Current research on virtual automotive crash safety aims to integrate Human Body Models (HBMs), which include – among other aspects – more realistic ribs. Experimental data have shown that human rib bone material varies significantly across the population [1]. However, prevailing material models within HBMs are deterministic and thus do not capture this inherent biological variability, which represents a major source of uncertainty. This research gap highlights the need to develop a material model that incorporates biological variability, along with a stochastic method to propagate material uncertainty through the FEM simulation to quantify the uncertainty of global quantities of interest. We propose an Artificial Neural Network-based elastoplastic model, integrated with the explicit FEM solver LS-DYNA, specifically developed for the rib cortical bone. At the model’s core is a Gated Recurrent Unit (GRU) which captures elastoplastic behavior, complemented by Transfer Learning layers to represent biological variability. Integration with the FEM solver is achieved through a User Material Subroutine (UMAT) [2]. Most common stochastic methods are non-intrusive and depend on performing multiple realizations of a deterministic FEM model, e.g., Monte Carlo Simulation [3]. In contrast, the approach presented here is intrusive and incorporates material uncertainty directly within the UMAT subroutine, requiring only a single augmented FEM simulation. The simulation results demonstrate stable behavior, reflecting both the numerical stability of the Neural Network as the material model throughout the simulation and the consistent propagation of material uncertainty within the explicit FEM framework. Nevertheless, further investigation is needed to ensure stability over longer simulation durations typical of crashworthiness assessments. Future work will leverage this model and methodology to address inverse problems, specifically by updating the initial probability distributions of material properties through comparison with experimental data.