Monash

Discovery

Predicting the severity of psychotic symptoms

03.10.2020

In a nutshell: Rather than considering schizophrenia symptoms as present or absent, researchers are using machine learning to predict their severity on a continuous scale.

View Paper Abstract

Predicting the severity of psychotic symptoms

Schizophrenia is a mental health disorder that affects how people think, feel and experience the world. It encompasses a wide range of symptoms, from social withdrawal and apathy to hallucinations and delusions.

Machine learning – a type of artificial intelligence – has been used to differentiate the brains of people with schizophrenia and those of healthy people. But one of the limitations to this approach is that it groups all schizophrenia patients together, regardless of how different their symptoms are. It also ignores the fact that healthy people can experience some of the same symptoms, too.

In a new approach, a team led by Brain Function CoE investigator Marta Garrido has used machine learning to analyse individual symptoms of schizophrenia.

The researchers recruited 80 people with schizophrenia to complete an auditory oddball task. This task tests how brain activity changes in response to an unexpected sound. The researchers recorded the participants’ brain activity and their performance during the task. They also obtained detailed information about the types of symptoms that each participant experienced, and their severity.

By combining this data with machine learning, the researchers could make individualized predictions about specific symptoms and their severity. They were able to predict with high accuracy symptoms such as hallucinations, difficulty paying attention, lack of motivation, and inability to feel pleasure.

They also mapped the brain regions that contributed to predicting each of these symptoms. Some of these regions are already known to be involved in schizophrenia or related behaviours. But other regions have not yet been linked to this disorder. They could represent new areas for future research. For example, they could be tested as potential targets for brain stimulation therapies.

Next steps:
The researchers plan to test if the same approach can be used to predict people at risk of developing psychosis.


Reference:
Taylor, J. A., Larsen, K. M., & Garrido, M. I. (2020). Multi-dimensional predictions of psychotic symptoms via machine learning. Human Brain Mapping, doi: 10.1002/hbm.25181


Republish this article:

We believe in sharing knowledge. We use a Creative Commons Attribution 4.0 International License, which allows unrestricted use of this content, subject only to appropriate attribution. So please use this article as is, or edit it to fit your purposes. Referrals, mentions and links are appreciated.

CIBF