This study evaluates the performance of optimization strategies across coagulation cascade models of varying complexity, distinguished by differences in the number of species and reactions.The results indicate that increasing model complexity significantly alters the optimization landscape, heightening susceptibility to local minima and convergence challenges. To mitigate these issues, we propose a hybrid optimization framework integrating gradient-based methods with evolutionary algorithms.When applied to synthetic numerical datasets, the approach demonstrates robust and reliable convergence. The strategy is further validated using real clinical and experimental thrombin generation data, confirming its practical utility in modeling physiological conditions and guiding treatment decisions.