A differentiation-based adaptive double threshold method for real time electrocardiogram R peak detection

Electrocardiogram (ECG) is a fundamental and important noninvasive cardiac signal. In current decade, it has been popular to acquire and analyze ECG by wearable devices with main purposes of long-term monitoring and potential cardiac event prediction. In this study, we aimed at developing a real time and accurate R peak detection method which may be useful for ECG analysis in the wearable devices. The method was proposed to detect R peaks using an adaptive double threshold judging strategy integrated with a speeding research operation both applied on the first derivative of filtered ECG with a low pass filter. The performance of the proposed method was then verified against R peak manual annotations on ECG data of 70 subjects (including 32 heart failure patients) collected in a previous bicycle exercise experiment. The results showed a high R peak detection accuracy (sensitivity: 99.80%; positive predictivity: 99.89%) and a short computation time (mean ± standard deviation is 0.0472 ± 0.0039 s per 40 sec ECG data at a sampling rate of 200 Hz). The results in this study indicated that the proposed method is promising to be used for real time ECG R peak detection in wearable devices in the future.