Page 89 - The-5th-MCAIT2021-eProceeding

Machine Learning in Predicting Cardiovascular Diseases Using

ECG Signal

a
b
a*
Talal A.A. Abdullah , M. Soperi Mohd Zahid , Khaleel Husain
a Department of Computer & Information Science, Universiti Teknologi PETRONAS, Seri Iskandar, 32610, Malaysia.
b Institute of Health and Analytics, Universiti Teknologi PETRONAS, Seri Iskandar, 32610, Malaysia.
* Email: msoperi.mzahid@utp.edu.my

Abstract
Statistics have shown that Cardiovascular diseases (CVDs) are the leading cause of death in Malaysia and worldwide. In
medicine, evidence-based practice is considered the guiding principle of clinical practice integrating individual clinical
expertise and best external evidence in making clinical decisions. However, the demands of diagnoses or generating a
prediction for large, heterogeneous populations is nearly impossible for traditional evidence-based methods to keep up with
the latest trials and studies in healthcare. Machine learning can be used to discover patterns and associations within massive
datasets to assist diagnoses and predict future outcomes. Since the last decades, several methods were reported for automatic
ECG beat classifications. In this work, we present a survey of the current state-of-the-art methods used to detect CVDs using
ECG signals. It includes the feature selection and machine learning approaches used for automatic detection and decision-
making process.

Keywords: Survey; CVD; Healthcare; ECG; Machine Learning.

1. Introduction

Cardiovascular Diseases (CVDs) takes place in the category of fatal diseases resulting in death around the
world (Nazlı, Gültepe, & Altural). According to the world health organization, 17.3 million people died from
cardiovascular diseases (CVDs) each year, which estimates 31 per cent of all death worldwide (Mendis, Puska,
Norrving, Organization, & others, 2011). The rates are remarkably higher in the Middle East, Asia, and Russia
than in the rest of the world (Alizadehsani, Abdar, et al., 2019; Zipes, Libby, Bonow, Mann, & Tomaselli, 2018).
In CVDs, angiography is widely used in cardiology by clinicians as it considered the most precise method (Kim
& Choi, 2015; Kim et al., 2015; Tsipouras et al., 2008). It is, however, an invasive and costly procedure, and it
may lead to various complications (Alizadehsani et al., 2018; Mnih et al., 2015). Electrocardiogram (ECG/EKG)
is another most common diagnostic tool for recording physiological heart activities during a specific period. It
is estimated that over 300 million ECGs are recorded worldwide per year (Holst, Ohlsson, Peterson, &
Edenbrandt, 1999), and the number keeps growing. ECG is a non-invasive and uncostly tool that can aid in
diagnosing many cardiovascular abnormalities, such as atrial fibrillation (AF), premature contractions of the
atria (PAC) or ventricles (PVC), congestive heart failure (CHF), and myocardial infarction (MI) (Hong, Zhou,
Shang, Xiao, & Sun, 2020). However, ECG signals are noisy sensitive and need a human expert to understand
what they mean and are most likely to be misunderstood. Therefore, computeraided interpretation of ECGs has
become more crucial, especially in low-income and middle-income countries where experienced cardiologists
are scarce (Organization & others, 2014). Although expert features can automatically be extracted using some
computer-based programs, they are still insufficient. The main reason is that they are limited by human expert
knowledge and data quality (Guglin & Thatai, 2006; Schläpfer & Wellens, 2017; Shah & Rubin, 2007).
Therefore, researchers are working on alternative methods to extract ECG signal features that do not require an
explicit feature extraction by human experts such as machine learning and deep learning. Machine Learning
(ML) is tremendously used recently in many areas, such as speech recognition and image processing. The
revolution in industrial technology proves the great success of machine learning and its applications in analyzing

E- Proceedings of The 5th International Multi-Conference on Artificial Intelligence Technology (MCAIT 2021) [76]
Artificial Intelligence in the 4th Industrial Revolution