Design of a real-time morphology-based anomaly detection method from ECG streams

Anomaly detection from ECG stream is a key step leading to a significant success of the remote and auto-triggered cardiac event monitoring system. This effort requires an online processing and efficient analysis on the real-time data. Moreover, its computational complexity should be kept low so that the detection algorithm can be implemented even on a small computing device used in sensor network. In this paper, we present a novel fast and effective approach to identify abnormalities based on differences of heart beat morphologies. Our approach is inspired from time-series data mining techniques and statistical outlier detection methods. The experimental results overall (open public QT database) demonstrate high quality performance. In particular, it obtains 0.971, 0.995 and 0.994, on an average, for of sensitivity, specificity and accuracy for the respective performance metrics.