Background-Cardiopulmonary resuscitation (CPR) creates artifacts on the ECG and, with automated defibrillators, a pause in CPR is mandatory during rhythm analysis. The rate of return of spontaneous circulation (ROSC) is reduced with increased duration of this hands-off interval in rats. We analyzed whether similar hands-off intervals in humans with ventricular fibrillation causes changes in the ECG predicting a lower probability of ROSC. Methods and Results-The probability of ROSC after a shock was continually determined from ECG signal characteristics for up to 20 seconds of 634 such hands-off intervals in patients with ventricular fibrillation. In hands-off intervals with an initially high (40% to 100%) or median (25% to 40%) probability for ROSC, the probability was gradually reduced with time to a median of 8% to 11% after 20 seconds (PϽ0.001). In episodes with a low initial probability (0% to 25%; median, 5%), there was no further reduction with time. Key Words: cardiopulmonary resuscitation Ⅲ defibrillation Ⅲ electrocardiography Ⅲ heart arrest Ⅲ Fourier analysis E arly defibrillation provides the single best option for restoring spontaneous circulation (ROSC) in patients with ventricular fibrillation (VF). 1 With automated external defibrillators, a "hands-off" interval without chest compressions must occur during the period of rhythm analysis and capacitor charging before the delivery of an electric shock. 1 Sato et al 2 reported a reduced rate of ROSC with increased duration of this interval without myocardial perfusion in rats, and they concluded that minimal hands-off delays should improve the effectiveness of automated defibrillators. Conclusions-TheWe wanted to evaluate whether the probability for ROSC is reduced during a hands-off interval in patients. This cannot be studied by giving multiple shocks during the interval and recording the ROSC rate, because by definition this is a no-shock interval. A predictor for ROSC must therefore be used, and some ECG signal characteristics during VF correlate with defibrillation success rate. [3][4][5][6][7] We developed a factor indicating the percent chance of ROSC after the shock, P ROSC (v), that is based on VF signal characteristics from 868 defibrillation attempts in 156 patients. 8,9 In the present study, we studied possible changes in the ECG during hands-off intervals in patients with out-of-hospital cardiac arrest. MethodsIn an observational, prospective study from Oslo, Norway, 10 we used data registered in the Heartstart 3000 defibrillator (Laerdal Medical) together with regular Utstein registration of prehospital cardiac arrest data. 11 We evaluated the rhythm before and after the delivery of each defibrillation shock. Details have been described elsewhere. 10 The Regional Committee for Research Ethics and the Norwegian Data Inspectorate approved the study.The hands-off intervals ( Figure 1) preceding 868 shocks (of which 87 caused ROSC and 781 failed to cause ROSC) in 156 patients were determined from the cardiopulmonary resuscitation (CPR) artifa...

The ECG contained information predictive of shock therapy. This could reduce the delivery of unsuccessful shocks and thereby the duration of unnecessary "hands-off" intervals during cardiopulmonary resuscitation. The low specificity and positive predictive value indicate that other features should be added to improve performance.

We have made an important step toward making classification of resuscitation rhythms more efficient in the sense of minimal feedback from human experts.

Early recognition of ventricular fibrillation (VF) and electrical therapy are key for the survival of out-of-hospital cardiac arrest (OHCA) patients treated with automated external defibrillators (AED). AED algorithms for VF-detection are customarily assessed using Holter recordings from public electrocardiogram (ECG) databases, which may be different from the ECG seen during OHCA events. This study evaluates VF-detection using data from both OHCA patients and public Holter recordings. ECG-segments of 4-s and 8-s duration were analyzed. For each segment 30 features were computed and fed to state of the art machine learning (ML) algorithms. ML-algorithms with built-in feature selection capabilities were used to determine the optimal feature subsets for both databases. Patient-wise bootstrap techniques were used to evaluate algorithm performance in terms of sensitivity (Se), specificity (Sp) and balanced error rate (BER). Performance was significantly better for public data with a mean Se of 96.6%, Sp of 98.8% and BER 2.2% compared to a mean Se of 94.7%, Sp of 96.5% and BER 4.4% for OHCA data. OHCA data required two times more features than the data from public databases for an accurate detection (6 vs 3). No significant differences in performance were found for different segment lengths, the BER differences were below 0.5-points in all cases. Our results show that VF-detection is more challenging for OHCA data than for data from public databases, and that accurate VF-detection is possible with segments as short as 4-s.