Medical News Observer

Key Points:

• The machine learning classifier to predict psychosis demonstrated an 85% accuracy rate in training and 73% in validation, particularly by analyzing cortical surface areas.
• This technology offers the promise of earlier identification of those at high risk for psychosis, which is crucial for timely intervention and improving patient outcomes.
• As part of ongoing efforts within the Brain/MINDS Beyond project, researchers focus on refining this tool, emphasizing its clinical applicability.

 

An international consortium, including researchers from the University of Tokyo, has developed a machine-learning tool capable of predicting the onset of psychosis through structural magnetic resonance imaging (MRI) brain scans. This method has shown promise in distinguishing between healthy individuals and those who are at a high risk of experiencing a psychotic episode. It paves the way for continuous efforts to improve and modify this technology for wider clinical use as a part of Japan’s national Brain/MINDS Beyond project.

Machine learning and MRI to predict psychosis study

Utilizing T1-weighted brain MRI scans from 1,165 clinical high-risk individuals and 1,029 healthy controls across 21 global locations, the study harmonized data on subcortical volume, cortical thickness, and surface area, adjusting for age and sex. 

A training dataset was used to build a classifier, which achieved 85% accuracy in training and 73% in independent confirmatory datasets. The classifier was particularly effective in identifying clinically high-risk individuals who developed psychosis later based on regional cortical surface areas. 

Implications

This advancement holds significant promise for clinical settings, offering a pathway to earlier intervention. Early intervention is often linked to favorable outcomes and minimal impact on the quality of life. This approach addresses a crucial gap in mental health care, especially given that only a fraction of those clinically deemed at high risk go on to experience psychosis. 

The development of this tool represents a crucial advancement in mental health research. It utilizes baseline MRI scans to identify structural brain differences in individuals at high risk of psychosis before the onset of symptoms. The researchers emphasize the importance of additional research to confirm the clinical applicability of the classifier.

References

• Zhu, Yinghan, Norihide Maikusa, Joaquim Radua, Philipp G. Sämann, Paolo Fusar-Poli, Ingrid Agartz, Ole A. Andreassen, et al. 2024. “Using Brain Structural Neuroimaging Measures to Predict Psychosis Onset for Individuals at Clinical High-Risk.” Molecular Psychiatry, February, 1–13. https://doi.org/10.1038/s41380-024-02426-7
• “Predicting Psychosis Before It Occurs.” n.d. The University of Tokyo. Accessed February 13, 2024. https://www.u-tokyo.ac.jp/focus/en/press/z0508_00332.html.
• Zhu, Yinghan. (2023) 2024. “Yh-Zhu/MolPsy_2024_ENIGMA.” Python. https://github.com/yh-zhu/MolPsy_2024_ENIGMA.