One-class support vector machine for joint variable selection and detection of postural balance degradation

The study of the static posture is of great interest for the analysis of the deficit of the control of balance. A method of balance analysis is to use a platform of forces which makes it possible to extract displacement of the centre of pressure (COP). The parameters extracted from COP time series prove like variables keys to supervise the degradation of balance. However, the irrelevance and\or the redundancy of some of them make difficult an effective detection of degradation. The objective of this paper is the implementation of a method of detection (SVDD) and of a procedure of selection of the relevant parameters able to detect a degradation of balance. The selected criterion of selection is the maximization of the area AUC under the curve ROC.