Abstract:
Heart failure is considered as one of the major reasons of death worldwide. The amount of deaths from heart failure surpasses the amount of deaths resulting from any other causes. This point makes heart disease as one of the most dangerous disease resulting in human deaths worldwide. Latest studies have concentrated on the usage of machine learning and data mining techniques to build predictive models that are capable to forecast the occurrence of heart failure. The presence of medical data leads to the need of smart data mining tools in order to extract valuable knowledge. Scientists have been using several statistical analysis and data mining techniques to increase the disease diagnosis accuracy in medicinal healthcare. Numerous researchers have presented various data mining techniques for heart disease diagnosis. Using a single data mining technique shows an acceptable level of accuracy for disease prediction. In recent times, more investigation is carried out in the direction of hybrid models which demonstrate incredible enhancement in cardiac vascular disease prediction accuracy. The purpose of the suggested research is to predict the heart disease in a patient more accurately. Therefore, we have suggested a Decision Support System’sframework to help the medical practitioners, doctors and decision makersto collect and understand information and construct a ground work for effective decisionmaking. The suggested framework will play a constructively significant role in medical field and hence likely to increase the quality of medication.The suggested model will overcome the conventionalperformance limitations by applying an ensemble of three diverse classifiers (SVM, LR and NB). Efficiency of the suggested ensemble technique is examined by comparison of results withnumerous renowned classifiers as well as ensemble methods. The experimental assessment expresses thatthe suggested framework dealt with all sorts of attributes and accomplished greaterprediction accuracy. For analysis, the data sets (Statlog and SPECTF) are collected from UCI data repository [14] and [25].
