pith. sign in

arxiv: 2409.01962 · v3 · submitted 2024-08-21 · 📡 eess.SP · cs.CV· cs.HC· cs.LG

Attentive Dilated Convolution for Automatic Sleep Staging using Force-directed Layout

Md Jobayer , Md Mehedi Hasan Shawon , Tasfin Mahmud , Md. Borhan Uddin Antor , Arshad M. Chowdhury This is my paper

Pith reviewed 2026-05-23 21:36 UTC · model grok-4.3

classification 📡 eess.SP cs.CVcs.HCcs.LG
keywords sleep stagingEEGconvolutional neural networkvisibility graphforce-directed layoutdeep learningattention mechanismautomated classification
0
0 comments X

The pith

AttDiCNN uses force-directed visibility graph layouts on EEG to reach 98-99% sleep staging accuracy with 1.4 million parameters.

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

The paper presents AttDiCNN as an automated sleep stage classifier from EEG signals. It converts the signals into force-directed layouts derived from visibility graphs to represent key spatial-temporal features and reduce the impact of data heterogeneity. The architecture combines a localized spatial feature extraction module, a spatio-temporal long retention module, and a global averaging attention module to process the data efficiently. Reported results show top accuracies of 98.56 percent on EDFX, 99.66 percent on HMC, and 99.08 percent on NCH while using only 1.4 million parameters, positioning the model as a candidate for clinical automated staging.

Core claim

The AttDiCNN model, which applies force-directed layouts from visibility graphs to EEG signals and processes them through localized spatial feature extraction, spatio-temporal long retention, and global averaging attention modules, achieves state-of-the-art accuracies of 98.56 percent, 99.66 percent, and 99.08 percent on the EDFX, HMC, and NCH datasets respectively while maintaining a parameter count of 1.4 million.

What carries the argument

The force-directed layout based on the visibility graph, which converts EEG time series into graph representations to capture the most significant spatial-temporal information for the subsequent convolutional processing.

If this is right

  • The approach could support reliable automated diagnosis of sleep disorders in clinical environments with varying recording equipment.
  • The 1.4 million parameter size would allow deployment on devices with limited compute resources.
  • The visibility graph preprocessing step could extend to other biomedical time-series tasks that require spatial-temporal feature capture.
  • The modular design separating spatial extraction, long-term retention, and attention could serve as a template for other signal classification networks.

Where Pith is reading between the lines

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

  • Visibility graph layouts may act as a reusable preprocessing technique that improves robustness when training on mixed EEG sources.
  • High cross-dataset performance implies reduced need for per-dataset retraining or heavy data augmentation in sleep staging pipelines.
  • The method's efficiency could be tested in real-time streaming EEG applications for continuous patient monitoring.

Load-bearing premise

The force-directed layout based on the visibility graph captures the most significant spatial-temporal information from the EEG signals for reliable sleep stage classification across heterogeneous datasets.

What would settle it

Testing the trained AttDiCNN on a new, independent EEG sleep dataset and obtaining accuracies below those of existing published methods would falsify the performance claim.

Figures

Figures reproduced from arXiv: 2409.01962 by Arshad M. Chowdhury, Md. Borhan Uddin Antor, Md Jobayer, Md Mehedi Hasan Shawon, Tasfin Mahmud.

Figure 1
Figure 1. Figure 1: An overview of the proposed system. It begins with acquiring EEG data from the devices and converting it to visibility graphs. It is then passed to [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: This illustration depicts the conversion process from the raw EDF data into time-series data. The process can be divided into three sections: filtering [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Sample undirected natural VG with 10 vertices equivalent to 10 series [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: This illustration illustrates generating an FDL graph from time-series data. The conversion process can be divided into three steps: normalizing the [PITH_FULL_IMAGE:figures/full_fig_p004_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Multihead attention network composed of multiple attention layers [PITH_FULL_IMAGE:figures/full_fig_p005_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: The illustration of the proposed model architecture. The network is composed of three distinct blocks: LSFE, S2TLR, and G2A. LSFE extracts [PITH_FULL_IMAGE:figures/full_fig_p006_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Performance metrics (accuracy and loss) plotted over epochs during training and validation for the EDFX dataset, using batch sizes of 32, 64, 128, [PITH_FULL_IMAGE:figures/full_fig_p009_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Representation of two error metrics for our model’s performance [PITH_FULL_IMAGE:figures/full_fig_p010_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Reliability diagram of the model for the three datasets with batch sizes ranging from 32 to 1024, including six statistical metrics: accuracy, kappa, [PITH_FULL_IMAGE:figures/full_fig_p012_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Confusion matrix displaying the leading three models with batch [PITH_FULL_IMAGE:figures/full_fig_p012_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: Kernel weight distribution comparison between the G2A module and the LSFE module to assess computational overhead. [PITH_FULL_IMAGE:figures/full_fig_p013_11.png] view at source ↗
read the original abstract

Sleep stages play an important role in identifying sleep patterns and diagnosing sleep disorders. In this study, we present an automated sleep stage classifier called the Attentive Dilated Convolutional Neural Network (AttDiCNN), which uses deep learning methodologies to address challenges related to data heterogeneity, computational complexity, and reliable and automatic sleep staging. We employed a force-directed layout based on the visibility graph to capture the most significant information from the EEG signals, thereby representing the spatial-temporal features. The proposed network consists of three modules: the Localized Spatial Feature Extraction Network (LSFE), Spatio-Temporal-Temporal Long Retention Network (S2TLR), and Global Averaging Attention Network (G2A). The LSFE captures spatial information from sleep data, the S2TLR is designed to extract the most pertinent information in long-term contexts, and the G2A reduces computational overhead by aggregating information from the LSFE and S2TLR. We evaluated the performance of our model on three comprehensive and publicly accessible datasets, achieving state-of-the-art accuracies of 98.56%, 99.66%, and 99.08% for the EDFX, HMC, and NCH datasets, respectively, while maintaining a low computational complexity with 1.4 M parameters. Our proposed architecture surpasses existing methodologies in several performance metrics, thereby proving its potential as an automated tool for clinical settings.

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

2 major / 2 minor

Summary. The manuscript introduces Attentive Dilated Convolutional Neural Network (AttDiCNN) for automatic 5-class sleep staging from single-channel EEG. It converts signals to a force-directed layout derived from visibility graphs to encode spatial-temporal structure, then processes them via three modules (LSFE for localized spatial features, S2TLR for long-term spatio-temporal retention, and G2A for global averaging attention) to reduce complexity. On the EDFX, HMC, and NCH public datasets the model is reported to reach 98.56%, 99.66%, and 99.08% accuracy respectively while using only 1.4 M parameters, outperforming prior methods.

Significance. If the headline accuracies prove robust under subject-independent partitioning and the force-directed visibility-graph representation demonstrably generalizes across heterogeneous recordings, the work would constitute a notable advance in both accuracy and efficiency for automated sleep staging, with direct relevance to clinical deployment.

major comments (2)
  1. [Abstract / Experimental Results] Abstract and Experimental Results section: the central claim that AttDiCNN + force-directed layout attains 98–99% accuracy rests on the evaluation protocol. No information is supplied on whether folds are subject-independent (no subject overlap between train and test), on the number of folds, or on per-fold standard deviations. Standard subject-independent 5-class staging on these datasets rarely exceeds ~85–88%; without explicit confirmation of inter-subject partitioning the reported numbers cannot be taken as evidence of architectural superiority.
  2. [Methods] Methods section describing the visibility-graph layout and LSFE/S2TLR/G2A modules: the manuscript must show that the force-directed embedding plus the three modules extract features that are not merely subject-specific artifacts. Given the heterogeneous datasets, an ablation that isolates the contribution of the layout versus standard time-series or spectrogram inputs is required to support the claim that this representation captures the “most significant spatial-temporal information.”
minor comments (2)
  1. [Abstract] Abstract: the sentence “while maintaining a low computational complexity with 1.4 M parameters” should be accompanied by a direct comparison (FLOPs or inference latency) to the cited baselines so that the efficiency claim can be evaluated.
  2. [Methods] Notation: the acronyms LSFE, S2TLR and G2A are introduced without an explicit expansion on first use; a table or paragraph listing the module names and their functions would improve readability.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We appreciate the referee's comments, which highlight important aspects for strengthening the manuscript. We will make revisions to address both major comments as detailed below.

read point-by-point responses

  1. Referee: [Abstract / Experimental Results] Abstract and Experimental Results section: the central claim that AttDiCNN + force-directed layout attains 98–99% accuracy rests on the evaluation protocol. No information is supplied on whether folds are subject-independent (no subject overlap between train and test), on the number of folds, or on per-fold standard deviations. Standard subject-independent 5-class staging on these datasets rarely exceeds ~85–88%; without explicit confirmation of inter-subject partitioning the reported numbers cannot be taken as evidence of architectural superiority.

    Authors: We agree that the evaluation protocol is not detailed in the manuscript. We will revise the Experimental Results section to explicitly describe the cross-validation setup, including the use of subject-independent partitioning, the number of folds, and the per-fold standard deviations. This will provide the necessary context to evaluate the reported accuracies. revision: yes

  2. Referee: [Methods] Methods section describing the visibility-graph layout and LSFE/S2TLR/G2A modules: the manuscript must show that the force-directed embedding plus the three modules extract features that are not merely subject-specific artifacts. Given the heterogeneous datasets, an ablation that isolates the contribution of the layout versus standard time-series or spectrogram inputs is required to support the claim that this representation captures the “most significant spatial-temporal information.”

    Authors: We acknowledge that an ablation study is needed to demonstrate the specific contribution of the force-directed visibility graph representation. We will add an ablation analysis in the revised manuscript comparing the proposed layout to standard time-series and spectrogram inputs, showing the performance gains attributable to this representation across the heterogeneous datasets. revision: yes

Circularity Check

0 steps flagged

No significant circularity; empirical results on public datasets

full rationale

The paper describes an AttDiCNN architecture with LSFE, S2TLR, and G2A modules applied to force-directed visibility-graph layouts of EEG signals. Reported accuracies are presented as empirical outcomes of training and testing on the EDFX, HMC, and NCH public datasets. No equations, fitted parameters, or self-citations are shown that reduce any performance claim to a definitional identity or to a prediction forced by construction from the inputs. The derivation chain consists of standard neural-network design choices followed by dataset evaluation and is therefore self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract-only review; no explicit free parameters, axioms, or invented entities are identifiable. The central claim rests on the unstated assumption that the visibility-graph layout plus the three modules produce generalizable features.

pith-pipeline@v0.9.0 · 5810 in / 1163 out tokens · 32349 ms · 2026-05-23T21:36:57.252466+00:00 · methodology

discussion (0)

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

Reference graph

Works this paper leans on

63 extracted references · 63 canonical work pages · 1 internal anchor

  1. [1]

    Sleeping hours: What is the ideal number and how does age impact this?

    J.-P. Chaput, C. Dutil, and H. Sampasa-Kanyinga, “Sleeping hours: What is the ideal number and how does age impact this?” Nature and Science of Sleep , 2018. DOI: 10.2147/nss.s163071

    work page doi:10.2147/nss.s163071 2018

  2. [2]

    Sleep Physiology,

    H. R. Colten, B. M. Altevogt, and I. o. M. ( C. o. S. M. a. Research, “Sleep Physiology,” en, in Sleep Disorders and Sleep Deprivation: An Unmet Public Health Problem , National Academies Press (US), 2006. [Online]. Available: https://www.ncbi.nlm.nih.gov/books/NBK19956/

    work page 2006

  3. [3]

    Which is more important for health: Sleep quantity or sleep quality?

    J. Kohyama, “Which is more important for health: Sleep quantity or sleep quality?” Children, 2021. DOI: 10.3390/children8070542

    work page doi:10.3390/children8070542 2021

  4. [4]

    The role of sleep in recovery following ischemic stroke: A review of human and animal data,

    S. B. Duss, A. Seiler, M. H. Schmidt, et al. , “The role of sleep in recovery following ischemic stroke: A review of human and animal data,” Neurobiology of Sleep and Circadian Rhythms, Sleep, Circadian Disruption, and Neurodegenerative Disease, vol. 2, pp. 94–105, Jan. 2017, ISSN : 2451-9944. DOI: 10.1016/j.nbscr.2016.11.003

    work page doi:10.1016/j.nbscr.2016.11.003 2017

  5. [5]

    Chapter 2 - Normal Human Sleep: An Overview,

    M. A. Carskadon and W. C. Dement, “Chapter 2 - Normal Human Sleep: An Overview,” in Principles and Practice of Sleep Medicine (Fourth Edition), M. H. Kryger, T. Roth, and W. C. Dement, Eds., Philadelphia: W.B. Saunders, Jan. 2005, pp. 13–23, ISBN : 978-0-7216- 0797-9. DOI: 10.1016/B0-72-160797-7/50009-4

    work page doi:10.1016/b0-72-160797-7/50009-4 2005

  6. [6]

    Feasibility of wearable devices and machine learning for sleep classification in children with Rett syndrome: A pilot study,

    M. Migovich, A. Ullal, C. Fu, S. U. Peters, and N. Sarkar, “Feasibility of wearable devices and machine learning for sleep classification in children with Rett syndrome: A pilot study,” en, DIGITAL HEALTH, vol. 9, p. 20 552 076 231 191 622, Jan. 2023, ISSN : 2055-2076, 2055-

    work page 2023

  7. [7]

    DOI: 10.1177/20552076231191622

    work page doi:10.1177/20552076231191622

  8. [8]

    Identification of Sleep Patterns via Clustering of Hypnodensities,

    J. R. Mirth, C. L. Felton, C. R. Haider, et al. , “Identification of Sleep Patterns via Clustering of Hypnodensities,” in 2023 45th Annual International Conference of the IEEE Engineering in Medicine & Biology Society (EMBC), Sydney, Australia: IEEE, Jul. 2023, pp. 1–4, ISBN : 9798350324471. DOI: 10.1109/EMBC40787.2023.10340905

    work page doi:10.1109/embc40787.2023.10340905 2023

  9. [9]

    doi: 10.1155/2018/6534041

    B. Zhang, T. Lei, H. Liu, and H. Cai, “EEG-Based Automatic Sleep Staging Using Ontology and Weighting Feature Analysis,” en, Compu- tational and Mathematical Methods in Medicine , vol. 2018, pp. 1–16, Sep. 2018, ISSN : 1748-670X, 1748-6718. DOI: 10.1155/2018/6534041

    work page doi:10.1155/2018/6534041 2018

  10. [10]

    Convolution- and Attention-Based Neural Network for Automated Sleep Stage Classification,

    T. Zhu, W. Luo, and F. Yu, “Convolution- and Attention-Based Neural Network for Automated Sleep Stage Classification,” en, International Journal of Environmental Research and Public Health, vol. 17, no. 11, p. 4152, Jan. 2020, ISSN : 1660-4601. DOI: 10.3390/ijerph17114152

    work page doi:10.3390/ijerph17114152 2020

  11. [11]

    MS-HNN: Multi-Scale Hierarchical Neural Network With Squeeze and Excitation Block for Neonatal Sleep Staging Using a Single-Channel EEG,

    H. Zhu, L. Wang, N. Shen, et al., “MS-HNN: Multi-Scale Hierarchical Neural Network With Squeeze and Excitation Block for Neonatal Sleep Staging Using a Single-Channel EEG,” IEEE Transactions on Neural Systems and Rehabilitation Engineering, vol. 31, pp. 2195–2204, 2023, ISSN : 1534-4320, 1558-0210. DOI: 10.1109/TNSRE.2023.3266876

    work page doi:10.1109/tnsre.2023.3266876 2023

  12. [12]

    Multi-scale ResNet and BiGRU automatic sleep staging based on attention mechanism,

    C. Liu, Y . Yin, Y . Sun, and O. K. Ersoy, “Multi-scale ResNet and BiGRU automatic sleep staging based on attention mechanism,” en, PLOS ONE , vol. 17, no. 6, S. V E, Ed., e0269500, Jun. 2022, ISSN : 1932-6203. DOI: 10.1371/journal.pone.0269500

    work page doi:10.1371/journal.pone.0269500 2022

  13. [13]

    A Novel Approach for Sleep Arousal Disorder Detection Based on the Interaction of Physiological Signals and Metaheuristic Learning,

    A. Badiei, S. Meshgini, and K. Rezaee, “A Novel Approach for Sleep Arousal Disorder Detection Based on the Interaction of Physiological Signals and Metaheuristic Learning,” en, Computational Intelligence and Neuroscience, vol. 2023, N. Iqbal, Ed., pp. 1–18, Jan. 2023, ISSN : 1687-5273, 1687-5265. DOI: 10.1155/2023/9379618

    work page doi:10.1155/2023/9379618 2023

  14. [14]

    EV ALUATING MACHINE LEARNING MODELS FOR PREDICTING SLEEP DISORDERS IN A LIFESTYLE AND HEALTH DATA CONTEXT,

    G. Airlangga, “EV ALUATING MACHINE LEARNING MODELS FOR PREDICTING SLEEP DISORDERS IN A LIFESTYLE AND HEALTH DATA CONTEXT,” JIKO (Jurnal Informatika dan Kom- puter), vol. 7, no. 1, pp. 51–57, Apr. 2024, ISSN : 2656-1948, 2614-

    work page 2024

  15. [15]

    DOI: 10.33387/jiko.v7i1.7870

    work page doi:10.33387/jiko.v7i1.7870

  16. [16]

    Temporal Feature Extraction and Machine Learning for Classification of Sleep Stages Using Telemetry Polysomnography,

    U. Lal, S. Mathavu Vasanthsena, and A. Hoblidar, “Temporal Feature Extraction and Machine Learning for Classification of Sleep Stages Using Telemetry Polysomnography,” en,Brain Sciences, vol. 13, no. 8, p. 1201, Aug. 2023, ISSN : 2076-3425. DOI: 10.3390/brainsci13081201

    work page doi:10.3390/brainsci13081201 2023

  17. [18]

    An effective hybrid feature selection using entropy weight method for automatic sleep staging,

    W. Wang, J. Li, Y . Fang, Y . Zheng, and F. You, “An effective hybrid feature selection using entropy weight method for automatic sleep staging,” Physiological Measurement, vol. 44, no. 10, p. 105 008, Oct. 2023, ISSN : 0967-3334, 1361-6579. DOI: 10.1088/1361-6579/acff35

    work page doi:10.1088/1361-6579/acff35 2023

  18. [19]

    Feature ranking and rank aggregation for automatic sleep stage classification: A comparative study,

    S. Najdi, A. A. Gharbali, and J. M. Fonseca, “Feature ranking and rank aggregation for automatic sleep stage classification: A comparative study,” BioMedical Engineering OnLine , vol. 16, no. 1, p. 78, Aug. 2017, ISSN : 1475-925X. DOI: 10.1186/s12938-017-0358-3

    work page doi:10.1186/s12938-017-0358-3 2017

  19. [20]

    Electrodermal activity patterns in sleep stages and their utility for sleep versus wake classification,

    A. Herlan, J. Ottenbacher, J. Schneider, D. Riemann, and B. Feige, “Electrodermal activity patterns in sleep stages and their utility for sleep versus wake classification,” en, Journal of Sleep Research , vol. 28, no. 2, e12694, Apr. 2019, ISSN : 0962-1105, 1365-2869. DOI: 10.1111/jsr.12694

    work page doi:10.1111/jsr.12694 2019

  20. [21]

    Quantitative analysis of wrist electrodermal activity during sleep,

    A. Sano, R. W. Picard, and R. Stickgold, “Quantitative analysis of wrist electrodermal activity during sleep,” International Journal of Psychophysiology, vol. 94, no. 3, pp. 382–389, Dec. 2014, ISSN : 0167-

    work page 2014

  21. [22]

    DOI: 10.1016/j.ijpsycho.2014.09.011

    work page doi:10.1016/j.ijpsycho.2014.09.011 2014

  22. [23]

    Evaluation of an automated single-channel sleep staging algorithm,

    R. Kaplan, Y . Wang, K. Loparo, and M. Kelly, “Evaluation of an automated single-channel sleep staging algorithm,” en, Nature and Science of Sleep , p. 101, Sep. 2015, ISSN : 1179-1608. DOI: 10.2147/ NSS.S77888

    work page 2015

  23. [24]

    A Systematic Review of Sensing Technologies for Wearable Sleep Staging,

    S. A. Imtiaz, “A Systematic Review of Sensing Technologies for Wearable Sleep Staging,” en, Sensors, vol. 21, no. 5, p. 1562, Jan. 2021, ISSN : 1424-8220. DOI: 10.3390/s21051562

    work page doi:10.3390/s21051562 2021

  24. [25]

    A Deep Learning Model for Automated Sleep Stages Classification Using PSG Sig- nals,

    O. Yildirim, U. B. Baloglu, and U. R. Acharya, “A Deep Learning Model for Automated Sleep Stages Classification Using PSG Sig- nals,” en, International Journal of Environmental Research and Public Health, vol. 16, no. 4, p. 599, Jan. 2019, ISSN : 1660-4601. DOI: 10. 3390/ijerph16040599

    work page 2019

  25. [26]

    The Promise of Sleep: A Multi-Sensor Approach for Accurate Sleep Stage Detection Using the Oura Ring,

    M. Altini and H. Kinnunen, “The Promise of Sleep: A Multi-Sensor Approach for Accurate Sleep Stage Detection Using the Oura Ring,” en, Sensors, vol. 21, no. 13, p. 4302, Jan. 2021, ISSN : 1424-8220. DOI: 10.3390/s21134302

    work page doi:10.3390/s21134302 2021

  26. [27]

    A Survey on Recent Advances in Machine Learning Based Sleep Apnea Detection Systems,

    A. Ramachandran and A. Karuppiah, “A Survey on Recent Advances in Machine Learning Based Sleep Apnea Detection Systems,” en, Healthcare, vol. 9, no. 7, p. 914, Jul. 2021, ISSN : 2227-9032. DOI: 10.3390/healthcare9070914

    work page doi:10.3390/healthcare9070914 2021

  27. [28]

    Joint classification and prediction cnn framework for automatic sleep stage classification,

    H. Phan, F. Andreotti, N. Cooray, O. Y . Ch ´en, and M. D. V os, “Joint classification and prediction cnn framework for automatic sleep stage classification,” IEEE Transactions on Biomedical Engineering , 2019. DOI: 10.1109/tbme.2018.2872652

    work page doi:10.1109/tbme.2018.2872652 2019

  28. [29]

    Ensemble SVM Method for Automatic Sleep Stage Classification,

    E. Alickovic and A. Subasi, “Ensemble SVM Method for Automatic Sleep Stage Classification,”IEEE Transactions on Instrumentation and Measurement, vol. 67, no. 6, pp. 1258–1265, Jun. 2018, ISSN : 1557-

    work page 2018

  29. [30]

    DOI: 10.1109/TIM.2018.2799059

    work page doi:10.1109/tim.2018.2799059 2018

  30. [31]

    Competitive multi- verse optimization with deep learning based sleep stage classification,

    A. M. Hilal, A. Al-Rasheed, J. S. Alzahrani, et al., “Competitive multi- verse optimization with deep learning based sleep stage classification,” Computer Systems Science and Engineering, 2023. DOI: 10.32604/csse. 2023.030603

    work page doi:10.32604/csse 2023

  31. [32]

    Automatic Sleep Staging Employing Convo- lutional Neural Networks and Cortical Connectivity Images,

    P. Chriskos, C. A. Frantzidis, P. T. Gkivogkli, P. D. Bamidis, and C. Kourtidou-Papadeli, “Automatic Sleep Staging Employing Convo- lutional Neural Networks and Cortical Connectivity Images,” IEEE Transactions on Neural Networks and Learning Systems, vol. 31, no. 1, pp. 113–123, Jan. 2020, ISSN : 2162-2388. DOI: 10.1109/TNNLS.2019. 2899781

    work page doi:10.1109/tnnls.2019 2020

  32. [33]

    Dynamic Information Flow Based on EEG and Diffusion MRI in Stroke: A Proof-of-Principle Study,

    O. G. Filatova, Y . Yang, J. P. A. Dewald, et al., “Dynamic Information Flow Based on EEG and Diffusion MRI in Stroke: A Proof-of-Principle Study,” English, Frontiers in Neural Circuits, vol. 12, Oct. 2018, ISSN : 1662-5110. DOI: 10.3389/fncir.2018.00079

    work page doi:10.3389/fncir.2018.00079 2018

  33. [34]

    From time series to complex networks: The visibility graph,

    L. Lacasa, B. Luque, F. Ballesteros, J. Luque, and J. C. Nu ˜no, “From time series to complex networks: The visibility graph,” en,Proceedings of the National Academy of Sciences , vol. 105, no. 13, pp. 4972–4975, Apr. 2008. DOI: 10.1073/pnas.0709247105

    work page doi:10.1073/pnas.0709247105 2008

  34. [35]

    An algorithm for drawing general undi- rected graphs,

    T. Kamada and S. Kawai, “An algorithm for drawing general undi- rected graphs,” en, Information Processing Letters , vol. 31, no. 1, pp. 7–15, Apr. 1989. DOI: 10.1016/0020-0190(89)90102-6

    work page doi:10.1016/0020-0190(89)90102-6 1989

  35. [36]

    A heuristic for graph drawing,

    P. Eades, “A heuristic for graph drawing,” Congressus numerantium, vol. 42, no. 11, pp. 149–160, 1984

    work page 1984

  36. [38]

    Monocular visual-inertial odometry in low-textured environments with smooth gradients: A fully dense direct filtering approach,

    F. Engelmann, T. Kontogianni, and B. Leibe, “Dilated Point Convo- lutions: On the Receptive Field Size of Point Convolutions on 3D Point Clouds,” in2020 IEEE International Conference on Robotics and Automation (ICRA), Paris, France: IEEE, May 2020, pp. 9463–9469, ISBN : 9781728173955. DOI: 10.1109/ICRA40945.2020.9197503

    work page doi:10.1109/icra40945.2020.9197503 2020

  37. [39]

    Dˆ2Conv3D: Dynamic Dilated Convolutions for Object Segmentation in Videos,

    C. Schmidt, A. Athar, S. Mahadevan, and B. Leibe, “Dˆ2Conv3D: Dynamic Dilated Convolutions for Object Segmentation in Videos,”

    work page

  38. [43]

    Analysis of a sleep-dependent neuronal feedback loop: The slow- wave microcontinuity of the eeg,

    B. Kemp, A. Zwinderman, B. Tuk, H. Kamphuisen, and J. Oberye, “Analysis of a sleep-dependent neuronal feedback loop: The slow- wave microcontinuity of the eeg,” IEEE Transactions on Biomedical Engineering, vol. 47, no. 9, pp. 1185–1194, 2000. DOI: 10.1109/10. 867928

    work page doi:10.1109/10 2000

  39. [44]

    H. Lee, B. Li, Y . Huang, Y . Chi, and S. Lin, NCH Sleep DataBank: A Large Collection of Real-world Pediatric Sleep Studies with Longi- tudinal Clinical Data . DOI: 10.13026/P2RP-SG37

    work page doi:10.13026/p2rp-sg37

  40. [45]

    A large collection of real-world pediatric sleep studies,

    H. Lee, B. Li, S. DeForte, et al. , “A large collection of real-world pediatric sleep studies,” en, Scientific Data, vol. 9, no. 1, p. 421, Jul. 2022, Number: 1 Publisher: Nature Publishing Group, ISSN : 2052-

    work page 2022

  41. [46]

    DOI: 10.1038/s41597-022-01545-6

    work page doi:10.1038/s41597-022-01545-6

  42. [47]

    Inter-database validation of a deep learning approach for automatic sleep scoring,

    D. Alvarez-Estevez and R. M. Rijsman, “Inter-database validation of a deep learning approach for automatic sleep scoring,” en, PLOS ONE, vol. 16, no. 8, e0256111, Aug. 2021, Publisher: Public Library of Science, ISSN : 1932-6203. DOI: 10.1371/journal.pone.0256111

    work page doi:10.1371/journal.pone.0256111 2021

  43. [48]

    Alvarez-Estevez and R

    D. Alvarez-Estevez and R. Rijsman, Haaglanden Medisch Centrum sleep staging database . DOI: 10.13026/T79Q-FR32

    work page doi:10.13026/t79q-fr32

  44. [49]

    SMOTE: Synthetic Minority Over-sampling Technique

    N. V . Chawla, K. W. Bowyer, L. O. Hall, and W. P. Kegelmeyer, “SMOTE: Synthetic Minority Over-sampling Technique,” 2011. DOI: 10.48550/ARXIV .1106.1813

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv 2011

  45. [50]

    A Coefficient of Agreement for Nominal Scales,

    J. Cohen, “A Coefficient of Agreement for Nominal Scales,” en, Educational and Psychological Measurement , vol. 20, no. 1, pp. 37– 46, Apr. 1960, ISSN : 0013-1644, 1552-3888. DOI: 10 . 1177 / 001316446002000104

    work page 1960

  46. [51]

    Interrater reliability: The kappa statistic,

    M. L. McHugh, “Interrater reliability: The kappa statistic,” Biochemia Medica, pp. 276–282, 2012, ISSN : 18467482. DOI: 10 . 11613 / BM . 2012.031

    work page 2012

  47. [52]

    4s-SleepGCN: Four-Stream Graph Convolutional Networks for Sleep Stage Classification,

    M. Li, H. Chen, Y . Liu, and Q. Zhao, “4s-SleepGCN: Four-Stream Graph Convolutional Networks for Sleep Stage Classification,” IEEE Access, vol. 11, pp. 70 621–70 634, 2023, ISSN : 2169-3536. DOI: 10. 1109/ACCESS.2023.3294410

    work page arXiv 2023

  48. [53]

    A Residual Based Attention Model for EEG Based Sleep Staging,

    W. Qu, Z. Wang, H. Hong, et al., “A Residual Based Attention Model for EEG Based Sleep Staging,”IEEE Journal of Biomedical and Health Informatics, vol. 24, no. 10, pp. 2833–2843, Oct. 2020, ISSN : 2168-

    work page 2020

  49. [54]

    DOI: 10.1109/JBHI.2020.2978004

    work page doi:10.1109/jbhi.2020.2978004 2020

  50. [55]

    ADAST: Attentive Cross- Domain EEG-Based Sleep Staging Framework With Iterative Self- Training,

    E. Eldele, M. Ragab, Z. Chen, et al. , “ADAST: Attentive Cross- Domain EEG-Based Sleep Staging Framework With Iterative Self- Training,” IEEE Transactions on Emerging Topics in Computational Intelligence, vol. 7, no. 1, pp. 210–221, Feb. 2023, ISSN : 2471-285X. DOI: 10.1109/TETCI.2022.3189695

    work page doi:10.1109/tetci.2022.3189695 2023

  51. [56]

    An Attention-Based Deep Learning Approach for Sleep Stage Classification With Single-Channel EEG,

    E. Eldele, Z. Chen, C. Liu, et al., “An Attention-Based Deep Learning Approach for Sleep Stage Classification With Single-Channel EEG,” IEEE Transactions on Neural Systems and Rehabilitation Engineering, vol. 29, pp. 809–818, 2021, ISSN : 1558-0210. DOI: 10.1109/TNSRE. 2021.3076234

    work page doi:10.1109/tnsre 2021

  52. [57]

    An Attention-Guided Spatiotemporal Graph Convolutional Network for Sleep Stage Classification,

    M. Li, H. Chen, and Z. Cheng, “An Attention-Guided Spatiotemporal Graph Convolutional Network for Sleep Stage Classification,” en, Life, vol. 12, no. 5, p. 622, May 2022, ISSN : 2075-1729. DOI: 10.3390/ life12050622

    work page 2022

  53. [58]

    Sleep stage classification in EEG signals using the clustering approach based prob- ability distribution features coupled with classification algorithms,

    W. Al-Salman, Y . Li, A. Y . Oudah, and S. Almaged, “Sleep stage classification in EEG signals using the clustering approach based prob- ability distribution features coupled with classification algorithms,” Neuroscience Research, vol. 188, pp. 51–67, Mar. 2023, ISSN : 0168-

    work page 2023

  54. [59]

    DOI: 10.1016/j.neures.2022.09.009

    work page doi:10.1016/j.neures.2022.09.009 2022

  55. [60]

    An intelligent model involving multi-channels spectrum patterns based features for auto- matic sleep stage classification,

    S. Abdulla, M. Diykh, S. Siuly, and M. Ali, “An intelligent model involving multi-channels spectrum patterns based features for auto- matic sleep stage classification,” en, International Journal of Medical Informatics, vol. 171, p. 105 001, Mar. 2023, ISSN : 13865056. DOI: 10.1016/j.ijmedinf.2023.105001

    work page doi:10.1016/j.ijmedinf.2023.105001 2023

  56. [61]

    An Improved Neural Network Based on SENet for Sleep Stage Classification,

    J. Huang, L. Ren, X. Zhou, and K. Yan, “An Improved Neural Network Based on SENet for Sleep Stage Classification,” IEEE Journal of Biomedical and Health Informatics , vol. 26, no. 10, pp. 4948–4956, Oct. 2022, ISSN : 2168-2208. DOI: 10.1109/JBHI.2022.3157262

    work page doi:10.1109/jbhi.2022.3157262 2022

  57. [62]

    A Deep Learning Method Approach for Sleep Stage Classification with EEG Spectrogram,

    C. Li, Y . Qi, X. Ding, J. Zhao, T. Sang, and M. Lee, “A Deep Learning Method Approach for Sleep Stage Classification with EEG Spectrogram,” en, International Journal of Environmental Research and Public Health , vol. 19, no. 10, p. 6322, Jan. 2022, ISSN : 1660-

    work page 2022

  58. [63]

    DOI: 10.3390/ijerph19106322

    work page doi:10.3390/ijerph19106322

  59. [64]

    Automatic Sleep Stage Classification With Single Channel EEG Signal Based on Two-Layer Stacked En- semble Model,

    J. Zhou, G. Wang, J. Liu, et al., “Automatic Sleep Stage Classification With Single Channel EEG Signal Based on Two-Layer Stacked En- semble Model,” IEEE Access, vol. 8, pp. 57 283–57 297, 2020, ISSN : 2169-3536. DOI: 10.1109/ACCESS.2020.2982434

    work page doi:10.1109/access.2020.2982434 2020

  60. [65]

    IEEE Access8, 199523–199538 (2020) https://doi.org/10.1109/ACCESS

    M. Abdollahpour, T. Y . Rezaii, A. Farzamnia, and I. Saad, “Transfer Learning Convolutional Neural Network for Sleep Stage Classification Using Two-Stage Data Fusion Framework,” IEEE Access , vol. 8, pp. 180 618–180 632, 2020, ISSN : 2169-3536. DOI: 10.1109/ACCESS. 2020.3027289

    work page doi:10.1109/access 2020

  61. [66]

    Simplifying Multimodal With Single EOG Modality for Automatic Sleep Staging,

    Y . Zhou, S. Zhao, J. Wang, et al., “Simplifying Multimodal With Single EOG Modality for Automatic Sleep Staging,” IEEE Transactions on Neural Systems and Rehabilitation Engineering , vol. 32, pp. 1668– 1678, 2024, ISSN : 1558-0210. DOI: 10.1109/TNSRE.2024.3389077

    work page doi:10.1109/tnsre.2024.3389077 2024

  62. [67]

    Performance Assessment of Automatic Sleep Stage Classification Using Only Partial PSG Sensors,

    I. Choi and W. Sung, “Performance Assessment of Automatic Sleep Stage Classification Using Only Partial PSG Sensors,” in 2022 IEEE Biomedical Circuits and Systems Conference (BioCAS) , ISSN: 2163- 4025, Oct. 2022, pp. 670–674. DOI: 10 . 1109 / BioCAS54905 . 2022 . 9948578

    work page 2022

  63. [68]

    Lee and A

    H. Lee and A. Saeed, Automatic Sleep Scoring from Large-scale Multi- channel Pediatric EEG, arXiv:2207.06921 [cs, eess], Oct. 2022. DOI: 10.48550/arXiv.2207.06921. Supplementary Materials Sleep Stage 1Sleep Stage 2Sleep Stage 3Sleep Stage 4Sleep Stage ?Sleep Stage RSleep Stage W (a) 0 10000 20000 30000 40000 50000Number of Samples 7229 8107 4799 1454 100 ...

    work page doi:10.48550/arxiv.2207.06921 2022