Premium
Automated Ventricular Tachyarrhythmia Recognition
Author(s) -
IGEL DAVID A.,
WILKOFK BRUCE L.
Publication year - 1997
Publication title -
journal of cardiovascular electrophysiology
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.193
H-Index - 138
eISSN - 1540-8167
pISSN - 1045-3873
DOI - 10.1111/j.1540-8167.1997.tb00804.x
Subject(s) - qrs complex , medicine , ventricular tachycardia , cardiology , tachycardia , ventricular fibrillation , discriminant function analysis , electrocardiography , discriminant , pattern recognition (psychology) , artificial intelligence , machine learning , computer science
Automated Ventricular Tachyarrhythmia Recognition. Introduction : Cardiac monitoring devices such as external cardioverter defibrillators and ICU computerized ECG monitoring systems sense individual QRS complexes to detect and subclassify ventricular tachyarrhythmias. Many algorithms that evaluate ECG morphology mark individual QKS complexes so that specific waveform characteristics can he measured. QRS sensors miss a percentage of electrogram events, especially during fibrillatory rhythms; thus, a morphology algorithm that helps subclassify arrhythmias without marking individual electrogram events may be more robust and improve arrhythmia detection accuracy. Methods and Results : Four nonlinear dynamics calculations were evaluated for detecting ventricular tachyarrhythmias and for subclassifying monomorphic ventricular tachycardia (MVT) and polymorphic ventricular tachycardia (PVT). Five‐second epochs of normal (NML, n = 48), MVT (n = 58), and PVT (n = 75) rhythms were presented to a statistical discriminant function based on cycle length (CL), and its performance was compared to two other discriminant functions, one consisting of the nonlinear dynamics calculations and one consisting of a combination of all variables. The discriminant function based on nonlinear dynamics calculations and CL detected 100% of the ventricular tachyarrhythmias, and subclassified more (P < 0.001) MVT and PVT arrhythmias (90%) than that based on CL alone (71%). Conclusions : The nonlinear dynamics measurements used in this study significantly increased the suhclassification accuracy of the CL‐based discriminant function, and they were calculated from ECG signals without marking individual QRS complexes. Therefore, arrhythmia detectors that use nonlinear dynamics measurements may commit fewer classification errors due to QRS undersensing and aid therapy decisions when circumstances suggest that QRS sensing is inaccurate.