Rural road safety remains a critical yet underexplored challenge in India's transportation system, where connectivity improvements under the Pradhan Mantri Gram Sadak Yojana (PMGSY) have been accompanied by increasing accident risks. Despite large-scale infrastructure development, the absence of structured, data-driven safety evaluation frameworks limits effective prioritization of high-risk segments. This study integrates regression analysis with a Relative Importance Index (RII) based multi-criteria framework to assess and prioritize safety risks on PMGSY rural roads in Sinnar Taluka, Nashik District. Accident data from 110 reported cases, field surveys and Indian Roads Congress (IRC) standards were used to evaluate critical parameters including sight distance, shoulder width, sharp curves, roadside environment, pavement condition and residential access. Regression analysis identified sight distance, blind turns and residential access density as the most influential predictors of accidents, while RII-based ranking reflected expert perceptions of safety severity. To address recent methodological advancements, a baseline artificial neural network (ANN) model was additionally employed as a comparative benchmark to examine consistency in segment prioritization. The integrated regression-RII approach enables a comprehensive, data and perception-driven assessment of rural road safety. Findings support targeted interventions, such as geometric realignment, roadside protection and access control, providing a replicable framework for policymakers and planners to enhance safety performance and promote sustainable rural mobility.
Kalekar et al. (Thu,) studied this question.