Unlocking Chronic Pain: A Network Science Approach to Identify fMRI Biomarkers

Chronic pain is a persistent condition that affects physical, emotional, and cognitive functioning, yet its underlying neural mechanisms remain difficult to characterize. In this study, we introduce a two-stage network-based analytical framework using rs-fMRI to identify functional connectivity alterations associated with chronic pain. Our approach integrates (i) group-level network estimation using sparse covariance models to capture shared patterns of reorganization and (ii) individualized sparse covariance models to detect region-specific differences between chronic pain patients and healthy controls. Using both the Pauli (subcortical) and MSDL (cortical) atlases, the group-level analysis revealed marked differences in global network properties including increased degree (8.125 vs. 7.375), density (0.542 vs. 0.492), and reduced modularity (0.218 vs. 0.442) in chronic pain. At the individual-level, ROI-based statistical testing identified significant differences in the Visual cortex (degree:p= 0.04; strength:p= 0.046) and a marginal effect in the Middle Default Mode Network (p= 0.047 after FDR correction). These results demonstrate that chronic pain is characterized by both widespread group-level network reorganization and focal region-specific disruptions. The proposed framework highlights key neurobiological markers of chronic pain and provides a reproducible graph-theoretic approach for studying clinical brain network alterations.