Optimum estimation of the 95 percent effective defibrillation dose

Optimum Bayesian algorithms were compared with three currently used approaches to estimating the leading edge voltage which successfully defibrillates 95% of the time (ED95). Two different design methods were used for the Bayesian techniques. Five dogs were investigated using the Bayesian algorithms, the up-down technique, the Bourland method, the step-down method, and, for reference, logistic regression. There was excellent agreement between the computer-predicted and measured performances. Both the computer predictions and the animal experimentation confirmed that the Bayesian method is significantly better than the three currently used methods for estimating the ED95.