Short-time fourier transform based ensemble classifiers for detection of atrial fibrillation from ECG datasets

Abstract

Atrial Fibrillation (Afib) is a common cardiac arrhythmia characterized by irregular and often rapid heart rate, leading to inefficient blood pumping from the atria. It increases the risk of stroke, heart failure, and other heart-related complications. Afib is often associated with symptoms like palpitations, shortness of breath, and fatigue. Diagnosis typically involves electrocardiography (ECG) to detect irregular electrical activity in the heart. Treatment options range from medication to procedures like catheter ablation, aimed at restoring normal heart rhythm and reducing associated risks. Soft computing methods can aid in automating the classification of cardiovascular diseases, assisting clinicians in diagnosing arrhythmias. In this research paper, ensemble classifiers are employed for the classification of Atrial Fibrillation based on ECG datasets. When utilizing the Catboost Classifier in conjunction with the STFT-based GEO implementation, the results indicate an average perfect classification rate of approximately 99%, an error rate of 1%, and a kappa coefficient of 0.9689% for detection of Afib.

Keywords

• adaboost, cardiovascular disease, catboost, ecg, xgboost