Abstract:
Cyclic and low-magnitude loading is considered effective in arresting the bone loss as it promotes osteogenesis (i.e. new bone formation) at the sites of elevated normal strain magnitude. In silico models assumed normal strain as the stimulus to predict the sites of new bone formation. These models, however, may fail to fit the amount of newly formed bone. Loading parameters such as strain, frequency, and loading cycle decide the amount of new bone formation. The models did not incorporate this information. In fact, there is no unifying relationship to quantify the amount of new bone formation as a function of loading parameters. Therefore, the present work aims to establish an empirical relationship between loading parameters and a new bone formation parameter i.e. mineral apposition rate (MAR). A neural network model is used to serve the purpose. Loading parameters are supplied as input, whereas, MAR served as output. The model is trained and tested with experimental data. The model establishes an empirical relationship to estimate MAR as a function of loading parameters. The model's predictions of MAR align with in vivo experimental results. The model's response is analyzed which indicates that the bone adaptation characteristics are successfully captured in the relationship. The relationship established may be incorporated further to improve qualitative and quantitative prediction capabilities of computer models. These findings can be extended in future to design and develop effective biomechanical strategies such as prophylactic exercise to cure bone loss.
