In recent years the use of artificial neural networks for classification and analysis of kinematic and kinetic characteristics of human locomotion has greatly increased. This happens in an attempt to overcome the limitations of traditional dynamic analysis and to find new clinical indicators for interpreting quick and objectively the large amount of information obtained in a gait lab. One of the most widely used neural networks for human gait analysis is the self-organizing or Kohonen map, based on unsupervised learning without prior definition of the formed natural groups. Among the advantages of using this type of neural network is the data dimensionality reduction, with minimal loss of information content, and the grouping of them in function of their similarities. Taking into account this, in this work an application case of a Kohonen map for clustering of locomotion kinetic characteristics in normal and Parkinson's disease individuals is presented. The results indicate that the groups identified by the map are consistent with the classification carried out by experts in function of traditional gait dynamic analysis, showing the potential of this technique for distinguishing between a population of individuals with normal gait and with gait disorders of different etiology.
Human gait; Parkinson's disease; Artificial neural network; Clustering
