The underlying thermomechanical effects that occur at the grain scale during fast slip rates in fault gouges, such as flash heating, can lead to frictional weakening of geological faults, with important implications for earthquake dynamics. Accurate numerical modeling of these effects can therefore deepen our understanding of dynamic fault weakening, an area in which significant knowledge gaps persist. The Discrete Element Method (DEM) is well suited for this purpose due to its ability to capture microscale effects in granular materials with high fidelity. However, the DEM has a very high computational cost, which makes it unfeasible for simulating large-scale problems. Therefore, multiscale strategies combining the DEM with more efficient continuum-based numerical methods offer a promising solution. This presentation highlights the progress of an ongoing research project aiming to develop a scale-bridging numerical framework to simulate the thermomechanical phenomena that drive fault strength deterioration. At the microscale, we propose a DEM model that captures the essential physics of flash heating: sharp and short-lived temperature spikes localized at particle contact interfaces. The proposed model relies on calibrated parameters that are obtained by fitting the model to experimental data from the literature by means of Bayesian uncertainty quantification. The Boundary Element Method (BEM) is adopted as the continuum-based method for the multiscale framework, and the coupling with the DEM follows a concurrent approach previously developed for mechanical problems where each method is employed in a different region of the domain. This presentation introduces a proof-of-concept extension of this coupled methodology to one-dimensional thermomechanical problems, where preliminary results indicate the framework's potential for efficiently handling large or semi-infinite domains.