An application of neural networks for distinguishing gait patterns on the basis of hip-knee joint angle diagrams

In this study neural networks were applied to perform automated diagnosis of gait patterns. The three conditions of gait used were normal gait, a simulation of leg length difference, and a simulation of leg weight difference. Kinematic temporal changes were recorded by an on-line motion recording system. Hip-knee joint angle diagrams were obtained from eight subjects under the three conditions. After pre-processing, the hip-knee joint angle diagrams were presented to neural networks, which learned to distinguish the three conditions. Subsequent to training, unknown gait patterns were presented to the neural networks, which assigned those patterns into the right class with a correct assignment ratio of 83.3%. The results suggest that neural networks could be applied successfully in the automated diagnosis of gait disorders in a clinical context.