Analysis of various techniques for ECG signal in healthcare, past, present, and future

Abstract

Cardiovascular diseases are the primary reason for mortality worldwide. As per WHO survey report in 2019, 17.9 million people died due to CVDs, accounting for 32% of all global deaths. Among these, heart attacks and strokes were responsible for 85%, whereas CVDs caused 38% of the premature deaths (under age of 70) affected by non-communicable diseases. The rate of death can be delayed and may be prevented by efficiently analyzing the ECG signals (i.e., captured by a non-invasive method) at the early stage of the disease. QRS complex in ECG provides pivotal information about the heart diseases. Many researchers have analyzed the ECG signal by traditional approach and machine learning methods for identifying the heart disorders. Performance of these techniques depend on accurate detection of different parameters (such as: P-, Q-, R-, S-, T-waveforms, QRS complex duration, R-peak, PR-interval, and RR-interval) from the ECG signals. This review paper provides a detail discussion and comparison of various ECG analysis techniques along with their pros and cons. It summarizes the ECG capturing method, databases available for disease detection & classification, and performance measures used by the researchers. Based on these, a future road map is suggested for real time ECG analysis (for identifying the heart related conditions) captured from the wearable devices and suggested the precautionary steps by the artificial system and experts. This method will help in identifying the co-relation of heart disorders with other body organs (such as: retina and brain parts) by analyzing ECG, fundus image, and magnetic resonance imaging (MRI) of human brain.

Keywords

• ECG
• QRS complex
• Arrhythmia
• Machine learning
• Heart disorder from fundus image