Understanding the complex behavior of fractured rock systems is critical for applications in energy development, geological sequestration, and tunnel construction. Microscale fracture surface morphology influences flow and mechanical behaviors, while upscaling frameworks. Despite progress in hydro-mechanical and thermo-hydro-mechanical coupling models, two-way mechanical-chemical interactions remain underexplored. Discrete fracture networks offer a robust statistical framework for modeling subsurface fracture systems. Advances in machine learning have accelerated the simulation and optimization of fractured geothermal systems, addressing the computational limitations of high-fidelity models. These methods support multi-objective design, enhance life cycle assessments, and provide insights into optimal geothermal management strategies. Fractured rocks serve as preferential pathways for fluid flow and heat transport, significantly influencing permeability and mechanical stability. However, the inherent complexity of coupled thermo-hydro-mechanical-chemical processes in these systems presents major challenges. Nonlinear fracture mechanics, stress perturbations, and chemical interactions drive dynamic changes in fracture connectivity and permeability, further complicated by recursive feedback mechanisms. By integrating numerical tools, machine learning techniques, and advanced discrete fracture network models, the fractured rock system could be optimized and clearly analyzed.