Daftar Isi: PENERAPAN MODEL CONVOLUTION NEURAL NETWORK (CNN) UNTUK MENGENALI POLA BEAT EKG

Main Authors:	FAUZA, ANNISA NUR, Desiani, Anita, Maiyanti, Sri Indra
Format:	Thesis NonPeerReviewed Book
Bahasa:	ind
Terbitan:	, 2019
Subjects:	QA Mathematics QA1-939 Mathematics QA75 Electronic computers. Computer science
Online Access:	http://repository.unsri.ac.id/2431/1/RAMA_44201_08011281520094.pdf http://repository.unsri.ac.id/2431/2/RAMA_44201_08011281520094_TURNITIN.pdf http://repository.unsri.ac.id/2431/3/RAMA_44201_08011281520094_0011127702_0004077205_01_front_ref.pdf http://repository.unsri.ac.id/2431/4/RAMA_44201_08011281520094_0011127702_0004077205_02.pdf http://repository.unsri.ac.id/2431/5/RAMA_44201_08011281520094_0011127702_0004077205_03.pdf http://repository.unsri.ac.id/2431/6/RAMA_44201_08011281520094_0011127702_0004077205_04.pdf http://repository.unsri.ac.id/2431/7/RAMA_44201_08011281520094_0011127702_0004077205_05.pdf http://repository.unsri.ac.id/2431/8/RAMA_44201_08011281520094_0011127702_0004077205_06_ref.pdf http://repository.unsri.ac.id/2431/

Daftar Isi:

Data on arrhythmia patients in the form of ECGs have large amounts of data. The data can be used to find out arrhythmias by recognizing normal and abnormal beat patterns. ECG data from 47 arrhythmia patients in the MIT-BIH Arrhyhtmia Database had a duration of 30 minutes for each patient or 650,000 signal points. The results of beat segmentation were 109,452 beats with 252 signal lengths. The algorithm used for normal and abnormal beat pattern recognition is CNN. CNN is commonly used for data in large dimensions. This study uses the percentage split method with a percentage of 33% which means 73,332 for training data and 36,120 for test data. The results of the accuracy obtained were 97.24%. Precision for normal and abnormal beat labels is 98.69% and 90.57%. Recall for normal and abnormal labels is 97.97% and 93.75%. It can be concluded that the application of the CNN model is very good for EKG beat pattern recognition.