Solving training issues in the application of the wavelet transform to precisely analyze human body acceleration signals

Abstract

Within the field of medical informatics, the analysis of human body acceleration signals to examine gait patterns can provide valuable information for multiple health-related applications. In this paper, we study the suitability of the wavelet transform for the analysis of body acceleration signals, and propose useful guidelines to solve existing issues in this field (such as the need for training), thus enabling a smooth application of this signal processing tool in medical environments. Making use of these guidelines, we have successfully tested our approach to analyze body acceleration signals, delivering a rich characterization of different gait patterns, without the need for training.