Application of Superposition Principle Method to proximal femur FE models generated from a statistical anatomy atlas

Testing new osteoporosis drugs requires a high level of evidence in phase III of clinical trials, and considering the expensiveness of the clinical endpoints currently in use, an emerging and promising option is In Silico Trials. The study of La Mattina et al. (2022) has made an important contribution to the development of this technique, which involved the use of an Active Shape and Appearance Modelling (ASAM) model. However, the computational cost of the implemented non-linear contact FE model implies the use of a supercomputer for simulating such a population scale. Recently, a method was developed in the study of Ziaeipoor et al. (2020), namely the Superposition Principle Method Squared (SPM2), proved to be compact, accurate and more computationally efficient than common FE models when computing bone strains in virtual populations generated via ASAM. In the present work, SPM2 was implemented and applied to the statistical femur model of La Mattina et al. for predicting femoral strains associated to sideways falls. SPM2 has proven to be a compact and accurate method under the FE and ASAM model linearity assumptions, as well as when the forces included in the model were applied on the bone surface. The application of SPM2 to the side-fall contact FE model has shown several limitations derived from modeling the hip contact force as a concentrated force in the center of the femur head. SPM2 still needs precise guidelines on its applicability conditions. Despite these limitations, it demonstrated a potential gain in computational efficiency for predicting bone strain in a massive statistical population context, avoiding expensive FE simulations once the training has been performed. The obtained results will advance In Silico Trials and personalized health care applications.