Ensuring workers’ safety is a critical component of social sustainability in the construction industry. Accidents that occur while workers are walking on construction sites constitute a significant portion of overall accidents, yet they are often overlooked in conventional task-oriented safety risk assessments. This study proposes novel Accident-During-Walk (ADW) risk indices, hierarchical and data-driven metrics designed to quantify workers’ accident risk during walks. The indices are built on Association Rule Mining and utilize structured accident data, accounting for both environmental and work-related attributes. By integrating these indices with project-specific work schedules and worker allocation plans, this study establishes an automated method for daily and weekly look-ahead ADW risk monitoring aligned with construction progress. Case studies on two construction projects validate the discriminative power of the proposed method. The results demonstrate that the indices effectively capture risk fluctuations driven by concurrent multi-trade operations and environmental severity. Notably, the analysis reveals counterintuitive patterns where adverse weather conditions paradoxically reduce risk values by constraining worker mobility, a nuance often missed by static assessments. Ultimately, this framework serves as a data-driven decision-support tool, enabling safety managers to transition from uniform inspections to targeted interventions during high-risk periods, thereby fostering a safer and more socially sustainable construction environment.