A Bayesian up-down estimator of the 80% effective defibrillation dose

An implantable cardioverter-defibrillator (ICD) can prevent sudden cardiac death. However, the ICD stimulus strength must be set appropriately for each patient. Current protocols for estimating the appropriate stimulus strength require many attempts at defibrillation (VF episodes), and/or estimate relatively ineffective doses, such as the ED50 (the 50% effective defibrillation dose). A new optimum, Bayesian up-down procedure is presented which accurately estimates the ED80 (the 80% effective defibrillation dose) using only four VF episodes. Because the ED80 is more effective than the ED50, it should be a better clinical tool for setting the stimulus strength of an ICD. Data from 23 humans show that stimuli given at the Bayesian up-down estimated ED80 were effective 79.3% of the time. These results suggest the Bayesian up-down protocol is an accurate means of estimating the ED80 in humans