The current state of the art approach in the numerical simulation of dilute particle-laden flow at high velocities is to describe the fluid-particle phases by the Euler-Lagrange approach. Since the dispersed phase is dilute, a one-/two-way coupling of both phases is sufficient. Hence, the only contact forces particles encounter are through interactions with solid walls, which are handled via, e.g., rebound models. The idea of a rebound model is to predict a new particle trajectory based on the surface’s material and the properties of the impinging particle. In general, those models are empirically derived, sometimes physically motivated and often equipped with tuning parameters to match the corresponding measured particle rebound, while the actual stochastic nature of the rebound and particle breakage are typically neglected. However, this affects the resulting particle trajectories and is particularly critical at high impact velocities. Hence, further improvements were proposed to consider the higher statistical moments of the measurements or the energy losses due to particle breakage. Recently, a data-driven approach based on artificial neural networks was proposed to equip a rebound model with these effects and to allow the consideration of further physical constraints. To this end, state of the art methods from function approximation, more precisely, deep dense neural networks are employed. The networks are trained through a supervised learning approach, where the neural network maps the impacting particles’ characteristics to its new particle trajectory after rebound. The particle breakage is based on a fracture probability distribution to determine if a particle breaks. In this talk, we present this data-driven framework and illustrate its predictive performance by the use of experimental measurements of the statistical, high-velocity rebound of sand particles. Moreover, we discuss the advantages and constraints of such a data-driven approach.