AN EFFICIENT CORONARY HEART DISEASE PREDICTION BY SEMI PARAMETRIC EXTENDED DYNAMIC BAYESIAN NETWORK WITH OPTIMIZED CUT POINTS

Dynamic Bayesian Network (DBNs) is the general tool for enhancing the dependencies between the variables evolving in time and it’s used to represent the complex stochastic processes to study their properties or make predictions based on the future behavior. The coronary heart disease (CHD) is considered as the one of the deadliest human diseases worldwide. The accurate prediction of CHD is very complex to be prevented and the treatment for it seems difficult. In early work, the TA methods with DBNs have been applied for the prognosis of the risk for coronary heart disease (CHD). The deviation of temporal abstractions from data is used for building DBN structure to predict CHD. However this approach cannot handle complex temporal abstractions due to irregular time intervals. The cut-off values decided for temporal abstraction is the another issue in this work. In order to overcome this issue in this paper proposed the technique used for regularizing the irregular time interval in extended dynamic Bayesian networks (DBNs) with temporal abstraction for coronary heart disease prediction. The proposed technique provides the global optimal solutions to assure the learning temporal solutions which provide observation of same irregularly spaced time points and the semi parametric subclass of the DBN proposed to allow further adaption of the irregular nature of the available data. The cut off value is searched from the domain expert knowledge base through the firefly optimization algorithm.