International Journal on Recent and Innovation Trends in Computing and Communication ISSN: 2321-8169 Volume: 5 Issue: 8 XXX – YYY _______________________________________________________________________________________________ 99 IJRITCC | August 2017, Available @ http://www.ijritcc.org _______________________________________________________________________________________ Prediction of Heart Disease using Machine Learning Algorithms: A Survey Himanshu Sharma, Department of Computer Engineering and Applications, National Institute of Technical Teachers’ Training and Research. hs13867@gmail.com M A Rizvi, Department of Computer Engineering and Applications, National Institute of Technical Teachers’ Training and Research. marizvi@nitttrbpl.ac.in Abstract: According to recent survey by WHO organisation 17.5 million people dead each year. It will increase to 75 million in the year 2030[1].Medical professionals working in the field of heart disease have their own limitation, they can predict chance of heart attack up to 67% accuracy[2], with the current epidemic scenario doctors need a support system for more accurate prediction of heart disease. Machine learning algorithm and deep learning opens new door opportunities for precise predication of heart attack. Paper provideslot information about state of art methods in Machine learning and deep learning. An analytical comparison has been provided to help new researches’ working in this field. Keywords: Machine learning, Heart Disease, Naïve Bayes, Decision Tree, Neural Network, SVM and Deep Learning. __________________________________________________*****_________________________________________________ I. INTRODUCTION Heart disease has created a lot of serious concerned among researches; one of the major challenges in heart disease is correct detection and finding presence of it inside a human. Early techniques have not been so much efficient in finding it even medical professor are not so much efficient enough in predicating the heart disease[3]. There are various medical instrumentsavailable in the market for predicting heart disease there are two major problems in them, the first one is that they are very much expensive and second one is that they are not efficiently able to calculate the chance of heart disease in human. According to latest survey conducted by WHO, the medical professional able to correctly predicted only 67% of heart disease [2]so there is a vast scope of research in area of predicating heart disease in human. With advancement in computer science has brought vast opportunities in different areas, medical science is one of the fields where the instrument of computer science can be used. In application areas of computer science varies from metrology to ocean engineering. Medical science also used some of the major available tools in computer science; in last decade artificial intelligence has gained its moment because of advancement in computation power. MachineLearning is one such tool which is widely utilized in different domains because it doesn’t require different algorithm for different dataset. Reprogrammable capacities of machine learning bring a lot of strength and opens new doors of opportunities for area like medical science. In medical science heart disease is one of the major challenges; because a lot of parameters and technicality is involve for accurately predicating this disease. Machine learning could be a better choice for achieving high accuracy for predicating not only heart disease but also another diseases because this vary tool utilizes feature vector and its various data types under various condition for predicating the heart disease, algorithms such as Naive Bayes, Decision Tree, KNN, Neural Network, are used to predicate risk of heart diseases each algorithm has its speciality such as Naive Bayes used probability for predicating heart disease, whereas decision tree is used to provide classified report for the heart disease, whereas the Neural Network provides opportunities to minimise the error in predication of heart disease. All these techniques are using old patient record for getting predication about new patient. This predication system for heart disease helps doctors to predict heart disease in the early stage of disease resulting in saving millions of life. This survey paper is dedicated for wide scope survey in the field of machine learning technique in heart disease. Later part of this survey paper will discusses about various machine learning algorithm for heart disease and their relative comparison on the various parameter.It also shows future prospectus of machine learning algorithm in heart disease. This paper also does a deep analysis on utilization of deep learning in field of predicting heart disease. II. LITERATURE REVIEW Different researchers have contributed for the development of this field. Predication of heart disease based on machine learning algorithm is always curious case for researchers recently there is a wave of papers and research material on this area. Our goal in this chapter is to bring out all state of art work by different authors and researchers. Marjia Sultana, Afrin Haider and Mohammad ShorifUddin[4] have illustrated about how the datasets available for heart disease are generally a raw in nature which is highly redundant and inconsistent. There is a need of pre-processing of these data sets; in this phase high dimensional data set is reduced to low data set. They also