TRUST-AI4D: TRUSTworthy AI models in Diabetes

The project is made possible by funding from the Trond Mohn Research Foundation (TMF). The goal is to create easy-to-use  artificial intelligence (AI) tools for medical doctors and patients to enable early detection of deterioration in organs' function that often suffer in patients with diabetes such as the kidneys, eyes, nerves and heart.

This poject is a collaboration together with Professor Iain Johnston from the Department of Mathematics, UiB, to exchange different areas of expertise to make this happen.

Diabetes epidemic is an epidemic of diabetes complications

Diabetes in adults is characterized by dysregulated sugar metabolism that can cause serious health problems. In people with diabetes, their bodies are not able to handle sugar correctly, and this might damage the blood vessels and nerves in the body, from the tiny ones to the big ones. 

This might lead to heart problems, stroke in the brain, blindness in the eyes that takes away people sight, problems with kidney function that can lead to need a machine to do kidney work (dialysis), or even losing a limb because of problems with loss of sensitivity and blood circulation problems. These health problems can shorten people's lives and cost a lot of money to manage to both patients and healthcare. There are medicines that help control the blood sugar levels, but there are not a cure and people respond to them differently.

How can we get better at recognizing who is at risk for developing diabetic complications?

The group's research suggests that high blood glucose level is not the only significant reason why people with diabetes have three to four fold elevated risk for other health problems, other factors are at play as well. Some of these factors might act in synergy with high blood sugar and worsen the disease, others can actually protect some people.

For example, they have learned that what happens to a baby in the womb, even before it's born, can affect the chances of getting a serious eye disease called proliferative diabetic retinopathy if they get diabetes later in life. It seems that factors that impact this early stage might make the blood vessels and nerves in the eyes extra sensitive to damage, so they get hurt more when blood sugar is high for a long time.

Each of us is unique, with our own genetic blueprint. Unraveling which genes are behind these silent health problems in diabetes is quite a challenge. This means we need to understand the early influences and body's molecules long before diabetes is diagnosed, as well as prevent these problems throughout someone’s life.

Can innovative AI solutions give us a better understanding of how diabetes worsens over time, and can help us take a better care?

We are now seeing incredible advances in smart computer programs that can analyse medical records and DNA code to pinpoint who might be at danger to have serious diabetes problems. These programs are complex, and even doctors are still learning how to use them.

But this need to understand our genes is not just for inventing new medicines. It's also about mapping the course of diabetes so we can see exactly when and where the body starts to suffer. For example, the British researchers recently published a new smart system that makes a decision whether a person with type 2 diabetes might do better with one type of medicine over another. They have built a tool that guides doctors in choosing between two medications: SGLT2 inhibitors and DPP4 inhibitors. Interestingly, they found that the people who would benefit most from SGLT2 drugs are not always the ones who are currently recommended to take them according to standard guidelines. This highlights the importance of developing intelligent computer programs that can support doctors and patients in making decisions who will benefit most from which treatments. These programs can help us understand diseases better and could even help reduce healthcare costs by preventing diabetes complications.

Photo:

Diabetologia

Link to graphical abstract: https://diabetologia-journal.org/editors-choice/genetics-of-diabetes%E2%80%91associated-microvascular-complications-published-online-14-07-2023/