Apparatus and method for detecting artifact of blood pressure signal, and computer readable recording medium storing the method

Abstract

A method for detecting an artifact of a blood pressure signal includes measuring consecutive signals of blood pressure in real time, dividing the consecutive signals into a plurality of unit waveforms by searching systolic starting points and peak points from the measured signals, determining two outputs by applying N consecutive unit waveforms among the plurality of unit waveforms as an input to a deep belief network (DBN), and determining whether the signal is an artifact or a normal signal by using the two determined outputs. According to the present disclosure, it is possible to minimize various false alarms of bio-signal streams which are measured in real time by automatically eliminating ABP artifacts using a deep belief network (DBN) which is one of deep neural network (DNN) models capable of learning causes and shapes of various types of artifacts together.