Role of Feature Selection in Building High Performance Heart Disease Prediction Systems

In the last few years, there has been a tremendous rise in the number of deaths due to heart diseases all over the world. In low- and middle-income countries, heart diseases are usually not detected in early stages which makes the treatment difficult. Early diagnosis can help significantly in preventing these diseases. Machine learning-based prediction systems offer a cost-effective and efficient way to diagnose these diseases in an early stage. Research is being carried out to increase the performance of these systems. Redundant and irrelevant features in the medical dataset deteriorate the performance of prediction systems. In this paper, an exhaustive study has been done to improve the performance of the prediction systems by applying 4 feature selection algorithms. Experimental results prove that the use of feature selection algorithms provides a substantial increase in accuracy and speed of execution of the prediction system. The prediction system proposed in this study shall prove to be a great help to prevent heart diseases by enabling the medical practitioners to detect heart diseases in early stages