An Autoregressive Model of the BAEP Signal for Hearng-Threshold Testing

A new objective method for hearing threshold testing, parametric modeling, is proposed in this paper. The fifth-order autoregressive models are estimated for the averaged BAEP records at a series of stimulus levels. The normalized prediction error enor is chosen as the hearing threshold measure. For normal hearing subjects, the values of enor are very different between the control condition and the click stimulus. For abnormal patients with different hearing loss problems, the estimated hearing thresholds in terms of enor show a good agreement with the physicians´ assessment.