AI on the Pulse: Real-Time Health Anomaly Detection with Wearable and Ambient Intelligence

Bardh Prenkaj; Davide Gabrielli; Paola Velardi; Stefano Faralli

REVIEW 5 major objections 6 minor 1 cited by

AI on the Pulse: Real-Time Health Anomaly Detection with Wearable and Ambient Intelligence

T0 review · 5 major / 6 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read A universal time-series model, fine-tuned to treat anomaly detection as masked imputation, flags personalized health anomalies from smartwatch and ambient data, beating 12 rival detectors by about 22% F1.

desk verdict A credible applied adaptation of UniTS with a real deployment, but the headline F1 gain is provisional given the asymmetric benchmark and missing ablation. read the letter →

arxiv 2508.03436 v1 pith:Q26VS5Y7 submitted 2025-08-05 cs.LG

Davide Gabrielli , Bardh Prenkaj , Paola Velardi , Stefano Faralli This is my paper

classification cs.LG

keywords AI-drivenhealthcareanomalydetectionwearablesensorsambientintelligenceexplainableAItime-seriesanalysisUniTSstress

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper argues that continuous home health monitoring does not require clinical-grade sensors or labeled classifications. It presents a system that learns each patient's own baseline from smartwatch and ambient-sensor data and flags deviations as anomalies in real time. On four stress-detection benchmarks, the fine-tuned universal time-series model outperforms twelve state-of-the-art anomaly detectors by about 22% average F1. In a three-month home deployment with six elderly patients, a senior geriatrician confirmed 30 of the 32 system-flagged anomalies as genuine, with the other two traced to sensor errors. If these numbers hold, consumer wearables plus ambient sensors could support personalized early-warning alerts for home-care teams.

What carries the argument

The central machinery is UniTS, a universal multivariate time-series transformer, repurposed for contextual anomaly detection. The paper reformulates anomaly detection as masked imputation: a sliding window is split into past tokens (visible) and future tokens (masked), contextual tokens from auxiliary signals are concatenated along time and feature axes, and prompt tokens are prepended to tell the model which anomaly type to expect. A dual multi-head self-attention over time and variable dimensions, a dynamic feed-forward network, and a Tower module reconstruct the masked future segment; the reconstruction error on the masked target features, thresholded with a peak-over-threshold method, is the anomaly score. This lets the model learn dependencies such as 'elevated heart rate is normal if the patient just walked,' which is what makes the system personalize rather than apply population thresholds.

What would settle it

Exhaustively label every minute of a held-out week from the @HOME patients (or a comparable home dataset) as normal or anomalous by a clinician, then run the system and compute recall as well as precision; if many unflagged events are clinically meaningful while flagged ones stay accurate, the paper's implied detection capability fails even though its alarm precision holds.

Watch

Extended reading notes

Core claim

On its own terms, the paper's central claim is that anomaly detection in patient monitoring can be converted from a classification problem into an imputation problem and solved by a fine-tuned universal time-series transformer. The authors take UniTS, a foundation model for multivariate time series, and extend it so that the future values of target signals (heart rate, heart-rate variability, and in the home deployment respiration, blood pressure, oxygen saturation, steps, and environmental readings) are masked while past values and contextual tokens are preserved; prompt tokens encode the type of anomaly being looked for. Reconstruction error on the masked targets becomes the anomaly score. The same recipe works on high-fidelity ECG and on noisy consumer wearable optical pulse signals, and the model adapts to each patient's personal baseline, so a stable but abnormal-looking blood pressure for that patient is not flagged. The deployment evidence is the paper's clinching claim: in @HOME, 30 of 32 alarms were confirmed by a senior geriatrician as clinically meaningful true positives, with the remaining two attributed to sensor issues.

Load-bearing premise

The load-bearing premise is that a reading below the 3rd or above the 97th percentile of a patient's own signal is a genuine health anomaly, and that checking only the alarms the system raised, without checking the periods it stayed silent on, measures detection ability.

Editorial extensions

If this is right

A single foundation model can serve as the core of a multi-signal home monitoring system, rather than one bespoke detector per vital sign.
Consumer smartwatch data can substitute for clinical ECG in stress-anomaly detection, since UniTS reaches comparable F1 on WESAD BVP (0.800) and ECG (0.793).
Anomaly alerts can be delivered with machine-generated clinical explanations, which the evaluating geriatrician rated significant (at least 3.0 on a 0-5 scale) for every anomaly type seen.
Real-time personalized monitoring can run on lightweight, non-invasive hardware: the @HOME pilot used consumer smartwatches with five-day battery life sampling every minute, plus ambient room sensors.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

Extrapolating beyond the paper: the benchmark's ground truth (below the 3rd or above the 97th percentile) collapses the target to 'statistical extreme,' so a re-run with clinician-labeled event timestamps would show whether the 22% F1 gain survives when the target is clinically defined.
The same imputation-error mechanism should transfer to other chronic conditions and to single-patient longitudinal monitoring, because it requires only a personal baseline, not labeled patient cohorts.
The @HOME precision of 93.75% measured no false negatives; a natural next deployment would have the clinician periodically review unflagged windows to estimate recall, which the paper explicitly notes it did not do.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Desk Editor's Note

read the letter

Colleague,

The takeaway: this is a real applied paper with a deployed system, but the headline 22% F1 edge over a dozen baselines should be treated as provisional. The core novelty is modest — a contextual-masking and prompt-token adaptation of the existing UniTS foundation model, applied to wearable/ambient health monitoring — and the evaluation has two load-bearing gaps.

What it does well: the system is not a toy. It runs on consumer smartwatches and environmental sensors, has been in a home study with six neurodegenerative patients for three months, and a senior geriatrician confirmed 30 of 32 flagged anomalies as true positives with plausible clinical significance. The authors are transparent that they only reviewed flagged positives and did not search for false negatives; that is an honest limitation statement. The benchmark covers four datasets and 12 baselines, which is broad, and the code is public.

The soft spots are in the benchmark. First, the ground-truth anomalies are defined as points below the 3rd or above the 97th percentile — a distributional-extreme rule, not a clinical event label. The model is trained to reconstruct future masked values; high reconstruction error will naturally align with extremes, so part of the F1 gain may simply reflect the model learning the label-generation rule. Second, the baselines were run with default hyperparameters, while UniTS was fine-tuned with a specific window, embedding, and learning rate. For percentile-based labels, threshold choice and hyperparameter tuning can move F1 substantially, so the 22% gap is not a like-for-like comparison. Third, there is no ablation of stock UniTS versus the proposed masked/prompt version. The Table 1 row labeled 'UniTS [8]' is the authors' fine-tuned model, so we cannot tell how much of the gain comes from the adaptation and how much from fine-tuning a strong foundation model. On WESAD, the improvement over the second-best is only 2.8–4%, which is consistent with the idea that the gains are concentrated where the protocol is most favorable.

The @HOME result is a useful precision story — the system does not spam alerts and those it issues are clinically meaningful — but it is not a recall story. With 32 alerts and one expert, it cannot support the broad claim of reliable detection.

Who should read this: anyone building on UniTS for anomaly detection, or working on consumer-grade wearable monitoring. It is not a new paradigm, but it is a plausible engineering path. Worth a serious referee, but the authors should be asked for an ablation, a symmetric baseline protocol with tuned thresholds, and an explicit treatment of recall in the deployment.

My recommendation: send it to peer review with major-revision expectations, not a desk reject.

Referee Report

5 major / 6 minor

Summary. The paper proposes 'AI on the Pulse', a real-time health anomaly detection system built by fine-tuning UniTS with contextual masking and prompt tokens, and evaluates it on HR/HRV stress data from DREAMER, HCI, and WESAD (ECG and BVP). The authors report that their method surpasses 12 established anomaly detection baselines by roughly 22% average F1 across these four settings. They also describe a deployment, '@HOME', in which six elderly patients wore consumer smartwatches and ambient sensors for three months; 32 system alerts were reviewed by a senior geriatrician, who confirmed 30 as true positives and attributed the remaining 2 to sensor errors. The system further generates LLM-based explanations for clinicians. The central claim is that consumer-grade wearable and ambient devices, combined with a fine-tuned foundation model, provide clinically meaningful personalized anomaly detection without clinical-grade equipment.

Significance. If the reported results are robust, the work has practical significance: it demonstrates a real deployment of a time-series foundation model for continuous health monitoring with consumer hardware, and it addresses interpretability through LLM-generated explanations. The authors provide code, use multiple public datasets, and include an actual pilot deployment, which are strengths. The headline 22% F1 improvement, however, rests on the percentile-based anomaly definition in Section 4.3, on default hyperparameters for all baselines, and on an evaluation protocol that does not measure false negatives in the deployment. The contribution of the proposed UniTS adaptation over stock UniTS is not isolated by an ablation. The deployment result is encouraging but statistically thin, with 32 alerts, a single assessor, and no recall estimate. The paper would be a solid demonstration of an outlier-detection system, but the stronger clinical-anomaly claim needs additional evidence.

major comments (5)

[Section 4.3] The ground-truth definition is not clinically validated: anomalies are defined as points below the 3rd or above the 97th percentile of the HR/HRV series. Since the proposed model is trained to reconstruct normal patterns and to flag high reconstruction error, its score is constructed to align with distributional extremes. Reporting F1 against this label may therefore measure agreement with the percentile rule rather than detection of clinically meaningful stress events. The authors should either use known stress labels (e.g., WESAD's stress conditions) or demonstrate that percentile-based extremes correspond to clinically annotated events.
[Section 4.3, Table 1] The benchmark is asymmetric: the proposed model is fine-tuned and its anomaly threshold is set through the reconstruction-error distribution, while all 12 baselines are run 'using their default hyperparameters' with no mention of threshold calibration. F1 in anomaly detection is highly sensitive to the threshold, so this protocol can substantially depress baseline scores. The authors should tune each baseline's threshold on a validation split (or report F1 across a threshold sweep) and, ideally, tune baseline hyperparameters, before claiming a 22% average improvement.
[Table 1 and Section 3.2] There is no ablation of the proposed adaptation against stock UniTS. Table 1 labels the row 'UniTS', but the text describes a fine-tuned UniTS with contextual masking and prompt tokens. Without a comparison to the original UniTS using the same fine-tuning, evaluation protocol, and percentile labels, the reported gain cannot be attributed to the paper's contextual masking and prompt-token contributions rather than to fine-tuning a strong foundation model.
[Section 4.4 and Table 2] The @HOME evaluation does not support the claim of detection capability. Only system-flagged positives are reviewed by the geriatrician, so false negatives are never counted; the 93.75% positive confirmation rate therefore says nothing about recall or missed health events. The authors should report the total number of alerts, the number of true/false positives, an estimate of missed events (e.g., through retrospective chart review or by running the system on labeled historical data), and a longer monitoring period with more than six patients.
[Section 4.4, Table 2] The clinical validation statistics are weak: 32 alerts over three months for six patients, with two alerts classified as sensor errors, and a single senior geriatrician as the sole assessor. The reported significance and criticality scores are averages with small sample sizes and no inter-rater reliability or uncertainty measure. This is a pilot feasibility observation, not a validation of clinical utility, and the wording should be softened accordingly.

minor comments (6)

[Section 4.1] The text contains a typo: 'wether' should be 'whether'.
[Figure 1] The figure header reads 'Hear Rate Variability'; it should be 'Heart Rate Variability'.
[References] References [7] and [6] point to the same CVPR paper but are formatted inconsistently; the same holds for [15]/[14] and [23]/[22].
[Section 4.2] The sentence 'As an example, Figure 2 shows data from two patients... the pre-pilot patient's systolic/diastolic blood pressure is consistently flagged as slightly or highly abnormal' seems to describe severity shading produced by a previous system; it is unclear whether this shading comes from the proposed model or from an external rule-based system, and the caption should be clarified.
[Section 4.3] The evaluation uses 5-fold CV on an 80:20 train/test split, but the text does not specify how the 3rd/97th percentile thresholds are computed within each fold or whether the test-set percentile is used, which can cause label leakage. Please state whether percentile thresholds are derived from training data only.
[Section 4.3] The phrase 'For WESAD, the only detectable anomaly was stress since no contextual data was provided' is unclear: if stress labels are available, why are anomalies defined by percentiles instead of the stress condition? This should be explained.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the central benchmark is a held-out evaluation against independently defined quantile labels, and the author self-citations are not load-bearing.

full rationale

The central claim (Table 1) is a standard benchmark: ground-truth anomalies are defined independently of the model as points below the 3rd or above the 97th percentile (Section 4.3), and UniTS is fine-tuned on an 80:20 split with a reconstruction loss (Sections 3.2 and 4.3); F1 is then computed on held-out windows. No equation equates the model's anomaly score to the quantile rule, and no threshold is fitted to the test labels, so the ~22% improvement is an empirical, falsifiable result rather than a tautology. The two author self-citations ([6] and [18]) appear only in related-work and problem-definition context and carry no load-bearing uniqueness or ansatz argument. The @HOME validation reports precision on system-flagged events only, as the paper states 'Only true and false positives were considered, as false negatives require exhaustive review'; this limits recall claims but is an evaluation-scope limitation, not a circular derivation. Similarly, running baselines with default hyperparameters is a protocol-fairness concern, not a circularity. No derivation step reduces to its own input by construction.

Assumptions & free parameters 6 free parameters · 5 assumptions · 0 invented entities

The system's central claims rest on several unvalidated domain assumptions and hand-chosen hyperparameters, balanced by the use of an existing foundation model and public datasets. No new physical entities are postulated.

free parameters (6)

Anomaly label percentile = 3rd/97th percentile
Used to define ground-truth anomalies in Section 4.3; this distributional cutoff is chosen by the authors and not clinically validated.
UniTS benchmark window size = 5 samples
Fine-tuning window for UniTS on stress datasets (Section 4.3).
@HOME window size = 16 samples (8 past, 8 masked future)
Contextual imputation window for the deployment experiments (Section 4.3).
Learning rate = 5 x 10^-4
Fine-tuning hyperparameter for UniTS (Section 4.3).
Training epochs = 20 for benchmark, 2 for @HOME
Chosen by hand; no sensitivity analysis reported.
Embedding dimension = 128
Model capacity choice for UniTS fine-tuning (Section 4.3).

assumptions (5)

domain assumption Reconstruction error on masked future values is a valid anomaly score
The method uses imputation error as the anomaly signal; no theoretical or clinical justification is given for this mapping.
ad hoc to paper Points below the 3rd or above the 97th percentile represent meaningful anomalies
This definition generates the benchmark labels in Section 4.3 and is not tied to verified clinical events.
domain assumption UniTS pre-trained representations transfer to wearable health data at reduced sampling rates
The system relies on a foundation model trained on different time-series tasks; transfer is assumed rather than demonstrated beyond the reported benchmarks.
domain assumption One-minute consumer wearable sampling preserves clinically meaningful physiological deviations
The @HOME dataset uses minute-level sampling; the paper does not show this resolution is sufficient for detecting the targeted health anomalies.
domain assumption LLM-generated explanations are clinically meaningful and appropriate
No evaluation of explanation quality beyond a single expert's subjective ratings is provided.

how reviews work

0 comments

Cite this review

Pith. "Pith review of AI on the Pulse: Real-Time Health Anomaly Detection with Wearable and Ambient Intelligence." pith.science (2026). https://pith.science/paper/Q26VS5Y7

@misc{pith2026250803436,
  author       = {Pith},
  title        = {Pith review of: AI on the Pulse: Real-Time Health Anomaly Detection with Wearable and Ambient Intelligence},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/Q26VS5Y7}},
  note         = {Machine review of arXiv:2508.03436}
}

read the original abstract

We introduce AI on the Pulse, a real-world-ready anomaly detection system that continuously monitors patients using a fusion of wearable sensors, ambient intelligence, and advanced AI models. Powered by UniTS, a state-of-the-art (SoTA) universal time-series model, our framework autonomously learns each patient's unique physiological and behavioral patterns, detecting subtle deviations that signal potential health risks. Unlike classification methods that require impractical, continuous labeling in real-world scenarios, our approach uses anomaly detection to provide real-time, personalized alerts for reactive home-care interventions. Our approach outperforms 12 SoTA anomaly detection methods, demonstrating robustness across both high-fidelity medical devices (ECG) and consumer wearables, with a ~ 22% improvement in F1 score. However, the true impact of AI on the Pulse lies in @HOME, where it has been successfully deployed for continuous, real-world patient monitoring. By operating with non-invasive, lightweight devices like smartwatches, our system proves that high-quality health monitoring is possible without clinical-grade equipment. Beyond detection, we enhance interpretability by integrating LLMs, translating anomaly scores into clinically meaningful insights for healthcare professionals.

Figures

Figures reproduced from arXiv: 2508.03436 by the authors.

**Figure 2.** @HOME data samples: pre-pilot (top) and pilot [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗

Discussion (0). Sign in to comment.

Forward citations

Cited by 1 Pith paper

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

HealthSLM-Bench: Benchmarking Small Language Models for Mobile and Wearable Healthcare Monitoring
cs.AI 2025-09 conditional novelty 5.0 of 10

Small language models rival large ones on several wearable health prediction tasks, with large efficiency gains, but suffer from class imbalance and poor calorie regression.

Reference graph

Works this paper leans on

32 extracted references · 31 canonical work pages · cited by 1 Pith paper

[8]

Shanghua Gao, Teddy Koker, Owen Queen, Tom Hartvigsen, Theodoros Tsiligkaridis, and Marinka Zitnik. 2025. UniTS: A unified multi-task time series model. Advances in Neural Information Processing Systems 37 (2025), 140589– 140631

work page 2025
[1]

Menatalla Abououf, Shakti Singh, Rabeb Mizouni, and Hadi Otrok. 2023. Ex- plainable AI for event and anomaly detection and classification in healthcare monitoring systems. IEEE Internet of Things Journal 11, 2 (2023), 3446–3457

work page 2023
[2]

Julien Audibert, Pietro Michiardi, Frédéric Guyard, Sébastien Marti, and Maria A Zuluaga. 2020. USAD: Unsupervised anomaly detection on multivariate time series. In ACM SIGKDD. 3395–3404

work page 2020
[3]

Gerald Bieber, Thomas Kirste, and Michael Gaede. 2014. Low sampling rate for physical activity recognition. In Proc. of Int. Conf. on PErvasive Technologies Related to Assistive Environments (Rhodes, Greece) (PETRA ’14). Association for Computing Machinery, Article 15, 8 pages

work page 2014
[4]

Nivedita Bijlani, Ramin Nilforooshan, and Samaneh Kouchaki. 2022. An unsu- pervised data-driven anomaly detection approach for adverse health conditions in people living with dementia: Cohort study. JMIR aging 5, 3 (2022), e38211

work page 2022
[5]

Ailin Deng and Bryan Hooi. 2021. Graph neural network-based anomaly detection in multivariate time series. In AAAI, Vol. 35. 4027–4035

work page 2021
[6]

Alessandro Flaborea, Bardh Prenkaj, Bharti Munjal, Marco Aurelio Sterpa, Dario Aragona, Luca Podo, and Fabio Galasso. 2023. Are we certain it’s anomalous?. In CVPR. 2896–2906

work page 2023
[7]

Alessandro Flaborea, Bardh Prenkaj, Bharti Munjal, Marco Aurelio Sterpa, Dario Aragona, Luca Podo, and Fabio Galasso. 2023. Are we certain it’s anomalous?. In Proc. of IEEE/CVF Conf. on Computer Vision and Pattern Recognition . 2897–2907

work page 2023

Show all 32 references

[9]

Alexander Geiger, Dongyu Liu, Sarah Alnegheimish, Alfredo Cuesta-Infante, and Kalyan Veeramachaneni. 2020. Tadgan: Time series anomaly detection using generative adversarial networks. In IEEE Int. Conf. on Big Data . IEEE, 33–43

2020
[10]

Heilman and Stephen E

Kenneth M. Heilman and Stephen E. Nadeau. 2022. Emotional and Neuropsy- chiatric Disorders Associated with Alzheimer’s Disease. Neurotherapeutics 19 (2022), 99 – 116

2022
[11]

Kyle Hundman, Valentino Constantinou, Christopher Laporte, Ian Colwell, and Tom Soderstrom. 2018. Detecting Spacecraft Anomalies using LSTMs and Non- parametric Dynamic Thresholding. In ACM SIGKDD. 387–395

2018
[12]

Stamos Katsigiannis and Naeem Ramzan. 2017. DREAMER: A Database for Emotion Recognition Through EEG and ECG Signals From Wireless Low-Cost Off-the-Shelf Devices. IEEE Journal of Biomedical and Health Informatics 22, 1 (2017), 98–107

2017
[13]

Robert Kraut, Han Li, and Haiyi Zhu. 2022. Mental Health During the COVID-19 Pandemic: Impacts of Disease, Social Isolation, and Financial Stressors. PLOS ONE 17 (2022), e0277562

2022
[15]

Dan Li, Dacheng Chen, Lei Shi, Baihong Jin, Jonathan Goh, and See-Kiong Ng
[16]

Simon Luo, Jianlong Zhou, Henry Been-Lirn Duh, and Fang Chen. 2017. BVP Feature Signal Analysis for Intelligent User Interface. InProc. of the 2017 CHI Conf. Extended Abstracts on Human Factors in Computing Systems (Denver, Colorado, USA) (CHI EA ’17). 1861–1868

2017
[17]

In ICANN

MAD-GAN: Multivariate Anomaly Detection for Time Series Data with Generative Adversarial Networks. In ICANN
[18]

Bardh Prenkaj and Paola Velardi. 2023. Unsupervised detection of behavioural drifts with dynamic clustering and trajectory analysis. IEEE TKDE 36, 5 (2023), 2257–2270

2023
[19]

Parvin, S

P. Parvin, S. Chessa, M. Kaptein, and F. Paternò. 2019. Personalized Real-Time Anomaly Detection and Health Feedback for Older Adults. Journal of Ambient Intelligence and Smart Environments 11 (2019), 453–469

2019
[20]

Alban Siffer, Pierre-Alain Fouque, Alexandre Termier, and Christine Largouet
[21]

Philip Schmidt, Attila Reiss, Robert Duerichen, Claus Marberger, and Kristof Van Laerhoven. 2018. Introducing wesad, a multimodal dataset for wearable stress and affect detection. In ACM ICMI. 400–408

2018
[22]

Ya Su, Youjian Zhao, Chenhao Niu, Rong Liu, Wei Sun, and Dan Pei. 2019. Robust anomaly detection for multivariate time series through stochastic recurrent neural network. In ACM SIGKDD. 2828–2837

2019
[23]

Ya Su, Youjian Zhao, Chenhao Niu, Rong Liu, Wei Sun, and Dan Pei. 2019. Robust anomaly detection for multivariate time series through stochastic recurrent neural network. In Proc. of ACM SIGKDD Int. Conf. on Knowledge Discovery & Data Mining. 2828–2837

2019
[24]

Mohammad Soleymani, Jeroen Lichtenauer, Thierry Pun, and Maja Pantic. 2011. A Multimodal Database for Affect Recognition and Implicit Tagging. IEEE TAC 3, 1 (2011), 42–55

2011
[25]

Shreshth Tuli, Giuliano Casale, and Nicholas R Jennings. 2022. Tranad: Deep transformer networks for anomaly detection in multivariate time series data. In VLDB Endowment, Vol. 15. 1201–1214

2022
[26]

Qiuejie Xie, Qiming Feng, Tianqi Zhang, Qingqiu Li, Yuejie Zhang, Rui Feng, and Shang Gao. 2024. Human Simulacra: A Step toward the Personification of Large Language Models. ArXiv abs/2402.18180 (2024)

2024 arXiv
[27]

Georgios Taskasaplidis, Dimitris Fotiadis, and Panagiotis Bamidis. 2024. Review of Stress Detection Methods Using Wearable Sensors. IEEE Access 12 (2024), 38219–38246

2024
[29]

Yuxin Zhang, Yiqiang Chen, Jindong Wang, and Zhiwen Pan. 2021. Unsupervised deep anomaly detection for multi-sensor time-series signals. IEEE TKDE 35, 2 (2021), 2118–2132

2021
[30]

Chuxu Zhang, Dongjin Song, Yuncong Chen, Xinyang Feng, Cristian Lumezanu, Wei Cheng, Jingchao Ni, Bo Zong, Haifeng Chen, and Nitesh V Chawla. 2019. A deep neural network for unsupervised anomaly detection and diagnosis in multivariate time series data. In AAAI, Vol. 33. 1409–1416

2019
[31]

Bo Zong, Qi Song, Martin Renqiang Min, Wei Cheng, Cristian Lumezanu, Daeki Cho, and Haifeng Chen. 2018. Deep autoencoding gaussian mixture model for unsupervised anomaly detection. In Int. Conf. on Learning Representations

2018
[33]

Hang Zhao, Yujing Wang, Juanyong Duan, Congrui Huang, Defu Cao, Yunhai Tong, Bixiong Xu, Jing Bai, Jie Tong, and Qi Zhang. 2020. Multivariate time-series anomaly detection via graph attention network. In Proc. of ICDM. IEEE, 841–850

2020
[2017]

In Proceedings of the 23rd ACM SIGKDD international conference on knowledge discovery and data mining

Anomaly detection in streams with extreme value theory. In Proceedings of the 23rd ACM SIGKDD international conference on knowledge discovery and data mining. 1067–1075
[2019]

In International Conference on Artificial Neural Networks

Mad-gan: Multivariate anomaly detection for time series data with gen- erative adversarial networks. In International Conference on Artificial Neural Networks. Springer, 703–716

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.

[1] [8]

Shanghua Gao, Teddy Koker, Owen Queen, Tom Hartvigsen, Theodoros Tsiligkaridis, and Marinka Zitnik. 2025. UniTS: A unified multi-task time series model. Advances in Neural Information Processing Systems 37 (2025), 140589– 140631

work page 2025

[2] [1]

Menatalla Abououf, Shakti Singh, Rabeb Mizouni, and Hadi Otrok. 2023. Ex- plainable AI for event and anomaly detection and classification in healthcare monitoring systems. IEEE Internet of Things Journal 11, 2 (2023), 3446–3457

work page 2023

[3] [2]

Julien Audibert, Pietro Michiardi, Frédéric Guyard, Sébastien Marti, and Maria A Zuluaga. 2020. USAD: Unsupervised anomaly detection on multivariate time series. In ACM SIGKDD. 3395–3404

work page 2020

[4] [3]

Gerald Bieber, Thomas Kirste, and Michael Gaede. 2014. Low sampling rate for physical activity recognition. In Proc. of Int. Conf. on PErvasive Technologies Related to Assistive Environments (Rhodes, Greece) (PETRA ’14). Association for Computing Machinery, Article 15, 8 pages

work page 2014

[5] [4]

Nivedita Bijlani, Ramin Nilforooshan, and Samaneh Kouchaki. 2022. An unsu- pervised data-driven anomaly detection approach for adverse health conditions in people living with dementia: Cohort study. JMIR aging 5, 3 (2022), e38211

work page 2022

[6] [5]

Ailin Deng and Bryan Hooi. 2021. Graph neural network-based anomaly detection in multivariate time series. In AAAI, Vol. 35. 4027–4035

work page 2021

[7] [6]

Alessandro Flaborea, Bardh Prenkaj, Bharti Munjal, Marco Aurelio Sterpa, Dario Aragona, Luca Podo, and Fabio Galasso. 2023. Are we certain it’s anomalous?. In CVPR. 2896–2906

work page 2023

[8] [7]

Alessandro Flaborea, Bardh Prenkaj, Bharti Munjal, Marco Aurelio Sterpa, Dario Aragona, Luca Podo, and Fabio Galasso. 2023. Are we certain it’s anomalous?. In Proc. of IEEE/CVF Conf. on Computer Vision and Pattern Recognition . 2897–2907

work page 2023