A pediatric shock advice algorithm based on the regularity of the detected beats

Abstract

Automated external defibrillators (AED) detect fatal ventricular arrhythmias: ventricular fibrillation (VF) and ventricular tachycardia (VT). We have developed an algorithm based on the regularity of the detected beats to accurately discriminate VF from nonshockable rhythms in pediatric patients.The beat detection method is based on a preprocessing band pass filter (5-35 Hz) followed by a nonlinear energy operator (NEO). The discrimination algorithm uses three parameters: the number of detected beats, the coefficient of variation of the interval between beats and the content around the zero line of the output of NEO. The values of these parameters were used in a decision tree that identified irregular shockable rhythms (VF), and slow and fast regular rhythms, classified as nonshockable. VT was excluded in the design of the algorithm because it is often a regular but shockable rhythm. The algorithm was tested on a database of 1091 records (959 nonshockable, 62 VF and 70 VT) from 650 pediatric patients. The specificity was 99.7% and the VF sensitivity was 96.6%. 33% of the VT windows were identified as shockable, 65.2% as fast nonshockable and 1.8% as slow nonshockable. The regularity of the detected beats can accurately discriminate VF from nonshockable rhythms. However, an additional stage to discriminate fast nonshockable rhythms from fast and regular VT is needed for a shock advice algorithm.