Gait features analysis using artificial neural networks - testing the footwear effect

The aim of this paper is to provide the methods for automatic detection of the difference in gait features depending on a footwear.

Artificial neural networks were applied in the study. The gait data were recorded during the walk with different footwear for testing and validation of the proposed method. The gait properties were analyzed considering EMG (electromyography) signals and using two types of artificial neural networks: the learning vector quantization (LVQ) classifying network, and the clustering competitive network.

Obtained classification and clustering results were discussed. For comparative studies, velocities of the leg joint trajectories, and accelerations were used. The features indicated by neural networks were compared with the conclusions formulated analyzing the above mentioned trajectories for ankle and knee joints.

The matching between experimentally recorded joint trajectories and the results given by neural networks was studied. It was indicated what muscles are most influenced by the footwear, the relation between the footwear type and the muscles work was concluded.