Primerjava sodobnih metod strojnega učenja na primeru diagnosticiranja EKG signalov

| | SLO | ENG | Cookies and privacy

Bigger font | Smaller font

First page > Show document

Show document

Title:	Primerjava sodobnih metod strojnega učenja na primeru diagnosticiranja EKG signalov
Authors:	ID Zavratnik, Jure (Author) ID Verber, Domen (Mentor) More about this mentor...
Files:	MAG_Zavratnik_Jure_2026.pdf (1,67 MB) MD5: 4733704932E6FA394058A65E5D0F70E4
Language:	Slovenian
Work type:	Master's thesis/paper
Typology:	2.09 - Master's Thesis
Organization:	FERI - Faculty of Electrical Engineering and Computer Science
Abstract:	V magistrskem delu primerjamo sodobne pristope strojnega učenja za večoznačno diagnosticiranje 12-odvodnih EKG signalov na podatkovni zbirki Georgia 12-lead ECG Challenge. Posvetimo se modelu random forest ter globokim arhitekturam konvolucijskih in rezidualnih nevronskih mrež, ocenjenim z makro in mikro metrikami, ter razložljivim z metodama SHAP in Grad-CAM. Rezultati kažejo, da ResNet najzanesljiveje prepozna kompleksne vzorce, CNN nudi dobro razmerje med natančnostjo, hitrostjo in velikostjo modela, medtem ko RF služi kot referenca. Kombinacija visoke uspešnosti, dobre razložljivosti in majhne računske zahtevnosti predstavlja ključni korak k zanesljivejši uporabi strojnega učenja v kliničnem okolju.
Keywords:	strojno učenje, EKG signali, medicinska diagnostika, klasifikacija, metrike
Place of publishing:	Maribor
Year of publishing:	2026
PID:	20.500.12556/DKUM-98957
Publication date in DKUM:	24.09.2026
Views:	96
Downloads:	1
Metadata:
Categories:	KTFMB - FERI
:	Copy citation

Average score:	(0 votes)
Your score:	Voting is allowed only for logged in users.
Share:

Hover the mouse pointer over a document title to show the abstract or click on the title to get all document metadata.

License:	CC BY 4.0, Creative Commons Attribution 4.0 International

Link:	http://creativecommons.org/licenses/by/4.0/
Description:	This is the standard Creative Commons license that gives others maximum freedom to do what they want with the work as long as they credit the author.
Licensing start date:	21.07.2026

Language:	English
Title:	Comparison of modern machine learning methods for ECG signal diagnosis
Abstract:	In this master's thesis, we compare modern machine learning approaches for the multi-label diagnosis of 12-lead ECG signals using the Georgia 12-lead ECG Challenge database. We focus on the random forest model and deep architectures of convolutional and residual neural networks, which are evaluated using macro and micro metrics, and explained using SHAP and Grad-CAM methods. The results show that ResNet most reliably recognizes complex patterns, CNN offers a good balance between accuracy, speed, and model size, while RF serves as a reference baseline. The combination of high performance, good explainability, and low computational complexity represents a key step toward the more reliable use of machine learning in a clinical environment.
Keywords:	machine learning, ECG signals, medicinal diagnostics, classification, metrics

Leave comment

You must log in to leave a comment.

Comments (0)

0 - 0 / 0

There are no comments!

Back