Severity-dependent alterations of EEG microstate dynamics in obsessive-compulsive disorder
Xinyue Zhang1, Rongrong Zhu2, Qihui Guo2
1Department of Psychological and Cognitive Sciences, Tsinghua University, Beijing, China.
Progress in Neuro-Psychopharmacology & Biological Psychiatry
|April 12, 2026
Summary
Obsessive-compulsive disorder (OCD) severity is linked to distinct electroencephalogram (EEG) microstate patterns. Machine learning successfully differentiated severe from non-severe OCD cases, highlighting neurophysiological heterogeneity.
Area of Science:
- Neuroscience
- Psychiatry
- Computational Neuroscience
Background:
- Elevated obsessive-compulsive disorder (OCD) symptom severity correlates with significant clinical burden and psychiatric risks.
- Large-scale network dynamics underlying OCD severity remain poorly understood.
- Electroencephalogram (EEG) microstate analysis provides insights into rapid neural network fluctuations, but its application in OCD is limited.
Purpose of the Study:
- To investigate the relationship between OCD severity and large-scale brain network dynamics using EEG microstate analysis.
- To identify distinct EEG microstate profiles associated with different levels of OCD symptom severity.
- To explore the utility of machine learning in classifying OCD severity based on neurophysiological markers.
Main Methods:
- Resting-state EEG data were collected from 101 participants (65 with OCD, 36 healthy controls).
- Participants with OCD were stratified into severe and non-severe groups based on Yale-Brown Obsessive-Compulsive Scale (YBOCS) scores.
- EEG microstate parameters were compared across groups, and machine learning models were applied to identify severity-related features.
Main Results:
- Distinct microstate patterns were observed based on OCD severity.
- Non-severe OCD patients showed altered microstates B and C; severe patients exhibited deficits in microstate D and elevated microstate A.
- Machine learning successfully distinguished severe from non-severe OCD cases with 69.76% balanced accuracy and an AUC of 0.78.
Conclusions:
- OCD severity is associated with distinct, severity-dependent microstate profiles, not a linear network disruption.
- The successful classification of OCD severity subgroups via machine learning underscores the neurophysiological heterogeneity of the disorder.
- Precision psychiatry approaches should consider tailored strategies based on specific network states across different disease stages.
