Hybrid Approach for Prediction of Cardiovascular Disease Using Class Association Rules and MLP

Abstract

:  In data mining classification techniques are used to predict group membership for data instances. These techniques are capable of processing a wider variety of data and the output can be easily interpreted. The aim of any classification algorithm is the design and conception of a standard model with reference to the given input. The model thus generated may be deployed to classify new examples or enable a better comprehension of available data.  Medical data classification is the process of transforming descriptions of medical diagnoses and procedures used to find hidden information. Two experiments are performed to identify the prediction accuracy of Cardiovascular Disease (CVD).A hybrid approach for classification is proposed in this paper by combining the results of the associate classifier and artificial neural networks (MLP).  The first experiment is performed using associative classifier to identify the key attributes which contribute more towards the decision by taking the 13 independent attributes as input. Subsequently classification using Multi Layer Perceptrons (MLP) also performed to generate the accuracy of prediction using all attributes. In the second experiment, identified key attributes using associative classifier are used as inputs for the feed forward neural networks for predicting the presence or absence of CVD.