In this thesis, a risk assessment framework based on the Pressure-State-Response (PSR) model is proposed for the risk assessment problem of crossing the mined-out region in tunnel construction. The subjective weights of the indicators are determined by the G1 method, the objective weights are determined by the improved CRITIC method combined with the Random Forest (RF) machine learning method, and the combined weights are calculated using the game-theoretic combination assignment method. On this basis, a two-dimensional cloud model was constructed to merge the risk frequency and consequence degree to comprehensively assess tunnel construction's safety risk across the mined-out region. Taking the construction project of Huayingshan Tunnel construction in Linshui County of Sichuan Province as an example, the risk cloud eigenvalues of each assessment indicator were derived through expert scoring and weight calculation, and the risk cloud diagram was drawn using MATLAB software for analysis. The results show that the risk of Huayingshan Tunnel construction is medium risk, in which environmental and geological safety, construction safety, and structural stability are the main risk sources. This thesis provides a new method for the safety assessment of tunnel construction across the mined-out region, which can effectively assess and control the safety risk during tunnel construction.