AI-Powered Retinal Scans May Unlock Early Detection of Heart and Brain Disease, Two Landmark Studies Show
核心洞察
Two independent studies published June 16 demonstrate that AI analysis of routine eye scans can predict risk factors for cardiovascular and neurological diseases, including Alzheimer's, heart failure (搜索), and Parkinson's.
University of Manchester researchers developed "Ret-AAE" using UK Biobank data from over 68,000 people, linking retinal features to genes involved in neurodegenerative pathways and cardiovascular function.
University of Florida scientists showed that machine learning analysis of retinal photographs from over 40,000 patients accurately predicted Alzheimer's risk factors such as blood pressure, smoking, alcohol use, and insomnia.
Two landmark studies published on June 16 have independently demonstrated that artificial intelligence applied to routine eye scans could transform how clinicians detect early signs of heart and brain disease, potentially years before symptoms emerge. The research, conducted by teams at the University of Manchester and the University of Florida, points toward a future where simple, widely available retinal imaging could serve as a powerful population-level screening tool.
Manchester Study Links Eye Features to Whole-Body Health
Researchers at the University of Manchester developed an AI tool called "Ret-AAE" to explore connections between the eye and systemic disease risk using UK Biobank data from over 68,000 volunteers. The study, published in Nature Cardiovascular Research, found that the associations between the eye and body are remarkably broad.
The team identified that the appearance of the eye is linked to risk of heart failure (搜索), high blood pressure, heart attack, Parkinson's disease (搜索), dementia (搜索), and more. Two widely available scan types were analyzed: 3D optical coherence tomography (OCT) scans of the inner retinal lining, and colour fundus photographs (CFP) of the back of the eye. Both are routinely captured by high-street optometrists across the UK, with several million scans already performed annually.
"The eye can reveal a remarkably broad picture of whole-body health, offering a way to identify those at risk of heart and brain disease before they occur," said lead author Dr. Tom Julian, Medical Research Council Clinical Research Training Fellow at the University of Manchester and Manchester Royal Eye Hospital.
The two scan types revealed complementary signals: OCT was more strongly linked to neurological traits, while CFP showed broader associations with cardiovascular traits. Genetic analyses demonstrated that eye features are linked to genes involved in neurodegenerative disease pathways, including those related to Parkinson's disease (搜索), dementia (搜索), and broader neurodegeneration. Physiological analyses connected eye features to blood pressure, blood vessel stiffness, and heart function.
Radiomic analyses — which convert medical images into measurable data — showed associations between eye features and brain size, as well as subtle changes in brain tissue structure detected via MRI. The researchers also identified connections between retinal features and fat-related molecules in the blood, potentially linking the eye to general metabolic health.
Some patterns detected by the AI were influenced by cataracts or natural differences in eye colour, indicating that researchers may need to account for age and ethnicity when analyzing images.
"While more work is needed before these tests could arrive on the high street, we hope and believe that routine eye tests will one day be used as part of health screening for disease prevention," said Dr. Panos Sergouniotis, Wellcome Clinician Scientist and Senior Lecturer at the University of Manchester, who oversaw the work alongside Professor Alejandro Frangi.
Florida Study Predicts Alzheimer's Risk Factors from Retinal Photographs
In a complementary study published in the Journal of Alzheimer's Disease (搜索), researchers led by Ruogu Fang, Ph.D., professor of biomedical engineering at the University of Florida, demonstrated that machine learning analysis of retinal photographs from more than 40,000 patients in a UK-based databank could accurately predict many of the most common risk factors associated with developing Alzheimer's disease.
"We know that Alzheimer's disease (搜索) develops over decades, but most of the diagnostic tools focus on late stage pathology when it is too late to intervene," Fang said. "By looking at novel biomarkers, like retinal health, we offer new opportunities to identify patients at risk, offer appropriate tests and encourage them to develop healthy lifestyles to mitigate their risk."
The AI model accurately predicted biological characteristics including sex and blood pressure, as well as lifestyle factors associated with Alzheimer's development such as smoking, alcohol use, and insomnia. The researchers noted that while many of these factors are captured in medical charts, those records are often incomplete, and some — like alcohol consumption and smoking — rely on unreliable self-reports.
"With the assistance of AI, we are now able to identify subtle retinal variations that were formerly overlooked across thousands of subjects, which may function as reliable indicators of future disease risk," said Seowung Leem, doctoral student at UF and first author of the publication.
The team was able to identify specific regions of the retina associated with Alzheimer's risk factors, including the arteries and optical nerve. Fang emphasized that retinal imaging functions "less as a surrogate questionnaire and more as an integrated biological sensor of cumulative risk," noting that retinal morphology could provide measurable indicators of neurovascular integrity highly relevant to Alzheimer's disease (搜索) vulnerability.
Fang's group had previously established that retinal photographs can detect active cases of Alzheimer's disease (搜索). The new findings extend this capability to identifying early risk factors, potentially enabling interventions — including protective lifestyle changes, certain medications, or brain training — before irreversible brain damage occurs.
Converging Evidence for Accessible Screening
Both studies underscore the near-ubiquity and low cost of retinal imaging compared to more expensive technologies like MRI. The Manchester research was supported by funders including the Medical Research Council, the Wellcome Trust, the British Heart Foundation, the Royal Academy of Engineering, and the NIHR Manchester Biomedical Research Centre. The Florida study received support in part from the National Science Foundation.
"Using scans available on every high street, an eye test may become much more than a way to check your glasses prescription," said Professor Frangi, who is also RAEng Chair and Digital Infrastructure Programme Co-Lead at the NIHR Manchester Biomedical Research Centre.