Multilevel brain functional heterogeneity in major depressive disorder based on normative modeling of multicenter resting-state EEG.

BACKGROUND: Depression exhibits substantial neurobiological heterogeneity. Conventional group-level EEG analyses often fail to identify reproducible biomarkers, limiting objective diagnosis and treatment. This study employed a normative modeling framework to characterize individual EEG abnormalities by quantifying deviations from a healthy reference distribution. METHOD: This study integrated multicenter resting-state EEG data from 1163 participants (participants with major depressive disorder [MDD] = 369; healthy controls [HC] = 794). Normative models were constructed using Dortmund Vital Study data (HC = 608). In the test set, individual deviations in time- and frequency-domain EEG features were quantified at both scalp and source levels. Abnormal patterns were subsequently characterized across electrode, regional, connectivity, and network levels. Group differences were assessed using permutation testing with FDR correction. RESULTS: Normative modeling revealed that EEG abnormalities in MDD were predominantly characterized by highly individualized deviation patterns, with limited overlap at the group level. Significant abnormalities were mainly observed in prefrontal regions and the orbitofrontal cortex (OFC), particularly in beta-band relative power. At the network level, only the limbic subnetwork remained significant after multiple-comparison correction, with approximately 25% of MDD patients showing significant deviations. CONCLUSION: Multilevel EEG analysis based on normative modeling reveals pronounced individualized neurofunctional abnormalities in MDD and identifies relatively stable alterations in the OFC and limbic network. This framework provides a robust basis for characterizing neurobiological heterogeneity and supporting individualized precision diagnosis in depression.