A Novel Transformer-Based Anomaly Detection Approach for ECG Monitoring Healthcare System

Anomaly detection plays a crucial role across various domains, including healthcare, where identifying deviations from normal patterns can lead to early intervention and improved outcomes. In healthcare, such as in ECG analysis, detecting anomalous signals is essential for timely diagnosis and treatment, as it can help identify potentially life-threatening conditions that might otherwise go unnoticed. In this work, by focusing on ECG anomaly detection as an illustrative healthcare application, we propose to use a transformer-based variational autoencoder network together with a MEWMA-SVDD control chart to achieve anomaly detection. By employing this approach, we can effectively control the false alarm rate, aligning with the intended goal of minimizing unnecessary alerts. Our proposed framework not only excels in terms of accuracy but also reduces the false alarm rate, making it a favorable choice compared to existing methods.