HealthCAT: An Interpretable Encoder-only Transformer Framework for Health Indicator Prediction and Temporal Interpretation of Wearable Sensor Data

Haejin Lee; Joshua Y. Kim; Kalina Yacef; Xiaotong Yu

REVIEW 5 major objections 5 minor 55 references

HealthCAT pairs a transformer's health predictions with time-step-level explanations that survive aggressive masking.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

T0 review · deepseek-v4-flash

2026-08-01 04:19 UTC pith:MAX7KXJF

load-bearing objection Useful application of a known interpretability method to wearable sensor time series, but the unexplained predictive gap over its own transformer backbone is the main weakness. the 5 major comments →

arxiv 2607.27635 v1 pith:MAX7KXJF submitted 2026-07-30 cs.AI cs.LG

HealthCAT: An Interpretable Encoder-only Transformer Framework for Health Indicator Prediction and Temporal Interpretation of Wearable Sensor Data

Xiaotong Yu , Joshua Y. Kim , HaeJin Lee , Kalina Yacef This is my paper

classification cs.AI cs.LG

keywords interpretable deep learningwearable sensorstransformertime-step attributionclass activationhealth predictionmasking evaluationtime-series interpretability

verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

The paper proposes HealthCAT, a framework that predicts health indicators from wrist-worn sensor data and, in the same pass, assigns each minute or hour of the recording a class-specific importance score. Its central claim is that these time-step-level scores are genuinely informative—not just plausible-looking—because masking out the low-scoring time steps hurts predictions far less than masking random ones. On two real-world datasets, adolescent physical-activity and adult sleep, the model beat LSTM, GRU, and plain Transformer baselines by up to 17% in F1 and 12% in accuracy. If the claim holds, HealthCAT would let researchers see when, during the day, a person's behaviour pushes a prediction toward a healthy or unhealthy outcome, moving wearable analytics beyond aggregated daily totals to individual-level temporal patterns.

Core claim

HealthCAT integrates an Encoder-only Transformer with an Attentive Class Activation Token (AttentiveCAT) that produces, per time step and per class, an importance score derived from the hidden representation, the gradient of the class logit with respect to it, and the head-averaged self-attention weights. The paper reports that on the physical-activity dataset (206 adolescents, healthy-weight status) and the sleep dataset (100 adults, apnea-hypopnea index), the model outperformed three deep-learning baselines, and in cumulative masking experiments the time steps selected by AttentiveCAT retained significantly more predictive value than randomly selected time steps across all masking conditio

What carries the argument

The central object is the AttentiveCAT score: for each encoder layer l and time step t, it is the product of the layer's head-averaged attention weights α^l_t and the class activation token CAT^l_t, where CAT^l_t is the element-wise product of the hidden representation h^l_t with the gradient of the class logit with respect to h^l_t. Summed over layers and pooled over hidden dimensions, each time step receives one scalar class-specific score. This combines two signals: the gradient tells how much the model's decision changes with that time step's representation, while the attention weights tell how much the model listens to that time step. The scores are then projected onto domain features (

Load-bearing premise

The whole interpretation stack assumes that the intensity cut-points and bout-length thresholds adopted from other populations correctly classify physical activity intensity in Pacific Island adolescents; if those thresholds are off, the time-step scores and the masking results would describe threshold artifacts rather than true behaviour.

What would settle it

Recalibrate the intensity cut-points and bout-length thresholds on a sample of the target adolescent population, then rerun the masking experiments; if the AttentiveCAT-vs-random advantage disappears, the claim that the identified time steps are predictively informative is an artifact of the chosen thresholds.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

The masking protocol provides a reusable, quantitative template for validating any future temporal-interpretability method, not just this one.
Individual-level analytics: a clinician or coach can inspect, for a specific person, which hours of moderate, light, or sedentary activity support the predicted health category.
The framework is flexible across temporal resolutions—hourly features for physical activity, minute-level for sleep—so the same code path can be applied to other wearable sensors and health indicators.
The contrastive participant examples suggest that temporal placement of activity matters: people with similar daily totals can show very different importance patterns, and HealthCAT makes that difference visible.

Where Pith is reading between the lines

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

The masking comparison is against random selection, not against ground-truth behavioural relevance; a clinician-labelled study would be the natural next step to test whether the highlighted minutes are ones a human expert would also flag.
Because the features inherit fixed activity-intensity thresholds borrowed from other populations, a sensitivity analysis that recalibrates those thresholds for the target group would reveal whether the interpretations are threshold-dependent or stable.
If these scores generalise, a natural application is just-in-time interventions: the model could point to the hours of the day where a person's routine has the most leverage for a healthier outcome.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Desk Editor's Note

Useful application of a known interpretability method to wearable sensor time series, but the unexplained predictive gap over its own transformer backbone is the main weakness.

read the letter

First, know this: the paper imports AttentiveCAT from NLP into wearable health time series, wraps it in a clean three-stage framework, and evaluates the resulting time-step importance scores with a masking-based faithfulness test. That is a legitimate, useful contribution. It is not a paradigm shift, but it moves a known method into a domain where fine-grained temporal explanations are genuinely needed.

What the paper does well: the preprocessing is clearly described; the two real-world datasets (PA and DREAMT) are appropriate; the masking protocol is a sensible way to check whether the scores point at time steps the model actually relies on; and the domain-adapted visualization showing contributions by time-of-day and intensity level is a nice practical touch. The authors also acknowledge the small cohort and the lack of contextual variables in their limitations. The writing is honest and mostly careful.

The soft spots, in order of importance. The biggest issue is the predictive gain over the vanilla Transformer. AttentiveCAT is a post hoc scoring layer; it does not change the forward pass or the training objective. So if HealthCAT uses the same encoder as the Transformer baseline, the two should give identical predictions. Table II shows a gap of about 0.06 F1 on both datasets. The paper claims this is due to 'sensor modality aware preprocessing and Encoder-only structure,' but the baseline presumably uses the same preprocessing and structure. The authors need to specify exactly what differs — different hyperparameters, a different training scheme, or a modified architecture. Without that, the headline improvement over baselines is confounded.

Second, the masking evaluation is a self-consistency check, not external validation. Showing that top-ranked time steps retain more predictive information than random time steps tells you the scores reflect the model's decision process. It does not tell you those time steps are the ones that actually matter for the physiological outcome. That is fine as a faithfulness test, but the paper should not oversell it as evidence about 'when and how physical activity patterns occur' in the world.

Third, the stress-test concern about intensity thresholds is legitimate: the MVPA/LPA/SED cut-points come from a calibration study in children and are applied without revalidation to Pacific Island adolescents. If body composition or activity patterns shift the boundaries, the hourly bout features — and every attribution built on them — inherit that bias. This doesn't sink the paper, but it should be stated as a limitation rather than hidden.

Minor points: the abstract says 'significantly more predictive value than random selection across all masking conditions,' which holds for the HealthCAT rows but not for all baseline-model comparisons in the tables. The reference list has [50] and [51] as the same paper. And the t-statistic is reported with df=100 rather than 99.

Who should read it: people working on explainable AI for time series or wearable health informatics. It is a solid application paper with a useful evaluation protocol, worth serious review. I would ask the authors to resolve the predictive-gap issue and recalibrate or at least discuss the thresholds before publication.

Referee Report

5 major / 5 minor

Summary. The paper proposes HealthCAT, an Encoder-only Transformer combined with AttentiveCAT class-activation scores, to produce class-specific, time-step-level importance scores for wearable sensor time series. The framework is evaluated on two real-world datasets: a physical activity dataset of 206 Pacific Island adolescents (predicting healthy weight status) and the DREAMT sleep dataset of 100 adults (predicting AHI-based sleep health). The authors report that HealthCAT outperforms vanilla Transformer, GRU, and LSTM baselines by up to 17% in F1 score, and that AttentiveCAT-guided masking retains more predictive information than random masking. They also present hour-of-day visualizations of MVPA/LPA/SED contributions for two example participants.

Significance. If the central claims are correct, the paper provides a practical step beyond aggregated wearable summaries: class-specific, time-step-level interpretations that can be mapped onto domain-relevant cycles. The use of two real-world datasets and a quantitative masking protocol are strengths, and the domain-adapted visualization is a useful template for health researchers. However, the current evaluation is mostly internal: the masking experiments demonstrate that the attribution scores track the model's own decisions, not that the identified time steps correspond to externally meaningful behavioral or clinical phenomena. The claim of being the 'first validated time-step-level interpretation' is therefore stronger than the evidence supports. The main predictive comparison is also confounded because the architectural difference between HealthCAT and the vanilla Transformer baseline is not specified.

major comments (5)

[§III-C / Table II] The paper never specifies how HealthCAT differs architecturally from the vanilla Transformer baseline. The only new component described, AttentiveCAT, is a post hoc scoring mechanism (Eqs. 3–5) and cannot alter training or predictions. Yet HealthCAT improves F1 by 0.063 on PA and 0.074 on DREAMT over the Transformer. HealthCAT's classification head uses the concatenated final-layer representation h̄ = [h^L_1;…;h^L_w]; if the baseline uses a [CLS]-style head or different pooling, that is an architectural difference unrelated to the proposed interpretability layer. Preprocessing differences are also unspecified. Without a controlled variant, RQ1's superiority claim is confounded and not reproducible.
[§III-E / Table III] The masking experiment tests internal model fidelity, not external behavioural relevance. Showing that steps ranked by AttentiveCAT retain more predictive value than random steps only establishes that the scores track the model's own decision process; it does not validate that these time steps are the ones that matter for the health outcome, since no external ground truth is available. The abstract's phrase 'validated time-step-level interpretability' and 'predictively informative' therefore overstates the evidence. In addition, the protocol is ambiguous: 'zeroing out ... bottom 25%, 50%, and 75%' is inconsistent with the non-monotonic HealthCAT results (e.g., 75% masking F1 0.7578 > 25% masking F1 0.7277 in Table III). Clarify whether masking keeps top-k or removes bottom-k, and discuss ranking stability.
[§III-C, Eq. (4)] Equation (4) is not well-defined as written. α^l_t is described only as head-averaged self-attention weights, but it is not stated whether it is a scalar, a vector over tokens, or a matrix; CAT^{c,l}_t is a d-dimensional vector. The product α^l_t · CAT^{c,l}_t therefore lacks a defined dimension. The original AttentiveCAT (ref. [20]) sums over token positions weighted by attention. Please provide the explicit summation over source tokens and define all indices.
[§III-E / Table III] The paper reports AttentiveCAT-guided masking for GRU and LSTM baselines, but AttentiveCAT as defined in Eqs. (3)–(5) requires Transformer self-attention weights. No description is given of how CAT or attention scores are computed for recurrent models. If gradients with respect to GRU/LSTM attention weights were used, the implementation is nontrivial and must be specified. As it stands, the cross-model rows of Table III cannot be reproduced or interpreted.
[§III-B / §IV-C] The intensity and bout thresholds for MVPA/LPA/SED are taken from [50]/[51] without recalibration for this cohort of Pacific Island adolescents (10–16 years). Because the multivariate features are bout counts derived from these thresholds, and because the domain-adapted visualizations in Fig. 2 are built from those features, a systematic threshold mismatch would propagate into both predictions and interpretations. The paper neither acknowledges this risk nor provides a sensitivity analysis (e.g., varying cut-points or bout lengths). This is important for the central interpretive claim about 'when and how physical activity patterns occur.'

minor comments (5)

[Appendix A vs §III-E] Appendix A states 'AdamW as optimiser,' while Section III-E states Rectified Adam (RAdam). This inconsistency must be resolved for reproducibility.
[References] References [50] and [51] are the same paper (Phillips, Parfitt, and Rowlands 2013); one duplicate should be removed or renumbered.
[§IV-B] The statistical reporting is slightly incorrect: a paired t-test with 100 trials has df = 99, not 100 (e.g., t(100)=2.016). Also, no multiple-comparison correction is applied across the 12 model-masking comparisons, and several comparisons are not significant; the abstract's 'across all masking conditions (p<0.05)' is too broad.
[Fig. 2] The caption does not define the color scale, the bucket width ('10 minutes duration within an hour' appears in text but not the caption), or the units of the importance scores. Add axis legends and a colorbar.
[Availability] The paper says 'The source code will be published on GitHub' but provides no repository link. Add a link or state an anonymized availability policy for review.

Circularity Check

0 steps flagged

No significant circularity: the masking experiment is a model-faithfulness check, not a fitted prediction; no load-bearing self-citation.

full rationale

The central interpretability claim is that time steps ranked by AttentiveCAT carry more predictive value than random selection (Abstract; Section III-E). AttentiveCAT scores are defined in Eqs. (3)-(5) as a fixed function of the trained model's class-logit gradients and head-averaged attention weights; they are not fitted to the masking outcome. The masking protocol ranks time steps by these scores and compares retained F1 with random masking (Table III), which is a standard faithfulness test for attribution methods. Passing this test confirms that the scores reflect the model's own decision dependence, not that the identified time steps are causally relevant to health; the latter would require external ground truth. This is a limitation of external validity, not a circular derivation. The thresholds from [50]/[51] are an external calibration choice and a possible population-mismatch risk, but they are not derived from the target results. No self-citations of the present authors are load-bearing, and no uniqueness theorem or ansatz is smuggled in via self-citation. Score 1 reflects the mild self-referential nature of validating explanations with the same model that produces them, while the actual derivation chain is self-contained against the random control.

Axiom & Free-Parameter Ledger

3 free parameters · 4 axioms · 0 invented entities

The central claim rests on imported calibration thresholds for activity intensity, on the assumption that attention-weighted gradients faithfully quantify time-step contributions, and on the reliability of the health labels. No new physical entities are introduced; the main numerical choices are model hyperparameters selected by grid search and the visualization bucket width.

free parameters (3)

Transformer hyperparameters (layers, heads, model dimension, batch size, learning rate) = Layers {3,6,8}, heads {2,4,8}, dim {16,32,64,128}, batch {8,16,32}, lr {1e-3,5e-4,1e-4}; selected per dataset by 5-fold
Chosen on training data via cross-validation; affect predictive performance and therefore the time-step importance scores.
Activity intensity and bout thresholds = 3 s bouts for MVPA/LPA, 60 s for SED; intensity cut-points from [50]/[51]
Imported from prior calibration, not re-estimated for Pacific Island adolescents; directly determine which time steps are labeled MVPA, LPA, or SED.
Visualization bucket width = 10 minutes within an hour
Chosen by hand in Section III-D; affects the visualized V^{c}_{k,r,b} values but not the model itself.

axioms (4)

domain assumption Established GENEActiv intensity and bout thresholds accurately classify MVPA/LPA/SED in Pacific Island adolescents
Section III-B applies thresholds from [50]/[51] without recalibration; any misclassification propagates into the time-step features and AttentiveCAT scores.
ad hoc to paper Head-averaged attention weights preserve the pairwise relationships needed for AttentiveCAT scoring
Section III-C averages attention over heads 'to simplify calculation' and asserts relationships are preserved, with no sensitivity analysis.
domain assumption HWS and AHI are reliable ground-truth health indicators
Targets are derived from FITNESSGRAM standards and clinical AHI cut-offs; label noise and measurement error are not modeled.
standard math Gradient of the class logit with respect to final-layer hidden states equals the classification weight W^c_{o,t}
Used in Equations 3–5; holds for the chosen linear head on concatenated representations, but this design choice is not discussed.

pith-pipeline@v1.3.0-daily-deepseek · 15984 in / 17773 out tokens · 171283 ms · 2026-08-01T04:19:40.364748+00:00 · methodology

0 comments

read the original abstract

Wearable sensors continuously capture fine-grained multivariate time-series data, providing opportunities to model behavioural patterns associated with health outcomes. However, existing deep learning methods prioritise predictive accuracy over interpretability, limiting their application in health research. In this study, we present HealthCAT, a flexible framework that integrates an Encoder-only Transformer with an Attentive Class Activation Token (AttentiveCAT) to generate class-specific, time-step-level interpretations. These interpretations can be mapped back onto behavioural cycles that are relevant to the domain (e.g., time-of-day), supporting individual-level analysis of wearable sensor data. We evaluated HealthCAT using two real-world wearable sensor datasets (306 participants in total). HealthCAT outperformed deep learning baselines by up to 17\% in F1-score and 12\% in accuracy on both datasets ($p<0.05$). In masking experiments, the time steps identified by HealthCAT carried significantly more predictive value than random selection across all masking conditions ($p<0.05$), indicating that the identified time steps are predictively informative. By coupling predictive performance with validated time-step-level interpretability, HealthCAT moves wearable sensor analysis beyond aggregated metrics towards temporal patterns that support health monitoring, behavioural pattern analysis, and intervention design in health research. The significance of this work is that it enables accurate prediction of health indicators from wearable sensor data while providing insights into when and how physical activity patterns occur, rather than relying solely on aggregated summary measures.

Figures

Figures reproduced from arXiv: 2607.27635 by Haejin Lee, Joshua Y. Kim, Kalina Yacef, Xiaotong Yu.

**Figure 1.** Figure 1: For processing accelerometer data, when the focus is on capturing physical activity, we compute the gravity-subtracted Signal Vector Magnitude (SVMgs), as defined in Equation 1, and categorise intensity levels based on previously established thresholds [50]. The other option is to compute the Standard Signal Vector Magnitude (SVM) 2 when the gravity component is not required for movement analysis, such as… view at source ↗

discussion (0)

Reference graph

Works this paper leans on

55 extracted references · 5 linked inside Pith

[1]

A deep learning approach to on-node sensor data analytics for mobile or wearable devices,

D. Ravi, C. Wong, B. Lo, and G.-Z. Yang, “A deep learning approach to on-node sensor data analytics for mobile or wearable devices,”IEEE journal of biomedical and health informatics, vol. 21, no. 1, pp. 56–64, 2016

2016
[2]

Wearable accelerometer-derived physical activity and incident disease,

S. Khurshid, L.-C. Weng, V . Nauffal, J. P. Pirruccello, R. A. Venn, M. A. Al-Alusi, E. J. Benjamin, P. T. Ellinor, and S. A. Lubitz, “Wearable accelerometer-derived physical activity and incident disease,” NPJ Digital Medicine, vol. 5, no. 1, p. 131, 2022

2022
[3]

Using fitness trackers and smartwatches to measure physical activity in research: analysis of consumer wrist-worn wearables,

A. Henriksen, M. Haugen Mikalsen, A. Z. Woldaregay, M. Muzny, G. Hartvigsen, L. A. Hopstock, and S. Grimsgaard, “Using fitness trackers and smartwatches to measure physical activity in research: analysis of consumer wrist-worn wearables,”Journal of medical Internet research, vol. 20, no. 3, p. e110, 2018

2018
[4]

Braem, W

C. Braem, W. Pasman, T. van den Broek, M. Caspers, F. Jagers, U. Yavuz, H. Hermens, P. Veltink, and S. Wopereis, “The association of physical activity, heart rate and sleep from an activity tracker with weight loss during a 6-month personalized combined lifestyle intervention: a retrospective analysis,”BMC Digital Health, vol. 3, no. 1, p. 8, 2025

2025
[5]

Smart wearables for the detection of cardiovascular diseases: a sys- tematic literature review,

M. Moshawrab, M. Adda, A. Bouzouane, H. Ibrahim, and A. Raad, “Smart wearables for the detection of cardiovascular diseases: a sys- tematic literature review,”Sensors, vol. 23, no. 2, p. 828, 2023

2023
[6]

Overview of artificial intelligence–driven wearable devices for diabetes: scoping review,

A. Ahmed, S. Aziz, A. Abd-Alrazaq, F. Farooq, and J. Sheikh, “Overview of artificial intelligence–driven wearable devices for diabetes: scoping review,”Journal of medical Internet research, vol. 24, no. 8, p. e36010, 2022

2022
[7]

Detection of atrial fibrillation in a large population using wearable devices: the fitbit heart study,

S. A. Lubitz, A. Z. Faranesh, C. Selvaggi, S. J. Atlas, D. D. McManus, D. E. Singer, S. Pagoto, M. V . McConnell, A. Pantelopoulos, and A. S. Foulkes, “Detection of atrial fibrillation in a large population using wearable devices: the fitbit heart study,”Circulation, vol. 146, no. 19, pp. 1415–1424, 2022

2022
[8]

Effectiveness of wearable activity trackers to increase physical activity and improve health: a systematic review of systematic reviews and meta-analyses,

T. Ferguson, T. Olds, R. Curtis, H. Blake, A. J. Crozier, K. Dankiw, D. Dumuid, D. Kasai, E. O’Connor, R. Virgaraet al., “Effectiveness of wearable activity trackers to increase physical activity and improve health: a systematic review of systematic reviews and meta-analyses,” The Lancet Digital Health, vol. 4, no. 8, pp. e615–e626, 2022

2022
[9]

Gptcoach: Towards llm-based physical activity coaching,

M. J ¨orke, S. Sapkota, L. Warkenthien, N. Vainio, P. Schmiedmayer, E. Brunskill, and J. A. Landay, “Gptcoach: Towards llm-based physical activity coaching,” inProceedings of the 2025 CHI Conference on Human Factors in Computing Systems, 2025, pp. 1–46

2025
[10]

Be active: Physical activity,

World Health Organization, “Be active: Physical activity,” 2025, accessed: 2025-08-12. [Online]. Available: https://www.who.int/ initiatives/behealthy/physical-activity

2025
[11]

Xsleepnet: Multi-view sequential model for automatic sleep staging,

H. Phan, O. Y . Ch ´en, M. C. Tran, P. Koch, A. Mertins, and M. De V os, “Xsleepnet: Multi-view sequential model for automatic sleep staging,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 44, no. 9, pp. 5903–5915, 2021

2021
[12]

U-sleep: resilient high-frequency sleep staging,

M. Perslev, S. Darkner, L. Kempfner, M. Nikolic, P. J. Jennum, and C. Igel, “U-sleep: resilient high-frequency sleep staging,”NPJ digital medicine, vol. 4, no. 1, p. 72, 2021

2021
[13]

Lightweight transformer for federated human activity recognition in home care,

A. Raza, K. P. Tran, L. Koehl, X. Zeng, K. Benzaidi, S. Li, S. Hotham, and K. D. Tran, “Lightweight transformer for federated human activity recognition in home care,”Available at SSRN 5021631, 2021

2021
[14]

Capslstm-based human activity recognition for smart healthcare with scarce labeled data,

P. Khan, Y . Kumar, and S. Kumar, “Capslstm-based human activity recognition for smart healthcare with scarce labeled data,”IEEE Trans- actions on Computational Social Systems, vol. 11, no. 1, pp. 707–716, 2022

2022
[15]

Sdigru: Spatial and deep features integration using multilayer gated recurrent unit for human activity recognition,

T. Ahmad and J. Wu, “Sdigru: Spatial and deep features integration using multilayer gated recurrent unit for human activity recognition,” IEEE Transactions on Computational Social Systems, vol. 11, no. 1, pp. 973–985, 2024

2024
[16]

Wearable sensor signals: An overview of the ai models most commonly applied to time series data analysis,

L. Verde and G. Sannino, “Wearable sensor signals: An overview of the ai models most commonly applied to time series data analysis,” Connected e-Health: Integrated IoT and Cloud Computing, pp. 147– 163, 2022

2022
[17]

Spatial-temporal masked autoencoder for multi-device wearable human activity recognition,

S. Miao, L. Chen, and R. Hu, “Spatial-temporal masked autoencoder for multi-device wearable human activity recognition,”Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies, vol. 7, no. 4, pp. 1–25, 2024

2024
[18]

Autoaughar: automated data augmentation for sensor-based human activity recognition,

Y . Zhou, H. Zhao, Y . Huang, T. R ¨oddiger, M. Kurnaz, T. Riedel, and M. Beigl, “Autoaughar: automated data augmentation for sensor-based human activity recognition,”Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies, vol. 8, no. 2, pp. 1–27, 2024

2024
[19]

A transformer-based framework for multivariate time series representation learning,

G. Zerveas, S. Jayaraman, D. Patel, A. Bhamidipaty, and C. Eickhoff, “A transformer-based framework for multivariate time series representation learning,” inProceedings of the 27th ACM SIGKDD conference on knowledge discovery & data mining, 2021, pp. 2114–2124

2021
[20]

Attcat: Explaining transformers via attentive class activation tokens,

Y . Qiang, D. Pan, C. Li, X. Li, R. Jang, and D. Zhu, “Attcat: Explaining transformers via attentive class activation tokens,”Advances in neural information processing systems, vol. 35, pp. 5052–5064, 2022

2022
[21]

Assessing physical activ- ity/behavior of adolescents living in the pacific with accelerometer data: 231 geneactiv records in new caledonia,

G. Wattelez, S. Frayon, and O. Galy, “Assessing physical activ- ity/behavior of adolescents living in the pacific with accelerometer data: 231 geneactiv records in new caledonia,”Data in Brief, vol. 58, p. 111228, 2025

2025
[22]

Dreamt: Dataset for real- time sleep stage estimation using multisensor wearable technology,

K. Wang, J. Yang, A. Shetty, and J. Dunn, “Dreamt: Dataset for real- time sleep stage estimation using multisensor wearable technology,” PhysioNet https://doi. org/10.13026/62AN-CB28, 2024

doi:10.13026/62an-cb28 2024
[23]

Deep convlstm with self-attention for human activity decoding using wearable sensors,

S. P. Singh, M. K. Sharma, A. Lay-Ekuakille, D. Gangwar, and S. Gupta, “Deep convlstm with self-attention for human activity decoding using wearable sensors,”IEEE Sensors Journal, vol. 21, no. 6, pp. 8575–8582, 2020

2020
[24]

Gated transformer networks for multivariate time series classification,

M. Liu, S. Ren, S. Ma, J. Jiao, Y . Chen, Z. Wang, and W. Song, “Gated transformer networks for multivariate time series classification,”arXiv preprint arXiv:2103.14438, 2021

Pith/arXiv arXiv 2021
[25]

Trans- formers in time series: A survey,

Q. Wen, T. Zhou, C. Zhang, W. Chen, Z. Ma, J. Yan, and L. Sun, “Trans- formers in time series: A survey,”arXiv preprint arXiv:2202.07125, 2022

Pith/arXiv arXiv 2022
[26]

Shapeformer: Shapelet transformer for multivariate time series classification,

X.-M. Le, L. Luo, U. Aickelin, and M.-T. Tran, “Shapeformer: Shapelet transformer for multivariate time series classification,” inProceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, 2024, pp. 1484–1494

2024
[27]

Long short-term memory,

A. Graves, “Long short-term memory,”Supervised sequence labelling with recurrent neural networks, pp. 37–45, 2012

2012
[28]

Empirical evaluation of gated recurrent neural networks on sequence modeling,

J. Chung, C. Gulcehre, K. Cho, and Y . Bengio, “Empirical evaluation of gated recurrent neural networks on sequence modeling,”arXiv preprint arXiv:1412.3555, 2014. AUTHORet al.: TITLE 11

Pith/arXiv arXiv 2014
[29]

Attention is all you need,

A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. N. Gomez, Ł. Kaiser, and I. Polosukhin, “Attention is all you need,”Advances in neural information processing systems, vol. 30, 2017

2017
[30]

Deepsleepnet: A model for automatic sleep stage scoring based on raw single-channel eeg,

A. Supratak, H. Dong, C. Wu, and Y . Guo, “Deepsleepnet: A model for automatic sleep stage scoring based on raw single-channel eeg,”IEEE transactions on neural systems and rehabilitation engineering, vol. 25, no. 11, pp. 1998–2008, 2017

1998
[31]

Cardiologist-level arrhythmia detection with convolutional neural networks,

P. Rajpurkar, A. Y . Hannun, M. Haghpanahi, C. Bourn, and A. Y . Ng, “Cardiologist-level arrhythmia detection with convolutional neural networks,”arXiv preprint arXiv:1707.01836, 2017

Pith/arXiv arXiv 2017
[32]

A transformer based approach for activity detection,

G. Sharma, A. Dhall, and R. Subramanian, “A transformer based approach for activity detection,” inProceedings of the 30th ACM International Conference on Multimedia, 2022, pp. 7155–7159

2022
[33]

Semi-supervised learning for wearable-based momentary stress detection in the wild,

H. Yu and A. Sano, “Semi-supervised learning for wearable-based momentary stress detection in the wild,”Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies, vol. 7, no. 2, pp. 1–23, 2023

2023
[34]

Self-management of long-term conditions by integrat- ing artificial intelligence with wearable devices and internet-of-thing technology: A review,

G. Anand, A. M. Kalra, M. M. Baig, H. GholamHosseini, E. Ullah, and W. Chen, “Self-management of long-term conditions by integrat- ing artificial intelligence with wearable devices and internet-of-thing technology: A review,”Journal of Engineering and Science in Medical Diagnostics and Therapy, vol. 9, no. 1, p. 011106, 2026

2026
[35]

Sleepecg-net: explainable deep learning approach with ecg for pediatric sleep apnea diagnosis,

C. Garc ´ıa-Vicente, G. C. Guti´errez-Tobal, F. Vaquerizo-Villar, A. Mart´ın- Montero, D. Gozal, and R. Hornero, “Sleepecg-net: explainable deep learning approach with ecg for pediatric sleep apnea diagnosis,”IEEE Journal of Biomedical and Health Informatics, vol. 29, no. 2, pp. 1021– 1034, 2024

2024
[36]

Exploring the applications of explainability in wearable data analytics: Systematic literature review,

Y . Abdelaal, M. Aupetit, A. Baggag, and D. Al-Thani, “Exploring the applications of explainability in wearable data analytics: Systematic literature review,”Journal of Medical Internet Research, vol. 26, p. e53863, 2024

2024
[37]

Sleep-see-through: explainable deep learning for sleep event detection and quantification from wearable somnography,

M. Rossi, D. Sala, D. Bovio, C. Salito, G. Alessandrelli, C. Lom- bardi, L. Mainardi, and P. Cerveri, “Sleep-see-through: explainable deep learning for sleep event detection and quantification from wearable somnography,”IEEE journal of biomedical and health informatics, vol. 27, no. 7, pp. 3129–3140, 2023

2023
[38]

En- hancing the locality and breaking the memory bottleneck of transformer on time series forecasting,

S. Li, X. Jin, Y . Xuan, X. Zhou, W. Chen, Y .-X. Wang, and X. Yan, “En- hancing the locality and breaking the memory bottleneck of transformer on time series forecasting,”Advances in neural information processing systems, vol. 32, 2019

2019
[39]

Sleeptransformer: Automatic sleep staging with interpretability and un- certainty quantification,

H. Phan, K. Mikkelsen, O. Y . Ch´en, P. Koch, A. Mertins, and M. De V os, “Sleeptransformer: Automatic sleep staging with interpretability and un- certainty quantification,”IEEE Transactions on Biomedical Engineering, vol. 69, no. 8, pp. 2456–2467, 2022

2022
[40]

Grad-cam: Visual explanations from deep networks via gradient-based localization,

R. R. Selvaraju, M. Cogswell, A. Das, R. Vedantam, D. Parikh, and D. Batra, “Grad-cam: Visual explanations from deep networks via gradient-based localization,” inProceedings of the IEEE international conference on computer vision, 2017, pp. 618–626

2017
[41]

A unified approach to interpreting model predictions,

S. M. Lundberg and S.-I. Lee, “A unified approach to interpreting model predictions,”Advances in neural information processing systems, vol. 30, 2017

2017
[42]

” why should i trust you?

M. T. Ribeiro, S. Singh, and C. Guestrin, “” why should i trust you?” explaining the predictions of any classifier,” inProceedings of the 22nd ACM SIGKDD international conference on knowledge discovery and data mining, 2016, pp. 1135–1144

2016
[43]

Cardiologist-level interpretable knowledge-fused deep neural network for automatic arrhythmia diagnosis,

Y . Jin, Z. Li, M. Wang, J. Liu, Y . Tian, Y . Liu, X. Wei, L. Zhao, and C. Liu, “Cardiologist-level interpretable knowledge-fused deep neural network for automatic arrhythmia diagnosis,”Communications medicine, vol. 4, no. 1, p. 31, 2024

2024
[44]

Grad-sam: Explaining transformers via gradient self-attention maps,

O. Barkan, E. Hauon, A. Caciularu, O. Katz, I. Malkiel, O. Armstrong, and N. Koenigstein, “Grad-sam: Explaining transformers via gradient self-attention maps,” inProceedings of the 30th ACM International Conference on Information & Knowledge Management, 2021, pp. 2882– 2887

2021
[45]

C-lime: A consistency-oriented lime for time-series health-risk predictions,

T. Ito, K. Ochiai, and Y . Fukazawa, “C-lime: A consistency-oriented lime for time-series health-risk predictions,” inKnowledge Management and Acquisition for Intelligent Systems, H. Uehara, T. Yamaguchi, and Q. Bai, Eds. Cham: Springer International Publishing, 2021, pp. 58–69

2021
[46]

Geneactiv: Raw data wearable for clinical & public health research,

L. Dodge, “Geneactiv: Raw data wearable for clinical & public health research,” Apr. 2025. [Online]. Available: https://activinsights. com/digital-health-technologies/professional-wearables/geneactiv/

2025
[47]

T. C. Institute,FITNESSGRAM® Healthy Fitness Zone Performance Standards. California Department of Education, jan 2018. [Online]. Available: https://resources.finalsite. net/images/v1680962646/lbusdk12caus/agqlptdn6dksijblxbmb/ FITNESSGRAM-Healthy-Fitness-Zone-Performance-Standard.pdf

2018
[48]

Empatica e4 wristband for research,

E. Inc., “Empatica e4 wristband for research,” 2025. [Online]. Available: https://www.empatica.com/en-int/research/e4/

2025
[49]

A. A. of Sleep Medicineet al., “Sleep-related breathing disorders in adults: recommendations for syndrome definition and measurement techniques in clinical research. the report of an american academy of sleep medicine task force,”Sleep, vol. 22, pp. 667–689, 1999

1999
[50]

Calibration of the genea accelerometer for assessment of physical activity intensity in children,

L. R. S. Phillips, G. Parfitt, and A. V . Rowlands, “Calibration of the genea accelerometer for assessment of physical activity intensity in children,”Journal of Science and Medicine in Sport, vol. 16, no. 2, pp. 124–128, 2013

2013
[51]

Calibration of the genea accelerometer for assessment of physical activity intensity in children,

L. R. Phillips, G. Parfitt, and A. V . Rowlands, “Calibration of the genea accelerometer for assessment of physical activity intensity in children,” Journal of science and medicine in sport, vol. 16, no. 2, pp. 124–128, 2013

2013
[52]

On the variance of the adaptive learning rate and beyond,

L. Liu, H. Jiang, P. He, W. Chen, X. Liu, J. Gao, and J. Han, “On the variance of the adaptive learning rate and beyond,”arXiv preprint arXiv:1908.03265, 2019

Pith/arXiv arXiv 1908
[53]

Dose-response associations between accelerometry measured physical activity and sedentary time and all cause mortality: systematic review and harmonised meta-analysis,

U. Ekelund, J. Tarp, J. Steene-Johannessen, B. H. Hansen, B. Jefferis, M. W. Fagerland, P. Whincup, K. M. Diaz, S. P. Hooker, A. Chernofsky et al., “Dose-response associations between accelerometry measured physical activity and sedentary time and all cause mortality: systematic review and harmonised meta-analysis,”bmj, vol. 366, 2019

2019
[54]

Clustering of lifestyle risk factors for non-communicable diseases in 304,779 adolescents from 89 countries: A global perspective,

R. Uddin, E.-Y . Lee, S. R. Khan, M. S. Tremblay, and A. Khan, “Clustering of lifestyle risk factors for non-communicable diseases in 304,779 adolescents from 89 countries: A global perspective,”Preventive medicine, vol. 131, p. 105955, 2020

2020
[55]

Healthprism: a visual analytics system for exploring children’s physical and mental health profiles with multimodal data,

Z. Jiang, H. Chen, R. Zhou, J. Deng, X. Zhang, R. Zhao, C. Xie, Y . Wang, and E. C. Ngai, “Healthprism: a visual analytics system for exploring children’s physical and mental health profiles with multimodal data,”IEEE Transactions on Visualization and Computer Graphics, vol. 30, no. 1, pp. 1205–1215, 2023. 12 IEEE TRANSACTIONS AND JOURNALS TEMPLA TE APPEN...

2023

[1] [1]

A deep learning approach to on-node sensor data analytics for mobile or wearable devices,

D. Ravi, C. Wong, B. Lo, and G.-Z. Yang, “A deep learning approach to on-node sensor data analytics for mobile or wearable devices,”IEEE journal of biomedical and health informatics, vol. 21, no. 1, pp. 56–64, 2016

2016

[2] [2]

Wearable accelerometer-derived physical activity and incident disease,

S. Khurshid, L.-C. Weng, V . Nauffal, J. P. Pirruccello, R. A. Venn, M. A. Al-Alusi, E. J. Benjamin, P. T. Ellinor, and S. A. Lubitz, “Wearable accelerometer-derived physical activity and incident disease,” NPJ Digital Medicine, vol. 5, no. 1, p. 131, 2022

2022

[3] [3]

Using fitness trackers and smartwatches to measure physical activity in research: analysis of consumer wrist-worn wearables,

A. Henriksen, M. Haugen Mikalsen, A. Z. Woldaregay, M. Muzny, G. Hartvigsen, L. A. Hopstock, and S. Grimsgaard, “Using fitness trackers and smartwatches to measure physical activity in research: analysis of consumer wrist-worn wearables,”Journal of medical Internet research, vol. 20, no. 3, p. e110, 2018

2018

[4] [4]

Braem, W

C. Braem, W. Pasman, T. van den Broek, M. Caspers, F. Jagers, U. Yavuz, H. Hermens, P. Veltink, and S. Wopereis, “The association of physical activity, heart rate and sleep from an activity tracker with weight loss during a 6-month personalized combined lifestyle intervention: a retrospective analysis,”BMC Digital Health, vol. 3, no. 1, p. 8, 2025

2025

[5] [5]

Smart wearables for the detection of cardiovascular diseases: a sys- tematic literature review,

M. Moshawrab, M. Adda, A. Bouzouane, H. Ibrahim, and A. Raad, “Smart wearables for the detection of cardiovascular diseases: a sys- tematic literature review,”Sensors, vol. 23, no. 2, p. 828, 2023

2023

[6] [6]

Overview of artificial intelligence–driven wearable devices for diabetes: scoping review,

A. Ahmed, S. Aziz, A. Abd-Alrazaq, F. Farooq, and J. Sheikh, “Overview of artificial intelligence–driven wearable devices for diabetes: scoping review,”Journal of medical Internet research, vol. 24, no. 8, p. e36010, 2022

2022

[7] [7]

Detection of atrial fibrillation in a large population using wearable devices: the fitbit heart study,

S. A. Lubitz, A. Z. Faranesh, C. Selvaggi, S. J. Atlas, D. D. McManus, D. E. Singer, S. Pagoto, M. V . McConnell, A. Pantelopoulos, and A. S. Foulkes, “Detection of atrial fibrillation in a large population using wearable devices: the fitbit heart study,”Circulation, vol. 146, no. 19, pp. 1415–1424, 2022

2022

[8] [8]

Effectiveness of wearable activity trackers to increase physical activity and improve health: a systematic review of systematic reviews and meta-analyses,

T. Ferguson, T. Olds, R. Curtis, H. Blake, A. J. Crozier, K. Dankiw, D. Dumuid, D. Kasai, E. O’Connor, R. Virgaraet al., “Effectiveness of wearable activity trackers to increase physical activity and improve health: a systematic review of systematic reviews and meta-analyses,” The Lancet Digital Health, vol. 4, no. 8, pp. e615–e626, 2022

2022

[9] [9]

Gptcoach: Towards llm-based physical activity coaching,

M. J ¨orke, S. Sapkota, L. Warkenthien, N. Vainio, P. Schmiedmayer, E. Brunskill, and J. A. Landay, “Gptcoach: Towards llm-based physical activity coaching,” inProceedings of the 2025 CHI Conference on Human Factors in Computing Systems, 2025, pp. 1–46

2025

[10] [10]

Be active: Physical activity,

World Health Organization, “Be active: Physical activity,” 2025, accessed: 2025-08-12. [Online]. Available: https://www.who.int/ initiatives/behealthy/physical-activity

2025

[11] [11]

Xsleepnet: Multi-view sequential model for automatic sleep staging,

H. Phan, O. Y . Ch ´en, M. C. Tran, P. Koch, A. Mertins, and M. De V os, “Xsleepnet: Multi-view sequential model for automatic sleep staging,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 44, no. 9, pp. 5903–5915, 2021

2021

[12] [12]

U-sleep: resilient high-frequency sleep staging,

M. Perslev, S. Darkner, L. Kempfner, M. Nikolic, P. J. Jennum, and C. Igel, “U-sleep: resilient high-frequency sleep staging,”NPJ digital medicine, vol. 4, no. 1, p. 72, 2021

2021

[13] [13]

Lightweight transformer for federated human activity recognition in home care,

A. Raza, K. P. Tran, L. Koehl, X. Zeng, K. Benzaidi, S. Li, S. Hotham, and K. D. Tran, “Lightweight transformer for federated human activity recognition in home care,”Available at SSRN 5021631, 2021

2021

[14] [14]

Capslstm-based human activity recognition for smart healthcare with scarce labeled data,

P. Khan, Y . Kumar, and S. Kumar, “Capslstm-based human activity recognition for smart healthcare with scarce labeled data,”IEEE Trans- actions on Computational Social Systems, vol. 11, no. 1, pp. 707–716, 2022

2022

[15] [15]

Sdigru: Spatial and deep features integration using multilayer gated recurrent unit for human activity recognition,

T. Ahmad and J. Wu, “Sdigru: Spatial and deep features integration using multilayer gated recurrent unit for human activity recognition,” IEEE Transactions on Computational Social Systems, vol. 11, no. 1, pp. 973–985, 2024

2024

[16] [16]

Wearable sensor signals: An overview of the ai models most commonly applied to time series data analysis,

L. Verde and G. Sannino, “Wearable sensor signals: An overview of the ai models most commonly applied to time series data analysis,” Connected e-Health: Integrated IoT and Cloud Computing, pp. 147– 163, 2022

2022

[17] [17]

Spatial-temporal masked autoencoder for multi-device wearable human activity recognition,

S. Miao, L. Chen, and R. Hu, “Spatial-temporal masked autoencoder for multi-device wearable human activity recognition,”Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies, vol. 7, no. 4, pp. 1–25, 2024

2024

[18] [18]

Autoaughar: automated data augmentation for sensor-based human activity recognition,

Y . Zhou, H. Zhao, Y . Huang, T. R ¨oddiger, M. Kurnaz, T. Riedel, and M. Beigl, “Autoaughar: automated data augmentation for sensor-based human activity recognition,”Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies, vol. 8, no. 2, pp. 1–27, 2024

2024

[19] [19]

A transformer-based framework for multivariate time series representation learning,

G. Zerveas, S. Jayaraman, D. Patel, A. Bhamidipaty, and C. Eickhoff, “A transformer-based framework for multivariate time series representation learning,” inProceedings of the 27th ACM SIGKDD conference on knowledge discovery & data mining, 2021, pp. 2114–2124

2021

[20] [20]

Attcat: Explaining transformers via attentive class activation tokens,

Y . Qiang, D. Pan, C. Li, X. Li, R. Jang, and D. Zhu, “Attcat: Explaining transformers via attentive class activation tokens,”Advances in neural information processing systems, vol. 35, pp. 5052–5064, 2022

2022

[21] [21]

Assessing physical activ- ity/behavior of adolescents living in the pacific with accelerometer data: 231 geneactiv records in new caledonia,

G. Wattelez, S. Frayon, and O. Galy, “Assessing physical activ- ity/behavior of adolescents living in the pacific with accelerometer data: 231 geneactiv records in new caledonia,”Data in Brief, vol. 58, p. 111228, 2025

2025

[22] [22]

Dreamt: Dataset for real- time sleep stage estimation using multisensor wearable technology,

K. Wang, J. Yang, A. Shetty, and J. Dunn, “Dreamt: Dataset for real- time sleep stage estimation using multisensor wearable technology,” PhysioNet https://doi. org/10.13026/62AN-CB28, 2024

doi:10.13026/62an-cb28 2024

[23] [23]

Deep convlstm with self-attention for human activity decoding using wearable sensors,

S. P. Singh, M. K. Sharma, A. Lay-Ekuakille, D. Gangwar, and S. Gupta, “Deep convlstm with self-attention for human activity decoding using wearable sensors,”IEEE Sensors Journal, vol. 21, no. 6, pp. 8575–8582, 2020

2020

[24] [24]

Gated transformer networks for multivariate time series classification,

M. Liu, S. Ren, S. Ma, J. Jiao, Y . Chen, Z. Wang, and W. Song, “Gated transformer networks for multivariate time series classification,”arXiv preprint arXiv:2103.14438, 2021

Pith/arXiv arXiv 2021

[25] [25]

Trans- formers in time series: A survey,

Q. Wen, T. Zhou, C. Zhang, W. Chen, Z. Ma, J. Yan, and L. Sun, “Trans- formers in time series: A survey,”arXiv preprint arXiv:2202.07125, 2022

Pith/arXiv arXiv 2022

[26] [26]

Shapeformer: Shapelet transformer for multivariate time series classification,

X.-M. Le, L. Luo, U. Aickelin, and M.-T. Tran, “Shapeformer: Shapelet transformer for multivariate time series classification,” inProceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, 2024, pp. 1484–1494

2024

[27] [27]

Long short-term memory,

A. Graves, “Long short-term memory,”Supervised sequence labelling with recurrent neural networks, pp. 37–45, 2012

2012

[28] [28]

Empirical evaluation of gated recurrent neural networks on sequence modeling,

J. Chung, C. Gulcehre, K. Cho, and Y . Bengio, “Empirical evaluation of gated recurrent neural networks on sequence modeling,”arXiv preprint arXiv:1412.3555, 2014. AUTHORet al.: TITLE 11

Pith/arXiv arXiv 2014

[29] [29]

Attention is all you need,

A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. N. Gomez, Ł. Kaiser, and I. Polosukhin, “Attention is all you need,”Advances in neural information processing systems, vol. 30, 2017

2017

[30] [30]

Deepsleepnet: A model for automatic sleep stage scoring based on raw single-channel eeg,

A. Supratak, H. Dong, C. Wu, and Y . Guo, “Deepsleepnet: A model for automatic sleep stage scoring based on raw single-channel eeg,”IEEE transactions on neural systems and rehabilitation engineering, vol. 25, no. 11, pp. 1998–2008, 2017

1998

[31] [31]

Cardiologist-level arrhythmia detection with convolutional neural networks,

P. Rajpurkar, A. Y . Hannun, M. Haghpanahi, C. Bourn, and A. Y . Ng, “Cardiologist-level arrhythmia detection with convolutional neural networks,”arXiv preprint arXiv:1707.01836, 2017

Pith/arXiv arXiv 2017

[32] [32]

A transformer based approach for activity detection,

G. Sharma, A. Dhall, and R. Subramanian, “A transformer based approach for activity detection,” inProceedings of the 30th ACM International Conference on Multimedia, 2022, pp. 7155–7159

2022

[33] [33]

Semi-supervised learning for wearable-based momentary stress detection in the wild,

H. Yu and A. Sano, “Semi-supervised learning for wearable-based momentary stress detection in the wild,”Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies, vol. 7, no. 2, pp. 1–23, 2023

2023

[34] [34]

Self-management of long-term conditions by integrat- ing artificial intelligence with wearable devices and internet-of-thing technology: A review,

G. Anand, A. M. Kalra, M. M. Baig, H. GholamHosseini, E. Ullah, and W. Chen, “Self-management of long-term conditions by integrat- ing artificial intelligence with wearable devices and internet-of-thing technology: A review,”Journal of Engineering and Science in Medical Diagnostics and Therapy, vol. 9, no. 1, p. 011106, 2026

2026

[35] [35]

Sleepecg-net: explainable deep learning approach with ecg for pediatric sleep apnea diagnosis,

C. Garc ´ıa-Vicente, G. C. Guti´errez-Tobal, F. Vaquerizo-Villar, A. Mart´ın- Montero, D. Gozal, and R. Hornero, “Sleepecg-net: explainable deep learning approach with ecg for pediatric sleep apnea diagnosis,”IEEE Journal of Biomedical and Health Informatics, vol. 29, no. 2, pp. 1021– 1034, 2024

2024

[36] [36]

Exploring the applications of explainability in wearable data analytics: Systematic literature review,

Y . Abdelaal, M. Aupetit, A. Baggag, and D. Al-Thani, “Exploring the applications of explainability in wearable data analytics: Systematic literature review,”Journal of Medical Internet Research, vol. 26, p. e53863, 2024

2024

[37] [37]

Sleep-see-through: explainable deep learning for sleep event detection and quantification from wearable somnography,

M. Rossi, D. Sala, D. Bovio, C. Salito, G. Alessandrelli, C. Lom- bardi, L. Mainardi, and P. Cerveri, “Sleep-see-through: explainable deep learning for sleep event detection and quantification from wearable somnography,”IEEE journal of biomedical and health informatics, vol. 27, no. 7, pp. 3129–3140, 2023

2023

[38] [38]

En- hancing the locality and breaking the memory bottleneck of transformer on time series forecasting,

S. Li, X. Jin, Y . Xuan, X. Zhou, W. Chen, Y .-X. Wang, and X. Yan, “En- hancing the locality and breaking the memory bottleneck of transformer on time series forecasting,”Advances in neural information processing systems, vol. 32, 2019

2019

[39] [39]

Sleeptransformer: Automatic sleep staging with interpretability and un- certainty quantification,

H. Phan, K. Mikkelsen, O. Y . Ch´en, P. Koch, A. Mertins, and M. De V os, “Sleeptransformer: Automatic sleep staging with interpretability and un- certainty quantification,”IEEE Transactions on Biomedical Engineering, vol. 69, no. 8, pp. 2456–2467, 2022

2022

[40] [40]

Grad-cam: Visual explanations from deep networks via gradient-based localization,

R. R. Selvaraju, M. Cogswell, A. Das, R. Vedantam, D. Parikh, and D. Batra, “Grad-cam: Visual explanations from deep networks via gradient-based localization,” inProceedings of the IEEE international conference on computer vision, 2017, pp. 618–626

2017

[41] [41]

A unified approach to interpreting model predictions,

S. M. Lundberg and S.-I. Lee, “A unified approach to interpreting model predictions,”Advances in neural information processing systems, vol. 30, 2017

2017

[42] [42]

” why should i trust you?

M. T. Ribeiro, S. Singh, and C. Guestrin, “” why should i trust you?” explaining the predictions of any classifier,” inProceedings of the 22nd ACM SIGKDD international conference on knowledge discovery and data mining, 2016, pp. 1135–1144

2016

[43] [43]

Cardiologist-level interpretable knowledge-fused deep neural network for automatic arrhythmia diagnosis,

Y . Jin, Z. Li, M. Wang, J. Liu, Y . Tian, Y . Liu, X. Wei, L. Zhao, and C. Liu, “Cardiologist-level interpretable knowledge-fused deep neural network for automatic arrhythmia diagnosis,”Communications medicine, vol. 4, no. 1, p. 31, 2024

2024

[44] [44]

Grad-sam: Explaining transformers via gradient self-attention maps,

O. Barkan, E. Hauon, A. Caciularu, O. Katz, I. Malkiel, O. Armstrong, and N. Koenigstein, “Grad-sam: Explaining transformers via gradient self-attention maps,” inProceedings of the 30th ACM International Conference on Information & Knowledge Management, 2021, pp. 2882– 2887

2021

[45] [45]

C-lime: A consistency-oriented lime for time-series health-risk predictions,

T. Ito, K. Ochiai, and Y . Fukazawa, “C-lime: A consistency-oriented lime for time-series health-risk predictions,” inKnowledge Management and Acquisition for Intelligent Systems, H. Uehara, T. Yamaguchi, and Q. Bai, Eds. Cham: Springer International Publishing, 2021, pp. 58–69

2021

[46] [46]

Geneactiv: Raw data wearable for clinical & public health research,

L. Dodge, “Geneactiv: Raw data wearable for clinical & public health research,” Apr. 2025. [Online]. Available: https://activinsights. com/digital-health-technologies/professional-wearables/geneactiv/

2025

[47] [47]

T. C. Institute,FITNESSGRAM® Healthy Fitness Zone Performance Standards. California Department of Education, jan 2018. [Online]. Available: https://resources.finalsite. net/images/v1680962646/lbusdk12caus/agqlptdn6dksijblxbmb/ FITNESSGRAM-Healthy-Fitness-Zone-Performance-Standard.pdf

2018

[48] [48]

Empatica e4 wristband for research,

E. Inc., “Empatica e4 wristband for research,” 2025. [Online]. Available: https://www.empatica.com/en-int/research/e4/

2025

[49] [49]

A. A. of Sleep Medicineet al., “Sleep-related breathing disorders in adults: recommendations for syndrome definition and measurement techniques in clinical research. the report of an american academy of sleep medicine task force,”Sleep, vol. 22, pp. 667–689, 1999

1999

[50] [50]

Calibration of the genea accelerometer for assessment of physical activity intensity in children,

L. R. S. Phillips, G. Parfitt, and A. V . Rowlands, “Calibration of the genea accelerometer for assessment of physical activity intensity in children,”Journal of Science and Medicine in Sport, vol. 16, no. 2, pp. 124–128, 2013

2013

[51] [51]

Calibration of the genea accelerometer for assessment of physical activity intensity in children,

L. R. Phillips, G. Parfitt, and A. V . Rowlands, “Calibration of the genea accelerometer for assessment of physical activity intensity in children,” Journal of science and medicine in sport, vol. 16, no. 2, pp. 124–128, 2013

2013

[52] [52]

On the variance of the adaptive learning rate and beyond,

L. Liu, H. Jiang, P. He, W. Chen, X. Liu, J. Gao, and J. Han, “On the variance of the adaptive learning rate and beyond,”arXiv preprint arXiv:1908.03265, 2019

Pith/arXiv arXiv 1908

[53] [53]

Dose-response associations between accelerometry measured physical activity and sedentary time and all cause mortality: systematic review and harmonised meta-analysis,

U. Ekelund, J. Tarp, J. Steene-Johannessen, B. H. Hansen, B. Jefferis, M. W. Fagerland, P. Whincup, K. M. Diaz, S. P. Hooker, A. Chernofsky et al., “Dose-response associations between accelerometry measured physical activity and sedentary time and all cause mortality: systematic review and harmonised meta-analysis,”bmj, vol. 366, 2019

2019

[54] [54]

Clustering of lifestyle risk factors for non-communicable diseases in 304,779 adolescents from 89 countries: A global perspective,

R. Uddin, E.-Y . Lee, S. R. Khan, M. S. Tremblay, and A. Khan, “Clustering of lifestyle risk factors for non-communicable diseases in 304,779 adolescents from 89 countries: A global perspective,”Preventive medicine, vol. 131, p. 105955, 2020

2020

[55] [55]

Healthprism: a visual analytics system for exploring children’s physical and mental health profiles with multimodal data,

Z. Jiang, H. Chen, R. Zhou, J. Deng, X. Zhang, R. Zhao, C. Xie, Y . Wang, and E. C. Ngai, “Healthprism: a visual analytics system for exploring children’s physical and mental health profiles with multimodal data,”IEEE Transactions on Visualization and Computer Graphics, vol. 30, no. 1, pp. 1205–1215, 2023. 12 IEEE TRANSACTIONS AND JOURNALS TEMPLA TE APPEN...

2023