Keywords
noisy physiological time series
diffusion reconstruction
disentangled representation
anomaly detection
motion artifact removal
health event detection
Abstract
Wearable health devices generate continuous physiological time series from heart rate, skin temperature, oxygen saturation, electrodermal activity, accelerometer signals, sleep stages, and respiratory indicators. These signals are highly noisy because of motion artifacts, loose sensor contact, battery fluctuation, user activity changes, and device synchronization errors. Direct anomaly detection on raw wearable data may confuse motion-induced noise with clinically relevant abnormal changes. This study develops a disentangled physiological signal reconstruction method for noisy wearable time series anomaly detection. The proposed model uses a diffusion-guided reconstruction process to remove motion-related artifacts and recover stable physiological trajectories. A dual-branch disentanglement module separates activity-related variation from abnormal physiological deviation. An adaptive residual scoring function is then applied to detect abnormal health-related events from the denoised latent space. Experiments are conducted on a wearable monitoring dataset containing 18,600 users, 7 device types, 29 physiological and motion variables, and 30-second records collected over 180 days. The dataset contains 286 million synchronized records and 2,840 annotated abnormal episodes, including nocturnal oxygen desaturation, abnormal resting heart-rate elevation, irregular respiration, sensor-contact failure, and fever-related temperature increase. The proposed method shortens median detection delay from 42.3 minutes to 13.6 minutes compared with a temporal convolutional reconstruction baseline. False health alerts are controlled at 2.1 cases per user-month after motion-artifact separation. Diffusion reconstruction lowers mean absolute signal reconstruction error from 0.184 to 0.071 normalized units. The disentanglement module removes 3.8 million motion-contaminated windows from the abnormal candidate queue during evaluation. The model completes daily user-level assessment in 5.2 minutes on a single GPU. These findings indicate that denoising and representation disentanglement can improve robust anomaly detection in noisy wearable physiological time series.
References
Yin, J., Rao, H., & Huang, Y. (2026, March). Dynamic Modeling and Heterogeneity Analysis of Platform User Behavior Time Series. In 2026 International Conference on AI in Education Technology and Applications (AIETA) (pp. 30-33). IEEE.
Shankar, R. S., Shailesh, T., R, V. A., Chadaga, K., & Cenitta, D. (2025). Smart health monitoring: an interactive Raspberry Pi-based system for managing diabetes and hypertension in children. Journal of Cloud Computing, 14(1), 70.
Yang, J. (2026). Computational Analysis of How Digital Non-Clinical Communication Tools Influence Social Participation Among Older Adults in Community-Based Elderly Care. Available at SSRN 6682838.
Bharath, A., & Sheeba, G. M. (2026). SmartHealth-IoT: a simulation-based IoT remote patient monitoring framework using virtual wearable physiological data streams. Scientific Reports.
Chen, X., Xiao, H., Zeng, Z., Zhang, S., & Xiao, R. (2025). Fine-Grained Multivariate Time Series Anomaly Detection via Causal Inference. Knowledge-Based Systems, 114765.
Zhang, Z. (2026). A Study on Multimodal Product Understanding and Assembly Guidance Optimization in e-commerce for Complex Consumer Goods. Available at SSRN 6734559.
Mohammadi, V., & Bigham, B. S. (2026). A Trust-Aware Hybrid Unsupervised Framework for Robust ECG Anomaly Detection in IoMT Monitoring Networks.
Xiong, W., Guo, Y., & Zeng, Y. (2026). A Study on the Energy Efficiency of MEP Systems and Coordinated Dispatch Mechanisms for Multi-energy Systems in High-density Urban Buildings. Available at SSRN 7120518.
Hong, Z., Xu, T., & Chen, H. (2026). A Study on Test Coverage Mapping and Release Risk Prediction in Continuous Delivery of Cloud Services.
Chopannavaz, M. R., & Ghaderi, F. (2025). An Empirical Investigation of Reconstruction-Based Models for Seizure Prediction from ECG Signals. arXiv preprint arXiv:2504.08381.
Zhang, Y., Wang, J., & Gu, W. (2026, February). Research on the Construction and Effectiveness of a Computer-Aided Systematic Training Framework for Wind Farm O&M Quality Amidst High Personnel Turnover. In Proceedings of the 2nd International Conference on Digital Management and Information Technology (pp. 647-651).
Li, J., Zhang, Y., & Gu, Y. (2026). Chain-Hash Audit Framework with Trusted Anchoring for Tamper-Evident Evidence Lifecycle Management.
Zhang, Z., Tong, Y., & Gao, Y. (2026). Retrieval-Augmented Generation with Low-Latency Deployment for Vertical Domains Question Answering: A Case Study on Economic Resource Platforms.
Attar, E. T. (2025). Physiological responses to physical activity: A multi-sensor wearable study across activity intensity, age, gender, body weight, and BMI variability. Medicine, 104(40), e44725.
Qi, C., & Qiao, X. (2026). Efficient Data Sampling and Feature Selection Algorithms for Scalable Machine Learning Pipelines.
Su, D., & Dong, Y. (2026). Classroom-Based Assessment with Bayesian Learning Analytics for Instructional Decision-Making in ASD Inclusive Education.
Al Abdi, R., AlKaabi, S., Elsifi, S., & Yousaf, J. (2026). Mental stress detection using physiological sensors and artificial intelligence: A review. Sensors, 26(5), 1616.
Feldmann, A., Quaresima, V., Perrey, S., & Ferrari, M. (2026). Distinguishing muscle oxygen saturation from tissue oxygen saturation: Accuracy, limitations, and practical utility in muscle oximetry. Sports Medicine-Open, 12(1), 61.
Du, Y., Dong, Y., Liu, Q., & Chen, Z. (2026). Modal Reliability Assessment and Drift Early Warning in Visible-Infrared Target Tracking. Available at SSRN 7132381.
Hong, Z., Chen, H., & Xu, T. (2026). Real-Time Capacity Risk Forecasting for Low-Latency Financial Trading Cloud Platforms. Available at SSRN 7118180.
Yahya, M. A., Moya, A. R., & Ventura, S. (2025). Deep learning for multivariate time series anomaly detection: An evaluation of reconstruction-based methods. Artificial Intelligence Review, 58(12), 400.
Li, J., Ma, R., & Gu, Y. (2026). From Audit Requirements to Computable Software Architecture: A Rule-Engine and Evidence-Indexing Method for Trusted Digital Infrastructure.
Du, Y., Chen, Z., Liu, Q., & Dong, Y. (2026). Co-identification of Vibration, Structural Looseness, and Surface Damage in Industrial Equipment Operation Videos.
Ayaz, H., Sultan, K., Sheraz, M., & Chuah, T. C. (2025). From patterns to predictions: spatiotemporal mobile traffic forecasting using AutoML, TimeGPT and traditional models. Computers, 14(7), 268.
Xiong, W., & Guo, Y. (2026). A Study on the Phenomenon of Abnormal Load Concentration and Its Cross-system Transmission Characteristics in High-density Building Operations. Available at SSRN 7120661.
Fu, Y., Gui, H., Li, W., & Wang, Z. (2020, August). Virtual Material Modeling and Vibration Reduction Design of Electron Beam Imaging System. In 2020 IEEE International Conference on Advances in Electrical Engineering and Computer Applications (AEECA) (pp. 1063-1070). IEEE.
Habijan, M., Krpić, Z., Perić, J., & Galić, I. (2025). Diffusion Models for MRI Reconstruction: A Systematic Review of Standard, Hybrid, Latent and Cold Diffusion Approaches. Electronics, 15(1), 76.
Zhang, Y., Gu, W., & Wang, J. (2025, December). Research on First Article Inspection (FAI)-Driven Quality Assurance Methods for Wind Turbine Installation and Operation & Maintenance and Their Effect on Reliability Improvement. In Proceedings of the 2025 6th International Conference on Computer Science and Management Technology (pp. 1700-1705).
Azadi, A., & García-Peñalvo, F. J. (2025). Optimizing clinical decision support system functionality by leveraging specific human-computer interaction elements: insights from a systematic review. JMIR Human Factors, 12(1), e69333.
Chen, H., Ma, X., Mao, Y., & Ning, P. (2025). Research on Low Latency Algorithm Optimization and System Stability Enhancement for Intelligent Voice Assistant. Available at SSRN 5321721.
Jiao, Y., Wang, A., Zhao, B., & Shi, T. (2026). Quantitative Study on the Construction and Application Effectiveness of Graffiti Wall Painting Teaching Models in Public Space Contexts.
Srikanteswara, R., Mishra, S., Subbalakshmi, C., Tiwari, S., Soliman, N. F., & Hassan, M. M. (2026). Personalized Generative Attention Modeling for Robust Anomaly Detection and Real-Time Monitoring in Multimodal Wearable Health Signals. Biomedical Materials & Devices, 1-23.