pith. sign in

arxiv: 2606.26723 · v1 · pith:CY4RENWRnew · submitted 2026-06-25 · 📡 eess.SP · cs.LG· q-bio.QM

State-Specific Respiratory Signatures for Affective and Stress Recognition: Interpretable Respiratory Markers, Autocorrelation Lags, and Compact CNN Models

Andrei Velichko , Mehmet Tahir Huyut This is my paper

Pith reviewed 2026-06-26 03:07 UTC · model grok-4.3

classification 📡 eess.SP cs.LGq-bio.QM
keywords respiratory signalsstress recognitionaffective states1D-CNNautocorrelation lagsstate-specific markerswearable sensing
0
0 comments X

The pith

Respiratory signals yield state-specific markers where raw-signal CNNs reach 96.72 percent accuracy on stress detection while handcrafted features perform better on meditation and other states.

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

The paper reframes respiratory recognition as both a prediction task and an explanatory one by testing whether breathing patterns differ enough across stress, baseline, amusement, and meditation to support both high accuracy and transparent markers. It processes 60-second chest respiration segments under leave-one-subject-out validation, running compact 1D-CNNs on the raw waveform alongside physically grouped handcrafted features that cover timing, breath-to-breath variability, waveform statistics, spectral descriptors, and autocorrelation lags. The raw-signal CNN produces the highest scores for binary stress versus rest, while the feature models yield stronger results for the remaining states, especially meditation. This separation shows that the choice of representation depends on the target state. The work therefore supplies both a practical detector and a set of physiologically readable signatures.

Core claim

Using the chest respiratory channel of the WESAD dataset, 60 s windows, and leave-one-subject-out validation, raw-signal 1D-CNNs achieve 96.72 percent accuracy, 95.30 percent macro-F1, and 90.61 percent MCC for binary stress versus non-stress detection. Compact handcrafted feature models instead give higher MCC values for baseline (65.34 percent), amusement (35.69 percent), and meditation (88.65 percent). The feature space is partitioned into respiratory timing, variability, waveform statistics, spectral and time-frequency descriptors, and autocorrelation/nonlinear predictability descriptors, with autocorrelation transition lags introduced as markers of respiratory correlation scale.

What carries the argument

Autocorrelation transition lags (Zpm/Zmp) as interpretable markers of respiratory correlation scale, used together with physically grouped handcrafted signatures and raw-signal 1D-CNNs on 60 s windows.

If this is right

  • Raw-signal CNNs supply the strongest practical stress detector.
  • Interpretable respiratory signatures supply stronger markers for baseline, amusement, and especially meditation.
  • State-specific analysis shows that different respiratory properties dominate each affective condition.
  • Compact grouped-feature models deliver physiological transparency with competitive performance on several non-stress tasks.

Where Pith is reading between the lines

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

  • The same autocorrelation lags could be tested on other wearable signals such as heart-rate intervals to check whether correlation-scale effects generalize.
  • A hybrid model that routes the CNN branch to stress detection and the feature branch to the remaining states might raise overall multi-class accuracy.
  • Shorter or longer analysis windows than 60 s might shift which states are best captured by raw versus feature representations.
  • If the markers remain stable across sensor types they could support real-time affective-state feedback in consumer wearables.

Load-bearing premise

The patterns extracted from 60-second windows in this dataset will appear consistently in new users and everyday conditions without shifts from sensor placement or labeling differences.

What would settle it

Running the identical CNN and feature pipelines on an independent respiration dataset collected with different hardware or subjects and observing stress-detection accuracy fall below 80 percent.

Figures

Figures reproduced from arXiv: 2606.26723 by Andrei Velichko, Mehmet Tahir Huyut.

Figure 1
Figure 1. Figure 1: Respiratory data used in the study. The primary binary analysis uses stress [PITH_FULL_IMAGE:figures/full_fig_p008_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Compact workflow of the study. The WESAD chest RESP channel is converted [PITH_FULL_IMAGE:figures/full_fig_p010_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Representative distributions of selected single-feature stress markers used in [PITH_FULL_IMAGE:figures/full_fig_p023_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Best LOSO MCC as a function of the number of input features for compact [PITH_FULL_IMAGE:figures/full_fig_p024_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Final comparison between the raw-signal CNN refit and the best compact feature [PITH_FULL_IMAGE:figures/full_fig_p028_5.png] view at source ↗
read the original abstract

Respiratory activity is a direct and interpretable physiological channel for wearable stress and affective-state recognition, yet many studies emphasize classification accuracy without identifying which respiratory properties separate different states. This work reframes RESP-based recognition as a joint predictive and explanatory problem. Using the chest respiratory channel of the WESAD dataset, we analyze 60 s windows under leave-one-subject-out validation and combine two complementary branches: compact raw-signal one-dimensional convolutional neural networks (1D-CNNs) and physically grouped handcrafted respiratory signatures. The primary application task is binary stress versus non-stress detection, while baseline, stress, amusement, and meditation are additionally analyzed in a one-vs-rest setting to reveal state-specific respiratory markers. The feature space is organized into respiratory timing, breath-to-breath variability, waveform statistics, spectral/time-frequency descriptors, and autocorrelation/nonlinear predictability descriptors, with the raw 60 s signal treated as a sixth representation for the CNN branch. We introduce autocorrelation transition lags (Zpm/Zmp) as interpretable markers of respiratory correlation scale and separately evaluate exploratory FEG-Pro/Lyapunov-like descriptors. In the final CNN refit setting, the raw-signal model achieved the strongest stress-vs-rest performance, with accuracy 96.72 percent, macro-F1 95.30 percent, and MCC 90.61 percent. In contrast, compact feature models were stronger for baseline, with MCC 65.34 percent, amusement, with MCC 35.69 percent, and especially meditation, with MCC 88.65 percent. These results show that CNNs are most useful for the practical stress detector, whereas interpretable respiratory signatures provide stronger and more physiologically transparent state-specific markers for several non-stress conditions.

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 reframes respiratory-signal classification on the WESAD chest RESP channel as a joint predictive/explanatory task. Using 60 s windows and leave-one-subject-out (LOSO) validation, it compares compact 1D-CNNs operating on raw signals against handcrafted feature sets (timing, breath-to-breath variability, waveform statistics, spectral/time-frequency, autocorrelation/nonlinear descriptors) for binary stress-vs-rest detection and one-vs-rest classification of baseline, amusement, and meditation. The headline result is that the raw-signal CNN attains 96.72 % accuracy, 95.30 % macro-F1 and 90.61 % MCC “in the final CNN refit setting,” while the interpretable feature models are reported stronger for the non-stress states.

Significance. If the reported LOSO numbers are verifiably out-of-sample, the work supplies both a practical stress detector and physiologically grounded, state-specific respiratory markers (including the introduced autocorrelation transition lags) that could be useful for wearable affective computing. The explicit separation of CNN versus feature-model strengths across states is a constructive contribution.

major comments (2)
  1. [Abstract and §4] Abstract and §4 (Results): the performance figures 96.72 % / 95.30 % / 90.61 % are stated for the “final CNN refit setting” under the LOSO protocol. The manuscript must explicitly state whether these metrics were obtained on a fresh held-out partition after refitting on the union of all folds, or whether they reflect performance on the same data used for refitting. If the latter, the numbers do not constitute evidence of generalization and the central claim is unsupported.
  2. [§3] §3 (Methods): the 1D-CNN architecture, hyper-parameters, training schedule, and precise definition of the “final refit” procedure are not supplied at a level that permits independent verification or reproduction of the quoted accuracy/F1/MCC values.
minor comments (2)
  1. [Tables and §4] Table captions and §4 should report the number of subjects, total windows, and any statistical tests or confidence intervals accompanying the MCC and F1 figures.
  2. [§2] The autocorrelation transition lags (Zpm/Zmp) are introduced as novel markers; a short derivation or explicit formula in §2 would improve interpretability.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback on validation clarity and reproducibility. We address each major comment below and will revise the manuscript accordingly.

read point-by-point responses

  1. Referee: [Abstract and §4] Abstract and §4 (Results): the performance figures 96.72 % / 95.30 % / 90.61 % are stated for the “final CNN refit setting” under the LOSO protocol. The manuscript must explicitly state whether these metrics were obtained on a fresh held-out partition after refitting on the union of all folds, or whether they reflect performance on the same data used for refitting. If the latter, the numbers do not constitute evidence of generalization and the central claim is unsupported.

    Authors: The 'final CNN refit setting' metrics are obtained after refitting on the union of all LOSO folds and evaluating on that same combined data; they are not from a fresh held-out partition. The primary evidence of generalization remains the LOSO results reported throughout the paper. We agree the current phrasing risks implying additional out-of-sample validation and will revise the abstract and §4 to explicitly distinguish the refit numbers (as final-model performance) from the LOSO generalization results, which we will emphasize as the central claim. revision: yes

  2. Referee: [§3] §3 (Methods): the 1D-CNN architecture, hyper-parameters, training schedule, and precise definition of the “final refit” procedure are not supplied at a level that permits independent verification or reproduction of the quoted accuracy/F1/MCC values.

    Authors: We agree that the manuscript does not currently provide sufficient detail on the 1D-CNN architecture, hyperparameters, training schedule, or the exact 'final refit' procedure. We will expand §3 with a complete specification including layer counts and types, kernel sizes, strides, activation functions, dropout rates, optimizer, learning rate schedule, batch size, epoch limits, early stopping, and the precise sequence of operations for the post-LOSO refit. This will enable independent reproduction. revision: yes

Circularity Check

0 steps flagged

No significant circularity; empirical ML results on public dataset with standard LOSO validation

full rationale

The paper reports classification performance for 1D-CNN and handcrafted respiratory feature models on the WESAD chest RESP channel using 60 s windows and leave-one-subject-out validation. The quoted metrics (including the 96.72% accuracy in the 'final CNN refit setting') are presented as outcomes of this protocol without any mathematical derivation chain, self-definitional equations, or load-bearing self-citations that reduce the claimed results to fitted inputs by construction. The work is self-contained as an empirical evaluation; no steps match the enumerated circularity patterns.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract-only review yields no explicit free parameters, axioms, or invented entities; all modeling choices and data assumptions remain implicit.

pith-pipeline@v0.9.1-grok · 5863 in / 1062 out tokens · 38101 ms · 2026-06-26T03:07:11.413335+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

49 extracted references · 46 canonical work pages · 1 internal anchor

  1. [1]

    A concept model of mHealth sensorics for digital assistance of human cogni- tive resilience

    Meigal A, Gerasimova-Meigal L, Korzun D. A concept model of mHealth sensorics for digital assistance of human cogni- tive resilience. In: Proceedings of the 34th Conference of Open Innovations Association (FRUCT); 2023. p. 100–107. doi:10.23919/FRUCT60429.2023.10328167

    work page doi:10.23919/fruct60429.2023.10328167 2023

  2. [2]

    Breathing rhythms and emo- tions

    Homma I, Masaoka Y. Breathing rhythms and emo- tions. Experimental Physiology. 2008;93(9):1011–1021. doi:10.1113/expphysiol.2008.042424

    work page doi:10.1113/expphysiol.2008.042424 2008

  3. [3]

    How breath-control can change your life: a systematic review on psycho-physiological correlates of slow breathing

    Zaccaro A, Piarulli A, Laurino M, Garbella E, Menicucci D, Neri B, Gemignani A. How breath-control can change your life: a systematic review on psycho-physiological correlates of slow breathing. Frontiers in Human Neuroscience. 2018;12:353. doi:10.3389/fnhum.2018.00353. 38

    work page doi:10.3389/fnhum.2018.00353 2018

  4. [4]

    Hypothesis: pulmonary afferent activity pat- terns during slow, deep breathing contribute to the neural induc- tion of physiological relaxation

    Noble DJ, Hochman S. Hypothesis: pulmonary afferent activity pat- terns during slow, deep breathing contribute to the neural induc- tion of physiological relaxation. Frontiers in Physiology. 2019;10:1176. doi:10.3389/fphys.2019.01176

    work page doi:10.3389/fphys.2019.01176 2019

  5. [5]

    Inhalation/exhalation ratio modulates the effect of slow breathing on heart rate variability and relaxation

    Van Diest I, Verstappen K, Aubert AE, Widjaja D, Vansteenwegen D, Vlemincx E. Inhalation/exhalation ratio modulates the effect of slow breathing on heart rate variability and relaxation. Applied Psy- chophysiology and Biofeedback. 2014;39:171–180. doi:10.1007/s10484- 014-9253-x

    work page doi:10.1007/s10484- 2014

  6. [6]

    Scientific Reports

    Kral TRA, Weng HY, Mitra V, Imhoff-Smith T, Azemi E, Goldman R, RosenkranzMA,WuS,ChenA,DavidsonRJ.Slowerrespirationrateis associated with higher self-reported well-being after wellness training. Scientific Reports. 2023;13:17126. doi:10.1038/s41598-023-43176-w

    work page doi:10.1038/s41598-023-43176-w 2023

  7. [7]

    Inclusion of res- piratory frequency information in heart rate variability analysis for stress assessment

    Hernando A, Lazaro J, Gil E, Arza A, Garzon-Rey JM, Lopez-Anton R, de la Camara C, Laguna P, Aguilo J, Bailon R. Inclusion of res- piratory frequency information in heart rate variability analysis for stress assessment. IEEE Journal of Biomedical and Health Informat- ics. 2016;20(4):1016–1025. doi:10.1109/JBHI.2016.2553578

    work page doi:10.1109/jbhi.2016.2553578 2016

  8. [8]

    A comprehensive evalua- tion of linear and non-linear HRV parameters between paced breathing and stressful mental state

    Chand K, Chandra S, Dutt V. A comprehensive evalua- tion of linear and non-linear HRV parameters between paced breathing and stressful mental state. Heliyon. 2024;10:e32195. doi:10.1016/j.heliyon.2024.e32195

    work page doi:10.1016/j.heliyon.2024.e32195 2024

  9. [9]

    WE- SAD: a multimodal dataset for wearable stress and affect detection

    Schmidt P, Reiss A, Duerichen R, Marberger C, Van Laerhoven K. WE- SAD: a multimodal dataset for wearable stress and affect detection. In: Proceedings of the 20th ACM International Conference on Multimodal Interaction; 2018. p. 400–408. doi:10.1145/3242969.3242985

    work page doi:10.1145/3242969.3242985 2018

  10. [10]

    Stress detection with machine learning and deep learning using multimodal physiological data

    Bobade P, Vani M. Stress detection with machine learning and deep learning using multimodal physiological data. In: Proceedings of the 2020 Second International Conference on Inventive Research in Com- puting Applications (ICIRCA); 2020. p. 51–57

    2020

  11. [11]

    Classification of mental stress from wearable physiological sensors using image- encoding-based deep neural network

    Ghosh S, Kim S, Ijaz MF, Singh PK, Mahmud M. Classification of mental stress from wearable physiological sensors using image- encoding-based deep neural network. Biosensors. 2022;12(12):1153. doi:10.3390/bios12121153. 39

    work page doi:10.3390/bios12121153 2022

  12. [12]

    Resp-BoostNet: men- tal stress detection from biomarkers measurable by smartwatches us- ing boosting neural network technique

    Kumar S, Chauhan AR, Kumar A, Yang G. Resp-BoostNet: men- tal stress detection from biomarkers measurable by smartwatches us- ing boosting neural network technique. IEEE Access. 2024;12:149861– 149874. doi:10.1109/ACCESS.2024.3476777

    work page doi:10.1109/access.2024.3476777 2024

  13. [13]

    Respiratory signal and human stress: non- contact detection of stress with a low-cost depth sensing camera

    Shan Y, Li S, Chen T. Respiratory signal and human stress: non- contact detection of stress with a low-cost depth sensing camera. Inter- national Journal of Machine Learning and Cybernetics. 2020;11:1825–

    2020

  14. [14]

    doi:10.1007/s13042-020-01070-4

    work page doi:10.1007/s13042-020-01070-4

  15. [15]

    Enhancement of stress analysis performance using res- piration information

    Barik S, Pal S. Enhancement of stress analysis performance using res- piration information. In: Proceedings of the 2023 IEEE 3rd Applied Signal Processing Conference (ASPCON); 2023. p. 216–221

    2023

  16. [16]

    Physiologically explainable ensemble framework for stress classification via respiratory signals

    Yang C, Wei S, Li J, Liu C. Physiologically explainable ensemble framework for stress classification via respiratory signals. Technologies. 2025;13(9):411. doi:10.3390/technologies13090411

    work page doi:10.3390/technologies13090411 2025

  17. [17]

    Breath rate variability: a novel measure to study the meditation effects

    Soni R, Muniyandi M. Breath rate variability: a novel measure to study the meditation effects. International Journal of Yoga. 2019;12(1):45–54. doi:10.4103/ijoy.ijoy_27_17

    work page doi:10.4103/ijoy.ijoy_27_17 2019

  18. [18]

    Temporal variations in the pattern of breathing: techniques, sources, andapplicationstotranslationalsciences.TheJournalofPhys- iological Sciences

    Oku Y. Temporal variations in the pattern of breathing: techniques, sources, andapplicationstotranslationalsciences.TheJournalofPhys- iological Sciences. 2022;72:32. doi:10.1186/s12576-022-00847-z

    work page doi:10.1186/s12576-022-00847-z 2022

  19. [19]

    It’s about time- breathing dynamics modulate emotion and cognition

    Goheen J, Catal Y, MacPhee I, Call TM, Carson C, Ali R, Khan R, Weche K, Anderson JAE, Northoff G. It’s about time- breathing dynamics modulate emotion and cognition. Psychophysiol- ogy. 2025;62:e70149. doi:10.1111/psyp.70149

    work page doi:10.1111/psyp.70149 2025

  20. [20]

    Review on psychological stress detection using biosig- nals

    Giannakakis G, Grigoriadis D, Giannakaki K, Simantiraki O, Roniotis A, Tsiknakis M. Review on psychological stress detection using biosig- nals. IEEE Transactions on Affective Computing. 2022;13(1):440–460. doi:10.1109/TAFFC.2019.2927337

    work page doi:10.1109/taffc.2019.2927337 2022

  21. [21]

    Wearable- based affect recognition: a review

    Schmidt P, Reiss A, Duerichen R, Van Laerhoven K. Wearable- based affect recognition: a review. Sensors. 2019;19(19):4079. doi:10.3390/s19194079. 40

    work page doi:10.3390/s19194079 2019

  22. [22]

    A review on mental stress detection using wearable sensors and machine learning techniques

    Gedam S, Paul S. A review on mental stress detection using wearable sensors and machine learning techniques. IEEE Access. 2021;9:84045– 84066. doi:10.1109/ACCESS.2021.3085502

    work page doi:10.1109/access.2021.3085502 2021

  23. [23]

    Personalized stress de- tection using biosignals from wearables: a scoping review

    Bolpagni M, Pardini S, Dianti M, Gabrielli S. Personalized stress de- tection using biosignals from wearables: a scoping review. Sensors. 2024;24(10):3221. doi:10.3390/s24103221

    work page doi:10.3390/s24103221 2024

  24. [24]

    An overview of stress analysis based on phys- iological signals: systematic review of open datasetsand current trends

    Chatzaki C, Tsiknakis M. An overview of stress analysis based on phys- iological signals: systematic review of open datasetsand current trends. Sensors. 2025;25:7108. doi:10.3390/s25237108

    work page doi:10.3390/s25237108 2025

  25. [25]

    A sensitivity analysis of biophysiological responses of stress for wearable sensors in connected health

    Iqbal T, Redon-Lurbe P, Simpkin A, Elahi A, Ganly S, Wijns W, Shahzad A. A sensitivity analysis of biophysiological responses of stress for wearable sensors in connected health. IEEE Access. 2021;9:93567– 93579. doi:10.1109/ACCESS.2021.3082423

    work page doi:10.1109/access.2021.3082423 2021

  26. [26]

    A comparison of personalized and generalized approaches to emotion recognition using consumer wearable devices: machine learning study

    Li J, Washington PY. A comparison of personalized and generalized approaches to emotion recognition using consumer wearable devices: machine learning study. JMIR AI. 2023;3:e52171. doi:10.2196/52171

    work page doi:10.2196/52171 2023

  27. [27]

    Hierarchical extreme puzzle learning machine-based emotion recognition using multimodal physiological signals

    Pradhan A, Srivastava S. Hierarchical extreme puzzle learning machine-based emotion recognition using multimodal physiological signals. Biomedical Signal Processing and Control. 2023;83:104624. doi:10.1016/j.bspc.2023.104624

    work page doi:10.1016/j.bspc.2023.104624 2023

  28. [28]

    Deep learning-based automated emotion recognition using multimodal physiological signals and time-frequency methods

    Kumar S, Govarthan K, Aleem A, Gadda S, Ganapathy N, Fredo J, Ronickom A, Kumar P. Deep learning-based automated emotion recognition using multimodal physiological signals and time-frequency methods. IEEE Transactions on Instrumentation and Measurement. 2024;73:1–12. doi:10.1109/TIM.2024.3420349

    work page doi:10.1109/tim.2024.3420349 2024

  29. [29]

    Dynamic alignment and fusion of multimodal physiological patterns for stress recognition

    Zhang X, Wei X, Zhou Z, Zhao Q, Zhang S, Yang Y, Li R, Hu B. Dynamic alignment and fusion of multimodal physiological patterns for stress recognition. IEEE Transactions on Affective Computing. 2024;15(2):685–696. doi:10.1109/TAFFC.2023.3290177

    work page doi:10.1109/taffc.2023.3290177 2024

  30. [30]

    Machine and deep learning models for stress detection using multimodal physiological data

    Abdelfattah E, Joshi S, Tiwari S. Machine and deep learning models for stress detection using multimodal physiological data. IEEE Access. 2025;13:4597–4608. doi:10.1109/ACCESS.2024.3525459. 41

    work page doi:10.1109/access.2024.3525459 2025

  31. [31]

    A deep learning approach to stress recognition through multimodal physi- ological signal image transformation

    Yang S, Gao Y, Zhu Y, Zhang L, Xie Q, Lu X, Wang F, Zhang Z. A deep learning approach to stress recognition through multimodal physi- ological signal image transformation. Scientific Reports. 2025;15:12712. doi:10.1038/s41598-025-01228-3

    work page doi:10.1038/s41598-025-01228-3 2025

  32. [32]

    Hierarchical transformer with auxiliary learning for subject-independent respira- tion emotion recognition

    Wang Y, Xu C, Na W, Liu D, Yan J, Yao S, Wu Q. Hierarchical transformer with auxiliary learning for subject-independent respira- tion emotion recognition. IEEE Sensors Journal. 2025;25:31290–31301. doi:10.1109/JSEN.2025.3587271

    work page doi:10.1109/jsen.2025.3587271 2025

  33. [33]

    Towards wearable respiration monitor- ing: 1D-CRNN-based breathing detection in smart textiles

    Steinmetzer T, Michel S. Towards wearable respiration monitor- ing: 1D-CRNN-based breathing detection in smart textiles. Sensors. 2025;25:6832. doi:10.3390/s25226832

    work page doi:10.3390/s25226832 2025

  34. [34]

    Exploring edge ma- chine learning-based stress prediction using wearable devices

    Sim S-H, Paranjpe T, Roberts N, Zhao M. Exploring edge ma- chine learning-based stress prediction using wearable devices. In: Proceedings of the 21st IEEE International Conference on Ma- chine Learning and Applications (ICMLA); 2022. p. 1266–1273. doi:10.1109/ICMLA55696.2022.00203

    work page doi:10.1109/icmla55696.2022.00203 2022

  35. [35]

    Combining multiple tiny ma- chine learning models for multimodal context-aware stress recogni- tion on constrained microcontrollers

    Gibbs M, Woodward K, Kanjo E. Combining multiple tiny ma- chine learning models for multimodal context-aware stress recogni- tion on constrained microcontrollers. IEEE Micro. 2024;44(1):67–75. doi:10.1109/MM.2023.3329218

    work page doi:10.1109/mm.2023.3329218 2024

  36. [36]

    TinyStressNet: on-device stress assessment with wearable sensors on edge devices

    Jaiswal D, Mukhopadhyay S, Sharma V. TinyStressNet: on-device stress assessment with wearable sensors on edge devices. In: Pro- ceedings of the IEEE PerCom Workshops; 2024. p. 166–171. doi:10.1109/PERCOMWORKSHOPS59983.2024.10502631

    work page doi:10.1109/percomworkshops59983.2024.10502631 2024

  37. [37]

    BioEdgeNet: a com- pact deep residual network for stress recognition on edge de- vices

    Cvetkovic S, Stankovic S, Nikolic SV. BioEdgeNet: a com- pact deep residual network for stress recognition on edge de- vices. Biomedical Signal Processing and Control. 2025;102:107361. doi:10.1016/j.bspc.2024.107361

    work page doi:10.1016/j.bspc.2024.107361 2025

  38. [38]

    Deployment of TinyML-based stress clas- sification using computational constrained health wearable

    Abu-Samah A, Ghaffa D, Abdullah NF, Kamal N, Nordin R, Dela Cruz JC, Magwili G, Mercado RJ. Deployment of TinyML-based stress clas- sification using computational constrained health wearable. Electron- ics. 2025;14(4):687. doi:10.3390/electronics14040687. 42

    work page doi:10.3390/electronics14040687 2025

  39. [39]

    Investi- gating lightweight and interpretable machine learning models for ef- ficient and explainable stress detection

    Ghose D, Chatterjee A, Balapuwaduge IAM, Lin Y, Dash S. Investi- gating lightweight and interpretable machine learning models for ef- ficient and explainable stress detection. Frontiers in Digital Health. 2025;7:1523381. doi:10.3389/fdgth.2025.1523381

    work page doi:10.3389/fdgth.2025.1523381 2025

  40. [40]

    Breathing rhythm and patternandtheirinfluenceonemotion.AnnualReviewofNeuroscience

    Ashhad S, Kam K, Del Negro CD, Feldman JL. Breathing rhythm and patternandtheirinfluenceonemotion.AnnualReviewofNeuroscience. 2022;45:223–247. doi:10.1146/annurev-neuro-090121-014424

    work page doi:10.1146/annurev-neuro-090121-014424 2022

  41. [41]

    From lung to brain: respiration modulates neural and mental activity

    Goheen J, Anderson JAE, Zhang J, Northoff G. From lung to brain: respiration modulates neural and mental activity. Neuroscience Bul- letin. 2023;39:1577–1590. doi:10.1007/s12264-023-01070-5

    work page doi:10.1007/s12264-023-01070-5 2023

  42. [42]

    Brief structured respiration prac- tices enhance mood and reduce physiological arousal

    Balban MY, Neri E, Kogon MM, Weed L, Nouriani B, Jo B, Holl G, Zeitzer J, Spiegel D, Huberman A. Brief structured respiration prac- tices enhance mood and reduce physiological arousal. Cell Reports Medicine. 2023;4:100895. doi:10.1016/j.xcrm.2022.100895

    work page doi:10.1016/j.xcrm.2022.100895 2023

  43. [43]

    Ef- fect of breathwork on stress and mental health: a meta-analysis of randomised-controlled trials

    Fincham GW, Strauss C, Montero-Marin J, Cavanagh K. Ef- fect of breathwork on stress and mental health: a meta-analysis of randomised-controlled trials. Scientific Reports. 2023;13:432. doi:10.1038/s41598-022-27247-y

    work page doi:10.1038/s41598-022-27247-y 2023

  44. [44]

    Slow breathing for reducing stress: the effect of extending exhale

    Birdee G, Nelson K, Wallston K, Nian H, Diedrich A, Paranjape SY, Abraham R, Gamboa A. Slow breathing for reducing stress: the effect of extending exhale. Complementary Therapies in Medicine. 2023;73:102937. doi:10.1016/j.ctim.2023.102937

    work page doi:10.1016/j.ctim.2023.102937 2023

  45. [45]

    Robust estimation of respiratory variability uncovers correlates of limbic brain activity and transcutaneous cervical vagus nerve stimulation in the context of traumatic stress

    Gazi A, Wittbrodt M, Harrison AB, Sundararaj S, Gurel N, Nye J, Shah AJ, Vaccarino V, Bremner D, Inan OT. Robust estimation of respiratory variability uncovers correlates of limbic brain activity and transcutaneous cervical vagus nerve stimulation in the context of traumatic stress. IEEE Transactions on Biomedical Engineering. 2022;69(2):849–859. doi:10.1...

    work page doi:10.1109/tbme.2021.3108135 2022

  46. [46]

    Heart and breathing rate variations as biomarkers for anxiety detection

    Ritsert F, Elgendi M, Galli V, Menon C. Heart and breathing rate variations as biomarkers for anxiety detection. Bioengineering. 2022;9(11):711. doi:10.3390/bioengineering9110711

    work page doi:10.3390/bioengineering9110711 2022

  47. [47]

    Exploring respiratory parameters related to psy- chophysiological indexes of mental health

    Suzuki M, Sato H. Exploring respiratory parameters related to psy- chophysiological indexes of mental health. In: Proceedings of the IEEE 43 International Symposium on Medical Measurements and Applications (MeMeA); 2023. p. 1–6. doi:10.1109/MeMeA57477.2023.10171945

    work page doi:10.1109/memea57477.2023.10171945 2023

  48. [48]

    Breathing rate complexity features for in-the-wild stress and anxiety measurement

    Tiwari A, Narayanan SS, Falk TH. Breathing rate complexity features for in-the-wild stress and anxiety measurement. In: Proceedings of the 27th European Signal Processing Conference (EUSIPCO); 2019. p. 1–5. doi:10.23919/EUSIPCO.2019.8902700

    work page doi:10.23919/eusipco.2019.8902700 2019

  49. [49]

    FEG-Pro: Forecast-Error Growth Profiling for Finite-Horizon Instability Analysis of Nonlinear Time Series

    Velichko A, N’Gbo N, Carpentieri B, Shams M. FEG-Pro: forecast-error growth profiling for finite-horizon instability anal- ysis of nonlinear time series. arXiv:2605.17282 [nlin.CD]. 2026. doi:10.48550/arXiv.2605.17282. 44

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2605.17282 2026