Machine Learning-Based Cardiac Arrhythmia Detection in Electrocardiogram Signals

Abstract

The cardiovascular system is vital for human physiology, regulating blood circulation. Cardiovascular Diseases (CVDs), including cardiac arrhythmias, can disrupt the heartbeat rhythm, impacting blood circulation. Black-box computational modeling of this system can facilitate the development of novel methods and devices to assist in diagnosing and treating CVDs. Artificial Neural Networks (ANNs) represent an effective black-box approach. Implementation involves selecting a database, separating training and test sets, and defining the model structure. The MIT-BIH database is commonly utilized to train computational models to detect cardiac arrhythmias. However, preliminary results with the ANN model trained using MIT-BIH data failed to meet the expected objectives, presenting numerous challenges. Nonetheless, given its nascent stage, there remains potential for optimizations, rendering it a prospective tool for diagnosing cardiac arrhythmias.