Skripsi

PENINGKATAN KINERJA KLASIFIKASI GANGGUAN JANTUNG PADA SINYAL ELEKTROKARDIOGRAM BERBASIS TIME-FREQUENCY DOMAIN DAN MACHINE LEARNING

---

Electrocardiogram (ECG) signals represent the electrical activity of the heart and are used to record disorders such as arrhythmia and heart failure. Due to their non-stationary nature, ECG signals require a time-frequency domain approach to capture their dynamic characteristics more accurately. This study aims to develop and evaluate machine learning-based heart disorder classification models using features extracted from Short-Time Fourier Transform (STFT) and Continuous Wavelet Transform (CWT). The study was conducted on two cases, namely arrhythmia classification (AFIB and N) and heart failure classification (CHF and NSR). The research stages included signal preprocessing, feature extraction, and model training using Support Vector Machine (SVM), Random Forest, Decision Tree, XGBoost, Logistic Regression, and K-Nearest Neighbor (KNN). Model evaluation was performed using Accuracy, Precision, Sensitivity, Specificity, F1-score, and ROC-AUC metrics on both validation and test sets. The results showed that in the arrhythmia scenario, the best STFT model on the test set was SVM (kernel = RBF, C = 50, γ = auto) with an accuracy of 0.9744 and F1-score of 0.9730, while on the validation set it achieved an accuracy of 0.871795 and F1-score of 0.830522. Meanwhile, the best CWT model on the test set was SVM (kernel = Linear, C = 10, γ = scale) with an accuracy of 0.9744 and F1-score of 0.9731, while on the validation set it achieved an accuracy of 0.846154 and F1-score of 0.812088. In the heart failure case, two experiments were conducted using combined features (35 features) and separated features consisting of STFT (20 features) and CWT (15 features). Based on all testing scenarios, the best performance was achieved using CWT features (15 features) with the XGBoost model on the test set, obtaining an accuracy of 0.9572 and an AUC of 0.9955.

---

Availability

Detail Information

Series Title

-

Call Number

T2002302026

Publisher

Indralaya : Prodi Sistem Komputer, Fakultas Ilmu Komputer Universitas Sriwijaya., 2026

Collation

xxvi, 229 hlm.; ilus.; tab.; 29 cm.

Language

Indonesia

ISBN/ISSN

-

Classification

006.307

Content Type

Text

Media Type

-

Carrier Type

-

Edition

-

Subject(s)

Prodi Sistem Komputer
Model Deep Learning

Specific Detail Info

-

Statement of Responsibility

MI

Other version/related

File Attachment

Comments

---

You must be logged in to post a comment