Anomaly Detection in Patient Monitoring Time Series Data

Abstract

Abstract Massive volumes of multivariate time-series data are produced by the ongoing patient monitoring in intensive care units (ICUs), which records vital physiological indicators including heart rate, blood pressure, oxygen saturation and respiration rate etc. Finding early indicators of clinical deterioration and facilitating timely medical intervention depend on the timely and accurate detection of aberrant patterns in this data. However, threshold-based alerts and conventional rule-based systems frequently have high false alarm rates and little flexibility to accommodate the heterogeneity of individual patients. In order to tackle the crucial problem of automated anomaly identification in ICU patient monitoring data, this thesis suggests a deep learning-based system that provides explainability for clinical transparency in addition to detecting anomalies in multivariate time-series signals.Using the publicly available eICU Collaborative Research Database, we introduce an end-to-end anomaly detection methodology that combines an Encoder- Decoder-Discriminator (EDD) model, structured as a Variational Autoencoder (VAE). This architecture learns latent representations of normal physiological behavior to effectively identify deviations indicative of abnormal conditions. Both periodic and aperiodic vital sign recordings are preprocessed using a sliding window approach to extract temporal relationships. The model achieved good precision and recall in anomaly identification after being trained on data from 150 patients and validated on 50 patients who were not physically present. Classification criteria such confusion matrices, F1-score, and classification reports were used to thoroughly assess performance. We use SHAP (SHapley Additive exPlanations) to measure feature importance in order to meet the crucial need for model interpretability in clinical decision-making. This method improves physicians’ confidence and comprehension of AI-driven warnings by highlighting the physiological factors that have the biggest impact on the model’s anomaly predictions. The results show that our explainable AI method not only increases the accuracy of anomaly identification in intensive care units but also opens the door for the incorporation of reliable machine learning tools into clinical processes in real time. Future research will examine the usage of Temporal SHAP for deployment in real-time intensive care unit settings, integration with multimodal inputs (such as lab findings and notes), and time aware interpretability.