Institutional Repository of Vilnius University
Duomenų balansavimo strategijų įtaka EKG širdies ritmo pūpsnių klasifikavimui, siekiant nustatyti aritmiją /
Abstract
dc:descriptionThe main goal of the final thesis is to apply data balancing methods to a dataset and compare the influence of different balancing methods on the classification of ECG beats. The following tasks were addressed in the thesis: analysis of data balancing algorithms based on scientific literature research for ECG beats identification, implementation and evaluation of balancing algorithms, selection of classification methods based on scientific literature research, implementation of chosen methods, practical comparison of algorithms and classifiers to classify arrhythmia beats, and evaluation of the impact of data balancing and different classifiers on classification results. A convolutional neural network and a random forest classifier were developed. Data balancing was performed using the random oversampling, random undersampling, SMOTE, ADASYN, and SMOTEENN algorithms. The highest classification results were obtained by training a random forest classifier on balanced data using the ADASYN method. The classification accuracy for class 0, normal hearbeats, was 0.991, for class 1, atrial arrhythmia, was 0.992, and for class 2, ventricular arrhythmia, was 0.998.
Degree
thesis:*- Grantor dc:publisher
- Institutional Repository of Vilnius University
- Year dc:date
- 2023
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Vaitkevičiūtė, Gabrielė,
Rights
dc:rights- Statement dc:rights
-
- info:eu-repo/semantics/openAccess
- Language dc:language
- lit
Identifiers
dc:identifier.*- Repository record dc:identifier
- https://repository.vu.lt/VU:ELABAETD192843092&prefLang=en_US