Machine Algorithm for ECG-Based Heartbeat Monitoring and Arrhythmia Detection

Abstract

A cardiac arrhythmia is a condition where the heartbeat is irregular and either too fast or too slow. It
is caused by malfunctioning electrical impulses that synchronize the heartbeats. Some severe arrhythmia diseases
can result in sudden cardiac death. Therefore, the main objective of electrocardiogram (ECG) research is to
accurately identify arrhythmias as potentially fatal in order to provide an appropriate treatment and save lives.
Waveforms known as ECG signals (P, QRS, and T) show how the human heart moves electrically. Each
waveform's duration, structure, and the separations between its many peaks are used to detect cardiac issues. The
parameters of the AR signal model are then determined by applying the signals' autoregressive (AR) analysis to a
particular selection of signal attributes. The training dataset offers high connection categorization and heart issue
detection for every ECG signal by neatly separating groups of extracted AR features for three different ECG
types. To improve the evaluation of ECG data, a novel method based on fractional Fourier transform (FFT)
algorithms and two-event-related moving averages (TERMAs) is proposed. Researchers may find this paper
useful in analyzing the most advanced methods currently used for arrhythmia identification. Cross-database
training and testing with enhanced features is a feature of our proposed machine learning methodology.