Assessing the Performance of Highway Safety Manual (HSM) Predictive Models for Brazilian Multilane Highways

Mendes, Olga Beatriz Barbosa; Larocca, Ana Paula Camargo; Silva, Karla Rodrigues; Pirdavani, Ali · 2023 · Crossref

DOI: 10.3390/su151310474

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This study evaluates the transferability and performance of the Highway Safety Manual (HSM) predictive models when applied to rural multilane highways in Brazil. The research is motivated by the need to improve road safety strategies in emerging countries, where fatal crash proportions are rising and local Safety Performance Functions (SPFs) are often lacking. Specifically, the authors aim to determine if the HSM, originally developed for U.S. conditions, can accurately predict crash frequencies in Brazil after local calibration. Additionally, the study assesses the model’s resilience during the atypical traffic conditions of 2020, caused by the COVID-19 pandemic. The methodology involved analyzing five rural divided highways in São Paulo State, totaling 235.6 km. The researchers utilized crash data and Average Annual Daily Traffic (AADT) volumes collected via sensors for the years 2016 through 2020. They applied the HSM’s negative binomial regression models to calculate local calibration factors ($C_x$) by comparing observed crash counts with predicted frequencies. The study employed Goodness of Fit (GOF) measures, including Mean Absolute Percentage Error (MAPE) and Mean Absolute Deviance (MAD), to evaluate model accuracy. Furthermore, the authors tested the calibrated model’s predictive capability for 2020 by applying the $C_x$ derived from the previous four years (2016–2019) to the pandemic-era data. The results indicate that the HSM models require significant local calibration for Brazilian highways. The calculated calibration factor was 2.62 for all crash types and 2.35 for Fatal or Injury (FI) crashes, suggesting the uncalibrated HSM significantly underestimates crash frequencies in this context. GOF measures revealed that the models performed better for all crash types than for FI crashes alone. When applied to the 2020 data, the calibrated model estimated crash frequencies for all types that were approximately 10% higher than the observed counts. However, the calibrated prediction for FI crashes was very close to the observed counts, demonstrating the model’s robustness despite the drastic changes in traffic patterns and volumes during the pandemic. The findings confirm that while the HSM predictive models are not directly transferable to Brazilian highways without calibration, they are useful tools for identifying high-risk areas once adjusted with local $C_x$ factors. The study highlights the importance of local calibration to avoid misallocation of safety resources. By validating the model’s performance even under the disruptive conditions of the COVID-19 pandemic, the research supports the use of HSM-based methods for sustainable transportation planning and infrastructure investment decisions in Brazil. This contributes to broader efforts to reduce road fatalities and align with United Nations Sustainable Development Goals.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• regulatory evaluation

Information type

What kind of knowledge this paper contributes, grouped by family — independent of topic (what it is about) and method (how it was studied).

• Empirical Findings: crash risk outcomes