Predicting psychosis before onset

summary: Researchers have developed a machine learning tool to accurately identify individuals at high risk of psychosis through MRI brain scans. This innovative approach achieved an 85% accuracy rate with training and 73% with new data, offering a promising avenue for early intervention in psychosis and potentially improving treatment outcomes.

The study involved more than 2,000 participants from 21 locations around the world, highlighting the potential of this tool in a variety of clinical settings. By detecting structural differences in the brain before the onset of psychosis, this tool represents a major advance in psychiatric care, aiming for better prediction and prevention strategies.

Important facts:

1. Machine learning classifiers can use MRI brain scans to distinguish between those at high risk for psychosis and those who are not, with high accuracy.
2. Early identification of psychosis risk through MRI scans may lead to more effective interventions and reduce the impact on individuals' lives.
3. This study highlights the need for further development to ensure the applicability of the classifier across different datasets and clinical environments.

sauce: University of Tokyo

The onset of psychosis can be predicted before it occurs using machine learning tools that can classify brain MRI scans into healthy people and those at risk of psychotic episodes.

An international consortium, including researchers from the University of Tokyo, used the classifier to compare scans of more than 2,000 people from 21 locations around the world. Approximately half of the participants were clinically identified as being at high risk of developing psychosis.

Using the training data, the classifier differentiated between people who were not at risk and those who later experienced overt psychotic symptoms with 85% accuracy.

Using new data, accuracy was 73%. This tool may be useful in future clinical settings. Most people who experience psychosis fully recover, because early intervention usually has less negative impact on people's lives and produces better outcomes.

Anyone can experience a psychotic episode that involves delusions, hallucinations, or confused thinking. There is no single cause, but it can be caused by illness, injury, trauma, drug or alcohol use, medication, or genetic predisposition.

Although it can be scary and worrying, mental illness is treatable and most people recover. It can be difficult to identify young people who need help because the age when first symptoms are most common is during adolescence or early adulthood, when major changes are occurring in the brain and body.

“At most, only 30% of clinically high-risk people will later develop clear psychotic symptoms, but the remaining 70% will not,” says Shinsuke Koike, associate professor at the University of Tokyo's Graduate School of Arts and Sciences. explains.

“Clinicians therefore need help identifying people with persistent psychotic symptoms, using not only underlying signs such as changes in thinking, behavior and emotion, but also some biological markers. is.”

A consortium of researchers has teamed up to create a machine learning tool that uses brain MRI scans to identify people at risk of psychosis before they develop. Previous studies using brain MRI have suggested that structural differences occur in the brain after the onset of psychosis.

But this is reportedly the first time that brain differences have been identified in people who are at very high risk but have not yet experienced psychosis.

Teams from 21 different institutions in 15 countries brought together a large and diverse group of youth and young adult participants.

Koike said MRI studies of mental disorders can be difficult because differences in brain development and MRI equipment make it difficult to obtain highly accurate and comparable results. Additionally, in young people, it may be difficult to distinguish between changes due to neurotypical development and changes due to mental illness.

“Different MRI models have different parameters, which also affect the results,” Koike explained.

“Just like with cameras, different equipment and filming specifications create different images of the same scene, in this case a participant's brain. But we are correcting for these differences and trying to detect the onset of psychosis. We were able to create a well-tuned classifier to make predictions.”

Participants were divided into three groups of clinically high-risk people. People who did not develop mental illness. those with uncertain follow-up status (3 groups total he 1,165 people), and his fourth group of healthy controls for comparison (1,029 people). The researchers used the scans to train a machine learning algorithm that identified patterns in the participants' brain anatomy.

From these four groups, the researchers used an algorithm to categorize participants into two main target groups: healthy controls and a group at high risk of later developing overt psychotic symptoms.

In training, the tool showed 85% accuracy in classifying outcomes, but in final testing with new data, it was 73% accurate in predicting which participants were at high risk of developing psychosis.

Based on this result, the team believes that offering brain MRI scans to people clinically identified as at high risk could help predict future development of psychosis. .

“We need to test whether the classifier performs well on new datasets. Some of the software we used is best suited for fixed datasets, so we can test MRIs from new facilities and machines. We need to build a classifier that can classify reliably. This is a challenge that Japan's national neuroscience project called Brain/MINDS Beyond is working on right now,” Koike said.

“If this is successful, we will be able to create more robust classifiers on new datasets and apply them in real-world, routine clinical settings.”

Funding: This research was supported in part by AMED (grant numbers JP18dm0307001, JP18dm0307004, JP19dm0207069), JST Moonshot Research and Development (JPMJMS2021), JSPS KAKENHI (JP23H03877 and JP21H02851), the Takeda Science Foundation, and the Senshin Medical Research Foundation. I received a grant. This research was also supported by the University of Tokyo International Research Center for Neurointelligence (WPI-IRCN).

About this psychiatric research news

author: Joseph Krisher
sauce: University of Tokyo
contact: Joseph Krisher – University of Tokyo
image: Image credited to Neuroscience News

Original research: Open access.
“Using structural brain neuroimaging measures to predict the onset of psychosis in clinically high-risk individuals” by Shinsuke Koike et al. molecular psychiatry

abstract

Using structural brain neuroimaging measures to predict the onset of psychosis in clinically high-risk individuals

Machine learning approaches using structural magnetic resonance imaging (sMRI) may be beneficial for disease classification, but their ability to predict psychosis is largely unknown.

We created a model using CHR healthy controls (HC) and individuals who later developed psychosis (CHR-PS+) that were distinguishable from each other.

We also assessed whether CHR-PS+ patients could be distinguished from those who did not later develop psychosis (CHR-PS-) and those with uncertain follow-up (CHR-UNK). T1-weighted structural brain MRI scans from her 1165 patients with CHR (CHR-PS+, n= 144; CHR-PS-, n= 793; and CHR-UNK, n= 228), and 1029 HCs were obtained from 21 sites.

ComBat was used to harmonize measurements of subcortical volume, cortical thickness, and surface area data, and a general additive model was used to correct for nonlinear effects of age and sex. CHR-PS+ (n= 120) and HC (n= 799) Data from 20 sites served as the training dataset, which was used to build the classifier.

The remaining samples used external validation datasets to evaluate classifier performance (testing, independent validation, and independent groups). [CHR-PS- and CHR-UNK] data set). The accuracy of the classifier on the training dataset and independent validation dataset was 85% and 73%, respectively.

Regional cortical surface area measurements (including measurements from right superior frontal cortex, right superior temporal cortex, and bilateral insular cortex) strongly contributed to classifying CHR-PS+ from HC. CHR-PS- and CHR-UNK individuals were more likely to be classified as HC compared to CHR-PS+ (classification rate to HC: CHR-PS+, 30%; CHR-PS-, 73% , CHR-UNK, 80%).

We used multisite sMRI to train a classifier to predict the onset of psychosis in CHR patients. The results showed the possibility of predicting her CHR-PS+ in an independent sample.

This result suggests that baseline MRI scans of CHR patients may be useful in determining prognosis when considering brain development during adolescence.

Future prospective studies are needed to determine whether this classifier is actually useful in clinical practice.