Aberrant posterior cingulate connectivity classify first-episode schizophrenia from controls: A machine learning...

Area of Science:

• Neuroimaging
• Psychiatry
• Computational Neuroscience

Background:

• Posterior cingulate cortex (PCC) is integral to the default mode network (DMN).
• Altered PCC functional connectivity (FC) is linked to schizophrenia.
• The utility of PCC-based FC for classifying schizophrenia remains unevaluated.

Purpose of the Study:

• To investigate PCC-based functional connectivity (FC) patterns for distinguishing first-episode schizophrenia (FES) patients from healthy controls (HC).
• To evaluate the potential of PCC FC as a biological classifier for schizophrenia.

Main Methods:

• Utilized resting-state functional MRI data for a data-driven analysis of PCC-based region- and voxel-wise FC.
• Employed false discovery rate estimation to identify discriminative PCC FCs.
• Trained and validated a gradient boosting classifier on independent FES and HC datasets.

Main Results:

• FES patients showed reduced PCC connectivity with frontal, parahippocampal, anterior cingulate, and parietal regions, alongside hyperconnectivity with temporal regions.
• Region-wise PCC FC analysis achieved 72.28% accuracy in discriminating FES from HC in independent samples.
• Voxel-wise analysis yielded a classification accuracy of 68.72%.

Conclusions:

• First-episode schizophrenia is characterized by a mixed pattern of increased and decreased PCC-based connectivity.
• Predominant hypoconnectivity between PCC and DMN-associated regions may serve as a differential diagnostic feature.
• PCC-based FC alterations offer insights into the neuropathophysiology of schizophrenia.