Estimation of falling risk based on acceleration signals during initial gait

Abstract

In an aging society, falling risk of the elderly is one of big problems. In order to improve Quality Of Life (QOL) and curb increases in the care burden and medical costs, it is desirable to estimate and ameliorate falling risk through timely rehabilitation exercise. We propose a method of estimating the falling risk based on acceleration signals during initial gait. The risk is defined by a screening tool (Berg balance scale) utilized by physical therapists. In this method, the feature values are calculated by focusing on the variation of wave trajectory and horizontal symmetry due to unstable behavior during the initial transitional phase after starting time of the gait. Finally, in an experiment to confirm the efficacy of the proposed method, we gathered acceleration data at the waist of 17 subjects while they started walking after standing still. Then, the SVM (Support Vector Machine) classifiers to estimate the label of falling risk (3 classes: safe, caution-needed, and high-risk class) were trained and it was ascertained that F-values over 70% were achieved as the estimate accuracy.