Online peak detection in photoplethysmogram signals using sequential learning algorithm

Photoplethysmogram signals are becoming increasingly important for the detection of abnormalities in patients. Peak detection plays a significant role in diagnosis and monitoring using PPG signals. Although a copious number of methods are available for peak detection, none of them consider an online processing of the signal. In this paper we propose an online peak detection algorithm that tries to mimic the human cognitive model using a three-layered feedforward neural network trained using online sequential learning algorithm. The signals are processed sequentially without pre-processing or feature extraction, and result in an almost instantaneous detection of peaks.