Keywords
non-stationary time series
vital sign anomaly detection
Abstract
Intensive care units generate continuous multivariate physiological time series from heart rate, blood pressure, oxygen saturation, respiratory rate, body temperature, electrocardiographic signals, and medication-related records. These signals are highly non-stationary because patient conditions change with disease progression, treatment response, medication dosage, and clinical intervention. This study proposes a latent dependency fusion model for non-stationary vital sign anomaly detection in intensive care monitoring. The model first learns patient-specific temporal regimes through a gated sequence encoder. It then constructs latent dependency representations among physiological variables using cross-channel attention and dynamic correlation estimation. A residual-based anomaly scoring module is applied to detect abnormal physiological changes while reducing false alerts caused by normal post-treatment fluctuations. Experiments are conducted on an ICU monitoring dataset containing 18,400 patient stays, 64 physiological variables, and 1-minute observations collected over 21 months. The dataset includes 96 million timestamped records and 3,260 clinically annotated abnormal episodes, including sudden hypotension, oxygen desaturation, arrhythmia-related instability, abnormal respiratory deterioration, and post-operative shock signals. The proposed method reduces median warning delay from 27.5 minutes to 8.9 minutes compared with a single-channel temporal autoencoder. False alarms decrease to 3.4 alerts per patient-day. The Matthews correlation coefficient reaches 0.889 across mixed patient groups, and the average event-boundary deviation is reduced by 18.6 minutes. Online scoring processes 22,000 physiological points per second, with a median inference latency of 31 ms per patient window. These results indicate that latent dependency fusion can improve anomaly detection for non-stationary clinical time series in intensive care environments.
References
1. Yang, M., Wu, J., Tong, L., & Shi, J. (2025). Design of Advertisement Creative Optimization and Performance Enhancement System Based on Multimodal Deep Learning.
2. Angelucci, A., Greco, M., Cecconi, M., & Aliverti, A. (2025). Wearable devices for patient monitoring in the intensive care unit. Intensive care medicine experimental, 13(1), 26.
3. Jiao, Y., Wang, A., Zhao, B., & Shi, T. (2026). The Impact of Visual Language Strategies in Public Art Creation on Community Spatial Perception and Public Behavior.
4. Lagina, M., & Valley, T. S. (2024). Diagnosis and management of acute respiratory failure. Critical care clinics, 40(2), 235-253.
5. 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.
6. 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.
7. Alqudah, A. M., & Moussavi, Z. (2025). A Review of Deep Learning for Biomedical Signals: Current Applications, Advancements, Future Prospects, Interpretation, and Challenges. Computers, Materials & Continua, 83(3).
8. Yang, J. (2026). Stage‐Coupled Computational Framework for Stratified Accessibility and Equity Analysis in Community‐Based Elderly Care Services.
9. Zhang, Z. (2026). A Study on Return Optimization in E-commerce for Complex Consumer Goods Driven by Installation Information Quality. Available at SSRN 6734720.
10. Leone, A., Carluccio, A. M., Caroppo, A., Manni, A., & Rescio, G. (2025). A systematic review of surface electromyography in sarcopenia: muscles Involved, signal processing Techniques, significant Features, and artificial intelligence approaches. Sensors, 25(7), 2122.
11. Gui, H., Wang, B., Lu, Y., & Fu, Y. (2025). Computational Modeling-Based Estimation of Residual Stress and Fatigue Life of Medical Welded Structures.
12. Chen, F., Liang, H., Li, S., Yue, L., & Xu, P. (2025). Design of Domestic Chip Scheduling Architecture for Smart Grid Based on Edge Collaboration.
13. Döner, B., Ingolfsson, T. M., Benini, L., & Li, Y. (2026). LUNA: efficient and topology-agnostic foundation model for EEG signal analysis. Advances in Neural Information Processing Systems, 38, 70682-70708.
14. Roser, S. M., & Bouloux, G. F. (2022). Medical management and preoperative patient assessment. In Peterson’s Principles of Oral and Maxillofacial Surgery (pp. 19-51). Cham: Springer International Publishing.
15. Liang, R., Ye, Z., Liang, Y., & Li, S. (2025). Deep Learning-Based Player Behavior Modeling and Game Interaction System Optimization Research.
16. Li, Y., & Liu, S. (2026, May). A Study on Dynamic Optimization of Alerting Policies and Multi-Agent Decision-Making Mechanisms in Cloud Environments. In 2026 7th International Seminar on Artificial Intelligence, Networking and Information Technology (AINIT) (pp. 703-706). IEEE.
17. Lewandowska, K., Mędrzycka-Dąbrowska, W., Mikłas, M., & Wujtewicz, M. (2025). Improving alarm management to reduce alarm fatigue in critical care: a mixed-methods study. Intensive and Critical Care Nursing, 104254.
18. Xu, T., Zhang, J., & Zhu, W. (2026). Reproducible Modeling Pipelines and Cross-Window Stability in Subprime Auto Loan Credit Risk Assessment. Available at SSRN 6893861.
19. Zhou, Y., & Long, L. (2026). Causal Effect Evaluation of Personalized Reminder Strategies on Government Welfare Program Enrollment: A Propensity Score Matching Approach. Journal of Computing Innovations and Applications, 4(1), 106-116.
20. Heydari, V., & Nyarko, K. (2026). Alarm-Budgeted Event-Level Evaluation Of Ics Anomaly Detection: Lessons From Swat And Wadi. IEEE Access.
21. Pei, Z., Huang, Q., & Wang, S. (2026). When LLMs Develop Languages: Symbolic Communication for Efficient Multi-Agent Reasoning. arXiv preprint arXiv:2606.29354.
22. Moralez, G. M., Amado, F., Liu, V. X., Tan, S. C., Meyfroidt, G., Stevens, R. D., ... & Salluh, J. I. (2025). Data-driven quality of care in the ICU: A concise review. Critical care medicine, 53(12), e2720-e2728.
23. 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.
24. Qi, C., & Qiao, X. (2026, May). From Traditional Machine Learning to Large Language Models: The Evolution of Infrastructure from an Engineering Perspective. In 2026 6th International Symposium on Computer Technology and Information Science (ISCTIS) (pp. 503-506). IEEE.
25. Pinsky, M. R., Cecconi, M., Chew, M. S., De Backer, D., Douglas, I., Edwards, M., ... & Vincent, J. L. (2022). Effective hemodynamic monitoring. Critical Care, 26(1), 294.
26. Su, D., & Zhang, H. Developing a Data-Driven Computational Framework for Scalable Special Education Practices for Autism and Intellectual Disabilities in the US Public School System.
27. Diallo, A. R., Homri, L., & Dantan, J. Y. (2025). Reducing false alarms in fault detection: A comparative analysis between conformal prediction and classical methods applied to PCA and autoencoders. Journal of Process Control, 152, 103495.
28. Ma, Y. (2026). Industrial Financial Risk Prediction Model Based on Graph Neural Network and Knowledge Graph Inference. Journal of Circuits, Systems and Computers, 35(16), 2650103.
29. Gao, G., Gao, R., Lu, C., Gao, R., & Kuang, Y. (2026, March). Security Governance Methods and Quantitative Evaluation for Enterprise SMS and Verification Code Systems. In 2026 International Conference on Generative Artificial Intelligence and Information Security (GAIIS) (pp. 455-458). IEEE.
30. Luijten, M. A., Schalet, B. D., Roorda, L. D., Haverman, L., & Terwee, C. B. (2025). Reducing patient burden of PROMs in healthcare through advanced computerized adaptive testing stopping rules. Quality of Life Research, 1-10.
This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.
Copyright (c) 2026 Sarah Mitchell, James Walker (Author)