Predicting refibrillation from pre-shock waveforms in optimizing cardiac resuscitation

Ventricular fibrillation (VF) is a lethal cardiac arrhythmia that if untreated within minutes of its occurrence will lead to sudden cardiac death. Defibrillation using electric shocks is the only choice of treatment to restore the heart to normal rhythm especially in out-of-the-hospital VF incidents. Refibrillation (i.e., recurrence of VF) is a common and significant problem in cardiac resuscitation as it negatively impacts the survival rates. In such refibrillation cases administration of anti-arrhythmic drugs could improve the shock outcomes or prevent refibrillation. In cases of prolonged VF, cardio pulmonary resuscitation (CPR) prior to the shocks have been shown to improve the survival rates. The proposed work using wavelet analysis of the pre-shock VF electrograms attempts to predict the shock outcomes as successful, refibrillation, and unsuccessful categories. This feedback in real-time would be of immense assistance to the Emergency Medical Services (EMS) personnel in choosing the right combination of therapies (i.e., shock, CPR, pharmacology interventions) in improving the shock outcomes. Using a real-word database of 34 pre-shock VF electrograms obtained from Toronto area EMS personnel, the proposed method achieved classification accuracies of 76.5% and 75% for a two level binary classification of the three groups.