Analysis of Machine Learning Algorithms in Health Care to Predict Heart Disease

Analysis of Machine Learning Algorithms in Health Care to Predict Heart Disease

P. Priyanga, N. C. Naveen
DOI: 10.4018/978-1-5225-8185-7.ch010
(Individual Chapters)
No Current Special Offers


This article describes how healthcare organizations is growing increasingly and are the potential beneficiary users of the data that is generated and gathered. From hospitals to clinics, data and analytics can be a very powerful tool that can improve patient care and satisfaction with efficiency. In developing countries, cardiovascular diseases have a huge impact on increasing death rates and are expected by the end of 2020 in spite of the best clinical practices. The current Machine Learning (ml) algorithms are adapted to estimate the heart disease risks in middle aged patients. Hence, to predict the heart diseases a detailed analysis is made in this research work by taking into account the angiographic heart disease status (i.e. ≥ 50% diameter narrowing). Deep Neural Network (DNN), Extreme Learning Machine (elm), K-Nearest Neighbor (KNN) and Support Vector Machine (SVM) learning algorithm (with linear and polynomial kernel functions) are considered in this work. The accuracy and results of these algorithms are analyzed by comparing the effectiveness among them.
Chapter Preview


In the present world there are numerous logical innovations which are not precise but help the specialists in taking clinical choice. Heart disease prediction framework can help therapeutic experts in anticipating condition of heart, in light of the clinical information of patients nourished into the framework (Maglogiannis, Loukis, Zafiropoulos & Stasis, 2009). Around the world 12 million deaths happen consistently because of the heart sicknesses and this has been evaluated by the World Health Organization. Due to cardio vascular illness a large portion of the deaths happen and is creating nations the most important motivation in research. The various infections that influence the heart are encompassed by the term heart disease. In general, it is viewed as the essential explanation for deaths in grown-ups (Kang, Li & Wang, 2013). In the diverse nations including India, heart disease is the real reason for setbacks. In the United States it kills one individual at regular intervals.

The expression of cardiovascular sickness normally integrates the multiplicity of situation that manipulate the heart and the veins the way in which blood is pumped and travel through the body. Cardio Vascular Disease (CVD) is a serious ailment, inability and passing while Coronary Heart Disease (CHD) can take place by the reduction of blood and oxygen supply to the heart (Feng, Zhang, Chen, Hua & Ren, 2015). The CHD includes the myocardial areas of dead tissue, heart assaults and angina pectoris, or trunk agony. A heart assault occurs because of a sudden blockage of a coronary corridor because of blood coagulation. The trunk torments is the deficiency in heart muscle obtained by the blood. The different types of cardiovascular infection (Tay, Poh & Kitney, 2015) are hypertension, coronary supply route malady, valve coronary illness, stroke, or rheumatic fever/rheumatic coronary illness.

Currently, the seriousness of coronary illness in patients is analyzed by the use of four fundamental strategies like Trunk X-beams, coronary angiograms, electro cardiograms otherwise called ECG and exercise stretch tests. Regarding diagnosing coronary illness and sparing the lives of patients, time and symptomatic exactness at early stages are exceptionally essential (Yanqin Bai, Xiao Han, Tong Chen, & Hua Yu, 2015). In some cases, specialists may neglect to take precise choices while diagnosing the coronary illness of a patient. Along these lines coronary illness forecast frameworks which utilize ML calculations aid to get exact results. Early identification of coronary illness helps doctors in deciding the most viable treatment and improves the survival of patients (Mona Hafez, 2012).

To outline and break down any sort of datasets ML comprises of an immense number of calculations and can be either supervised or unsupervised. In supervised learning, information is prepared and anticipated in light of the preparation of dataset with a capacity to make view of preparing and testing uncertain examples. In unsupervised learning, framework stays untrained and unlabeled (Nambi et al., 2010).

Computational model based frameworks, created by utilizing ML procedures are presently considered as an extremely helpful method to foresee and analyze numerous illness. All around characterized ML techniques are adequate to fit successfully and proficiently to anticipate the disease (Antiochos, Marques-Vidal, McDaid, Waeber & Vollenweider, 2016). As contrast with conventional technique ML models needn't bother with profound information of insights of data. SVM, Naïve Bayes, Decision Tree and Artificial Neural Network (ANN) are classifier models of ML methods which are generally utilized as a part of medicinal services. Neural Network work successfully on large datasets and parameters that can subsequently be used to gauge illness or disease. In the meantime, SVM has turned out to be one of the best classifiers for making forecasts (Skretteberg et al., 2012) in the field of healthcare.

Complete Chapter List

Search this Book: