Artificial intelligence (AI) is leading a diabetes care revolution.
This field of computer science enables machines to mimic human intelligence by learning from data, recognising patterns and making decisions.
AI is already powering improvements for people with or at risk of diabetes today, with researchers working to make even bigger breakthroughs in the future. Here’s how.
Monitoring and management
Managing blood sugar levels can be challenging, but AI is making it easier and more automatic.
AI-enhanced continuous glucose monitors (CGM) now provide real-time blood sugar readings and predict future trends. By learning from someone's glucose data, systems can alert to dangerous highs or lows before they happen.
Hybrid closed loop systems go even further by linking a CGM with an insulin pump via an AI-powered algorithm. The AI acts as the system’s ‘brain’ to make decisions on how much insulin to give and when.
With our funding, researchers are now testing even more sophisticated algorithms that could fully automate insulin delivery, so the technology could think and act just like a real pancreas.
Preventing complications
Diabetes complications can be serious, but AI is helping to catch warning signs sooner.
AI is now used in diabetes eye screening to speed up and enhance analysis of eye photographs. It means thousands of images can be examined in minutes and is helping to spot signs of damage earlier, including those that might be missed by the human eye.
Researchers are also using AI to unlock deeper insights from eye screening photos. The retina at the back of the eye acts as a window into the body’s network of blood vessels – which are critical to the health of organs. By learning from and revealing hidden patterns in retina photos, AI could predict who’s at risk of complications like heart disease, kidney disease or foot problems. So people can get the right care and treatments earlier to prevent life-threatening complications.
Earlier diagnosis
AI holds the potential to detect diabetes before it even develops. AI can sift through vast amounts of data – like medical records or genetic information – to identify early signs of diabetes risk.
For example, our researchers have developed an AI tool that analyses children’s GP records to flag early warning signs of type 1 diabetes. It looks for combinations of clues such as urinary infections, bedwetting, family history, and antibiotic use to predict who might be diagnosed with type 1 diabetes in the next 90 days. This would allow for an earlier diagnosis and reduce the risk of dangerous diabetic ketoacidosis.
AI is also being explored to detect type 2 diabetes risk early. In one trial, researchers are using AI to analyse heart test (ECG) results and find subtle patterns that predict type 2 diabetes risk – up to 13 years in advance. This could help doctors to offer support to those at risk sooner, to delay or prevent a diagnosis.
Personalised care
No two people with diabetes are the same – and AI is helping tailor treatment to each individual, ensuring people receive the most effective treatments for them.
At the University of Exeter, researchers have trained AI using data from one million people in the UK. They created a tool that looks at simple characteristics, like sex and weight, to match each individual with type 2 diabetes to the blood glucose-lowering drug most likely to work for them. It would mean everyone could receive the best treatment to help keep their blood sugars in target range and minimise their risk of developing complications.
AI is also being used to analyse our DNA. In the future, this could help doctors to choose the most effective treatments based on a person’s unique genetic make-up.
Virtual support
Support and education play a huge role in living well with diabetes. AI chatbots and virtual assistants are being developed to offer people personalised advice outside of medical appointments.
At the University of West England, researchers are developing an AI chatbot designed to supplement type 2 diabetes education courses. It offers tailored support, helping people stay informed and supported to self-manage their condition.
There’s also research underway to see if AI chatbots can provide emotional support and help people manage the psychological side of the condition. Voice assistants are being adapted to help with diabetes management. The hope is that they could help remind people to take their insulin, suggest meal plans, or alert caregivers in emergencies.
Speeding up the search for a cure
AI is also helping scientists move faster than ever to find a cure. Researchers can now use AI to automate tasks that previously required enormous manual effort, like analysing thousands of potential drug combinations to find those that are most effective and speed up the discovery of breakthrough new treatments.
Through the Type 1 Diabetes Grand Challenge, Professor Sarah Richardson is using AI to analyse images of pancreas samples. By teaching AI to recognise insulin-producing beta cells and track how the immune system attacks them, her research is uncovering vital clues about type 1’s roots. This could pave the way for new immunotherapies that slow or stop the type 1 diabetes immune attack.
As research progresses, AI-driven innovations could mean earlier diagnosis, better treatments and a future where diabetes is easier to live with, and ultimately cured. We’re proud to support pioneering research that’s turning this vision a reality.
