SA-HGNN: Sample-Adaptive Hyperbolic Graph Neural Network for EEG-Based Depression Recognition

Sep 21, 2026

Uploaded by Eliana

Graph Neural Networks (GNNs) improve electroencephalography (EEG)-based depression recognition by modeling spatial functional connectivity, yet they struggle to represent the inherent hierarchical organization of brain networks in patients with depression. This paper presents Sample-Adaptive Hyperbolic Graph Neural Network (SA-HGNN), combining personalized sample-adaptive graph construction, hyperbolic graph convolution to capture latent hierarchy beyond Euclidean bottlenecks, and attention pooling to suppress redundant noisy EEG channels. Experiments on public EEG datasets validate robust performance across resting-state and task-related settings, with effective recovery of abnormal connectivity patterns.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages6

Reading Time15 min

AccessFree to download

Abstract

I. INTRODUCTION

Major Depressive Disorder (MDD) and EEG-based diagnosis

Why graph neural networks for functional connectivity

Hierarchical rich-club disruption and modeling challenges

Proposed SA-HGNN approach and contributions

SA-HGNN: Sample-Adaptive Hyperbolic Graph Neural Network for EEG-Based Depression Recognition Yang Li 1∗, Pan Hu 1∗, Yan Zhang 1 , Wenfan Yang 1 , Lianbo Guo 1 , and Tao Wu 1† arXiv :2607 .02063v 1 [ cs .LG] 2 Jul 2026 Abstract—Graph Neural Networks (GNNs) have been widely used to capture spatial functional connectivity patterns to improve electroencephalography (EEG)-based depression recognition performance. However, the functional connectivity of brain networks in patients with depression exhibits an inherent hierarchical structure, making it difficult to capture accurate connection patterns. To address these issues, this paper proposesa novel model named Sample-Adaptive Hyperbolic Graph Neural Network (SA-HGNN), which aims to accurately extract the authentic hierarchical structure of depression-affected brain networks. Specifically, the proposed model comprises three core modules. First, a Sample-Adaptive Graph Construction module dynamically constructs personalized brain network topologies to capture more complex spatial relationships within the brain network. Second, hyperbolic graph convolution is employed to overcome the representation bottlenecks of Euclidean space, leveraging hyperbolic geometry to precisely capture latent hierarchical relationships within the brain network. Finally, an Attention Pooling module adaptively filters out highly redundant noise channels in EEG signals, effectively mitigating the interference of inherent noise on the authentic hierarchical topology. Extensive experiments on public EEG datasets demonstrate the superior performance of our method across restingstate and task-related paradigms, validating its robustness to noise and efficacy in capturing abnormal functional connectivity patterns in brain networks of patients with depression. Index Terms–Major Depressive Disorder Recognition, Electroencephalography, Graph Neural Networks, Hyperbolic Geometry, Graph Pooling. I. INTRODUCTION Major Depressive Disorder (MDD) has become a leading global health crisis. According to the World Health Organization, over 332 million people suffer from MDD[1], leading to trillions of dollars in annual economic losses [2] . Given the insidious nature of its early symptoms, developing timely and objective automated diagnostic methods is critical. Due to its non-invasiveness, low cost, and high temporal resolution, Electroencephalography (EEG) is widely used to capture dynamic electrophysiological activities for MDD diagnosis [3] . Early deep learning approaches, such as Convolutional Neural Networks (CNNs) and Transformers, have shown strong capabilities in extracting spatiotemporal EEG features [4] . However, extensive neurophysiological studies indicate that MDD is fundamentally a disruption of whole-brain functional connectivity rather than a localized lesion, which is closely related to clinical symptoms [5] . Consequently, Graph Neural Networks (GNNs) have rapidly become the 1 School of Software Engineering, Huazhong University of Science and Technology, Wuhan, Hubei, China ∗ Equal contribution, † Corresponding author Euclidean space distance metric Brain region Measurement Euclidean distance Functional adjacency matrix Hyperbolic space distance metric (Ours) Measurement Brain region Hyperbolic distance Functional adjacency matrix (a) (b) Fig. 1: Measurement methods of brain functional connectivity: (a) Euclidean space metric. (b) Hyperbolic space metric (Ours) . mainstream approach, as they can naturally model the brain network and effectively capture these abnormal functional connectivity patterns [6] . Focusing on graph structure learning for MDD, recent GNN methods have achieved promising results by explicitly modeling EEG topological structures. For example, MASTGCN [7] utilizes adaptive spatial-temporal convolutions to capture dynamic EEG dependencies, while TFAGL [8] constructs dynamic local-global graphs to simulate regional brain collaborations.

FAQs

What problem does SA-HGNN address in EEG-based depression recognition?

It addresses the difficulty of capturing accurate hierarchical connection patterns in depression-affected brain functional connectivity using conventional Euclidean-space GNN modeling.

What are the three core modules of SA-HGNN?

It uses a Sample-Adaptive Graph Construction module, hyperbolic graph convolution to model latent hierarchy, and an Attention Pooling module to filter redundant noise channels in EEG signals.

How is the effectiveness of SA-HGNN validated?

Extensive experiments on public EEG datasets show superior performance across resting-state and task-related paradigms, demonstrating robustness to noise and efficacy in capturing abnormal functional connectivity patterns.

Sep 21, 2026

Uploaded by Eliana