Explainable Artificial Intelligence for Diagnosis of Cardiovascular Disease

Explainable Artificial Intelligence for Diagnosis of Cardiovascular Disease

Megha Bhushan, Abhishek Kukreti, Arun Negi
DOI: 10.4018/979-8-3693-2141-6.ch007
OnDemand:
(Individual Chapters)
Available
$37.50
No Current Special Offers
TOTAL SAVINGS: $37.50

Abstract

Cardiovascular disease (CVD) is among the top causes of mortality in today's world; according to the World Health Organisation (WHO), 17.9 million individuals worldwide have died from this illness, leading to 31% of all fatalities. Through early detection and alteration in lifestyle, more than 80% of deaths due to CVD can be avoided. The majority of CVD cases are identified in adults; however, the risk factors for its beginning develops at a younger age. Various machine learning and deep learning algorithms have been utilized to diagnose and predict different types of CVDs, resulting in the development of sophisticated and efficient risk classification algorithms for every patient with CVD. These models incorporate explainability modalities which can improve people's comprehension of how reasoning works, increase transparency, and boost confidence in the usage of models in medical practice. It can help in optimising the frequency of doctor visits and carrying out prompt therapeutic along with preventative interventions against CVD occurrences.
Chapter Preview
Top

1. Introduction

Cardiovascular Diseases (CVD) are illnesses which impact the heart and arteries of human beings (Singh & Bhushan, 2022). These illnesses may affect one or more parts of the cardiovascular system and/or the vessels that carry blood. A person may be asymptomatic (not feeling anything at all) or symptomatic (physically experiencing the disease). As per the reports, CVDs caused the deaths of 17.9 million people worldwide in 2019, which represents 32% of the total number of fatalities and also, 85% of these deaths were triggered by heart attacks or strokes. The major cause of mortality (42.1%) attributed to CVD in 2019 was Coronary Heart Disease (CHD), and followed by Heart Failure (HF) (9.6%), high blood pressure (11.0%), stroke (17.0%), disorders of the arteries (2.9%), and other CVD (17.4%) (Salah & Srinivas, 2022). Also, cardiovascular ailments can be avoided by managing behavioural risk factors such as cigarette use, food habits and obesity, inactivity, and problematic alcohol intake. Figure 1 depicts the data of deaths caused by various chronic diseases in previous years (Elflein, 2022).

Figure 1.

Deaths caused by various chronic diseases

979-8-3693-2141-6.ch007.f01
Top

2. Types Of Cardiovascular Diseases

Cardiovascular illnesses can take many different forms, which includes but is not restricted to (Bhushan et al., 2023):

Arrhythmia: Arrhythmia is an irregular heartbeat in which the heart beats too quickly, too slowly, or in an aberrant manner. It happens when the electrical impulses that control the heart's beat are interrupted (Mayo Clinic, 2023a). The heart possesses a unique electrical structure that coordinates contractions and ensures appropriate blood flow through the body. Heart injury, cardiovascular disease, hypertension, imbalances in electrolytes, stress, medicines, and certain substances such as alcohol, caffeine, or narcotics can all cause arrhythmia. Some arrhythmia has no known origin and can occur in individuals who are otherwise fit.

Heart failure: When the heart's muscles lack the ability to pump blood as effectively as it should, HF occurs. This frequently causes shortness of breath because blood regularly backs up and liquid accumulates in the lungs. Some heart problems can cause the heart to stiffen or become weak enough to fill up blood and circulate it adequately. These conditions include high blood pressure and heart artery narrowing (Mayo Clinic, 2023b).

Stroke: A stroke, also referred to as a Cerebrovascular Accident (CVA), is a health condition that takes place when the blood supply to a portion of the brain is interrupted or diminished, resulting in a lack of nutrients and oxygen to brain cells. Brain cells start to deteriorate with time (in min). It is a serious medical condition, thus obtaining care as soon as possible is critical (Mayo Clinic, 2022). Early intervention can reduce issues and brain damage. Stroke disabilities can be avoidable with the use of appropriate treatments.

Atrial fibrillation: Atrial Fibrillation (AF) is a kind of cardiac arrhythmia that affects the atria, or upper chambers of the heart (Mayo Clinic, 2021). The electrical impulses that govern the heart's beat become disordered in AF, resulting in a fast and irregular beating. Age, underlying heart illnesses such as high blood pressure, heart valve difficulties, HF, coronary artery disease, heart defects that are congenital, lung disorders, hyperthyroidism, and excessive drink or stimulant usage are all possible causes of AF.

Cardiac arrest: Cardiac arrest is an unexpected and sudden loss of cardiac function in which the heart ceases to properly pump blood. It is a health concern, if not addressed immediately, can be fatal. An arrhythmia, or electrical abnormality in the heart, is frequently the cause of cardiac arrest (Cleveland Clinic, 2022). The most prevalent arrhythmia linked to sudden cardiac arrest is Ventricular Fibrillation (VF), a condition in which the electrical signals of the heart become disorganized, resulting in inefficient pumping.

Top

3. Techniques Used

Some of the techniques used in existing works are as follow:

Key Terms in this Chapter

Explainable Artificial Intelligence: It refers to the ability of an AI system to provide clear and understandable explanations regarding its decisions or predictions.

Deep Neural Networks: A subgroup of ML models, motivated by the composition and operation of the human brain.

Deep Learning: By using neural networks to simulate human intellect, these algorithms seek to produce results that resemble those of the model.

SHapley Additive exPlanations: It is used to explain ML model output.

Decision Tree: The model divides the sample according to the tree's level, with the leaf nodes serving as the location for determining the target variable's mean and standard deviation.

eXtreme Gradient Boosting: It is a ML technique that falls under ensemble learning renowned for its effectiveness, speed, and top performance across a range of ML applications, especially those involving structured or tabular data.

Complete Chapter List

Search this Book:
Reset