Classification of the 12-lead electrocardiogram employing a framework of bi-group neural networks

A framework employing bi-group neural networks (BGNNs) is proposed to classify an unknown 12-lead electrocardiogram (ECG) into one from a possible six diagnostic classes. The framework was compared with a conventional approach of neural network classification, a decision tree and a classifier based on multiple regression. The proposed approach attained a correct classification level of 80% in comparison with 68%, 66.7% and 68% for the other methods, respectively