Machine Learning-Based Application for Long-Term Electrocardiogram Analysis

Abstract

Electrocardiogram (ECG) analyses can only be performed by health professionals whose demand for care is often greater than the availability. In this context, this work consists of the development of an application capable of processing long-lasting ECG signals to assist health professionals in making decisions. The application has an interactive interface that allows view the entire ECG signal in a single image generated by all overlapping cardiac cycles. The proposed application still has email communication between users with the objective of facilitating patient follow-up. The application was tested on three different ECG signals, one artificial and two real. The first signal was an artificial signal generated in software Matlab. The second ECG signal has normal sinus rhythm, available in the MIT-BIH normal sinus rhythm database. The third ECG sign diagnosed with arrhythmia can be found in the MIT-BIH arrhythmia database. The results obtained by the proposed method can be used to support decision-making in clinical practice.