A neural network system to classify simulated ECG rhythms

A neural network system to detect abnormal ECG rhythms is developed. An electronic ECG simulator is used to generate normal and abnormal ECGs, for training and testing the system. The Neural Works Professional II Plus neural network building package was used for this purpose. Also, the performance and efficiency of the neural network for ´delta rule backpropagation´ and ´Widrow-Hoffman rule´ is compared. The delta rule is found to be more efficient in classifying data not encountered in its training phase.