A Bayesian estimator of the minimum stimulus strength to induce ventricular fibrillation

The American standard for AC current leakage in the myocardium was recently raised from 10 microamps to 50 microamps to match European standards. A lack of fibrillation (VF) induction data from humans has made at difficult to support or reject the increase. This paper quantifies the amount of information which could be gained from humans during the course of a defibrillator implant by carefully controlling and analyzing the stimulus strengths used to induce VF. The authors propose a new, Bayesian up-down stimulation protocol to accurately estimate an AC stimulus strength with a low probability of inducing VF. This Bayesian protocol balances the conflicting goals of reducing testing and accurate estimation. The authors show that it is possible to estimate the stimulus strength that will induce VF 5% of the time with 3.9 /spl mu/A rms error using 10 VF induction attempts averaging 4.8 VF episodes.