In muscoloskeletal simulations the shoulder joint is often represented as a simple ball and socket joint. This means in this simulated shoulder cannot dislocate, regardless of shoulder reaction forces.
If such a model is used for estimation of muscle activation patterns during movement tasks, the simulation is prone to computing non physiological movement patterns, which would dislocate the shoulder in real life.
In this project we present a novel approach that incorporates reaction forces into the optimization of muscle activation patterns and hence leads to more realistic results.
Moreover, this optimization approach allows us to incorporate reaction force targets for isometric force tasks. Incorporating reaction force targets into the optimization, rather than using external forces, leads to more stable simulations and using our optimization approach the inclusion did not incur any increase in execution time and simulations run in real-time