Artificial neural network classifier based on kinetic parameters of human motion

As most of elderly encounter osteoporosis, falling can cause serious fractures in them. Kinetic signals contain useful information about the balance impairment of human during walking, however these details cannot be directly recognized by the observer The aim of this paper is to investigate artificial neural network model for classifying the kinetic pattern in to two groups: faller and non-faller. The kinetic parameters obtained by a six-channel force plate for 3 groups of volunteer as healthy young, healthy elderly and faller elderly. Data space is then normalized and rearranged as input data matrixes for a 3-layer feed forward neural network to classify the patterns. Neural network classifier is seen to be corrected in about 85% of the test cases.