Abstract 19179: Assessing Prehospital Resuscitation Quality Using Empirical Mode Decomposition Analysis of Automatic External Defibrillator Records
Methods: AED tracings during prehospital CPR sessions were obtained and digitized. An algorithm to automatically identify and reconstruct rhythms associated with chest compression from these AED tracings was developed using empirical mode decomposition (EMD). In the model, each mode of oscillation, termed intrinsic mode functions (IMFs), was decomposed sequentially from the original time series by a sifting process. The significant fluctuations associated with adequate pressing pressure could be reconstructed from the dominant IMF. Quality of CPR parameters in terms of no-flow time, no flow fraction, average compression rates, time of inadequate CPR (rate < 90 or > 120 /min) were calculated according to uniformed reporting template proposed by Kramer-Johnson. Quality of CPR parameters of the same tracings were assessed by experienced clinicians manually through reviewing the audio and graphic recordings of the CPR sessions.
Results: The first 5 minutes of prehospital CPR sessions among 50 asystole AED tracings underwent quality assessment by both EMD algorithm and manual review. Among the four quality parameters, the correlation between EMD algorithm vs. manual analysis were high for average compression rates (r=0.87, p=0.000), and time of inadequate CPR (r=0.79, p=0.000); and fair for no flow time (r=0.45, p=0.002) and no flow fraction (r=0.30, p=0.047).
Conclusions: Automatic algorithm analysis of widely available AED tracings provide a potential tool for assessing prehospital CPR quality.
- © 2010 by American Heart Association, Inc.