Heart Disease Prediction Using Hybrid Machine Learning Model Based on Decision Tree and Neural Network

Abstract

Cardiovascular disease is the leading cause of death in the world. Nowadays, tremendous amount of data is collected on heart disease. Investigating the data and obtaining insight using data mining can improve the detection and prevention rate, especially in early stages. So far, many researches are performed on data mining models for diagnoses. In this paper, we intend to present a model for the diagnosis of heart disease using a feature-based approach as a preprocessing step. The proposed solution include four main steps as preprocessing the data, selecting effective features, clustering by using the K-Means algorithm and proposing a hybrid model of decision tree and neural network to determine the disease. In selecting the effective features, we use three methods as Pearson correlation coefficient, information gain, and component analysis. The evaluation results confirm that the proposed hybrid model outperforms the existing methods by 0.97 accuracy.