Clavicle fracture prediction: simulation of shoulder lateral impacts with geometrically personalized finite elements models.

BACKGROUND Human body numerical models can help to develop protection devices against effects of road crashes. In the context of a side impact, a shoulder model able to predict shoulder injuries and more especially clavicle fracture would be helpful. METHODS A shoulder model derived from an existing finite element model of the human body representing an average male (50th percentile), HUMOS1, has been upgraded. An isolated clavicle model was assessed thanks to experimental corridors derived from dynamic tests up to failure. Then, the whole upgraded shoulder model was evaluated by comparison with results from experimental side impact tests on the shoulder. Eventually, the upgraded model was geometrically personalized toward the anthropometry of the subjects and its ability to simulate fractures was assessed. RESULTS The isolated clavicle model was assessed as validated. The upgraded 50th percentile shoulder model provided accurate results in the subinjurious domain. At higher velocities, the personalized models produced realistic shoulder injuries: clavicle fracture was accurately predicted in four cases of six, the model was conservative for the two other cases. CONCLUSION The upgraded shoulder model presented here was successfully submitted to a rigorous assessment process. Once geometrically personalized, it provided positive results for clavicle fracture prediction. As clavicle fracture is the major shoulder injury, this model could help the design of safety devices for shoulder protection. Furthermore, this study enhances the need for geometrical personalization methods when using finite element model for injury risk prediction.