Cardiovascular risk, read from the eye.
RetinaHeart AI analyses a standard retinal photograph with deep learning, reads the retinal blood vessels, and combines them with traditional risk factors to flag people who may need a cardiovascular check-up early.
The problem
The world's leading cause of death is often found too late.
Cardiovascular disease frequently develops without symptoms. Screening relies on clinic visits, blood tests and specialist time, and rising healthcare costs and longer lifespans are increasing the pressure on all of them.
people die from cardiovascular disease every year.
of all deaths worldwide are caused by cardiovascular disease.
studies were reviewed by our research showing deep learning can extract CVD risk from retinal images.
Sources: World Health Organization figures as cited in the founding research paper; study count from the paper's review.
How it works
One photo of the retina. One clear next step.
The retina is the only place in the body where small blood vessels can be seen directly and without an incision. Changes in those vessels reflect the health of the circulation.
Capture
A non-invasive retinal (fundus) photograph is taken in an eye clinic, optician or primary care setting.
Analyse
Deep learning models extract retinal-vessel characteristics linked in the literature to cardiovascular risk.
Combine
Vessel features are merged with traditional risk factors such as age, blood pressure and smoking in one prediction model.
Refer
The clinician receives a clear risk signal and decides whether a full cardiovascular work-up is needed.
Product
Built to support clinicians, not replace them.
RetinaHeart AI is a decision-support tool. Every result is reviewed and acted on by a qualified professional.
Non-invasive screening
No needles and no radiation, just a standard retinal photograph that takes seconds to capture.
Vessel-level analysis
The model focuses on retinal-vessel characteristics, the signal the reviewed studies tracked most closely.
Hybrid risk model
Imaging features and classic risk factors are scored together instead of in separate silos.
Standardised reporting
A consistent output format, following the paper's call to standardise methods across sites and studies.
Who it's for
Screening where patients already are.
Eye clinics & optometrists
Turn routine retinal photos into an additional cardiovascular safety check.
Primary care
Prioritise which patients need a cardiology referral first.
Screening programmes
Add a low-cost, scalable cardiovascular component to existing diabetic-eye screening.
The research
Grounded in a critical review of the evidence.
- Reviews 30 studies showing that deep learning can extract cardiovascular risk signals from retinal images.
- Identifies retinal-vessel characteristics as the key imaging marker to track.
- Recommends combining retinal-vessel imaging with traditional risk factors in one prediction model.
- Calls for standardised methods so results can be compared across studies and settings.
Lamkuche, Kouali, & Rumman (2025). Revolutionizing Cardiovascular Disease Prediction with Advanced Retinal Imaging and AI-Driven Approaches. In Business Resilience and Business Innovation for Sustainability, Studies in Systems, Decision and Control, vol. 587. Springer.
Team
Clinical science meets AI engineering.
Founded in 2025 by a student research team from two faculties.
Faculty of Allied Medical Sciences
Clinical relevance, retinal imaging workflow and risk-factor selection.
Faculty of Engineering & Information Technology
Deep learning models, image processing and the software platform.
FAQ
Questions we hear often.
Is RetinaHeart AI a diagnosis?
No. It provides a risk signal to support clinical judgement. Diagnosis and treatment decisions stay with qualified clinicians.
What equipment is needed?
A standard retinal (fundus) camera of the kind already used in eye clinics and diabetic-eye screening.
Why the retina?
It is the only place where small blood vessels can be viewed directly and non-invasively, and their characteristics have been linked to cardiovascular risk in the reviewed literature.
Can we run a pilot?
Yes. We are looking for clinics and screening programmes to validate the tool under real conditions.
Help us catch heart risk earlier.
We are partnering with eye clinics, primary care providers and screening programmes for pilot studies.