Multiscale modelling of the skeleton for the prediction of the risk of fracture

Abstract

Background The development of a multiscale model of the human musculoskeletal system able to accurately predict the risk of bone fracture is still a grand challenge . The aim of this paper is to present the Living Human Project, to describe the final system and to review the achievements obtained so far. The Living Human musculoskeletal supermodel is conceived as the interconnection of five interdependent sub-models: the continuum, the boundary condition, the constitutive equation, the remodelling history and the failure criterion sub-models. Methods Methods are available to develop accurate subject-specific finite element models of bones that can incorporate the subject’s tissue-density distribution and empirically derived constitutive laws. Anatomo-functional musculoskeletal models can be registered with gait analysis data to predict muscle and joint forces acting on the patient’s skeleton during gait. These are the boundary conditions for the continuum models that showed an average error of 12% in the prediction of the failure load. Still, the entire supermodel is defined as a collection of procedural macros to predict the risk of fracture and should be improved. Findings Even with these limitations, the organ-level model already found some clinically relevant applications, especially in the analysis of joint prostheses. Also, the body-organ level multiscale model finds some clinical applications in paediatric skeletal oncology. The tissue- and the cell-level models are not yet fully validated. Thus, they cannot be safely used in clinical applications. Interpretation The continuum sub-model is the most mature model available. More powerful methods are needed for the generation of anatomo-functional musculoskeletal models. Muscle force prediction should be improved, investigating new probabilistic approaches to identify the neuro-motor strategy. The changes of the tissue properties in the various regions of the skeleton and predictive remodelling models should be included. An adequate information technology infrastructure should be developed to support collaborative work and integration of different sub-models.