A survey on Machine Learning Classifiers and Big data for Accurate and Reliable Heart Disease Pre-diagnosis

Abstract

Since a decade, emergence of interdisciplinary computer technologies changed the pace of medical diagnosis systems by insisting up-to-date intelligence and supervision. These intellectual systems predict the future health problems by processing the current health information of patients, which helps in prevention of diseases rather than cure. Although the medical diagnosis systems are adequate intelligent in disease diagnosis, but they are still suffering in pre-diagnosis of diseases due to the complexity in processing of huge medical datasets. Recently introduced Data Mining techniques with Big Data processing environment expanded the horizons of medical diagnosis systems to process the high velocity medical data sets to diagnose the occurrence of diseases early. Today’s medical diagnosis systems, which are utilizing different data mining techniques like Decision Trees (DT), Support Vector Machines (SVM), Naïve Bayes (NB), Fuzzy Logics and K-Nearest Neighbor (KNN), are suffering from uncertainty, imprecision and complexity in processing. In this paper we are proposing a cross reference methodology to improvise the reliability and precision of diagnostic results and utilizing big data tools to diminish the complexity in processing huge sets of medical data. Most popular data mining techniques, which are participating in medical data processing with high accuracy are selected and cross referenced by our proposed framework to overcome uncertainty and imprecision. In order to process the high velocity medical datasets with several data mining techniques, this frame work outsources the data processing business to Apache Hadoop environment. Experiments on Cleveland medical dataset proved that the proposed cross reference methodology framework recorded high precision, recall and accuracy in results than its counterparts.