Use of artificial neural networks for assessing parameters of gait symmetry.

The study attempts to assess gait symmetry based on measurement of vertical component of ground reaction force (GRF) in lower limbs. The aim of the study was to compare the results of gait classification obtained by means of artificial neural networks (ANN) and authors' own quantitative index method. Twenty male and twenty female physiotherapy students participated in the study. Measurements were carried out by means of the Kistler force plate. The profiles of GRF were analysed using ANN which classifies the cases under one of four groups of asymmetry based on suitably prepared training set. Author's own index method was employed for quantitative assessment of the degree of gait asymmetry. The analysis of our symmetry index revealed that the difference between the cases classified by the network as symmetrical and other asymmetrical profiles was significant (p<0.001), which suggests the conformity of both methods.