Work

EEG microstate dynamics as candidate neurophysiological correlates of attention-deficit/hyperactivity disorder in children and adolescents: a cross-sectional study

Year: 2026

Type: article

Abstract: Attention-deficit/hyperactivity disorder (ADHD) affects approximately 3–7% of school-age children and lacks objective neurophysiological correlates in clinical practice. Resting-state EEG microstate analysis quantifies transient large-scale brain network dynamics and offers a candidate-marker framework. We recorded resting-state EEG from 121 children aged 6–15 years: 58 typically developing (TD; Group A) and 63 with ADHD (Group B), using a 24-channel system (Pz online reference, offline average re-referencing). Artefacts were removed by independent component analysis (extended Infomax) with ICLabel component classification. Four microstate classes (A-D) were identified by k-means clustering (global explained variance 54.74% at k = 4; plateau across k = 4–7). Twenty-four microstate parameters (duration, occurrence, coverage, transitions) were compared between groups using Welch t-tests with Benjamini-Hochberg false discovery rate (BH-FDR) correction. Within the ADHD subgroup, linear regression related microstate C occurrence to the IVA-CPT response control quotients. Three machine-learning classifiers (logistic regression, random forest, and support vector machine) were trained and evaluated. Microstate C per-second occurrence (d = -0.56, q = 0.040) and the C-to-D transition probability (d = -0.58, q = 0.040) were significantly reduced in ADHD after FDR correction, and remained significant after adjustment for age, sex, and IQ. Within the ADHD subgroup, microstate C occurrence showed positive but exploratory associations with IVA-CPT a response control quotients (same direction as the group difference), none significant after correction for the three tests (BH-FDR p = 0.051). Under repeated 5-fold cross-validation the microstate-only classifier achieved AUC = 0.744 (95% CI 0.650–0.830); an IVA-CPT-only model (a behavioural attention test) reached AUC = 0.913 and a combined model AUC = 0.930; microstate features added no significant increment over IVA-CPT alone. Reduced microstate C engagement and C-to-D transition probability distinguish children with ADHD from TD peers with medium effect sizes, and show exploratory within-group associations with attentional control (not significant after multiple-comparison correction), most consistent for the visual response control quotient, in the same direction as the group difference. EEG microstate parameters are candidate neurophysiological correlates that may complement clinical ADHD assessment; in this sample they did not add incremental discriminative value beyond the IVA-CPT. Replication, external and longitudinal validation are required before clinical translation.

Source: BMC Psychiatry

DOI: 10.1186/s12888-026-08555-3