Effective Diagnosis of Heart Disease Using Multilayer Feed Forward Neural Network using Back Propagation Algorithm and Associative Neural Network.

Cardiovascular disease is common and prevalent in most human community and is vital reasons of morbidity and mortality. It still remains a dreadful disease, hence speedy and accurate diagnosis of Heart Disease (HD) is essential. Artificial neural network provides a powerful tool to help the medical professionals to clinically analyze the disease and get the best solutions. The aim of this paper is to make best use of Multilayer Feed Forward Neural Network (MLFFNN) using Back Propagation (BP) algorithm and Associative Neural Network (ASNN) for the effective diagnosis of Heart Disease. The MLFFNN is configured with 13, 20 and 1 neurons in the input, hidden and output layer. The ASNN is configured with 13, 10 and 1 neurons in the input, hidden and output layer. 270 datasets are used for training. The performance of both Neural Networks is analyzed in training and predicting the heart diseases. MLFFNN predicts heart disease with the excellent correlation co-efficient (R 2 ) = 0.9410 and 0.9999 for training and testing. ASSN provides satisfactory results with the correlation co-efficient (R 2 ) = 0.9999 for training and 0.9460 for testing. The results are cross validated by Leave-One-Out (LOO) procedure. The performance of MLFFNN is compared with ASNN. MLFFNN predict heart disease with better accuracy than ASNN for test dataset. The results so obtained have proved that the intricate and significant task of heart disease prediction can be carried out precisely and effectively using this automated medical diagnosis system.