Normative Modeling of EEG Complexity Reveals Cortical Rigidity Associated with Pain Severity in Chronic Pain.

Background: Chronic pain is neurophysiologically heterogeneous and may be obscured by group-average EEG metrics. We hypothesized that chronic pain would be characterized by reduced cortical complexity relative to a healthy-control-derived normative reference, defined as the age- and sex-conditioned expected distribution of EEG features, and that larger negative deviations would be associated with greater pain intensity. Methods: Resting-state EEG from 147 participants, including 78 patients with chronic pain and 69 healthy controls, was analyzed. Complexity was quantified using Higuchi fractal dimension (HFD) across seven scalp regions. A Gaussian process regression normative model was trained exclusively on healthy controls, using age and sex as covariates, to estimate expected HFD values and predictive uncertainty. Individual deviations were expressed as W-scores, with more negative values indicating reduced complexity relative to the normative reference. In patients, regional W-scores were correlated with average pain intensity using Pearson correlations with false discovery rate correction. Robustness was examined after adjustment for age, sex, depression, and anxiety, and spectral power measures were evaluated as exploratory benchmarks. Results: More negative frontal HFD W-scores were associated with higher pain intensity (r = -0.545, P_FDR < 0.001). Significant HFD-pain associations were observed across multiple regions, with correlation coefficients ranging from -0.441 to -0.545 among FDR-significant HFD features. The parietal association also survived correction (r = -0.523, P_FDR < 0.001). Compared with the examined spectral power benchmarks, HFD-based deviations showed larger descriptive associations with pain intensity. The frontal HFD-pain association persisted after adjustment for age, sex, depression, and anxiety (r = -0.520, two-tailed P < 0.001). Exploratory subgroup analyses suggested larger negative latent deviation scores in neuropathic conditions, particularly PNP and PHN. Conclusion: Normative deviation mapping of EEG complexity may provide a candidate research marker for quantifying inter-individual variability in chronic pain severity. These findings support further validation of HFD-based normative modeling for mechanism-informed stratification, while requiring replication in larger, harmonized, and longitudinal datasets.