Cardiopulmonary resuscitation Ppattern evaluation based on ensemble empirical mode decomposition filter via non-linear approaches

Abstract

Out-of-hospital cardiac arrest (OHCA) is a critical cardiac disorder. The OHCA survival rate is still relatively low. Cardiopulmonary resuscitation (CPR) is very essential with the cardiac arrest. This study evaluates a non-linear approximation of the CPR given to patients, especially asystole patients. In order to clean the electrocardiography (ECG) signal which is collected by the automated external defibrillator (AED), the raw signal is filtered using ensemble empirical mode decomposition (EEMD), and the CPR-related IMFs are chosen to be evaluated. Sample entropy (SE), complexity index (CI), detrended fluctuation algorithm (DFA) and statistical analysis using Anova are utilized. The CPR evaluation compares the patient survival rates after two hours of the cardiac arrest. The CPR pattern of the 951 asystole patients are analyzed. In the CPR-related IMFs peak-to-peak interval analysis, for both classes, patient groups who are younger than or older than 60 years, does not have any significance. Furthermore, the amplitude difference evaluation, both classes do not have any significant difference for SE (p = 0.28) and DFA (p = 0.92) except for the CI (p = 0.028). The results show that patients group aged younger than 60 years have higher survival rate with high complexity of the CPR-IMFs amplitude differences.