Extended Classifier For The Photoplethysmographic Waveforms

The methods for filtering, peak analysis and segmentation of the photoplethysmographic signals and methods for unsupervised cluster analysis are considered. The support vector machine classifier is proposed using the automatically segmented waveforms for supervised learning. The four known morphological classes of pulse waves are extended with three intermediate classes to improve the model's flexibility. The feature selection problem is discussed. The use of statistical analysis of the pulse wave distribution across the morphological classes for cardiac activity monitoring is proposed.