Heart Disease Prediction Framework Using Soft Voting-Based Ensemble Learning Techniques

Introduction

The heart is the most important complex organ. In a nutshell, it controls blood circulation inside our bodies. Any cardiac abnormality might induce agony in plenty of other parts of the body (Sivabalakrishnan, 2019). Cardio Vascular Disease (CVD) is defined as any impairment in the regular beating of the heart. Coronary artery disorder is one of the leading causes of mortality in modern society. The myocardial disease can be caused by a sedentary lifestyle, smoking, drinking, and saturated calorie consumption, all of which can lead to hypertension (Dutta et al., 2020). As per the WHO survey, greater than 15 million people worldwide die each year because of heart problems. The latest WHO report published in 2020, 59.8 million fatalities worldwide transpired in 2018 because of myocardial infarction. Cardiovascular disease claimed the lives of 20.6 million individuals in 2015 (Anitha & Sridevi, 2019). Data gathering has been indicated by the WHO as having the opportunity to assist and diagnose the beginning stages of cardiac disease and deliver proper illness solutions. The key to avoiding cardio ailments is to maintain good lifestyle habits. CVDs include heart diseases, vascular diseases of the brain, and blood vessel diseases. One of the predominant sources of life-threatening impediments and fatality is cardiac disease. Heart disease treatment and therapy are highly challenging, especially in developing nations, due to a lack of effective diagnostic instruments, medical specialists, and other resources, all of which impede patient prognosis and treatment. The main contributing factors are insufficient preventive measures and a scarcity of trained or unpracticed medical personnel in the discipline (Latha & Jeeva, 2019). Although a large portion of heart disease is preventable, the number of instances continues to rise due to a shortage of preventive interventions. In today's digital era, the number of impersonal recommendations assist methods for cardiac disorder detection has been expanded by various researchers to facilitate and guarantee a good prognosis (El-Hasnony et al., 2022).

Figure 1.

Different factors affecting heart disease

The conduction system of the heart is controlled by a set of nodes, valves, and neurotransmitters in your atrium. Each time this happens, electric impulses are transmitted through it. These signals cause individual

portions of your heart to grow and constrict. Using different machine learning models implemented with python, this study aims at predicting how well the heart is functioning (Mohan et al., 2019). The results of an electrocardiogram (ECG) test tell us how well the heart is functioning by tracing the electrical activity within the heart. The electrical pulse normally travels from the sinoatrial node (Repaka et al., 2019), represented by the P wave, across the atrium, to the atrioventricular node, and finally through the ventricular septum, indicated by the SRQ path. The curve in T will normally follow the direction of the QRS complex. When it occurs in the opposite direction, it reflects cardiac disease (Dwivedi, 2018).

Recently, many research works are studied AI-based solution for prediction of heart diseases. (Ali et al., 2019) These existing schemes suffer from poor accuracy. Thus, this paper has focused on design of an effective framework for efficient prediction of heart diseases using a novel soft voting-based ensemble learning techniques. The main contribution of this paper is pointed out as follows:

• 1.

  Design of an effective heart disease prediction framework using soft voting-based ensemble learning technique.

• 2.

  Performance evaluation of the proposed model and its comparison with the state-of-art models is done using a benchmark dataset in terms of accuracy, precision, sensitivity, specificity, and F1-score.