Abstract P147: A New Criterion for Assessment of Arrhythmia Detection Algorithm Performance for Low Amplitude Signals : Asystole upper Threshold Value
Purpose : The majority of the lower amplitude rhythms encountered during automatic external defibrillator (AED) operations are, by order of decreasing magnitude : coarse VF, fine VF, asystole. As recommended by literature for rhythm recognition algorithms of AEDs : Coarse VF should be detected with more than 90% sensitivity if its amplitude is higher than 200μV. Asystole should be detected with more than 95% specificity. Asystole amplitude level is not defined in current standards. Each manufacturer should specify the amplitude threshold separating fine VF and asystole. The purpose of this study is to present a new way of quantifying the transitions of a VF detector from coarse VF to fine VF and from fine VF to asystole by defining thresholds.
Methods : Ten signals of different amplitudes are obtained using a reference VF signal (1mV median peak to peak) from an ECG simulator. The range of amplitudes from 80μV to 300μV is input to an AED and for each of 1,210 analysis periods, shock advice recommendation is recorded. The plot of sensitivity in function of VF amplitude is built using a logistic regression model. The value of VF amplitude corresponding to a 90% sensitivity is called VF lower threshold (VFLT). The value of asystole amplitude corresponding to a 95% specificity is called asystole upper threshold (AUT).
Results : VF lower threshold and asystole upper threshold are presented (95% Confidence Intervals) : VFLT=202μV (198, 207), AUT=150μV (145, 155) (see figure⇓).
Conclusion : This study proposes a new way of quantifying the transitions of a VF detector from coarse VF to fine VF and from fine VF to asystole by defining respectively VF lower threshold (VFLT) and asystole upper threshold (AUT).