Author : J.E.Nalavade, M.L.Gavali, N.D.Gohil and S.C.Jamale

Diagnosis of diseases is an important and intricate job in medicine. The identification of heart disease from diverse features is a no of layered problem that is not free from the wrong assumptions and is frequently accompanied by impulsive effects. Thus to exploit knowledge and experience number of specialists and clinical screening data of patients inserted in databases to assist the diagnosis procedure is regarded as a valuable option. This system work is the extension of our previous system with intelligent and effective heart attack prediction system by using neural network. A professional methodology for the extraction of easiest patterns from the heart disease warehouses for heart attack prediction has presented. Data warehouse is preprocessed in sequence to make it easy for the mining process. Processing gets finished, then heart disease warehouse is clustered with aid of K-means clustering algorithm, which will extract data, appropriate to heart attack from warehouse. The frequent patterns applicable to heart disease are mined with aid of the algorithm from data extracted. The patterns important to heart attack prediction are selected on basis of the significant weight. The neural network is well trained with selected significant patterns for effective heart attack prediction system. We have implemented the Multilayer Neural Network with Back-propagation training algorithm. Results obtained have illustrated that designed prediction system is capable of predicting the heart attack more effectively.

Keywords: Hidden layer, Back Propagation, Data Mining, Artificial Neural Network (ANN), K-means clustering.