Cardiac Arrhythmia Classifier

Ecemiş C., Avcu N., Sarı Z.

2nd International Conference on Applied Engineering, Konya, Türkiye, 10 Mart - 13 Kasım 2022, ss.489-492

  • Yayın Türü: Bildiri / Tam Metin Bildiri
  • Basıldığı Şehir: Konya
  • Basıldığı Ülke: Türkiye
  • Sayfa Sayıları: ss.489-492
  • Dokuz Eylül Üniversitesi Adresli: Evet

Özet

The irregularities in the heartbeat are called arrhythmias and can be an essential subject for heart health. The artifacts in the mechanical structure or the electrical signaling system of the heart result in these irregularities. Arrhythmias depending on the different artifacts, cause different patterns on electroencephalogram recordings of patients. Early diagnosis of cardiac arrhythmia is quite crucial to saving patient lives. The main goal of this study is to detect the presence of cardiac arrhythmia and classify it into sixteen different groups from the electrocardiogram (ECG) recordings. First, the classification performances of four well-known classification algorithms, namely, multilayer perceptron (MLP), support vector machines (SVM), random forest (RF), and K-means clustering, have been obtained using an imbalanced electrocardiogram (ECG) dataset. The classifier performance criteria such as accuracy and precision are used to determine the best classifier structure for each arrhythmia class. Then, to obtain a better classifier performance, an ensemble network which is a cascaded form of two RF classifiers, is constructed. The designed ensemble network consisting of two RF networks results in an accuracy of 73.63%.