Heart rate reliability for the Smart Rollator

In this paper, we propose an algorithm to quantify the reliability of heart rates estimated from electrocardiogram (ECG) data. Long-term ECG monitoring is becoming more prevalent with an increasing number of ambulatory monitors, including the Smart Rollator which is a rollator equipped with a series of minimally obtrusive sensors. The ECG reliability index enables the automatic pre-processing of data, to highlight or discard data that are corrupted by noise or other artifacts. Three reliability indices are proposed that are based on the assumption of a short-term invariant PQRST waveform. These indices use distance measures to quantify the reliability of the ECG, which are: 1) percent residual difference, 2) cross-correlation coefficient, and 3) a wavelet distance measure. The reliability indices are evaluated using real ECG data corrupted with three types of noise: 1) baseline wander, 2) electromyogram artifacts, and 3) motion artifact. All three reliability indices demonstrate an ability to track the accuracy of heart rate estimates derived from the ECG. Among the three indices investigated in this work, the wavelet distance measure provides the best performance.