Investigating Unique DUI Crash Characteristics in Rural Areas

Abstract: Driving under the influence (DUI) is a critical traffic safety concern because it significantly impairs a driver's judgment, coordination, and reaction times, leading to a higher likelihood of crashes and severe outcomes such as fatalities and serious injuries. Despite various measures, including lowering Blood Alcohol Concentration (BAC), DUI crashes and associated fatalities remain persistently high across the US. The issue of DUI is more problematic in rural areas, which requires further exploration. This study collected five years (2017-2021) of crash data in Louisiana to investigate the associated crash patterns. The study considered DUI (1,152) and non-DUI (27,966) crashes for analysis to provide in-depth details of the critical contributing factors and identify the differences between crash patterns between these two scenarios. A rule-based Machine Learning technique, Association Rules Mining (ARM), was used for analysis. Association rules were generated based on crash injury severity: fatal/incapacitating and non-capacitating/minor injuries. DUI crashes resulting in fatal or severe (KA) outcomes often involve unrestrained male drivers in head-on and single-vehicle, and dark-condition collisions with lane departures face higher fatal/severe risks. In contrast, restrained drivers on clear weather and on straight roads are less likely to experience severe DUI crashes. Non-DUI crash factors differ markedly, with head-on collisions at residential stop signs and right-angle curves linked to more severe outcomes. At the same time, restrained drivers in clear conditions on weekdays tend to have minor injuries. These findings highlight distinct patterns that inform targeted safety interventions for DUI and non-DUI scenarios.