EEG-SeeGraph: Interpreting functional connectivity disruptions in dementias via sparse-explanatory dynamic EEG-graph learning

Fengcheng Wu; Zhenxi Song; Guoyang Xu; Kaisong Hu; Zirui Wang; Yi Guo; Zhiguo Zhang

eess.SPMar 3, 2026

EEG-SeeGraph: Interpreting functional connectivity disruptions in dementias via sparse-explanatory dynamic EEG-graph learning

Authors: Fengcheng Wu, Zhenxi Song, Guoyang Xu, Kaisong Hu, Zirui Wang, Yi Guo, Zhiguo Zhang

Organizations: Harbin Institute of Technology, Shenzhen, China · Institute of Neurological Diseases, Shenzhen Bay Laboratory, Shenzhen · Institute of Neurological Diseases, Shenzhen Bay Laboratory, Shenzhen, China

Abstract

Robust and interpretable dementia diagnosis from noisy, non-stationary electroencephalography (EEG) is clinically essential yet remains challenging. To this end, we propose SeeGraph, a Sparse-Explanatory dynamic EEG-graph network that models time-evolving functional connectivity and employs a node-guided sparse edge mask to reveal the connections that drive diagnostic decisions, while remaining robust to noise and cross-site variability. SeeGraph comprises four components: (1) a dual-trajectory temporal encoder that models dynamic EEG with two streams, where node signals capture regional oscillations and edge signals capture interregional coupling; (2) a topology-aware positional encoder that derives graph-spectral Laplacian coordinates from the fused connectivity and augments node embeddings; (3) a node-guided sparse explanatory edge mask that gates the connectivity into a compact subgraph; and (4) a gated graph predictor that operates on the sparsified graph. The framework is trained using cross-entropy loss together with a sparsity regularizer on the mask, yielding noise-robust and interpretable diagnoses. The effectiveness of SeeGraph is validated on public and in-house EEG cohorts, including patients with neurodegenerative dementias and healthy controls, under both raw and noise-perturbed conditions. Its sparse, node-guided explanations highlight disease-relevant connections and align with established clinical findings on functional connectivity alterations, thereby offering transparent cues for neurological evaluation.

Figures & tables

Figure 1: Architecture of SeeGraph : dynamic EEG graphs are constructed, node and connectivity trajectories are temporally encoded, Laplacian positional encodings augment node embeddings, and a node-guided sparse mask yields an explanatory subgraph for a gated graph predictor. The objective combines diagnosis cross-entropy with a sparsity regularizer on the edge mask to learn compact, interpretable connectivity.

Method	SZPH Dataset ( China )						AHEPA Dataset ( Greece )
	Raw (no added noise)			Noisy			Raw (no added noise)			Noisy
	ACC	AUROC	F1	ACC	AUROC	F1	ACC	AUROC	F1	ACC	AUROC	F1
BIOT [ 27 ]	0.765	0.820	0.761	0.621	0.767	0.607	0.812	0.937	0.798	0.733	0.786	0.712
DCRNN [ 21 ]	0.760	0.819	0.779	0.661	0.740	0.737	0.738	0.819	0.719	0.651	0.734	0.657
EGCN [ 15 ]	0.734	0.726	0.718	0.645	0.643	0.660	0.648	0.769	0.614	0.602	0.736	0.522
CNN-LSTM	0.770	0.804	0.772	0.723	0.731	0.695	0.635	0.752	0.556	0.592	0.631	0.542

Table 1: Performance comparison on the SZPH and AHEPA datasets under normal and noisy conditions

Bands	SZPH Dataset ( China )						AHEPA dataset ( Greece )
	Raw (no added noise)			Noisy			Raw (no added noise)			Noisy
	ACC	AUROC	F1	ACC	AUROC	F1	ACC	AUROC	F1	ACC	AUROC	F1
$\delta$ (Delta)	0.693	0.795	0.716	0.663	0.765	0.664	NAN	NAN	NAN	NAN	NAN	NAN
$\theta$ (Theta)	0.702	0.867	0.621	0.815	0.830	0.804	0.739	0.897	0.719	0.725	0.876	0.710
$\alpha$ (Alpha)	0.792	0.817	0.773	0.729	0.816	0.745	0.881	0.969	0.876	0.832	0.937	0.851
$\beta$ (Beta)	0.828	0.887	0.818	0.785	0.862	0.764	0.827	0.948	0.820	0.789	0.895	0.763

Table 2: Diagnostic utility of EEG frequency bands for dementia diagnosis across two cohorts

Method	SZPH			AHEPA
Method	ACC	AUROC	F1	ACC	AUROC	F1
SeeGraph	0.874	0.951	0.875	0.841	0.953	0.834
w/o C-wise	0.855	0.938	0.862	0.823	0.904	0.805
w/o PE	0.837	0.905	0.841	0.824	0.918	0.829
w/o SR	0.770	0.864	0.761	0.812	0.937	0.798
w/o FFT	0.758	0.842	0.772	0.784	0.924	0.768

Table 3: Ablation study of SeeGraph

Figure 2: SeeGraph localizes salient fronto-temporal subnetworks via a node-guided sparse mask. AD exhibits temporal-centered ipsilateral clustering (red), while HC shows more interhemispheric and bilateral integration (blue); darker colors indicate higher salience.

Electroencephalography

Electroencephalography (EEG) studies brain-generated electrical activity to understand cognition and support diagnosis, monitoring, and brain-computer interfaces.

See all 180 papers

M-LINKX: Multiview Graph Learning for Brain Cognitive Disease Detection
Aug 14, 2026An Phan, Yufei Jin, Xingquan ZhuElectroencephalography Multi-View Learning
Dept. of Electrical Engineering and Computer Science, Florida Atlantic University, Boca Raton, FL 33431, USA
Task-guided Spatiotemporal Network with Diffusion Augmentation for EEG-based Dementia Diagnosis and MMSE Prediction
Apr 27, 2026Xiaoyu Zheng, Xu Tian, Bin Jiao +4Alzheimer's Disease Electroencephalography
Xinjiang Engineering Research Center of Big Data and Intelligent Software, School of Software, Xinjiang University, Urumqi 830091, China. · Hunan Provincial Key Lab on Bioinformatics, School of Computer Science and Engineering, Central South University, Changsha 410083, China. · Department of Neurology, Xiangya Hospital, Central South University, Changsha 410083, China. +1
Adaptive Multi-Expert Graph Transformer for Interpretable EEG-Based Diagnostics
Jul 20, 2026Maryam Rahimimovassagh, Md Elias Hossain, Ivan Garibay +1Electroencephalography Dynamic Graph Learning
Department of Industrial Engineering and Management Systems University of Central Florida, Orlando, FL, USA

Aug 14, 2026J/K move · Enter open · S save

M-LINKX: Multiview Graph Learning for Brain Cognitive Disease Detection

An Phan, Yufei Jin, Xingquan Zhu

Dept. of Electrical Engineering and Computer Science, Florida Atlantic University, Boca Raton, FL 33431, USA

Electroencephalography Multi-View Learning Brain Network Analysis

Open paper AskPDF arXiv

Electroencephalogram (EEG) is a non-invasive and relatively low-cost procedure that measures brain electricity for the detection of cognitive diseases. EEG-based classification of dementia-related conditions, including Alzheimer's disease (AD), mild cognitive impairment (MCI), and frontotemporal dementia (FTD), remains challenging because EEG signals are noisy, non-stationary, and vary across subjects. Segment-based learning provides a practical way to model long EEG recordings by converting them into fixed-length inputs. For each segment, discriminative information may be explored by using signals within each channel (i.e. electrode), as well as interactions between EEG channels. In this paper, we propose M-LINKX, a multi-view graph learning framework for EEG-based dementia classification. For each segment, we extract channel-level node features and construct multiple functional-connectivity (FC) graph views, where each view is defined by a specific combination of connectivity metric, frequency band, and topology filter, respectively. Instead of relying on message passing over the constructed graphs, M-LINKX follows a simple design in modeling node features and adjacency-based connectivity representations. The graph-view representations are fused using global trainable view weights, and subject-level prediction is obtained by averaging segment-level probabilities. Experiments on two three-class EEG datasets with different diagnostic groups, CAUEEG (HC/MCI/Dementia) and AHEAP (HC/AD/FTD), show that M-LINKX achieves the best subject-level performance under the main experimental settings. Our study suggests that multi-view functional connectivity can improve EEG-based dementia classification when integrated with an appropriate graph-learning architecture. Code and data are available at https://github.com/anphantt/MLINKX.

Abstract

Electroencephalography

Explore similar work

M-LINKX: Multiview Graph Learning for Brain Cognitive Disease Detection

Task-guided Spatiotemporal Network with Diffusion Augmentation for EEG-based Dementia Diagnosis and MMSE Prediction

Adaptive Multi-Expert Graph Transformer for Interpretable EEG-Based Diagnostics