Due to the complexity and large scale of road construction processes, safety assessment poses significant challenges. By introducing the Analytic Hierarchy Process (AHP) and BP neural network comprehensive evaluation method, we can effectively improve the efficiency of traditional algorithms, reduce the complexity of the calculation process, and minimize the deviation of the results. We use multiple evaluation indicators, including geological disasters, hazardous sources, hidden dangers, and compliance with safety standards, to construct a hidden evaluation model for the BP neural network through expert scoring and set analysis methods. This allows for a more precise evaluation of road construction safety. With accurate data analysis and measurements, we ensure the safety risks of road construction and obtain reliable assessments. Experimental results show that the relative error of the proposed algorithm is smaller than that of the traditional method, indicating the feasibility of integrating AHP and BP neural network for road construction safety assessment.