Bootstrap Prediction and Confidence Bands for Frequency Response Functions in Posturography

Abstract

The frequency response function (FRF) is an established way to describe the outcome of experiments in posture control literature. The FRF is an empirical transfer function between an input stimulus and the induced body segment sway profile, represented as a vector of complex values associated with a vector of frequencies. For this reason, testing the components of the FRF independently with Bonferroni correction can result in a too-conservative approach. Performing statistics on scalar values defined on the FRF, e.g., comparing the averages, implies an arbitrary decision by the experimenter. This work proposes bootstrap prediction and confidence bands as general methods to evaluate the outcome of posture control experiments, overcoming the foretold limitations of previously used approaches.