A Dual-Layer Normative Deviation System with Regime Mapping for Resting EEG
Description
A dual-layer resting EEG framework for subject-level biomarker mapping that separates magnitude of deviation from type of large-scale cortical organization. The system combines a multivariate Normative Deviation Index (NDI), a Universal Architecture Score (UAS), and a joint regime map. Evaluated on a TDBRAIN-derived labeled cohort of 911 subjects, the framework achieved AUC = 0.824 for clinical versus healthy, AUC = 0.821 for ADHD versus healthy, and AUC = 0.832 for MDD versus healthy. Reduced subject-level baselines were weaker, supporting the importance of multivariate organization relative to healthy reference structure. The joint UAS-NDI space revealed distinct regimes including healthy-like compact states, preserved-structure high-deviation states, structural-shift states, and reorganized states. This reframes resting EEG biomarker analysis from single-score classification toward structured deviation mapping.
That description is faithful to the manuscript abstract and results.
Files
A_Dual_Layer_Normative_Deviation_System_with_Regime_Mapping_for_Resting_EEG__Architecture__Optimization__and_Subject_Level_Inference.pdf
Files
(604.6 kB)
| Name | Size | Download all |
|---|---|---|
|
md5:d816cbbb6612d2e5777c70699b88698f
|
604.6 kB | Preview Download |