Development of a post-processing algorithm to classify rhythms detected as ventricular tachyarrhythmias by Implantable Cardioverter Defibrillators

Abstract

Implantable Cardioverter-Defibrillators (ICD) detect ventricular tachycardia /fibrillation (VT/VF) using atrial (A) and ventricular (V) electrograms (EGMs). ICD algorithms discriminate VT/VF from supraventricular tachycardias (SVTs), but misclassify some SVTs as VT/VF. Clinicians review detected episodes to identify true SVT episodes and guide appropriate clinical action. A post-processing, expert-system algorithm was developed to classify tachyarrhythmias detected and stored in ICD memory. The algorithm was designed to diagnose rhythms with V EGM and/or timing of A/ V events. Rhythms that did not fulfill the criteria were classified as Unknown. The algorithm was tested using a dataset of 469 episodes. The algorithm correctly classified 80% of the episodes with 99% accuracy. This accuracy may be sufficient that physician review may be required only for Unknown episodes.