☑ AI can diagnose ADHD with 99 percent certainty

For those who are diagnosed with ADHD, there is plenty of help available today. Everything from adapting work or schooling to medical treatment. But to get the help you need, you must first get the right diagnosis. Here, an AI developed by researchers at the University at Buffalo can help .

The researchers used machine learning where an AI had to go through lots of images of adults' brain activity, both those who had been diagnosed with ADHD and those who did not have ADHD. In this way, the AI ​​learned to see the types of brain activities that characterized ADHD.

The result when the AI ​​then had to make diagnoses on images it had not made was a little outstanding. The result over the whole group, the AI ​​was right in 99 percent of the cases. The fact that the study concerned adults who had already been diagnosed as children makes the researchers believe that there are activities in the brain that cause ADHD and that can be used to make reliable diagnoses.

- This suggests that connections in the brain act as stable biomarkers for ADHD, at least so far into childhood that an individual has adapted their behavior so that it becomes more difficult to detect the underlying problems, says Chris McNorgan at University at Buffalo and the study's lead author .

In other words, the AI ​​could help doctors and psychologists make a correct diagnosis even in cases that are difficult to determine with regular investigations. In addition, the AI ​​can point out where on the spectrum of ADHD a person is located.

Personal treatment

Not everyone suffers from ADHD in the same way and even one and the same person can have different ailments from one day to another. This can make it difficult to decide which treatment is best for each individual. However, the AI ​​can see patterns in fMRi images that people would find very difficult to see.

Simplified, it works so that the AI ​​can register how different areas of the brain communicate with each other. For example, if area A and area B communicate with each other, it may show that the person has ADHD. But not if both areas simultaneously communicate with area C.

By identifying a variety of such patterns, the AI ​​can not only determine if a person has ADHD but also where on the spectrum the person is and what treatment would work best. It can also open up new types of treatments.

"Because different networks in the brain are activated for humans on different parts of the spectrum, this method can make it possible to develop therapies that focus on specific networks," says Chris McNorgan.