Detection of postural balance degradation using fuzzy neural network

Abstract

Postural balance is often studied in order to understand the effect of sensory degradation with age. The aim of this study is to analyse the static and dynamic stabilogram signals to determine different features, which can be used to detect a degradation in equilibrium using the self-adaptive neurofuzzy inference systems (SANFIS). The main features are critical point interval (CPI), autoregressive moving average (ARMA) and area of a curve under the slope (Z-Area) that are identified from the stabilogram signals. The determined features of the stabilogram signals are used to detect and predict the degradation in postural balance using the fuzzy neural network. The selected features are randomly selected for training and testing during the classification and prediction in postural balance, where we have achieved average 95.3% accuracy of the result of classification and prediction of the degradation in equilibrium in 10 trials.