Summary of: Elberling, C., & Wahlgreen, O. (1985). Estimation of auditory brainstem response, ABR, by means of Bayesian inference. Scandinavian audiology, 14(2), 89–96. https://doi.org/10.3109/01050398509045928
This paper describes a method to recover the ABR by weighted averaging. The method is an effective technique to deal with the destructive effect of fluctuating, non-stationary background noise, and is based on a statistical approach called ‘Bayesian inference’. The contribution of the individual sweep (or block of sweeps) is weighted inversely proportional to the level of background noise during the acquisition of the sweep. Based on 50 sets of clinical recordings the weighted averaging method is evaluated. Weighted averaging is always as good as or better than traditional averaging, and in about 30% of the cases the weighted averaging improves the recovered ABR significantly over what is obtained by traditional averaging. In these cases the traditional averaging would require 50% more sweeps to be averaged in order to obtain the same precision of the ABR, and the variance of the wave V latency is improved by a factor of approximately two.
Related course: Beyond the Basics: Threshold ABR