AI can detect patterns in the brain associated with Alzheimer’s, schizophrenia and autism

The new artificial intelligence (AI) is able to detect mental health conditions by sifting through brain imaging data to find patterns associated with autism, schizophrenia and Alzheimer’s, and it can do so before symptoms appear.

The model was first trained on images of the brains of healthy adults and then shown to people with mental health problems, allowing it to detect tiny changes that go unnoticed by the human eye.

The sophisticated computer program was developed by a group of researchers led by the State of Georgia, who noted that it could one day detect Alzheimer’s disease in a person as young as 40 years old, about 25 years before the onset of symptoms.

Early detection of such illnesses would help patients receive treatment that could reduce the stress of mental illness.

The AI ​​was trained on a huge dataset of more than 10,000 people to understand functional magnetic resonance imaging (fMRI), which measures brain activity by detecting changes in blood flow.

Scroll down for video

AI is able to find patterns in brain scans that are linked to mental health issues. Here are the results for children diagnosed with autism

Once the AI ​​was able to read the basic fMRI, the team fed it datasets of more than 1,200 people diagnosed with mental illness — autism, schizophrenia, and Alzheimer’s disease.

And the system was able to identify different patterns for three mental illnesses.

The team notes that using fMRI to detect mental illness can be costly — a human needs to carefully review the data — but using AI greatly reduces costs and time.

Vince Calhoun, founding director of the TReNDS Center and one of the authors of the study, said in statement: “Even if we know from other tests or family history that someone is at risk for a disorder like Alzheimer’s, we still can’t predict exactly when that will happen.”

The model was first trained on images of the brains of healthy adults and then shown to people with mental health problems.  This allowed AI to understand the difference between people with and without mental illness.

The model was first trained on images of the brains of healthy adults and then shown to people with mental health problems. This allowed AI to understand the difference between people with and without mental illness.

“Brain imaging can narrow this time window by capturing relevant patterns as they occur before clinical disease is apparent.”

The use of AI to detect mental illness is nothing new – in April, one analyzed people’s conversations on the social media platform Reddit to determine if they had any problems.

A team of computer scientists at Dartmouth College in Hanover, New Hampshire, has begun training an AI model for analyzing social media texts.

The team chose Reddit to train their model as it has half a billion active users, all of whom regularly discuss a wide range of topics on a network of subreddits.

They focused on looking for the emotional intent in a post rather than the actual content, and found that it performed better over time at detecting mental health issues.

In their study, the researchers focused on what they call emotional disorders—major depression, anxiety, and bipolar disorder—which are characterized by distinct emotional patterns that can be tracked.

Early detection of such illnesses will help patients receive treatment that reduces or even eliminates mental illness.

They looked at data from users who reported having one of these disorders and users without any known psychiatric disorder.

They trained their AI model to label the emotions expressed in user posts and display emotional transitions between different posts.

The post can be tagged by the AI ​​as “joyful”, “anger”, “sad”, “fear”, “lack of emotion”, or a combination of these.

The map is a matrix showing how likely the user is to go from one state to another, such as from anger to a neutral state of emotionlessness.

Scientists have explained that different emotional disorders have their own characteristic patterns of emotional transitions.

By creating an emotional “fingerprint” of the user and comparing it with the established signs of emotional disorders, the model can detect them.

To confirm their results, they tested it on posts that were not used during the training and showed that the model accurately predicts which users may or may not have one of these disorders, and that it improves over time.