Texas Tech University researchers have developed a way to automatically diagnose epilepsy with an accuracy rate of 94 percent, by training a neural network to recognize the characteristic patterns in EEG data that indicate the patient is epileptic.