Curve fitting defibrillation success rate versus energy

The defibrillation energy versus success rate (SR) plots (DSREs) from ten dogs were fit to three types of curves (linear, exponential, and probit transformed linear). Seven to ten (8.60±0.84: mean±standard deviation) energies were used which span the DSRE curve. Ten episodes per energy were obtained for the outer energies and six for the middle, for a total of 70.0±8.4 episodes. The exponential model fit best (R=0.0944±0.013), followed by the probit model (R=0.926±0.051) and the linear model ( R=0.917±0.057). Points representing the maximum energy for 0% SR and the energy for 80% SR along with their 95% confidence intervals were predicted from the curve fits. The DSRE distribution was then bootstrapped with 100 replications for two to six samples per energy. This technique demonstrated that four samples per energy for a mean total of 34.4 shocks would suffice to reconstruct the DSRE curve with acceptable accuracy