Things we observed when watching people walk: Classification of gait patterns with self-organizing maps

Clinical observation and the evaluation of the various locomotive patterns in human walk is the subject of gait analysis. We suggest a novel approach to gait pattern classification as a tool for subsequent medical diagnosis and therapy considerations. Based on data collected from patients at an rehabilitation centre we achieve a clustering according to various gait malfunctions by using self-organizing maps. The major benefit of such an approach is that the learning process ends up with a classification of gait malfunctions. Relying exclusively on observable data, we do not face the need of the tremendous effort to define a biomechanical model of gait. From a biomedical point of view, our study demonstrates that the collected data from ground reaction force measurement platforms actually contains relevant information to classify gait patterns.