Methods and apparatus for classifying cardiac events with an implantable

Abstract

Methods and apparatus are provided for classifying cardiac events with an implantable cardiac device. Cardiac signals are stored in a buffer. When a significant cardiac event is confirmed, a corresponding segment of the stored cardiac signals is processed to generate a set of feature values. The set of feature values is compared to various sets of reference values. Each set of reference values preferably corresponds to a separate type of cardiac event, so that the cardiac device may classify the significant cardiac event by matching its set of feature values to a set of reference values. Feature values include the maximum positive and negative slew rates exhibited by the cardiac signal during the segment, the maximum positive and negative signal amplitudes, the times to reach the maximum positive and negative signal amplitudes, the area under the cardiac signal curve, the area under the cardiac signal curve above zero as compared to the area below zero, and the number of zero crossings made by the cardiac signal. Additional feature values are generated by simultaneously processing the cardiac signals measured along two different vector directions in the heart.