Using Retinal Photographs And AI, Scientists Detect Changes Associated With Alzheimer's

What if a simple eye exam could reveal warning signs of Alzheimer's disease almost a decade before symptoms appear? Scientists used artificial intelligence to analyze thousands of retinal photographs and discovered surprising clues hidden in the eyes.

Alzheimer's disease is the most common cause of dementia in the world, and currently, one of the greatest challenges in medicine is identifying who is at risk before symptoms appear. In recent decades, scientists have discovered that several factors, such as aging, high blood pressure, diabetes, smoking, depression, poor sleep quality, and lifestyle, are associated with the development of the disease.

Now, new research suggests that a simple photograph of the retina, the part of the eye responsible for capturing light, may contain important clues about these risk factors and perhaps even about future vulnerability to Alzheimer's.

The retina is considered an extension of the central nervous system. In other words, it is part of the same tissue that forms the brain. Therefore, many researchers believe that brain changes can leave visible marks in the eyes.

The big question was whether ordinary photographs of the fundus of the eye would be able to reveal signs related to factors that increase the risk of Alzheimer's disease. To answer this question, scientists turned to artificial intelligence.

Researchers used nearly 63,000 retinal photographs from over 44,000 participants in the UK Biobank, one of the world's largest biomedical databases. The images were analyzed using deep learning models, an advanced form of artificial intelligence capable of identifying extremely subtle patterns that often go unnoticed even by human experts.

The goal was to determine if artificial intelligence could predict, simply by observing the eyes, twelve factors known to be associated with the risk of Alzheimer's disease.

Among the factors analyzed were age, sex, body mass index, blood sugar levels, blood pressure, smoking, alcohol consumption, depression, sleep difficulties, socioeconomic status, and the age at which the person completed their studies. Instead of directly looking for signs of Alzheimer's, the scientists wanted to discover if the eyes carried information related to factors that increase the likelihood of developing the disease throughout life.

The results were surprising. Artificial intelligence managed to predict several of these factors with a significant level of accuracy. In many cases, its performance was superior to that of traditional methods that analyze only known anatomical measurements of the retina.

This suggests that there are complex patterns hidden in the images that reflect aspects of the overall health of the organism and the brain. In other words, the eyes seem to record information about biological aging, the state of blood vessels, and other processes that can influence brain health.

To understand how the system was arriving at its conclusions, the researchers also investigated which parts of the images attracted the most attention from the artificial intelligence. The analyses showed that two regions were particularly important: the optic nerve, responsible for transmitting visual information to the brain, and the blood vessels of the retina.

These structures are already known to reflect cardiovascular, metabolic, and neurological changes, which strengthens the idea that the retina can function as a kind of "window" to brain health.

In a further step, the scientists compared people who later developed Alzheimer's disease with similar individuals who remained disease-free. The photographs had been taken, on average, more than eight years before the onset of symptoms.

Interestingly, some of the patterns identified by artificial intelligence differed between the two groups, suggesting that certain retinal changes may arise many years before the clinical signs of the disease. Although the study did not develop a test capable of diagnosing Alzheimer's, the results indicate that the eyes may contain early clues about biological processes associated with the risk of the disease.

The authors conclude that simple photographs of the fundus of the eye can reveal valuable information about risk factors linked to Alzheimer's. It is not yet possible to diagnose the disease through this examination, but the research paves the way for the development of rapid, non-invasive, and low-cost screening tools.

In the future, a routine eye exam could help doctors identify people with greater vulnerability to Alzheimer's many years before the first symptoms appear, allowing for earlier preventative interventions.

READ MORE:

Prediction of Alzheimer’s disease risk factors from retinal images via deep learning: Development and validation of biologically relevant morphological associations in the UK Biobank

Seowung Leem, Yunchao Yang, Adam J. Woods, and Ruogu Fang

Journal of Alzheimer’s Disease. 2026; 0 (0).

DOI:10.1177/13872877261457650

Abstract: 

The systemic, metabolic, lifestyle factors have established associations with Alzheimer's disease (AD) through epidemiologic and AD-specific biomarker studies. Whether colored fundus photography (CFP) contains retinal structural signatures corresponding to these AD-related risk domains remains unclear. To determine whether deep learning (DL) models can predict 12 AD-related risk factors from CFP and to characterize the retinal structures underlying these predictions, thereby assessing whether CFP reflects pathways to AD vulnerability. Using 62,876 CFPs from 44,501 unique participants from the UK Biobank, DL models were trained to predict 12 factors linked to AD pathology or incidence: 6 categorical (sex, smoking, sleeplessness, economic status, alcohol use, depression) and 6 continuous (age, age at completing education, body mass index, systolic, diastolic blood pressure, HbA1c). Model performance, model saliency, and saliency-derived scores (CAM-Score) were evaluated and compared to retinal morphometry. The scores were also compared between incident-AD cases (average 8.55 years before onset) and matched controls. Predictive performance of DL ranged from AUROC between 0.5654 and 0.9480 for categorical factors and R2  between −0.0291 and 0.7620 for continuous factors, outperforming most of the morphometry-based machine learning models. Saliency-based score consistently highlighted biologically meaningful regions, particularly the optic nerve head and retinal vasculature. It also aligned with present morphometric variations. Several saliency-based scores differed significantly between incident AD and matched controls, suggesting potential overlap between retinal correlates of AD-related risk factors and preclinical AD-associated changes. CFP encodes retinal signatures linked to AD risk factors. Although not diagnostic, DL-derived retinal representations may uncover biologically meaningful risk-related structural changes mirroring the potential AD vulnerability.