Institutional Repository of Vilnius University
Klasterizavimo metodų taikymas širdies aritmijų identifikavimui naudojant elektrokardiogramų duomenis /
Abstract
dc:descriptionThe primary task of the thesis is to investigate the performance of the selected clustering methods in dividing the heart beats in an unbalanced and a balanced dataset into separate clusters, considering the ECG flow data of these beats and the class assigned to each beat, in order to check if the methods are suitable for arrhythmia identification. The hypothesis is "Is it possible to cluster heart beats into distinct clusters using clustering methods described in scientific studies?". The methods chosen for the study are the fuzzy k-means method, the self-organising map (SOM), and the hierarchical method using the Ward, single, average, and complete methods. The results of the tasks show that the most accurate results for the problem are achieved using the fuzzy k-means clustering method. When the results of this method were visualised using the T-SNE data visualisation method, the heart beats belonging to separate arrhythmia classes were separated and grouped into separate clusters. In terms of the silhouette measure, the best results were obtained using the fuzzy k-means method with an unbalanced dataset. The resulting value of the measure is 0.38.
Degree
thesis:*- Grantor dc:publisher
- Institutional Repository of Vilnius University
- Year dc:date
- 2023
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Gaidukevičius, Danielius,
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:ELABAETD192842364&prefLang=en_US