PREDICTION OF THE FEMURAL LOAD USING BONE ADAPTATION MODELS AND ARTIFICIAL NEURAL NETWORKS

Abstract

Bone adaptation models in combination with finite element modelling can be used to determine the density distribution that results from a given set of loads. Conversely, one may be interested in the loads that have resulted in a given density distribution. The possibility of determining the loads that have resulted in a given density distribution provides us with a non-invasive tool for estimation of musculoskeletal loading over a certain period of time. Unlike the forward problem, the solution of the inverse problem is not straightforward, because the governing nonlinear differential equations of the inverse problem are not known. This study presents a novel technique for estimation of the loads from the bone density distribution. Artificial Neural Networks (ANNs) are proposed as a tool for construction of a nonlinear mapping from the space of density distribution to the space of loading parameters.