Predict dementia risk with smart software

At present, the software developed by researchers Matthijs Biesbroek and Mirthe Coenen can only be used in patients who already have cognitive impairments, such as the onset of dementia. In the future, however, it should be possible, the researchers hope, to use this intelligent diagnostic even in people who do not yet have any disorders. For this, intelligent software needs to be further developed, whereby the use of AI (artificial intelligence) is also being considered.

Age damage and cognitive problems

The white matter in the brain is, so to speak, the “fiber-optic network” through which all information flows. The substance is made up of axons of nerve cells, enclosed by a fatty white substance, myelin, which protects the axons as a sort of insulating material. Old age is one of the important “enemies” of white matter. As a result, the tissue is damaged and cognitive problems can arise, among other things.

The consequences of these injuries can be different for each patient. The consequences that the patient experiences depend in particular on the location of the damage. “Sometimes people have a lot of white matter damage and don’t suffer from it. Other times, people function less well than before, while having relatively little damage to their white matter,” explains neurologist and researcher Matthijs Biesbroek.

Predict the risk of dementia

Thanks to it Research and intelligent software from the UMCU team, it is now possible to determine and predict what consequences white matter damage may have for individual patients. The researchers were first able to identify the main nodes of the white matter.

They then used this information to develop intelligent software that can analyze brain scans. This software is then able to predict which disorders, such as dementia, patients might develop in the future due to damage to the white matter.

“To determine which of these white matter pathways are important and which are less important, we needed a large group of patients,” says Matthijs. Both a brain scan and cognitive tests had to be available for those patients. In total, the Utrecht researchers used data from 3,525 patients from Europe, Asia and the United States.

Further development with AI

The software as developed can also be used for analysis of standard brain scans such as those done at the memory clinic. But, as already mentioned, the software is only suitable for analyzing people who already have problems.

The researchers are already working on further development of the software. Furthermore, it must be possible to carry out all process steps fully automatically. They even plan to apply AI (artificial intelligence) to give the program the ability to learn. The software is expected to be ready in early 2023.

The use of artificial intelligence applications and algorithms to improve and accelerate the diagnosis of brain problems, such as dementia, is nothing new. The first steps in this area were already taken years ago. One of the most recent initiatives is the international one AI-Mind project. He wants to develop smart digital tools with the help of artificial intelligence to screen brain connections and estimate the risk of dementia in people with mild cognitive impairment.