Sunderland AI Eye Tests May Help Detect Diabetes

A Sunderland project is using artificial intelligence during eye examinations to help identify possible signs of diabetes. The initiative forms part of broader efforts to determine whether routine retinal scans can reveal information about conditions affecting the rest of the body.

Sunderland’s AI health hub is also reported to be receiving a £12 million boost to develop eye-scan technology aimed at detecting diabetes and heart disease Source 5. The programme could support research into whether retinal images enable earlier referral and assessment.

The project is not intended to let AI diagnose patients independently. Its role is to support eye examinations and clinical decision-making. An image may identify a pattern requiring further investigation, but diagnosis normally depends on medical tests and professional assessment.

Available reports confirm the project’s purpose and funding announcement but do not provide diagnostic accuracy figures, implementation dates, detailed results or information about the exact software and imaging equipment.

How the Project Could Work

A typical pathway could involve five stages:

1. A patient attends an eye examination.
2. A retinal image or other eye scan is captured.
3. AI analyses the image for relevant patterns.
4. A healthcare professional reviews the result alongside the patient’s symptoms and medical history.
5. The patient receives further testing or referral if necessary.

The distinction between detection support and diagnosis is important. A flagged scan might lead to blood glucose testing, a specialist eye examination or cardiovascular risk assessment. It would not, by itself, prove that a patient has diabetes or heart disease.

The original report describes a Sunderland project using AI during eye tests to help detect diabetes, but the supplied summary provides no further technical or clinical details Source 1.

Why Retinal Images May Be Useful

Diabetes can damage blood vessels throughout the body. The retina contains a dense network of small vessels, and changes such as leakage, bleeding or other abnormalities may be associated with diabetic complications.

AI systems may examine features including:

• Blood-vessel patterns and structure.
• Small retinal haemorrhages.
• Areas of damage or abnormality.
• Patterns associated with eye disease or wider vascular risk.

These are general examples from retinal AI research. The available information does not confirm which features the Sunderland system will use, how its model was trained or which criteria will trigger an alert.

An eye scan cannot replace blood testing. Diabetes assessment may require blood glucose measurements, symptom review and other clinical information. Its potential value is providing an additional opportunity to identify people who need further assessment.

The Reported £12 Million Investment

The reported investment could support:

• AI model development and refinement.
• Collection and analysis of retinal images.
• Clinical research.
• Testing in real healthcare settings.
• Secure data infrastructure.
• Collaboration among universities, clinicians and technology developers.
• Staff training and implementation planning.

The source summary does not specify the complete funding structure, delivery timetable or organisations responsible for individual workstreams. It is therefore too early to say how the funding will be allocated or when the technology might enter routine use.

Possible Links to Heart Disease

The eye contains small blood vessels that may reflect aspects of wider vascular health. This creates the possibility that retinal images could contain markers associated with cardiovascular risk.

The reported funding is linked to eye-scan technology for detecting both diabetes and heart disease Source 5. This is a research and development goal, not evidence that the system can diagnose heart disease.

An AI alert would require confirmation through appropriate medical tests, such as blood pressure measurement, cardiovascular assessment or other investigations selected by a clinician.

Benefits and Limitations

Potential benefits include earlier referral, improved access to screening and support for clinical teams reviewing large numbers of images. Adding health-risk information to routine eye appointments could also reach people who do not regularly attend primary care appointments.

However, faster analysis does not automatically reduce waiting times. Staffing, referral capacity and access to follow-up tests would still affect patient outcomes. AI should support qualified professionals, not replace optometrists or doctors.

A negative result also requires careful interpretation. Symptoms, family history, obesity, high blood pressure and other risk factors may justify further assessment even when an image does not trigger an alert. No screening system identifies every case.

Performance may be affected by poor-quality images, existing eye conditions, age, ethnicity, health history and differences between imaging devices. A model may also reproduce errors or bias if its training data are incomplete or unrepresentative.

Patient Safety, Privacy and Governance

False positives can cause anxiety, unnecessary appointments and additional testing. False negatives can create false reassurance or delay diagnosis. Meaningful evaluation should report sensitivity, specificity, false-positive rates and false-negative rates, including performance across demographic groups.

Retinal images are sensitive health data. A responsible programme requires secure storage, controlled access, lawful data use and clear patient information. Patients should understand whether their images are used for care, research or algorithm development, and who reviews any AI-generated alert.

Human accountability must remain clear. Patients should know what the system can and cannot do, who makes follow-up decisions and how results are communicated.

How the Project Should Be Evaluated

Researchers should compare AI-supported assessments with expert reviews and established diagnostic tests. Evaluation should examine whether the project leads to:

• Earlier diabetes diagnoses.
• Earlier cardiovascular risk assessments.
• Better attendance at follow-up appointments.
• More effective disease management.
• Fewer avoidable complications.

Operational measures should include appointment times, clinician workload, referral volumes, waiting times and costs. Patient and staff feedback would show whether the technology fits clinical practice.

Technical accuracy alone does not prove that a system benefits patients.

Wider North East Research

Durham University researchers are involved in a separate £4.2 million eye-health research institute Source 7. The institute supports broader eye-health research, while Sunderland’s initiative focuses on AI-enabled eye scans and possible detection of diabetes and heart disease.

The available information does not confirm a direct operational partnership. The two initiatives should therefore be treated as separate unless a formal connection is announced.

What Happens Next?

Important unanswered questions include:

• When will the technology enter wider clinical use?
• Which patients will be eligible?
• Which scans and devices will be used?
• How will AI findings be verified?
• Who will arrange follow-up after an alert?
• How will patient data be protected?
• What evidence will be published?

The project’s promise is significant: a routine eye test could identify people who need further assessment for diabetes or cardiovascular risk. Its success will depend on accurate technology, professional oversight, fair performance across communities and reliable access to follow-up care.

FAQ

Can an eye test detect diabetes?

An eye examination can identify retinal changes associated with diabetes and may prompt further testing. Blood tests and clinical assessment are normally needed to confirm diabetes.

How will AI be used during the eye test?

AI is expected to analyse eye scans or retinal images for relevant patterns. A healthcare professional should interpret the result and decide whether follow-up is needed. The exact software, device and workflow have not been disclosed.

Can the project detect heart disease?

The planned technology is reported to support detection of diabetes and heart disease Source 5. An eye-image alert is not a confirmed diagnosis; further medical tests are required.

Will AI replace optometrists or doctors?

No. Professionals remain responsible for reviewing results, considering symptoms and medical history, explaining findings and arranging follow-up.

What is the reported funding?

The Sunderland AI health hub is reported to be receiving a £12 million boost linked to eye-scan technology for diabetes and heart disease detection. No detailed spending breakdown is available.

Is the project connected to Durham University’s institute?

Both initiatives are regional eye-health developments, but the available sources do not confirm a direct partnership or shared programme.