pith. sign in

arxiv: 2410.18103 · v2 · submitted 2024-10-08 · 📡 eess.SP · cs.AI· cs.LG

A Hybrid Graph Neural Network for Enhanced EEG-Based Depression Detection

Yiye Wang , Wenming Zheng , Yang Li , Hao Yang This is my paper

Pith reviewed 2026-05-23 19:33 UTC · model grok-4.3

classification 📡 eess.SP cs.AIcs.LG
keywords EEGdepression detectiongraph neural networkshybrid architecturebrain connectivityhierarchical poolingsignal processing
0
0 comments X

The pith

A hybrid graph neural network merges fixed and adaptive connections to capture both common and individualized depression patterns in EEG data.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper aims to improve EEG-based depression detection by addressing two gaps in existing GNN methods. It introduces a hybrid model with one branch using fixed graphs for common brain abnormalities across patients and another using adaptive graphs for unique patterns in each individual. The adaptive branch is further equipped with pooling and unpooling to capture hierarchical brain structures that differ between people. This combination is shown through experiments to reach state-of-the-art results on two public datasets, indicating that both types of patterns and the hierarchy are important for accurate detection.

Core claim

The central discovery is that a Hybrid GNN architecture, consisting of a Common Graph Neural Network branch with fixed connections and an Individualized Graph Neural Network branch with adaptive connections enhanced by a Graph Pooling and Unpooling Module, can effectively capture both common and individualized depression patterns as well as individualized hierarchical information from EEG signals, leading to improved detection performance.

What carries the argument

The Hybrid GNN (HGNN) architecture that integrates CGNN for fixed common patterns, IGNN for adaptive individualized patterns, and GPUM for hierarchical extraction.

If this is right

  • The hybrid approach allows the model to complement common and individualized pattern capture in brain networks.
  • Graph pooling and unpooling extracts hierarchical channel-to-region information that varies individually.
  • State-of-the-art performance is achieved on two public EEG depression datasets.
  • Previous methods are limited by focusing on only one type of connection or ignoring hierarchy.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • If the hybrid structure works, it could be adapted for detecting other mental health conditions with similar pattern mixtures in brain signals.
  • The emphasis on individualized hierarchy suggests potential for patient-specific models in clinical EEG analysis.
  • Testing the model on datasets with different demographics could reveal how well the common patterns generalize.

Load-bearing premise

That combining fixed and adaptive graph connections is necessary and sufficient to capture the mix of common and individualized depression patterns, and that the varying hierarchical structure holds key depression information.

What would settle it

Running the proposed model against versions without the IGNN branch or without the GPUM on the same two public datasets and finding no performance improvement would challenge the claim.

Figures

Figures reproduced from arXiv: 2410.18103 by Hao Yang, Wenming Zheng, Yang Li, Yiye Wang.

Figure 1
Figure 1. Figure 1: The framework of the proposed HybGNN 2.2 EEG Temporal Feature Extraction 1-D CNN has been widely used in multivariate time series, such as biomedical data, due to its competitive performance, as well as its real-time and low-cost hardware implementation compared to RNN-based methods [25]. We employ a 1-D CNN module to extract temporal dependencies within each electrode, which is operated on time dimension … view at source ↗
Figure 2
Figure 2. Figure 2: Hyperparameter optimization for the number of regions [PITH_FULL_IMAGE:figures/full_fig_p011_2.png] view at source ↗
read the original abstract

Graph neural networks (GNNs) are becoming increasingly popular for EEG-based depression detection. However, previous GNN-based methods fail to sufficiently consider the characteristics of depression, thus limiting their performance. Firstly, studies in neuroscience indicate that depression patients exhibit both common and individualized brain abnormal patterns. Previous GNN-based approaches typically focus either on fixed graph connections to capture common abnormal brain patterns or on adaptive connections to capture individualized patterns, which is inadequate for depression detection. Secondly, brain network exhibits a hierarchical structure, which includes the arrangement from channel-level graph to region-level graph. This hierarchical structure varies among individuals and contains significant information relevant to detecting depression. Nonetheless, previous GNN-based methods overlook these individualized hierarchical information. To address these issues, we propose a Hybrid GNN (HGNN) that merges a Common Graph Neural Network (CGNN) branch utilizing fixed connection and an Individualized Graph Neural Network (IGNN) branch employing adaptive connections. The two branches capture common and individualized depression patterns respectively, complementing each other. Furthermore, we enhance the IGNN branch with a Graph Pooling and Unpooling Module (GPUM) to extract individualized hierarchical information. Extensive experiments on two public datasets show that our model achieves state-of-the-art performance.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

3 major / 2 minor

Summary. The manuscript proposes a Hybrid Graph Neural Network (HGNN) for EEG-based depression detection. It combines a Common Graph Neural Network (CGNN) branch using fixed graph connections to capture common abnormal brain patterns with an Individualized Graph Neural Network (IGNN) branch using adaptive connections, augmented by a Graph Pooling and Unpooling Module (GPUM) to extract individualized hierarchical channel-to-region information. The central claim is that this architecture addresses limitations of prior GNN methods and achieves state-of-the-art performance on two public datasets by better modeling neuroscience-derived characteristics of depression.

Significance. If the empirical claims hold with proper validation, the work could advance EEG depression detection by explicitly incorporating both shared and subject-specific brain network features plus hierarchical structure, potentially leading to more accurate and interpretable models. The hybrid fixed/adaptive design and GPUM module represent a targeted architectural response to the stated neuroscience motivations.

major comments (3)
  1. [Abstract / §1] Abstract and §1 (motivation): The claim that the CGNN branch isolates common patterns while the IGNN+GPUM branch isolates individualized hierarchical patterns is load-bearing for the paper's interpretation, yet no branch-specific ablations, adjacency-matrix visualizations, or pattern-separation analyses are described to rule out the alternative that gains arise simply from added model capacity or ensemble effects.
  2. [Abstract] Abstract: The assertion of state-of-the-art performance on two public datasets supplies no quantitative metrics, baseline comparisons, statistical tests, or ablation tables; without these in the experimental section the central performance claim cannot be evaluated.
  3. [§3 / §4] §3 (method) and §4 (experiments): Graph-construction hyperparameters and adaptive-connection learning are fitted on the same datasets used for final evaluation, creating a circularity risk for the individualized-pattern claim; an external validation set or parameter-free baseline would be required to strengthen the result.
minor comments (2)
  1. [§3] Notation for the GPUM module (pooling/unpooling operations) should be defined with explicit equations to clarify how hierarchical information is preserved across individuals.
  2. [§4] The two public datasets should be named with citation and basic statistics (subject count, channel count, class balance) in the experimental setup.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the constructive comments. We address each major comment point by point below.

read point-by-point responses

  1. Referee: [Abstract / §1] Abstract and §1 (motivation): The claim that the CGNN branch isolates common patterns while the IGNN+GPUM branch isolates individualized hierarchical patterns is load-bearing for the paper's interpretation, yet no branch-specific ablations, adjacency-matrix visualizations, or pattern-separation analyses are described to rule out the alternative that gains arise simply from added model capacity or ensemble effects.

    Authors: We agree that explicit evidence separating the contributions of each branch would strengthen the neuroscience-based interpretation. The revised manuscript will add branch-specific ablations (CGNN-only, IGNN-only, IGNN without GPUM), visualizations of fixed versus learned adjacency matrices, and comparative analyses of branch-specific representations to demonstrate that performance gains reflect complementary modeling of common and individualized patterns rather than capacity or ensemble effects alone. revision: yes

  2. Referee: [Abstract] Abstract: The assertion of state-of-the-art performance on two public datasets supplies no quantitative metrics, baseline comparisons, statistical tests, or ablation tables; without these in the experimental section the central performance claim cannot be evaluated.

    Authors: Section 4 already contains the requested quantitative metrics, baseline comparisons, statistical tests, and ablation tables on both datasets. To make the abstract self-contained, the revision will incorporate key performance numbers and a summary of the comparisons. revision: partial

  3. Referee: [§3 / §4] §3 (method) and §4 (experiments): Graph-construction hyperparameters and adaptive-connection learning are fitted on the same datasets used for final evaluation, creating a circularity risk for the individualized-pattern claim; an external validation set or parameter-free baseline would be required to strengthen the result.

    Authors: Hyperparameters were selected via cross-validation on training folds, following standard practice for these public benchmarks. The revision will add a parameter-free baseline and expand discussion of the validation procedure to reduce circularity concerns. revision: partial

Circularity Check

0 steps flagged

No circularity; empirical architecture validated on held-out data

full rationale

The paper proposes a hybrid CGNN+IGNN+GPUM architecture motivated by external neuroscience observations about common/individualized patterns and hierarchical brain structure. It reports SOTA accuracy after training and testing on two public datasets. No derivation chain, equations, or first-principles result is presented that reduces by construction to fitted parameters, self-citations, or renamed inputs. The performance numbers are ordinary empirical outcomes of model fitting, not tautological predictions.

Axiom & Free-Parameter Ledger

1 free parameters · 2 axioms · 0 invented entities

The paper rests on standard supervised-learning assumptions plus two domain assumptions about brain networks; no new physical entities are introduced.

free parameters (1)
  • graph construction hyperparameters
    Fixed vs. adaptive edge weights and pooling ratios are chosen or learned from data to define the common and individualized graphs.
axioms (2)
  • domain assumption Depression patients exhibit both common and individualized brain abnormal patterns that can be separated by fixed versus adaptive graph connections.
    Invoked in the first motivation paragraph to justify the dual-branch design.
  • domain assumption Brain networks possess an individualized hierarchical structure from channel-level to region-level graphs that carries depression-relevant information.
    Invoked in the second motivation paragraph to justify the GPUM module.

pith-pipeline@v0.9.0 · 5756 in / 1403 out tokens · 29261 ms · 2026-05-23T19:33:45.847630+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. AI Models for Depressive Disorder Detection and Diagnosis: A Review

    cs.AI 2025-08 accept novelty 5.0

    A systematic review of AI for depressive disorder detection that introduces a novel hierarchical taxonomy organized by clinical task, data modality, and model class.

Reference graph

Works this paper leans on

49 extracted references · 49 canonical work pages · cited by 1 Pith paper · 1 internal anchor

  1. [1]

    Neuron28(2), 335–341 (2000)

    Fava, M., Kendler, K.S.: Major depressive disorder. Neuron28(2), 335–341 (2000)

    work page 2000

  2. [2]

    Nature reviews Disease primers 2(1), 1–20 (2016) 12 Y

    Otte, C., Gold, S.M., Penninx, B.W., Pariante, C.M., Etkin, A., Fava, M., Mohr, D.C., Schatzberg, A.F.: Major depressive disorder. Nature reviews Disease primers 2(1), 1–20 (2016) 12 Y. Wang et al

    work page 2016

  3. [3]

    Archives of general psychiatry66(7), 785–795 (2009)

    Seedat, S., Scott, K.M., Angermeyer, M.C., Berglund, P., Bromet, E.J., Brugha, T.S., Demyttenaere, K., De Girolamo, G., Haro, J.M., Jin, R., et al.: Cross-national associations between gender and mental disorders in the world health organization world mental health surveys. Archives of general psychiatry66(7), 785–795 (2009)

    work page 2009

  4. [4]

    IEEE Transactions on Affective Computing11(2), 272–283 (2017)

    Cummins, N., Sethu, V., Epps, J., Williamson, J.R., Quatieri, T.F., Krajewski, J.: Generalized two-stage rank regression framework for depression score prediction from speech. IEEE Transactions on Affective Computing11(2), 272–283 (2017)

    work page 2017

  5. [5]

    Neuropsychiatric Disease and Treatment pp

    Chen, J., Lv, Y.n., Li, X.b., Xiong, J.j., Liang, H.t., Xie, L., Wan, C.y., Chen, Y.q.,Wang,H.s.,Liu, P.,etal.: Urinarymetabolite signaturesforpredictingelderly stroke survivors with depression. Neuropsychiatric Disease and Treatment pp. 925– 933 (2021)

    work page 2021

  6. [6]

    Journal of adolescence33(3), 429–438 (2010)

    Calear, A.L., Christensen, H.: Systematic review of school-based prevention and early intervention programs for depression. Journal of adolescence33(3), 429–438 (2010)

    work page 2010

  7. [7]

    European neurology 73(5-6), 329–336 (2015)

    Acharya, U.R., Sudarshan, V.K., Adeli, H., Santhosh, J., Koh, J.E., Adeli, A.: Computer-aided diagnosis of depression using eeg signals. European neurology 73(5-6), 329–336 (2015)

    work page 2015

  8. [8]

    Ieee Access7, 92630–92641 (2019)

    Peng, H., Xia, C., Wang, Z., Zhu, J., Zhang, X., Sun, S., Li, J., Huo, X., Li, X.: Multivariate pattern analysis of eeg-based functional connectivity: A study on the identification of depression. Ieee Access7, 92630–92641 (2019)

    work page 2019

  9. [9]

    IEEE Journal of Biomedical and Health Informatics 26(7), 3466–3477 (2022)

    Zhang, B., Cai, H., Song, Y., Tao, L., Li, Y.: Computer-aided recognition based on decision-level multimodal fusion for depression. IEEE Journal of Biomedical and Health Informatics 26(7), 3466–3477 (2022)

    work page 2022

  10. [10]

    Computers in Biology and Medicine148, 105815 (2022)

    Zhu, J., Jiang, C., Chen, J., Lin, X., Yu, R., Li, X., Hu, B.: Eeg based depres- sion recognition using improved graph convolutional neural network. Computers in Biology and Medicine148, 105815 (2022)

    work page 2022

  11. [11]

    IEEE Journal of Biomedical and Health Informatics (2023)

    Shen, J., Zhang, Y., Liang, H., Zhao, Z., Zhu, K., Qian, K., Dong, Q., Zhang, X., Hu, B.: Depression recognition from eeg signals using an adaptive channel fusion method via improved focal loss. IEEE Journal of Biomedical and Health Informatics (2023)

    work page 2023

  12. [12]

    IEEE Transactions on Neural Systems and Rehabilitation Engineering 31, 356–365 (2022)

    Shen, J., Zhang, Y., Liang, H., Zhao, Z., Dong, Q., Qian, K., Zhang, X., Hu, B.: Exploring the intrinsic features of eeg signals via empirical mode decomposition for depression recognition. IEEE Transactions on Neural Systems and Rehabilitation Engineering 31, 356–365 (2022)

    work page 2022

  13. [13]

    IEEE trans- actions on affective computing13(4), 2106–2118 (2022)

    Chen, T., Guo, Y., Hao, S., Hong, R.: Exploring self-attention graph pooling with eeg-based topological structure and soft label for depression detection. IEEE trans- actions on affective computing13(4), 2106–2118 (2022)

    work page 2022

  14. [14]

    IEEE Transactions on Instrumentation and Measurement70, 1–13 (2021)

    Seal, A., Bajpai, R., Agnihotri, J., Yazidi, A., Herrera-Viedma, E., Krejcar, O.: De- prnet: A deep convolution neural network framework for detecting depression using eeg. IEEE Transactions on Instrumentation and Measurement70, 1–13 (2021)

    work page 2021

  15. [15]

    IEEE Access (2023)

    Sharma, G., Joshi, A.M., Gupta, R., Cenkeramaddi, L.R.: Depcap: a smart health- care framework for eeg based depression detection using time-frequency response and deep neural network. IEEE Access (2023)

    work page 2023

  16. [16]

    In: 2021 IEEE Inter- national Conference on Bioinformatics and Biomedicine (BIBM)

    Wang, D., Lei, C., Zhang, X., Wu, H., Zheng, S., Chao, J., Peng, H.: Identification of depression with a semi-supervised gcn based on eeg data. In: 2021 IEEE Inter- national Conference on Bioinformatics and Biomedicine (BIBM). pp. 2338–2345. IEEE (2021)

    work page 2021

  17. [17]

    In: 2023 IEEE Inter- national Conference on Bioinformatics and Biomedicine (BIBM)

    Shen, J., Chen, J., Ma, Y., Cao, Z., Zhang, Y., Hu, B.: Explainable depression recognition from eeg signals via graph convolutional network. In: 2023 IEEE Inter- national Conference on Bioinformatics and Biomedicine (BIBM). pp. 1406–1412. IEEE (2023) A Hybrid Graph Neural Network 13

    work page 2023

  18. [18]

    Chaos: An Inter- disciplinary Journal of Nonlinear Science25(2) (2015)

    Yu, M., Hillebrand, A., Tewarie, P., Meier, J., van Dijk, B., Van Mieghem, P., Stam, C.J.: Hierarchical clustering in minimum spanning trees. Chaos: An Inter- disciplinary Journal of Nonlinear Science25(2) (2015)

    work page 2015

  19. [19]

    Complexity2017 (2017)

    Li, X., Jing, Z., Hu, B., Zhu, J., Zhong, N., Li, M., Ding, Z., Yang, J., Zhang, L., Feng, L., et al.: A resting-state brain functional network study in mdd based on minimum spanning tree analysis and the hierarchical clustering. Complexity2017 (2017)

    work page 2017

  20. [20]

    Cui, W., Sun, M., Dong, Q., Guo, Y., Liao, X.F., Li, Y.: A multiview sparse dynamicgraphconvolution-basedregion-attentionfeaturefusionnetworkformajor depressivedisorderdetection.IEEETransactionsonComputationalSocialSystems (2023)

    work page 2023

  21. [21]

    Frontiers in Psychiatry14, 1125339 (2023)

    Venkatapathy, S., Votinov, M., Wagels, L., Kim, S., Lee, M., Habel, U., Ra, I.H., Jo, H.G.: Ensemble graph neural network model for classification of major depres- sive disorder using whole-brain functional connectivity. Frontiers in Psychiatry14, 1125339 (2023)

    work page 2023

  22. [22]

    Translational psychiatry9(1), 136 (2019)

    Jiang, X., Shen, Y., Yao, J., Zhang, L., Xu, L., Feng, R., Cai, L., Liu, J., Chen, W., Wang, J.: Connectome analysis of functional and structural hemispheric brain networks in major depressive disorder. Translational psychiatry9(1), 136 (2019)

    work page 2019

  23. [23]

    Advances in neural in- formation processing systems31 (2018)

    Ying, Z., You, J., Morris, C., Ren, X., Hamilton, W., Leskovec, J.: Hierarchical graph representation learning with differentiable pooling. Advances in neural in- formation processing systems31 (2018)

    work page 2018

  24. [24]

    In: international conference on machine learning

    Gao, H., Ji, S.: Graph u-nets. In: international conference on machine learning. pp. 2083–2092. PMLR (2019)

    work page 2083

  25. [25]

    Mechanical systems and signal processing 151, 107398 (2021)

    Kiranyaz, S., Avci, O., Abdeljaber, O., Ince, T., Gabbouj, M., Inman, D.J.: 1d convolutional neural networks and applications: A survey. Mechanical systems and signal processing 151, 107398 (2021)

    work page 2021

  26. [26]

    Journal of neural engineering15(5), 056013 (2018)

    Lawhern, V.J., Solon, A.J., Waytowich, N.R., Gordon, S.M., Hung, C.P., Lance, B.J.: Eegnet: a compact convolutional neural network for eeg-based brain– computer interfaces. Journal of neural engineering15(5), 056013 (2018)

    work page 2018

  27. [27]

    Cerebral cortex15(9), 1332–1342 (2005)

    Salvador, R., Suckling, J., Coleman, M.R., Pickard, J.D., Menon, D., Bullmore, E.: Neurophysiological architecture of functional magnetic resonance images of human brain. Cerebral cortex15(9), 1332–1342 (2005)

    work page 2005

  28. [28]

    IEEE Transactions on Affective Computing 11(3), 532–541 (2018)

    Song, T., Zheng, W., Song, P., Cui, Z.: Eeg emotion recognition using dynamical graph convolutional neural networks. IEEE Transactions on Affective Computing 11(3), 532–541 (2018)

    work page 2018

  29. [29]

    Advances in neural information pro- cessing systems 30 (2017)

    Vaswani,A.,Shazeer,N.,Parmar,N.,Uszkoreit,J.,Jones,L.,Gomez,A.N.,Kaiser, Ł., Polosukhin, I.: Attention is all you need. Advances in neural information pro- cessing systems 30 (2017)

    work page 2017

  30. [30]

    Semi-Supervised Classification with Graph Convolutional Networks

    Kipf, T.N., Welling, M.: Semi-supervised classification with graph convolutional networks. arXiv preprint arXiv:1609.02907 (2016)

    work page internal anchor Pith review Pith/arXiv arXiv 2016

  31. [31]

    Frontiers in neuroscience4, 7572 (2010)

    Meunier, D., Lambiotte, R., Bullmore, E.T.: Modular and hierarchically modular organization of brain networks. Frontiers in neuroscience4, 7572 (2010)

    work page 2010

  32. [32]

    arxiv 2020

    Cai, H., Gao, Y., Sun, S., Li, N., Tian, F., Xiao, H., Li, J., Yang, Z., Li, X., Zhao, Q., et al.: Modma dataset: A multi-modal open dataset for mental-disorder analysis. arxiv 2020. arXiv preprint arXiv:2002.09283

    work page arXiv 2020

  33. [33]

    IEEE Transactions on Neural Systems and Rehabilitation Engineering 29, 1977–1986 (2021) 14 Y

    Jia, Z., Lin, Y., Wang, J., Ning, X., He, Y., Zhou, R., Zhou, Y., Li-wei, H.L.: Multi- view spatial-temporal graph convolutional networks with domain generalization for sleep stage classification. IEEE Transactions on Neural Systems and Rehabilitation Engineering 29, 1977–1986 (2021) 14 Y. Wang et al

    work page 1977

  34. [34]

    PloS one 12(2), e0171409 (2017)

    Mumtaz, W., Xia, L., Mohd Yasin, M.A., Azhar Ali, S.S., Malik, A.S.: A wavelet- based technique to predict treatment outcome for major depressive disorder. PloS one 12(2), e0171409 (2017)

    work page 2017

  35. [35]

    In: Proceedings of the IEEE conference on computer vision and pattern recognition

    Hu, J., Shen, L., Sun, G.: Squeeze-and-excitation networks. In: Proceedings of the IEEE conference on computer vision and pattern recognition. pp. 7132–7141 (2018)

    work page 2018

  36. [36]

    In: Proceedings of the IEEE conference on computer vision and pattern recognition

    Chollet, F.: Xception: Deep learning with depthwise separable convolutions. In: Proceedings of the IEEE conference on computer vision and pattern recognition. pp. 1251–1258 (2017)

    work page 2017

  37. [37]

    IEEE transactions on neural networks and learning systems (2022)

    Grattarola, D., Zambon, D., Bianchi, F.M., Alippi, C.: Understanding pooling in graph neural networks. IEEE transactions on neural networks and learning systems (2022)

    work page 2022

  38. [38]

    Nature reviews neuroscience 10(3), 186–198 (2009)

    Bullmore, E., Sporns, O.: Complex brain networks: graph theoretical analysis of structural and functional systems. Nature reviews neuroscience 10(3), 186–198 (2009)

    work page 2009

  39. [39]

    Science335(6076), 1634–1636 (2012)

    Chen, C.H., Gutierrez, E., Thompson, W., Panizzon, M.S., Jernigan, T.L., Eyler, L.T., Fennema-Notestine, C., Jak, A.J., Neale, M.C., Franz, C.E., et al.: Hier- archical genetic organization of human cortical surface area. Science335(6076), 1634–1636 (2012)

    work page 2012

  40. [40]

    Science342(6158), 1238411 (2013)

    Park, H.J., Friston, K.: Structural and functional brain networks: from connections to cognition. Science342(6158), 1238411 (2013)

    work page 2013

  41. [41]

    Neuroscience & Biobehavioral Reviews56, 330–344 (2015)

    Mulders, P.C., van Eijndhoven, P.F., Schene, A.H., Beckmann, C.F., Tendolkar, I.: Resting-state functional connectivity in major depressive disorder: a review. Neuroscience & Biobehavioral Reviews56, 330–344 (2015)

    work page 2015

  42. [42]

    Neuropsychopharmacology41(7), 1822–1830 (2016)

    Kaiser, R.H., Whitfield-Gabrieli, S., Dillon, D.G., Goer, F., Beltzer, M., Minkel, J., Smoski, M., Dichter, G., Pizzagalli, D.A.: Dynamic resting-state functional con- nectivity in major depression. Neuropsychopharmacology41(7), 1822–1830 (2016)

    work page 2016

  43. [43]

    In: Proceedings of the AAAI Conference on Artificial Intelli- gence

    Song, T., Liu, S., Zheng, W., Zong, Y., Cui, Z.: Instance-adaptive graph for eeg emotion recognition. In: Proceedings of the AAAI Conference on Artificial Intelli- gence. vol. 34, pp. 2701–2708 (2020)

    work page 2020

  44. [44]

    In: 2022 International Joint Conference on Neural Networks (IJCNN)

    Xue, Y., Zheng, W., Zong, Y., Chang, H., Jiang, X.: Adaptive hierarchical graph convolutional network for eeg emotion recognition. In: 2022 International Joint Conference on Neural Networks (IJCNN). pp. 1–8. IEEE (2022)

    work page 2022

  45. [45]

    In: IJCAI

    Jia, Z., Lin, Y., Wang, J., Zhou, R., Ning, X., He, Y., Zhao, Y.: Graphsleepnet: Adaptive spatial-temporal graph convolutional networks for sleep stage classifica- tion. In: IJCAI. vol. 2021, pp. 1324–1330 (2020)

    work page 2021

  46. [46]

    IEEE Transactions on Neural Systems and Rehabilitation Engineering (2023)

    Luo, G., An, P., Li, Y., Hong, R., Chen, S., Rao, H., Chen, W.: Exploring adaptive graph topologies and temporal graph networks for eeg-based depression detection. IEEE Transactions on Neural Systems and Rehabilitation Engineering (2023)

    work page 2023

  47. [47]

    Molecular psychiatry24(6), 888–900 (2019)

    Beijers, L., Wardenaar, K.J., van Loo, H.M., Schoevers, R.A.: Data-driven biologi- cal subtypes of depression: systematic review of biological approaches to depression subtyping. Molecular psychiatry24(6), 888–900 (2019)

    work page 2019

  48. [48]

    IEEE Transactions on Neural Systems and Rehabilitation Engineering (2024)

    Klepl, D., Wu, M., He, F.: Graph neural network-based eeg classification: A survey. IEEE Transactions on Neural Systems and Rehabilitation Engineering (2024)

    work page 2024

  49. [49]

    Advances in neural information processing sys- tems 32 (2019)

    Paszke, A., Gross, S., Massa, F., Lerer, A., Bradbury, J., Chanan, G., Killeen, T., Lin, Z., Gimelshein, N., Antiga, L., et al.: Pytorch: An imperative style, high- performance deep learning library. Advances in neural information processing sys- tems 32 (2019)

    work page 2019