Computer assisted ECG analysis using neural network inference

Abstract

This paper discusses an interactive expert system for ECG interpretation and diagnosis of heart diseases using decision tables and neural networks. A laboratory prototype has been described. A feed forward artificial neural network is proposed to handle uncertainties in data and rules. The neural network is trained using a new optimization based algorithm which is tested for both single/multiple input-output training. The feasibility of self-contained low cost microcomputer based implementation has been established.