Risk prediction of heart disease based on swarm optimized neural network

Abstract

Heart disease (HD) remains the biggest cause of deaths worldwide. This shows the importance of HD prediction at the early stage. In this paper, multi-layer feedforward neural network (MLFFNN) optimized with particle swarm optimization (PSO) is adopted for HD prediction at the early stage using the patientâ€TMs medical record. The network parameters considered for optimization are the number of hidden neurons, momentum factor, and learning rate. The efficiency of the PSO optimized neural network (PSONN) is calculated using the records collected from standard Cleveland database and Real time clinical dataset. The results show the proposed system can predict the likelihood of HD patients in a more efficient and accurate way.