SINGLE-SWEEP ANALYSIS USING AN AUTOREGRESSIVE WITH EXOGENOUS INPUT (ARX) MODEL

Abstract

Single-sweep visual evoked potential analysis would be useful in clinical electrophysiology practice because it would make possible the evaluation of transient phenomena, but recording single-sweep visual evoked potentials is difficult because of the low signal-noise ratio. To increase this ratio we used a filter based on an autoregressive with exogenous input model. We studied a group of 12 diabetic patients matched with a control group of 14 normal subjects. The model, in most cases, allowed us to extrapolate the P100 component from each single sweep of visual evoked potential. The visual evoked potential values obtained by means of averaging were not significantly different in the groups studied, but single-sweep analysis showed different distribution of the P100 component amplitude. The preliminary results of our study evidenced differences in the amplitude and latency distribution of normal and diabetic subjects, thus confirming the power of this new technique and its ability to obtain some information that is masked by the averaging method.