Classification in postural maintenance based on stochastic process modeling
Résumé
This article contributes to the study of postural style, focusing on the issue of classifying subjects in terms of how they maintain posture. Here, we specifically tackle the statistical problem of classifying subjects sampled from a two-class population. Each subject (enrolled in a cohort of 54 participants) goes through two experimental protocols designed to evaluate potential deficits in maintaining posture. Measures of foot pressure are obtained at discrete times throughout the protocols. The classification procedure is a two-step procedure. In the first step, the data are modeled by parametric diffusion processes. The parameters can change at some unknown change points. Both parameters and change points are estimated from the data. In the second step, we use the V-fold cross-validation principle and the super-learning methodology to build two classifiers based on the parameters and change points estimators. We achieve a satisfactory 91% rate of correct classification.
Origine : Fichiers produits par l'(les) auteur(s)