Artificial neural networks for the classification of cardiac patient states using ECG and blood pressure data

Abstract

The aim of this paper is to look into the feasibility of using ECG and blood pressure data into a neural network for the classification of cardiac patient states. Both Back Propagation (BP) and Radial Basis function (RBF) networks have been used and a comparison of the performance of the two neural networks has been made. Various parameters extracted from the multimodal data have been used as input to the neural network and the diagnosis is made by classifying the output into three categories viz, Normal, Abnormal and Premature Ventricular Contraction (PVC). A performance comparison of the two neural networks has shown that RBF gives slightly higher classification accuracy compared to BP. The success of the implementation on limited input data has indicated the feasibility of fusing multimodal input data using neural network for better classification of cardiac patient states in an ICU setting.