Study of site-specific bone formation using a neural network model

It has been observed from experimental studies that bone formation due to mechanical loading is site-specific and non-uniform. The response is known to depend on the type of stimuli (e.g., strain gradient, strain magnitude, strain frequency etc.). This has led researchers to believe that a spatial relation exists between specific mechanical stimuli and sites of bone formation. Linear models have been proposed in the past to depict the underlying relationship. We propose an adaptive system to capture this characteristic from available experimental data (amount and nature of mechanical loading and site and quantity of bone formation). We observe that the proposed model supports the linear relationship between circumferential strain gradient and bone formation predicted by previous studies. However, a functional relationship between strain energy density and bone formation that is nonlinear in nature is discovered by the proposed system