View in EDS

Learning-based methods for spatial road safety analysis using in-vehicle telematics data: A systematic review.

Saved in:
Bibliographic Details
Title: Learning-based methods for spatial road safety analysis using in-vehicle telematics data: A systematic review.
Authors: Paradiso, Simone1 (AUTHOR) simone_paradiso@mail.ntua.gr, Ziakopoulos, Apostolos1 (AUTHOR), Yannis, George1 (AUTHOR)
Source: Journal of Safety Research. Jun2026, Vol. 97, p737-761. 25p.
Subjects: Automotive telematics, Deep learning, Evidence synthesis, Road safety measures, Machine learning, Feature extraction
Abstract: • This review concerns telematics-based spatial analyses through learning-based methods. • A PRISMA systematic review of 44 studies based on in-vehicle data is presented. • Data collection, features, spatial scales, modeling types & research gaps are discussed. • Telematics-based spatial analysis is rare; deep learning offers useful tools for nuances. • Gaps remain in transferability, multiscale evaluations and graph-based modeling. Introduction: Road safety monitoring is essential for achieving safer mobility. Traditionally studied with statistical methods, the field is evolving due to AI and the growing availability of in-vehicle telematics data. These developments open new possibilities in terms of spatial analysis, leading to telematics-informed spatial scales, moving beyond reliance solely on geometric characteristics of spatial units. Methods: This research presents a structured review of the existing literature at the intersection of Learning-Based methods, spatial analysis, and surrogate safety measures derived from in to vehicle telematics data. Following PRISMA guidelines, 44 studies were identified through keyword searches and supplemented by reference screening or manual search and afterwards analyzed. The studies were analyzed and narratively synthesized focusing on data collection methods, feature engineering processes, and their implications on the selection of the spatial scale. The methods employed across the selected studies range from traditional econometrics to cutting-edge deep learning techniques, with explainable machine learning techniques employed to ensure the interpretability of the methods used. Results: Key methodological challenges are discussed, such as the interrelation between data source and spatial scale, the combination of the spatial scale alongside the methodological framework. The selected features and the important ones identified across the studies were highlighted. Finally, discussion on recent deep learning methodologies and their core advantages are presented. These findings provide valuable insights into the direction of future research, improving data integration, deployment of AI models and the selection of the spatial scale. Conclusions: Advancements in AI and telematics data are reshaping road safety research, providing new tools to interpret safety analyses and generate actionable insights. Clarifying the relationships among data sources, feature selection, and spatial scale would strengthen the analytical framework and improve understanding for safer mobility. Practical Applications : Telematics-informed spatial safety assessments through AI models enable interventions tailored to actual driver behavior. [ABSTRACT FROM AUTHOR]
Copyright of Journal of Safety Research is the property of Pergamon Press - An Imprint of Elsevier Science and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)
Database: Engineering Source
Description
Abstract:• This review concerns telematics-based spatial analyses through learning-based methods. • A PRISMA systematic review of 44 studies based on in-vehicle data is presented. • Data collection, features, spatial scales, modeling types & research gaps are discussed. • Telematics-based spatial analysis is rare; deep learning offers useful tools for nuances. • Gaps remain in transferability, multiscale evaluations and graph-based modeling. Introduction: Road safety monitoring is essential for achieving safer mobility. Traditionally studied with statistical methods, the field is evolving due to AI and the growing availability of in-vehicle telematics data. These developments open new possibilities in terms of spatial analysis, leading to telematics-informed spatial scales, moving beyond reliance solely on geometric characteristics of spatial units. Methods: This research presents a structured review of the existing literature at the intersection of Learning-Based methods, spatial analysis, and surrogate safety measures derived from in to vehicle telematics data. Following PRISMA guidelines, 44 studies were identified through keyword searches and supplemented by reference screening or manual search and afterwards analyzed. The studies were analyzed and narratively synthesized focusing on data collection methods, feature engineering processes, and their implications on the selection of the spatial scale. The methods employed across the selected studies range from traditional econometrics to cutting-edge deep learning techniques, with explainable machine learning techniques employed to ensure the interpretability of the methods used. Results: Key methodological challenges are discussed, such as the interrelation between data source and spatial scale, the combination of the spatial scale alongside the methodological framework. The selected features and the important ones identified across the studies were highlighted. Finally, discussion on recent deep learning methodologies and their core advantages are presented. These findings provide valuable insights into the direction of future research, improving data integration, deployment of AI models and the selection of the spatial scale. Conclusions: Advancements in AI and telematics data are reshaping road safety research, providing new tools to interpret safety analyses and generate actionable insights. Clarifying the relationships among data sources, feature selection, and spatial scale would strengthen the analytical framework and improve understanding for safer mobility. Practical Applications : Telematics-informed spatial safety assessments through AI models enable interventions tailored to actual driver behavior. [ABSTRACT FROM AUTHOR]
ISSN:00224375
DOI:10.1016/j.jsr.2026.05.014