Strain-adaptive in silico modeling of bone adaptation--a computer simulation validated by in vivo micro-computed tomography data.

Computational models are an invaluable tool to test different mechanobiological theories and, if validated properly, for predicting changes in individuals over time. Concise validation of in silico models, however, has been a bottleneck in the past due to a lack of appropriate reference data. Here, we present a strain-adaptive in silico algorithm which is validated by means of experimental in vivo loading data as well as by an in vivo ovariectomy experiment in the mouse. The maximum prediction error following four weeks of loading resulted in 2.4% in bone volume fraction (BV/TV) and 8.4% in other bone structural parameters. Bone formation and resorption rate did not differ significantly between experiment and simulation. The spatial distribution of formation and resorption sites matched in 55.4% of the surface voxels. Bone loss was simulated with a maximum prediction error of 12.1% in BV/TV and other bone morphometric indices, including a saturation level after a few weeks. Dynamic rates were more difficult to be accurately predicted, showing evidence for significant differences between simulation and experiment (p<0.05). The spatial agreement still amounted to 47.6%. In conclusion, we propose a computational model which was validated by means of experimental in vivo data. The predictive value of an in silico model may become of major importance if the computational model should be applied in clinical settings to predict bone changes due to disease and test the efficacy of potential pharmacological interventions.