Multilayer perceptron neural network for detecting singular points of photoprethysmograms
Photoplethysmography is an important optical method in healthcare and contains a wealth of information about physiological dynamics. For a quantitative assessment of vascular tone indices, an effective and modern method of analyzing the relief of the photoplethysmogram within the pulse cycle is required. A new method for detecting singular points of the contour of photoplethysmograms, based on a model of a multilayer perceptronic neural network, is proposed. Demonstrated the successful operation of the network of the method for finding the systolic peak, dicrotic notch and diastolic peak. It is shown that the average relative error does not exceed 5 percent. The proposed method can be used in computer monitoring systems of vascular tone.
File with abstract
Vladislav Yurievich Gribkov
Volgograd State University
File with report