Abstract
At present, the number of articles on Heart Disease Detection (HDD) based on classification searched by Google Scholar search engine exceeds 17,000. The medical sector is one of the most important fields that benefit from ML. Heart diseases (HDs) are considered to be the leading cause of death worldwide, as it is difficult for doctors to predict them earlier. Therefore, the HDD is highly required. Today, the health sector contains huge data that has hidden information where this information can be considered as essential to make diagnostic decisions. In this paper, a new diagnostic model for the detection of HDs is on a multi-classifier applied to the heart disease dataset, which consists of 270 instances and 13 attributes. Our multi-classifier is composed of Artificial Neural Network (ANN), Naive Bays (NB), J48, and REPTree classifiers, which select the most accurate of them. In addition, the most effective feature on prediction is determined by applying feature selection using the "GainRatioAttributeEval" technique and "Ranker" method based on the full tainting set. Experimental results show that the NB classifier is the best, and our model yields over 85% accuracy using the WEKA tool.