Analysis of patient data to explore cardiovascular risk factors

Abstract

According to the World Health Organisation, cardiovascular diseases claim over 17.9 million lives yearly on a global scale. Hence, cardiovascular diseases are responsible for 32 percent of global deaths yearly. Furthermore, it is estimated that more than 50 percent of heart disease cases are only discovered after they have reached the critical stage of heart failure and stroke. However, early detection of these heart diseases can reduce the mortality rates of cardiovascular diseases. Scientists have suggested using machine learning algorithms to identify the risk factors. However, the unavailability of data has hindered the significant success of this approach. In this study, machine learning algorithms are used to identify the important features that should be monitored to prevent heart diseases by considering a dataset obtained from 1000 patients. The six machine learning algorithms used for this study are Logistic Regression, Support Vector Machine, k-nearest Neighbour, Decision Tree, Random Forest and Multi-layer Perception Classifier. The dataset consists of twelve features that are considered to be associated with heart disease and a target variable. The results from this study show that patients suffering from typical and atypical angina chest pain are prone to heart disease. Patients who exercise up the slope have a higher likelihood of living without heart disease. Among the six algorithms used, the MLP Multi-layer Perception Classifier outperforms all others by achieving a 99 percent accuracy. Moreover, the Random Forest algorithm follows with an accuracy of 98 percent.