pith. sign in

arxiv: 1907.07671 · v1 · pith:6BTVJRBBnew · submitted 2019-07-17 · 📡 eess.SP · cs.LG· stat.ML

Electroencephalography based Classification of Long-term Stress using Psychological Labeling

Sanay Muhammad Umar Saeed , Syed Muhammad Anwar , Humaira Khalid , Muhammad Majid , Ulas Bagci This is my paper

Pith reviewed 2026-05-24 20:38 UTC · model grok-4.3

classification 📡 eess.SP cs.LGstat.ML
keywords EEGstress classificationalpha asymmetrybio-markersupport vector machineexpert evaluationlong-term stressfrequency features
0
0 comments X

The pith

Expert evaluation of baseline EEG signals classifies long-term stress at 85.2 percent accuracy using alpha asymmetry.

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

The paper tests whether baseline EEG recordings can objectively separate people with long-term stress from controls. Two labeling approaches are compared: self-reported perceived stress scale scores versus expert psychological evaluation. Frequency features plus frontal and temporal alpha asymmetry are extracted from five channels, a t-test selects the significant ones, and support vector machine classification is applied. Accuracy reaches 85.20 percent only when expert labels are used, leading the authors to propose alpha asymmetry as a bio-marker under that labeling condition.

Core claim

Long-term stress can be classified from baseline EEG recordings at up to 85.20 percent accuracy when subjects are labeled by expert evaluation rather than perceived stress scale scores, with alpha asymmetry serving as the key feature for a support vector machine classifier.

What carries the argument

Alpha asymmetry computed from four frontal and temporal EEG channels, selected by t-test and supplied to an SVM classifier whose training labels come from expert evaluation instead of self-report scores.

Load-bearing premise

Expert evaluation supplies a reliable ground-truth label for long-term stress that is independent of the EEG features themselves.

What would settle it

A replication on new subjects in which expert-labeled groups show no statistical difference in alpha asymmetry or in which SVM accuracy falls below chance when the same expert labeling protocol is repeated.

Figures

Figures reproduced from arXiv: 1907.07671 by Humaira Khalid, Muhammad Majid, Sanay Muhammad Umar Saeed, Syed Muhammad Anwar, Ulas Bagci.

Figure 1
Figure 1. Figure 1: The proposed methodology for long-term human stress classification. [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: The sequence of events during the data acquisition process. [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: A histogram of PSS scores for participants showing labels assigned using the PSS based labelling method (green: control group, red: stress group, [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Box plots of features (a) Alpha asymmetry (b) beta (c) gamma (d)Alpha asymmetry (EE) (e) beta (EE) (f) gamma(EE), where EE represents the [PITH_FULL_IMAGE:figures/full_fig_p006_4.png] view at source ↗
read the original abstract

Stress research is a rapidly emerging area in thefield of electroencephalography (EEG) based signal processing.The use of EEG as an objective measure for cost effective andpersonalized stress management becomes important in particularsituations such as the non-availability of mental health facilities.In this study, long-term stress is classified using baseline EEGsignal recordings. The labelling for the stress and control groupsis performed using two methods (i) the perceived stress scalescore and (ii) expert evaluation. The frequency domain featuresare extracted from five-channel EEG recordings in addition tothe frontal and temporal alpha and beta asymmetries. The alphaasymmetry is computed from four channels and used as a feature.Feature selection is also performed using a t-test to identifystatistically significant features for both stress and control groups.We found that support vector machine is best suited to classifylong-term human stress when used with alpha asymmetry asa feature. It is observed that expert evaluation based labellingmethod has improved the classification accuracy up to 85.20%.Based on these results, it is concluded that alpha asymmetry maybe used as a potential bio-marker for stress classification, when labels are assigned using expert evaluation.

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 investigates classification of long-term stress from baseline five-channel EEG recordings. Two labeling methods are compared: perceived stress scale scores versus expert evaluation. Frequency-domain features plus frontal and temporal alpha/beta asymmetries are extracted; alpha asymmetry is computed from four channels. A t-test identifies statistically significant features, after which SVM classification is applied. The work reports that expert-evaluation labeling yields 85.20% accuracy and concludes that alpha asymmetry may serve as a biomarker for stress when labels are assigned by expert evaluation.

Significance. If the accuracy figure is obtained with feature selection nested inside cross-validation and with an adequate subject sample, the direct comparison of labeling methods would be a useful contribution toward objective, low-cost EEG-based stress assessment. The explicit focus on alpha asymmetry as a candidate biomarker under expert labeling is a clear, testable claim that could be followed up in subsequent work.

major comments (2)
  1. [Methods] The description of the t-test feature selection (Abstract and Methods) does not state whether the test was performed inside each cross-validation fold or on the pooled dataset before any train/test split. Application to the full dataset would condition feature selection on test data and render the reported 85.20% SVM accuracy optimistically biased.
  2. [Results] No information is supplied on the number of subjects, the cross-validation scheme (k-fold, leave-one-subject-out, etc.), or whether the expert evaluator was blinded to the EEG recordings. These details are required to evaluate whether the 85.20% accuracy and the biomarker conclusion are statistically reliable.
minor comments (2)
  1. [Abstract] Abstract contains typographical errors ('inthefield', missing hyphen in 'cost effective').
  2. [Results] The abstract states that 'support vector machine is best suited' but provides no comparative results against other classifiers; this claim should be supported by a table or explicit comparison in the Results section.

Simulated Author's Rebuttal

2 responses · 0 unresolved

Thank you for the constructive feedback on our manuscript. We address each major comment below and will revise the manuscript to improve clarity and rigor.

read point-by-point responses

  1. Referee: [Methods] The description of the t-test feature selection (Abstract and Methods) does not state whether the test was performed inside each cross-validation fold or on the pooled dataset before any train/test split. Application to the full dataset would condition feature selection on test data and render the reported 85.20% SVM accuracy optimistically biased.

    Authors: We agree that the manuscript does not specify the relationship between the t-test and cross-validation. We will revise the Methods section to describe the exact procedure used. If feature selection was performed on the full dataset, we will re-execute the analysis with nested feature selection inside each cross-validation fold and report any resulting changes to the accuracy figures. revision: yes

  2. Referee: [Results] No information is supplied on the number of subjects, the cross-validation scheme (k-fold, leave-one-subject-out, etc.), or whether the expert evaluator was blinded to the EEG recordings. These details are required to evaluate whether the 85.20% accuracy and the biomarker conclusion are statistically reliable.

    Authors: The manuscript does not include these details. We will add the number of subjects, specify the cross-validation scheme, and clarify the blinding status of the expert evaluator in the revised Results and Methods sections to allow proper assessment of statistical reliability. revision: yes

Circularity Check

0 steps flagged

No circularity: empirical ML pipeline with external labels

full rationale

The paper reports an empirical classification study: EEG features (including alpha asymmetry) are extracted, a t-test selects significant features, and SVM is trained to distinguish stress vs control groups using labels from either PSS scores or expert evaluation. No equations, derivations, or first-principles claims exist that reduce any reported accuracy or biomarker conclusion to a fitted parameter or self-referential definition by construction. The 85.20% figure is an observed classifier performance on externally labeled data, not a tautological renaming or self-citation load-bearing step. Standard ML procedures do not trigger the enumerated circularity patterns.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The central claim rests on the domain assumption that expert labels constitute valid ground truth and on the standard statistical assumption that t-test-selected features generalize; no free parameters or invented entities are introduced.

axioms (2)
  • domain assumption Expert evaluation accurately partitions subjects into long-term stress and control groups independent of the EEG signals
    The 85.2% accuracy figure is only meaningful if this labeling is treated as ground truth; the abstract provides no validation of the expert process.
  • domain assumption Frequency-domain power and alpha asymmetry are sufficient to distinguish the labeled groups
    Invoked when these features are extracted and fed to the classifier.

pith-pipeline@v0.9.0 · 5748 in / 1535 out tokens · 22856 ms · 2026-05-24T20:38:20.411752+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

45 extracted references · 45 canonical work pages

  1. [1]

    The stress syndrome,

    H. Selye, “The stress syndrome,” The American Journal of Nursing , pp. 97–99, 1965

    work page 1965

  2. [2]

    Neurobiology of early life stress: clinical studies

    C. Heim and C. B. Nemeroff, “Neurobiology of early life stress: clinical studies.” in Seminars in Clinical Neuropsychiatry , vol. 7, no. 2, 2002, pp. 147–159

    work page 2002

  3. [3]

    Chronic stress, acute stress, and depressive symptoms,

    K. A. McGonagle and R. C. Kessler, “Chronic stress, acute stress, and depressive symptoms,” American journal of community psychology , vol. 18, no. 5, pp. 681–706, 1990

    work page 1990

  4. [4]

    Psychological stress and disease,

    S. Cohen, D. Janicki-Deverts, and G. E. Miller, “Psychological stress and disease,” Jama, vol. 298, no. 14, pp. 1685–1687, 2007

    work page 2007

  5. [5]

    Stress and cardiovascular disease,

    A. Steptoe and M. Kivim ¨aki, “Stress and cardiovascular disease,” Nature Reviews Cardiology, vol. 9, no. 6, p. 360, 2012

    work page 2012

  6. [6]

    Can stress cause depression?

    H. Van Praag, “Can stress cause depression?” Progress in Neuro- Psychopharmacology and Biological Psychiatry, vol. 28, no. 5, pp. 891– 907, 2004

    work page 2004

  7. [7]

    Measurement of chronic stress,

    C. Hammen, E. D. Dalton, and S. M. Thompson, “Measurement of chronic stress,” The encyclopedia of clinical psychology , pp. 1–7, 2014

    work page 2014

  8. [8]

    A procedure for reducing errors in reports of life events,

    L. C. Sobell, T. Toneatto, M. B. Sobell, R. Schuller, and M. Maxwell, “A procedure for reducing errors in reports of life events,” Journal of Psychosomatic Research, vol. 34, no. 2, pp. 163–170, 1990

    work page 1990

  9. [9]

    Toward the standardization of life stress assessment: Definitional discrepancies and inconsistencies in methods,

    J. R. McQuaid, S. M. Monroe, J. R. Roberts, S. L. Johnson, G. L. Garamoni, D. J. Kupfer, and E. Frank, “Toward the standardization of life stress assessment: Definitional discrepancies and inconsistencies in methods,” Stress Medicine, vol. 8, no. 1, pp. 47–56, 1992

    work page 1992

  10. [10]

    A method of identifying chronic stress by eeg,

    H. Peng, B. Hu, F. Zheng, D. Fan, W. Zhao, X. Chen, Y . Yang, and Q. Cai, “A method of identifying chronic stress by eeg,” Personal and ubiquitous computing, vol. 17, no. 7, pp. 1341–1347, 2013. IEEE JOURNAL OF BIOMEDICAL AND HEALTH INFORMATICS 8

    work page 2013

  11. [11]

    Biosignal analysis to assess mental stress in automatic driving of trucks: Palmar perspiration and masseter electromyography,

    R. Zheng, S. Yamabe, K. Nakano, and Y . Suda, “Biosignal analysis to assess mental stress in automatic driving of trucks: Palmar perspiration and masseter electromyography,” Sensors, vol. 15, no. 3, pp. 5136–5150, 2015

    work page 2015

  12. [12]

    A hybrid scheme for drowsiness detection using wearable sensors,

    A. Mehreen, S. M. Anwar, M. Haseeb, M. Majid, and M. O. Ullah, “A hybrid scheme for drowsiness detection using wearable sensors,” IEEE Sensors Journal, pp. 1–1, 2019

    work page 2019

  13. [13]

    Human stress classification using eeg signals in response to music tracks,

    A. Asif, M. Majid, and S. M. Anwar, “Human stress classification using eeg signals in response to music tracks,” Computers in biology and medicine, 2019

    work page 2019

  14. [14]

    Selection of neural oscillatory features for human stress classification with single channel eeg headset,

    U. Saeed, S. Muhammad, S. M. Anwar, M. Majid, M. Awais, and M. Alnowami, “Selection of neural oscillatory features for human stress classification with single channel eeg headset,” BioMed research international, vol. 2018, 2018

    work page 2018

  15. [15]

    Emotion recognition in response to traditional and tactile enhanced multimedia using electroen- cephalography,

    A. Raheel, S. M. Anwar, and M. Majid, “Emotion recognition in response to traditional and tactile enhanced multimedia using electroen- cephalography,” Multimedia Tools and Applications , pp. 1–15, 2018

    work page 2018

  16. [16]

    A game player expertise level classification system using electroen- cephalography (eeg),

    S. Anwar, S. Saeed, M. Majid, S. Usman, C. Mehmood, and W. Liu, “A game player expertise level classification system using electroen- cephalography (eeg),” Applied Sciences, vol. 8, no. 1, p. 18, 2018

    work page 2018

  17. [17]

    Eeg signal processing,

    S. Sanei and J. A. Chambers, “Eeg signal processing,” 2007

    work page 2007

  18. [18]

    Towards multilevel mental stress assessment using svm with ecoc: an eeg approach,

    F. Al-shargie, T. B. Tang, N. Badruddin, and M. Kiguchi, “Towards multilevel mental stress assessment using svm with ecoc: an eeg approach,” Medical & biological engineering & computing , vol. 56, no. 1, pp. 125–136, 2018

    work page 2018

  19. [19]

    Fisch, Fisch and Spehlmann’s EEG Primer: Basic Principles of Digital and Analog EEG, 3e

    B. Fisch, Fisch and Spehlmann’s EEG Primer: Basic Principles of Digital and Analog EEG, 3e . Amsterdam: Elsevier, 1999

    work page 1999

  20. [20]

    What does the prefrontal cortex do in affect: perspec- tives on frontal eeg asymmetry research,

    R. J. Davidson, “What does the prefrontal cortex do in affect: perspec- tives on frontal eeg asymmetry research,” Biological psychology, vol. 67, no. 1-2, pp. 219–234, 2004

    work page 2004

  21. [21]

    Covariations of eeg asymmetries and emo- tional states indicate that activity at frontopolar locations is particularly affected by state factors,

    I. Papousek and G. Schulter, “Covariations of eeg asymmetries and emo- tional states indicate that activity at frontopolar locations is particularly affected by state factors,” Psychophysiology, vol. 39, no. 3, pp. 350–360, 2002

    work page 2002

  22. [22]

    Eeg correlates of the severity of posttraumatic stress symptoms: a systematic review of the dimensional ptsd literature,

    I. Lobo, L. C. Portugal, I. Figueira, E. V olchan, I. David, M. G. Pereira, and L. de Oliveira, “Eeg correlates of the severity of posttraumatic stress symptoms: a systematic review of the dimensional ptsd literature,” Journal of affective disorders , vol. 183, pp. 210–220, 2015

    work page 2015

  23. [23]

    Changes in eeg mean frequency and spectral purity during spontaneous alpha blocking,

    I. I. Goncharova and J. S. Barlow, “Changes in eeg mean frequency and spectral purity during spontaneous alpha blocking,” Electroencephalog- raphy and clinical neurophysiology , vol. 76, no. 3, pp. 197–204, 1990

    work page 1990

  24. [24]

    Machine learning framework for the detection of mental stress at multiple levels,

    A. R. Subhani, W. Mumtaz, M. N. B. M. Saad, N. Kamel, and A. S. Malik, “Machine learning framework for the detection of mental stress at multiple levels,” IEEE Access, vol. 5, pp. 13 545–13 556, 2017

    work page 2017

  25. [25]

    A pervasive approach to eeg-based depression detection,

    H. Cai, J. Han, Y . Chen, X. Sha, Z. Wang, B. Hu, J. Yang, L. Feng, Z. Ding, Y . Chen et al., “A pervasive approach to eeg-based depression detection,” Complexity, vol. 2018, 2018

    work page 2018

  26. [26]

    Efficient human stress detection system based on frontal alpha asymmetry,

    A. Baghdadi, Y . Aribi, and A. M. Alimi, “Efficient human stress detection system based on frontal alpha asymmetry,” in International Conference on Neural Information Processing . Springer, 2017, pp. 858–867

    work page 2017

  27. [27]

    The urban brain: analysing outdoor physical activity with mobile eeg,

    P. Aspinall, P. Mavros, R. Coyne, and J. Roe, “The urban brain: analysing outdoor physical activity with mobile eeg,” Br J Sports Med , vol. 49, no. 4, pp. 272–276, 2015

    work page 2015

  28. [28]

    Relative frontal brain asymmetry and cortisol release after social stress: The role of action orientation,

    R. D ¨using, M. Tops, E. L. Radtke, J. Kuhl, and M. Quirin, “Relative frontal brain asymmetry and cortisol release after social stress: The role of action orientation,” Biological psychology, vol. 115, pp. 86–93, 2016

    work page 2016

  29. [29]

    Electroencephalogram alpha asymmetry in geriatric depression,

    A. K. Kaiser, M. Doppelmayr, and B. Iglseder, “Electroencephalogram alpha asymmetry in geriatric depression,” Zeitschrift F ¨ur Gerontologie Und Geriatrie, vol. 51, no. 2, pp. 200–205, 2018

    work page 2018

  30. [30]

    Stress and eeg,

    S.-H. Seo and J.-T. Lee, “Stress and eeg,” in Convergence and hybrid information technologies. InTech, 2010

    work page 2010

  31. [31]

    Frontal midline theta oscilla- tions during mental arithmetic: effects of stress,

    M. G ¨artner, S. Grimm, and M. Bajbouj, “Frontal midline theta oscilla- tions during mental arithmetic: effects of stress,” Frontiers in behavioral neuroscience, vol. 9, p. 96, 2015

    work page 2015

  32. [32]

    Quantification of human stress using commercially available single channel eeg headset,

    S. M. U. SAEED, S. M. ANW AR, and M. Majid, “Quantification of human stress using commercially available single channel eeg headset,” IEICE Transactions on Information and Systems , vol. 100, no. 9, pp. 2241–2244, 2017

    work page 2017

  33. [33]

    Stress assessment by prefrontal relative gamma,

    J. Minguillon, M. A. Lopez-Gordo, and F. Pelayo, “Stress assessment by prefrontal relative gamma,” Frontiers in computational neuroscience, vol. 10, p. 101, 2016

    work page 2016

  34. [34]

    Supervised machine learning: A review of classification techniques,

    S. B. Kotsiantis, I. Zaharakis, and P. Pintelas, “Supervised machine learning: A review of classification techniques,” Emerging artificial intelligence applications in computer engineering , vol. 160, pp. 3–24, 2007

    work page 2007

  35. [35]

    World medical asso- ciation declaration of helsinki: ethical principles for medical research involving human subjects

    G. A. of the World Medical Association et al. , “World medical asso- ciation declaration of helsinki: ethical principles for medical research involving human subjects.” The Journal of the American College of Dentists, vol. 81, no. 3, p. 14, 2014

    work page 2014

  36. [36]

    Noninvasive neural prostheses using mobile and wireless eeg,

    C. Lin, L. Ko, J. Chiou, J. Duann, R. Huang, S. Liang, T. Chiu, T. Jung et al. , “Noninvasive neural prostheses using mobile and wireless eeg,” PROCEEDINGS-IEEE, vol. 96, no. 7, p. 1167, 2008

    work page 2008

  37. [37]

    Mental tasks classifications using s-transform for bci applications,

    V . Vijean, M. Hariharan, A. Saidatul, and S. Yaacob, “Mental tasks classifications using s-transform for bci applications,” inSustainable Uti- lization and Development in Engineering and Technology (STUDENT), 2011 IEEE Conference on . IEEE, 2011, pp. 69–73

    work page 2011

  38. [38]

    A brain-computer interface for classifying eeg correlates of chronic mental stress

    R. Khosrowabadi, C. Quek, K. K. Ang, S. W. Tung, and M. Heijnen, “A brain-computer interface for classifying eeg correlates of chronic mental stress.” in IJCNN, 2011, pp. 757–762

    work page 2011

  39. [39]

    Eeg based stress level identification,

    G. Jun and K. Smitha, “Eeg based stress level identification,” in Systems, Man, and Cybernetics (SMC), 2016 IEEE International Conference on . IEEE, 2016, pp. 003 270–003 274

    work page 2016

  40. [40]

    The processing of word stress: Eeg studies on task-related components,

    J. Knaus, R. Wiese, and U. Janßen, “The processing of word stress: Eeg studies on task-related components,” in Proceedings of the 16th International Congress of Phonetic Sciences , 2007, pp. 709–712

    work page 2007

  41. [41]

    Generator sources of eeg large waves elicited by mental stress of memory recall or mental calculation,

    K. Matsunami, S. Homma, X. Y . Han, and Y . F. Jiang, “Generator sources of eeg large waves elicited by mental stress of memory recall or mental calculation,” The Japanese journal of physiology , vol. 51, no. 5, pp. 621–624, 2001

    work page 2001

  42. [42]

    The effect of a natural- istic stressor on frontal eeg asymmetry, stress, and health,

    R. S. Lewis, N. Y . Weekes, and T. H. Wang, “The effect of a natural- istic stressor on frontal eeg asymmetry, stress, and health,” Biological psychology, vol. 75, no. 3, pp. 239–247, 2007

    work page 2007

  43. [43]

    The relation between affective style of stressor on eeg asymmetry and stress scale during multimodal task,

    S. Seo, Y . Gil, and J. Lee, “The relation between affective style of stressor on eeg asymmetry and stress scale during multimodal task,” in Convergence and Hybrid Information Technology, 2008. ICCIT’08. Third International Conference on , vol. 1. IEEE, 2008, pp. 461–466

    work page 2008

  44. [44]

    Beta-endorphin response to exercise and mental stress in patients with ischemic heart disease,

    P. F. Miller, K. C. Light, E. E. Bragdon, M. N. Ballenger, M. C. Herbst, W. Maixner, A. L. Hinderliter, S. S. Atkinson, G. G. Koch, and D. S. Sheps, “Beta-endorphin response to exercise and mental stress in patients with ischemic heart disease,” Journal of psychosomatic research, vol. 37, no. 5, pp. 455–465, 1993

    work page 1993

  45. [45]

    Long-term stability of cardiovascular and catecholamine responses to stress tests: an 18-year follow-up study,

    S. S. Hassellund, A. Flaa, L. Sandvik, S. E. Kjeldsen, and M. Rostrup, “Long-term stability of cardiovascular and catecholamine responses to stress tests: an 18-year follow-up study,” Hypertension, vol. 55, no. 1, pp. 131–136, 2010

    work page 2010