Revealing multiple biological subtypes of schizophrenia through a data-driven approach

Area of Science:

• Neuroscience
• Microbiome Research
• Psychiatry

Background:

• Schizophrenia (SZ) research has focused on brain imaging subtypes.
• The heterogeneity across multiple biological data types in SZ remains largely unexplored.

Purpose of the Study:

• To investigate the heterogeneity of schizophrenia by integrating brain imaging and gut microbiota data.
• To identify distinct biological subtypes of schizophrenia using a data-driven approach.

Main Methods:

• Utilized a data-driven model to classify brain imaging and gut microbiota data.
• Employed dot product fusion for brain-gut data integration.
• Correlated identified subtypes with clinical symptoms and cognitive performance.

Main Results:

• Identified independent brain subtypes (structural vs. functional alterations) linked to negative and positive symptoms, respectively.
• Discovered three gut microbiota subtypes (Collinsella, Prevotella, Streptococcus dominant) associated with symptoms.
• Revealed brain-gut interaction subtypes correlating with attention deficits and positive symptoms.
• Found brain and brain-gut subtypes strongly associated with clinical symptoms, while gut subtypes offered more cognitive biomarkers.

Conclusions:

• Demonstrated the potential to identify multiple, distinct biological subtypes in schizophrenia.
• Highlighted the utility of distinct biomarkers for each subtype.
• Suggests a future direction for personalized and precise treatment strategies for schizophrenia.