Robust and Accurate Anomaly Detection in ECG Artifacts Using Time Series Motif Discovery

Figure 6

Comparison of existing algorithm on two different settings in the number of data points used. (a) The whole 7,710 data points were run at one time and (b) the signal was split into separate runs, each with 2,570 data points. The dashed-line box frames the period of real anomaly ECG beat. The number in parentheses identifies the order of detected anomaly subsequence from existing algorithm. It is apparent that more false alarms are detected when longer ECG is used in the calculation.