AN UP-DOWN BAYESIAN, DEFIBRILLATION EFFICACY ESTIMATOR

In both the clinic and the laboratory, efficacy estimators are used to estimate the shock strength required to achieve a given defibrillatio n success rate. In the clinic, efficacy estimators are used to estimat e highly effective doses (i.e., the shock strength that defibrillates 95% of the time), in order to choose the setting for an ICD. Effcicacy estimators are used in the laboratory to compare defibrillation techn iques and configurations. Current efficacy estimators are inadequate b ecause they are either difficult to use, can only estimate the shock s trength that defibrillates 50% of the time, or do not yield desirable accuracy (low RMS error). This article presents a Bayesian estimation technique that forces the difference between successive test shock str engths (step-size) to be a fixed value after each measurement. Constra ining the difference dramatically reduces the computational complexity of the up-down Bayesian method. This new up-down Bayesian protocol ca n be used with up to 15 measurements to estimate the shock strength fo r any given success rate. Simulations show that the added constraint ( fixed step-size) only slightly increases the rms error, as compared to the optimum Bayesian protocol. Our simulations also show that protoco ls can be generated for shock strengths rounded to the nearest 1, 10, or 50 V, without a great increase in RMS error. Experimental results f rom a subset of all the simulations are reported from six animals, sho wing a < -2.4% difference between the simulated and measured errors.