Diabetes Treatment and AI. What you Need to Know.

Image by Tesa Robbins from Pixabay

The content on this page was current at the time of publication — please check the date/time stamp on this page.

What Is Diabetes?

• Diabetes is a long-term (chronic) condition where the level of sugar (glucose) in your blood is too high. nhsinform.scot
• There are several types, but the two most common are:
  • Type 1 diabetes — where the body does not produce insulin.
  • Type 2 diabetes — where the body does not use insulin properly, or doesn’t make enough. nhsinform.scot
• There is also pre-diabetes (sometimes called non-diabetic hyperglycaemia) — this is when blood sugar is higher than normal but not high enough to be diagnosed as diabetes. nhsinform.scot

How Common Is Diabetes in the UK?

• According to Diabetes UK, there are 4.6 million people in the UK with a diagnosis of diabetes, and millions more with pre-diabetes. Diabetes UK
  
• About 90% of diagnosed diabetes in the UK is Type 2; around 8% is Type 1, and the remaining ~2% are other, less common types. Diabetes UK
  
• In England, there’s a significant number of people who are undiagnosed — an ONS analysis suggests that around 30% of people with Type 2 diabetes are not aware they have it. Diabetes UK
  
• For pre-diabetes, NHS data shows hundreds of thousands more people are at risk: in 2023, over half a million more people in England were identified as pre-diabetic. NHS England

How Is Diabetes Traditionally Managed in the UK?

Diagnosis

• People are usually diagnosed via blood tests, such as HbA1c (a measure of average blood sugar over several months), or fasting glucose tests.
• For pre-diabetes, blood tests and risk assessments by GPs help identify people at risk.

Lifestyle Changes

• Managing or preventing Type 2 diabetes often starts with lifestyle: healthy eating, weight loss, physical activity, and reducing sugar intake.
• The NHS offers structured programmes (like the NHS Diabetes Prevention Programme) to support people at high risk. NHS England

Medication

• Depending on type and severity, people with diabetes may be prescribed insulin (Type 1) or medications such as metformin, sulfonylureas, or newer agents (Type 2) to control blood sugar.

Monitoring & Care

• People with diabetes typically have regular reviews with their GP or diabetes specialist.
• They receive a set of “care processes” each year: blood sugar check, blood pressure, cholesterol, kidney function, foot checks, eye screening etc. According to the National Diabetes Audit, uptake of these care checks has improved over time. GOV.UK
• Annual diabetic eye screening is offered via the NHS Diabetic Eye Screening Programme (DESP) to detect diabetic retinopathy.

How Is AI Currently Helping People with Diabetes?

AI is being used in several research and pilot projects across the UK to improve how diabetes is detected, monitored, and managed. Here are some examples:

1. Predicting Risk from ECGs
   • A new AI tool called AIRE-DM (developed by Imperial College) analyses ECG (heart) readings from routine scans to detect subtle changes that might indicate a risk of developing Type 2 diabetes, up to 10-13 years in advance. Imperial College Healthcare NHS Trust+1
   • Early detection like this could allow people to make lifestyle changes before the disease develops or progresses.
     
     
2. Diabetic Eye Disease (Retinopathy) Screening
   • AI systems such as EyeArt (by Eyenuk) can automatically analyse retinal (eye) images for signs of diabetic retinopathy — the algorithm is evaluated to be safe and cost-effective per the UK National Screening Committee. eyenuk.com
   • A newer model from King’s College London uses NHS retinal imaging data to predict who is at high risk of sight-threatening diabetic retinopathy up to 3 years ahead, potentially allowing for more personalised eye screening schedules. King's College London
   • NICE (UK) has also reviewed AI technologies (EyeArt, RetinaLyze, Retmarker) for diabetic retinopathy screening, noting their potential to reduce manual workload. NICE
      
3. Predicting Kidney Disease via Eye Images
   • Research from the University of Dundee (UK) shows AI can analyse routine diabetic eye screening photos (retinal images) to predict future risk of kidney disease in people with Type 2 diabetes, even before traditional kidney tests flag problems. Diabetes UK
     ​ 
4. Risk Prediction Models Using Machine Learning
   • Using data from the UK Biobank, researchers have built a prediction model for 10-year risk of Type 2 diabetes that only needs information that can be collected via a smartphone. arXiv
   • Another automated machine-learning framework (AutoPrognosis 2.0) used UK Biobank data to build a risk score for diabetes that may outperform some traditional clinical risk scores. arXiv

How Might AI Help People with Diabetes in the Future?

Here are some of the promising ways AI could change diabetes care in the UK in the coming years:

1. Very Early Risk Identification
   • AI tools that predict diabetes years before onset (e.g., from ECGs) could help people take preventive action long before they develop high blood sugar.
   • Combined with risk models, AI could personalise lifestyle advice based on an individual’s unique risk profile.
      
     
2. Individualised Eye Screening
   • Rather than everyone having the same annual eye scan, AI risk models could tailor screening intervals: people at low risk might need less frequent checks, while high-risk individuals could be seen sooner.
   • This would optimise NHS resources and reduce patient burden.
      
     
3. Early Detection of Complications
   • Using retinal images to predict kidney disease is a powerful example. In future, similar AI tools might predict other diabetic complications, like cardiovascular disease, nerve damage, or foot problems, enabling earlier intervention.
      
     
4. Continuous Monitoring Through Wearables
   • AI could be integrated with wearable devices (e.g., smartwatches or glucose monitors) to provide real-time risk alerts or trend warnings (such as rising glucose) — this could support both people with Type 1 and Type 2 diabetes.
      
     
5. Decision Support for Clinicians
   • AI could help clinicians decide which medications or treatment strategies might be most effective based on a patient’s risk profile, history, and predicted trajectories.
     ​ 
     
6. Precision Medicine
   • By combining genetic data, imaging (such as retinal scans), and clinical data, AI could help stratify people into subtypes of diabetes. This might guide more precise treatment (which drug, what dose, when to intensify).

Risks and Challenges of using AI in Diabetes Care

• Accuracy and Bias: AI models must be trained and validated on diverse UK populations; otherwise they could underperform for some demographic groups.
   
  
• Regulation and Safety: Not all AI tools are fully approved or deployed; they need rigorous clinical trials and regulatory oversight.
   
  
• Data Privacy: Using sensitive health data (ECG, retinal scans, genomic data) requires strong data protection, secure storage, and patient consent.
   
  
• Integration into NHS: Large-scale implementation can be difficult — tools need to integrate with NHS systems, workflows, and budgets.
  ​ 
  
• Over-reliance: Clinicians must be careful not to rely solely on AI predictions — AI should support, not replace, clinical judgement.

Summary

• Diabetes is common in the UK, especially Type 2, and many people remain undiagnosed.
• Traditional care involves blood tests, regular monitoring, lifestyle change, and treatment.
• AI is already being used in research and pilot projects — for example, to predict diabetes risk from ECG, to screen for eye disease, and even to forecast kidney disease from retinal images.
• In the future, AI could enable earlier detection, more personalised screening, better monitoring, and smarter treatment decisions.
• But for these benefits to become widespread, challenges around accuracy, regulation, data protection, and NHS integration must be addressed.

References

• NHS Inform – About diabetes nhsinform.scot
  
• Office for National Statistics / Diabetes UK – undiagnosed Type 2 diabetes Diabetes UK
  
• NHS England – people at risk of Type 2 diabetes (pre-diabetes figures) NHS England
  
• Imperial College Healthcare NHS Trust – AI prediction of Type 2 diabetes from ECGs Imperial College Healthcare NHS Trust
  
• NICE Medtech Innovation Briefing – AI in diabetic retinopathy screening NICE
  
• King’s College London – AI model for predicting diabetic retinopathy risk King's College London
  
• Diabetes UK / University of Dundee – AI predicting kidney disease from eye images Diabetes UK
  
• UK Biobank / Machine Learning risk model for Type 2 diabetes arXiv
  
• AutoPrognosis 2.0 – prognostic modelling for diabetes arXiv