sickle cell anemia.

The Potential of AI

Eydelman said that AI is poised to transform oculomics into a tool that can be widely accessible and scalable rather than a niche, specialist-driven capability. “By standardizing image capture and interpretation, reducing dependence on expert graders, and enabling deployment across diverse care settings, AI is already extending the reach of oculomics well beyond traditional ophthalmology environments,” she said.

Automated or guided image capture, including by robots, can help to minimize operator error, she said. “Workflow protocols in development could also help primary care providers and other clinicians who are not optometrists or ophthalmologists send results to the appropriate provider.”

“That progress alone will not be enough,” Eydelman added. “The impact of AI in oculomics will ultimately depend on whether outputs are clinically interpretable, rigorously validated in the intended population, seamlessly integrated into clinical workflows, and clearly linked to actionable decisions and measurable patient benefit.”

Harris foresees “a big shift” to where retinal cameras will be in the offices of general practitioners and other medical offices to broaden access to the devices and paired with AI algorithms to interpret the images. “We’re not there quite yet, but with the buildup in data and more scientific rigor, there could be a very accurate diagnoses toward many of these systemic diseases and central nervous system changes,” he said.

His group is performing two levels of studies of oculomics using AI. One is to data mine the relationship or presence of cardiovascular disease from images captured in ophthalmology clinics and building AI algorithms that can analyze the association between the abnormalities from the existing data sets from patient records. The second, which Harris said is even more accurate, is to build prediction models using AI that stem from prospective studies. These developments have the potential to make imaging “simpler and user friendly,” he said.

Many More Tears

Capturing quality images of the back of the eye has some limitations, Labriola said. “The patient has to be cooperative. You have to have clear media. You have to have access to the equipment, and then it does need the AI or physician interpretation,” she said.

Labriola’s experience in the clinic led her to study the tear fluid, a concept she called a liquid biopsy. Her group published a study in 2024 in which they identified 500 to 800 proteins in.5 microliters of tear fluid. A 2022 study identified increased expression of three biomarkers for breast cancer in samples of tear fluid taken from women with breast cancer.

“A tear drop has over 1300 proteins or biomarkers, but we have shown that about 800 of them could be reproducibly and reliably measured,” Labriola told Medscape Medical News. The tears are an “ultrafiltrate of the blood plasma,” she said. “It is really just the blood that gets filtered through the lacrimal glands to be a very clean and consistent fluid.” Urine is acidic, mucus has cellular debris, and plasma has a high level of albumin, all of which can interfere with testing and none of which affects the tear fluid, she added.

Tears have long been used as a biomarker for dry eye, but Labriola said she wanted to study the fluid of healthy individuals to see if it could contain proteins associated with other conditions. It could potentially replace an invasive test, she said. “It does seem that the tear fluid has biomarkers similar to what you would find in the cerebrospinal fluid.”

As with models that use retinal images to identify biomarkers, AI is a key component of analyzing the tear film, Labriola said. “I believe that there’s even a greater role for AI to be able to take all of that information together to give the user the final output,” she said. “If we can do this the right way, we could probably detect disease earlier in more patients and find it before that end organ disease sets in.”

Labriola reported having financial relationships with AbbVie, Genentech/Roche, Regeneron Pharmaceuticals and Belite Bio, and being chief executive and founder of Insight Technology and cofounder of Visionary Products.

Eydelman reported being president and chief executive of Salution Nexus.

Harris reported having financial relationships with AdOM, Qlaris, Cipla, SlitLed, and AEYE Health.

Richard Mark Kirkner is a medical journalist based in Philadelphia.