Detection of abnormal electrocardiograms using a neural network approach

The results of using an artificial neural network system (ANNS) for detection and recognition of abnormal electrocardiograms in heart-rate-variability studies are reported. The ANNS is trained initially by standard abnormal EKG patterns. Once the network has been trained, it detects abnormal EKGs in real time with less than 10% error. In this particular case the bidirectional associative memory model was used for the neural network.<
>