Experimental Study on Heart Disease Prediction Using Different Machine Learning Algorithms

Abstract

Heart disease which can also be referred to as the cardiovascular disease is one of the raising concerns in today’s world. It is one of the major health problems causing death among humans irrespective of the age group and therefore has made it necessary to look into different medical factors that are required to predict the same in advance using the collected historical datasets of various patients. Thus, we have used various machine learning algorithms to predict the potential of person, to suffer from a heart disease with high precision and reliability so that we can admonish the patient in advance and take the required precautionary measures. In this study, an existing heart disease dataset accessible from the UCI Machine Learning Repository is used as the primary dataset. The experimental analysis and comparative study between different algorithms, helps in deciding the best suitable algorithm for the given problem statement by using results obtained which are very competitive and can be used for identification and treatment. The proposed work predicts the probability of heart condition and ranks the patient's risk level based on different supervised learning algorithms such as K-Neighbors, AdaBoost, Gradient Boosting, etc. The test results portray that K-neighbors has the highest accuracy score of 91% when compared with other algorithms.