A
system and method based on electrocardiogram (ECG) complexity analysis for real-time detecting shockable
ventricular fibrillation (VF) and
ventricular tachycardia (VT), and discriminating them from non-shockable tachyarrhythmia (e.g.
supraventricular tachycardia (SVT) and
atrial fibrillation (AF)) and high-
frequency noise. In the disclosed invention, complexity measure CM (0 to 100), quantitatively characterizing the complexity nature of the non-linear dynamics underlying
cardiac arrhythmia, is extracted from the sensed patient
ECG signal using ECG complexity analysis. From the calculated complexity measure, by three thresholds (
low complexity threshold (LCT), mediate complexity threshold (MCT), and
high complexity threshold (HCT)), different kinds of tachyarrhythmia (i.e.
heart rate (HR) above a preset rate threshold) and high-
frequency noise are discriminated from each other: for non-shockable tachyarrhythmia, CM<=LCT; for VT, LCT<CM<=MCT; for VF, MCT<CM<=HCT; and for high-
frequency noise, HCT<CM. The disclosed
system and method can be used as a primary cardiac tachyarrhythmia detection scheme or as a
backup system to reconfirm
arrhythmia detection using conventional techniques.