Distilling Trial-Level EEG Variability into Participant Representations: Neurophysiologically Constrained Graph Learning in ADHD
- 1TU Dresden, Universitatsklinikum Carl Gustav Carus
Abstract
Extracting stable individual-level representations from noisy and variable trial-wise EEG remains a major challenge in computational cognitive neuroscience. We developed a spatiotemporal graph attention network to analyze Go/NoGo EEG from children and adolescents with and without ADHD. The framework combines ERP-informed temporal windows, phase-synchrony graph priors, and trial-to-participant transfer learning, making both information routing and representation geometry explicitly testable. The model learned reproducible attention trajectories, yet ADHD and control participants did not differ reliably in spatial attention allocation. Instead, group-related structure emerged in compact participant-level embeddings that distilled trial-level variability and retained discriminative information beyond convolutional representations. Performance exceeded a convolutional neural network comparison, but the central advance is methodological: neurophysiologically constrained graph learning can reveal where diagnostically relevant information emerges within the analytical pipeline. This provides a framework for developing auditable cognitive-neurophysiological markers of neurodevelopmental variation.
Related
- Distilling Neural Population Geometry into Transformers and GRUs for Motor Control
- Predictors of pulmonary embolism in chronic obstructive pulmonary diseases patients admitted for worsening respiratory symptoms: An individual participant data meta-analyses
- Participant-specific macromolecule information improves quantification of GABA-edited spectra at 3T
- Participant perspectives on a person-centred care model for mobile individuals in HIV care from the SEARCH Dynamic Choice Treatment trial in rural Kenya and Uganda
- Inconsistent ethics reporting in individual participant data meta-analyses: a call for updating PRISMA and PRISMA-IPD
BibTeX
@misc{tiwari2026distilling,
title = {{Distilling Trial-Level EEG Variability into Participant Representations: Neurophysiologically Constrained Graph Learning in ADHD}},
author = {R. Tiwari and C. Beste and N. Gholamipourbarough},
howpublished = {bioRxiv},
year = {2026},
month = oct,
doi = {10.64898/2026.10.01.755897},
url = {https://www.biorxiv.org/content/10.64898/2026.10.01.755897v1},
note = {CC BY-NC-ND},
}
RIS
TY - UNPB
TI - Distilling Trial-Level EEG Variability into Participant Representations: Neurophysiologically Constrained Graph Learning in ADHD
AU - R. Tiwari
AU - C. Beste
AU - N. Gholamipourbarough
PY - 2026
DA - 2026/10/01
PB - bioRxiv
DO - 10.64898/2026.10.01.755897
UR - https://www.biorxiv.org/content/10.64898/2026.10.01.755897v1
AB - Extracting stable individual-level representations from noisy and variable trial-wise EEG remains a major challenge in computational cognitive neuroscience. We developed a spatiotemporal graph attention network to analyze Go/NoGo EEG from children and adolescents with and without ADHD. The framework combines ERP-informed temporal windows, phase-synchrony graph priors, and trial-to-participant transfer learning, making both information routing and representation geometry explicitly testable. The model learned reproducible attention trajectories, yet ADHD and control participants did not differ reliably in spatial attention allocation. Instead, group-related structure emerged in compact participant-level embeddings that distilled trial-level variability and retained discriminative information beyond convolutional representations. Performance exceeded a convolutional neural network comparison, but the central advance is methodological: neurophysiologically constrained graph learning can reveal where diagnostically relevant information emerges within the analytical pipeline. This provides a framework for developing auditable cognitive-neurophysiological markers of neurodevelopmental variation.
ER -