Discrete Triple-Network Coupling Architectures Orthogonal to Psychiatric Diagnosis

Gaussian mixture modeling of triple-network resting-state coupling (DMN-CEN, DMN-SN, CEN-SN) in a transdiagnostic sample (healthy controls, ADHD, bipolar, schizophrenia) using fMRIPrep-preprocessed data with 13-regressor confound regression reveals two discrete architectures: a low-coupling mode (75.9%) and a high-coupling mode (24.1%). Diagnostic category does not predict cluster membership (chi-squared p=0.748). Coupling topology was orthogonal to psychiatric classification in this sample.