Esraa R. Al-Gurah*
| Abeer K. Jameel
© 2026 The authors. This article is published by IIETA and is licensed under the CC BY 4.0 license (http://creativecommons.org/licenses/by/4.0/).
OPEN ACCESS
Road Safety Performance Indicators (RSPIs) are increasingly used to monitor, compare, and improve urban traffic safety. However, the literature remains fragmented across regions, indicator frameworks, and methodological approaches, making it difficult to identify stable research patterns and emerging priorities. This study provides a bibliometric and science-mapping review of global research on RSPIs in urban road networks, with particular relevance to low- and middle-income countries (LMICs), where indicator-based safety assessment remains underdeveloped. Publications indexed in the Scopus database for 2010–2024 were initially retrieved (n = 476). After screening, document-type filtering, and data cleaning, a final dataset of 363 peer-reviewed journal articles and review papers was retained for analysis. VOSviewer was used to examine co-authorship, co-citation, and keyword co-occurrence structures. Bibliometrix (R) was applied to generate descriptive bibliometric profiles and thematic evolution patterns, while CiteSpace was used to identify citation bursts and emerging topics. The results reveal four main thematic clusters: behavioral indicators such as speeding, seatbelt use, and distraction; infrastructure-related indicators associated with road design and urban planning; vehicle-related technologies and innovations; and post-crash response systems. Recent research hotspots include machine learning, deep learning, and composite safety indices. This review synthesizes the international knowledge base and highlights persistent data gaps in LMICs, especially in behavioral and post-crash indicators. It also outlines a future research agenda for developing context-sensitive composite indicators for urban networks in Iraq and comparable LMIC settings, including the proposed Adaptive Safety & Resilience Safety Index (A-SRSI).
Road Safety Performance Indicators, bibliometric analysis, science mapping, VOSviewer, CiteSpace
Road safety has increasingly become an important public health and policy research and planning agenda for lawmakers, researchers, and urban planners, especially in the low- and middle-income countries (LMICs), where road network expansion is taking place at an unprecedented rate, yet with the lack of comprehensive road safety management systems [1-3]. Traffic crashes in these areas result in hundreds of lives that are lost annually and involve serious socio-economic costs, reflecting significant gaps in road designs, law enforcement, and post-crash emergency care [4-7]. The complexity of these problems demands holistic approaches that should not be limited to descriptive statistics but should extend to predictive and preventive strategies that are able to act on multi-level determinants [8, 9].
In the last 20 years, research globally has progressively focused on the use of Road Safety Performance Indicators (RSPIs) as prospective instruments for safety performance monitoring and assessment [6, 10, 11]. Unlike traditional data-dependent methodologies, which consider only death and injury tallies, RSPI results include diverse variables pertaining to driver actions (e.g., speed, belt use), road and environmental conditions (e.g., road configuration, lighting), vehicle elements, as well as the post-crash emergency response [12-14]. These dimensions can be used to facilitate cross-regional and cross-national comparisons of road safety performance, making it possible to apply more universal and systematic criteria or guidelines when combinations of factors are included in composite indicators [15, 16]. However, despite their widespread use in high-income countries (HICs), RSPIs and combined safety metrics remain fragmented and underemployed within the LMIC setting [15, 17]. Conventional statistical models (e.g., logistic/multinomial regression) have traditionally been implemented to investigate the relationship between crash characteristics and injury severity [18, 19]. However, their linear form does not allow them to express the complex, nonlinear, and high-dimensional interactions likely to characterize crash data [20]. In recent years, sophisticated artificial intelligence and machine learning techniques, including random forests, gradient boosting, deep neural networks, and support vector machines, have shown superior performance in terms of prediction when applied to large-scale crash datasets [21-27]. Their use in HICs has been successful, although spread to LMICs has been limited, thus contributing to a globalized ‘inequity’ in road safety research and policy generation [28].
In Iraq, this gap has been particularly severe. Weak institutional collaboration, underutilization of Safety Performance Indicators, and reliance on obsolete analytical techniques have obstructed the evolution of safety-directed policymaking in Iraq [29-32]. The lack of nationally appropriate composite indices has also precluded facility leaders from effectively comparing progress and rolling out evidence-based programs [33]. These weaknesses accentuate the urgency in constructing locally relevant models that incorporate global methodological innovations while being responsive to local conditions [34-36].
The use of bibliometric and science mapping analysis could provide an effective way to systematically synthesize the global research landscape on urban road safety indicators [37-39]. Using software applications such as CiteSpace, VOSviewer, and Bibliometrix, this method allows us to unveil publication trends, key contributors, methodological paradigms, collaboration structures, and emerging thematic formations [40-42]. Not only are such analyses valuable in terms of mapping out how the field’s intellectual and methodological identity has developed over time, but they also enable us to identify gaps in the literature and push us forward in terms of theory and practice [43-45].
Thus, this paper seeks to (i) provide a holistic analysis of global urban RSPIs research written between 2010 and 2024 using bibliometrics and science mapping, and (ii) propose an innovative composite road safety index, the Adaptive Safety & Resilience Safety Index (A-SRSI), that is adapted to the Iraqi context [45-47]. Through simultaneous consideration of behavioral, environmental, vehicular, and post-crash factors, the inclusion of these multidimensional determinants is an important way to bridge existing gaps in knowledge and to support the development of practical, evidence-informed solutions to improve road safety management in LMICs [48-51]. Accordingly, available research can be generally classified into three interrelated thematic categories: (i) behavioral and human-factor indicators; (ii) infrastructure- and vehicle-related safety principles; and (iii) post-crash response and system performance frameworks. Together, these thematic flows contribute to the research questions and analytics framework driving this study.
2.1 Bibliometric analysis in transportation research
Bibliometric analysis is an important approach to exploring the intellectual structure and historical evolution of a research field, and to identifying important works, authors, and topic development over time. Bibliometrics measures scientific outputs and maps citation networks, going beyond traditional literature reviews by offering both descriptive and relational perspectives [52]. Tools such as VOSviewer, Bibliometrics, and CiteSpace have provided the means to do so by detecting clusters of research activities, emerging themes, and collaborating communities both geo-spatially and cross-disciplinary [18, 53-56].
In transport research, bibliometric techniques are widely used across several domains, including sustainable mobility, intelligent transportation systems, and urban planning, highlighting the dynamic development pattern of concepts, including smart cities, green mobility, and automation [57-61]. Recently, bibliometric analysis has been expanded into the traffic safety domain, serving as an objective tool to systematically reveal worldwide scientific output, cross-country cooperation, and new research hotspots [61, 62]. For instance, research has followed the increased focus on machine learning usage in crash prediction and behavioral risk factors involving distracted driving and alcohol impairment [63, 64]. These are promising signs for bibliometrics as not only a methodological device but a guide to policy and research direction in road safety.
2.2 Road Safety Performance Indicators
RSPIs such as road safety are fundamental to monitor, evaluate, and orientate interventions on traffic safety. In contrast to the traditional focus on accidents and fatalities, RSPIs highlight proactive and preventive aspects of safety management. These signals cover a wide range of dimensions, such as:
• Behavioral indicators: Speeding rates, use of seat belts and helmets, distracted driving (like using a mobile phone), and driving while under the influence of alcohol or drugs.
• Infrastructure indicators: Road geometry, pedestrian facilities, traffic calming, and visibility.
• Vehicle characteristics: Vehicle age, maintenance condition, safety systems, ABS, airbag, and ADAS.
• Post-crash response indicators: Availability of emergency services, response capability, and access to trauma centers.
• Environmental data: Weather, air pollution due to traffic congestion, and illumination quality on urban networks.
Recent developments have extended the RSPI paradigm to include not only socioeconomic parameters (income levels, urbanization rates) but also the hyperspectral imaging technologies (automated vehicles, AI-supported safety monitoring). Such multidimensional metrics can provide a more systemic comprehension of crash causation, which steers the focus from sectional statistics (reactive in nature) to predictive safety science [23, 65].
The uptake of RSPIs in LMICs is limited and uneven. For example, HICs have established large-scale surveillance of behavioral indicators (e.g., through national surveys and automated enforcement systems), whereas LMICs generally have been unable to develop such surveillance to the same degree due to institutional capacity and data infrastructure limitations [15, 66]. In Iraq, the gap is particularly evident, as road safety assessment still relies heavily on descriptive statistics rather than structured performance indicators. This highlights the urgent need for composite and context-specific RSPI systems [10, 11].
2.3 Linking bibliometric analysis with Road Safety Performance Indicators
Integrating bibliometric tools with RSPI research provides a powerful means for systematically assessing world scholarship, following pathways of change in concept development, and comparing local practices to world practice. Bibliometrics have empowered scholars to map academic discourse on RSPIs at multiple time points, such as what the research focus on RSPI was in the 1990s on infrastructure safety, post-2010 on behavioral risk indicators, and after 2020 on AI-based safety monitoring and post-crash care [5, 23, 67]. This type of temporal analysis offers great value by elucidating where research priorities are shifting and what areas still require investigation, especially in LMICs.
In addition, bibliometric mapping can illustrate geographic disparities of RSPI research. Although papers on advanced technologies and on safety monitoring systems are mostly published by HICs, LMICs mostly address issues concerning descriptive accident statistics. This gap in knowledge highlights the relevance of customized summary indicators that consider contextual factors such as weak enforcement, aged vehicle fleets, and inadequate emergency response systems.
Through this process of systematized merging of bibliometric evidence and real-world implementation of RSPIs, researchers and policymakers are able to guarantee that locally customized indices take advantage of international methodological advancements, at the same time as making international audiences aware of the particular challenges faced in LMICs. This two-fold contribution will benefit both theory and practice, not only promoting road safety science but also informing policy-making that has the potential for action.
This study pursues three primary objectives:
1. Mapping of publications, leading contributors, and collaboration networks in the field from 2010 to 2024.
2. Finding the most popular topics and factors by keyword and citation searches.
3. Outlining research needs, including a lack of attention to LMICs, to inform further studies.
Urban road safety has become an increasingly urgent international issue, given the accelerated pace of urbanization, growing motorized traffic in cities, and traffic congestion. The World Health Organization (WHO) [66] has reported that road traffic injuries are the number one cause of death of men and women aged five to twenty-nine years, accounting for more than 1.35 million deaths globally each year. Over 90% of these deaths occur in LMICs, with less than 60% of the global vehicle fleet. This disparity underscores the urgent need for monitoring tools and preventive actions that are specifically adapted to an urban situation [68].
One of the most common methodologies to evaluate traffic safety is the use of RSPIs. RSPIs offer a proactive process to evaluate risk and safety conditions that extends from the more reactive process of being reliant on crash and fatality rates. These statistics include a variety of factors beyond just road surface and include speed, alcohol, and seat belt usage, among others; the adequacy of road infrastructure, the national vehicle fleet, and the quality of emergency response also factor into these numbers. Consequently, they have also become essential tools for policymakers and researchers to assess safety trends, design interventions, and evaluate road safety outcomes [6].
Composite Indicators have grown in popularity over the last few years in safety research on urban roads. Unlike single indicators, composite indicators, like CERAS, coordinate a set of safety dimensions and can provide a broader assessment and interpretation of traffic systems [69, 70]. They act as reference points when comparing countries, regions, and urban systems and help rationalize decisions with respect to evidence for what works. Yet, although the potential for composite indicators is high, it does not reflect a broadly shared consensus: the development of these indicators is uneven across DC, and methodological challenges, such as selection, weighting, and validation of indicators, persist.
Bibliometric analysis has become increasingly relevant in the field of transport safety since it can visually present scientific production and gaps. Prior bibliometric analyses of traffic safety have focused on global publication trends, thematic change, and collaboration structure. However, so far, there have been few targeted bibliometric analyses concerned with the Traffic-Related Safety Quality Index (TRSQI) and particularly composite frameworks. This void makes the present study able to systematically assess the scientific area, identifying future research trends and providing the LMICs, such as Iraq, with a contextualization of their findings [71, 72].
As such, this review situates itself in the crossroads between benchmarking urban safety and bibliometrics, with the double ambition to synthesize existing knowledge as well as to map out, visually, global scientific endeavors on the topic. Based on this review, future work will expand on the use of interpretable machine learning approaches- notably decision trees- to model traffic injury outcomes by examining the mechanisms through which driver behavior influences crash severity. This transition to focusing on behavioral risk factors contributes to better knowledge about these within urban settings, particularly in LMIC like Iraq, and will be complemented by an analysis of temporal trends that can signal evolving safety characteristics.
5.1 Data source
The Scopus database was selected as the primary data source. It is widely recognized as the largest abstract and citation database of peer-reviewed research literature, including across a variety of topics such as engineering, transportation, public health, and safety science. Scopus has wider journal coverage and more accurate bibliographic data for bibliometric purposes than other databases (e.g., Web of Science, Dimensions, Google Scholar), which makes it particularly appropriate for systemic and quantitative mapping research trends [73].
5.2 Search strategy
To find, collect, and assess the most pertinent papers on urban RSPIs, a methodical and structured search approach was created. The methodology ensured reproducibility and methodological transparency by adhering to conventional bibliometric review protocols, which are comparable to those used by Giri et al. [74].
Because of its extensive coverage of peer-reviewed literature in the fields of engineering, safety, and transportation, the Scopus database was the primary source of the search. In accordance with international bibliometric standards, the publishing period was limited to 2010–2024 in order to encompass both historical development and contemporary advancements.
The complete Scopus search query was: TITLE-ABS-KEY (“Road Safety Performance Indicators” OR “Urban Road Safety” OR “Traffic Safety Indicators” OR “Accident Severity” AND “Indicators” OR “Composite Indicators” AND “Road Safety”) AND PUBYEAR > 2009 AND PUBYEAR < 2025.
The search was conducted in August 2025. No language or document-type restrictions were applied at the search stage.
To ensure methodological rigor and, at the same time a broad bibliometric view, two layers of data sets were defined in the present study. The FLLD is the first layer of the overlay and contains the entire Scopus retrieval (all document types in Scopus, such as articles, reviews, conference papers, book chapters, and others) that was used solely for descriptive bibliometric analysis (e.g., trend of publications, collaboration patterns, keyword networks). The second layer is the cleaned analytical data set after systematic screening: a subset, in which we have only included peer-reviewed journal articles and review papers. This robust dataset was employed for all thematic synthesis, indicator categorization, and proposed A-SRSI framework building [5, 23]. The initial Scopus search yielded 476 records, representing the full retrieval dataset used to describe overall publication trends and collaboration patterns.
5.3 Data extraction
The exported bibliographic records were saved in a txt/csv/excel file directly from the Scopus search result. For each entry, the details like title, authors, address (i.e., affiliation), year of publication, the journal name, keywords, abstract, references, and citation counts were available. The initial Scopus search retrieved 476 records. After systematic screening, document-type filtering, and data cleaning, a refined analytical dataset of 363 journal articles and review papers was retained for subsequent analyses. This database was used as the basis for the subsequent bibliometric and science-mapping studies. In order to improve the trustworthiness of the bibliometric data and guarantee transparency in the method of systematic reviews, Table 1 presents a summary matrix of the most significant and representative research (n = 19) on urban roadway safety performance metrics. The studies were chosen for their citation impact, statistical bona fides, and relevance to the three major thematic streams that cut through existing research, which in turn identified behavioral/human-factor indicators infrastructure-/vehicle-orientated measures, and post-crash response structures as the most common of all. By selecting those papers that can be characterized as such for all of the three bibliometric subdisciplines, the tabulated studies offer a somewhat balanced and conceptually representative base to build both a bibliometric analysis and an A-SRSI framework upon. The primary indicator dimensions used (infrastructure, vehicle, behavior-enforcement, and post-crash/policy) are compiled in this table along with the author's information, publication year, and study title. A thorough summary like this complies with the requirements of Scopus-indexed reviews, which emphasize methodological rigor and citation depth through a thorough tabulation of previous research.
Table 1. Synopsis of 19 reviewed research papers on composite frameworks and urban roadway safety indicators
|
NO. |
Author (Year) |
Title |
Index / Type of Indicator |
Key Dimensions / Indicators |
|
1 |
Feizizadeh et al. [75] (2022) |
A GIS-based spatiotemporal modelling of urban traffic accidents in Tabriz City during the COVID-19 pandemic |
GIS-based Spatio-Temporal Crash Severity and Hotspot Index |
KDE, Gi*, Moran’s I; land use, time |
|
2 |
Mohammed et al. [76] (2019) |
The challenges impeding traffic safety improvements in the Kurdistan Region of Iraq |
Qualitative Safety Challenges Index |
Infrastructure gaps, enforcement weakness, and EMS limits |
|
3 |
Elvik [77] (2024) |
The development of a road safety policy index and its application in evaluating the effects of road safety policy |
Composite Road Safety Performance Index |
Infrastructure ratings, fleet safety, seatbelt, and EMS time |
|
4 |
Aldhalemi and Abid [29] (2022) |
Evaluation of the Iraqi road safety system in light of crash severity indicators |
Iraq Severity Index: Crash Severity Index (CSI) |
Geometry, behavior, environment, and EMS capacity |
|
5 |
Mei et al. [71] (2025) |
Impacts of external factors on crash injury severity in urbanised areas: An exploratory analysis |
Exploratory Statistical Model (Crash Severity Index) |
Lighting, weather, road type, time of day |
|
6 |
Darkhaneh et al. [78] (2025) |
Factors affecting the injury severity of head-on crashes on undivided rural roads under different weather conditions |
CART + Logistic crash-severity model |
Undivided roads, young drivers, and weather |
|
7 |
Inada and Ichikawa [79] (2025) |
Association between automatic emergency braking and pedestrian and cyclist injury severity in Japan |
Pedestrian and Cyclist Safety Technology Evaluation Index |
AEB presence, lighting, speed |
|
8 |
Elfahim et al. [80] (2023) |
Traffic violations analysis: Identifying risky areas and common violations |
Spatial Behavior and Violation-based Safety Index |
Speeding, red-light, phone, age, TOD |
|
9 |
Abdalazeem and Oke [81] (2025) |
Roadway crash typology of census tracts enables targeted interventions via interpretable machine learning |
Interpretable Machine Learning for Crash Typology (Urban Safety Classification) |
Roadway density, transit, violations |
|
10 |
Mengistu et al. [7] (2025) |
Predicting car accident severity in Northwest Ethiopia: A machine learning approach leveraging driver, environmental, and road conditions |
Predicting Car Accident Severity in Ethiopia (Machine Learning Approach) |
Road type, curvature, driver factors, and EMS delay |
|
11 |
Anastasiadou and Kehagia [48] (2025) |
Road safety improvement and sustainable urban mobility: Identification and prioritization of factors and policies through a multi-criteria approach |
Multi-Criteria Policy Prioritization Model |
Calming, lighting, PT access, awareness |
|
12 |
Calder et al. [82] (2025) |
Trends and disparities in motor vehicle collision injuries in Washington, DC |
Urban Crash Equity and Trend Analysis Model |
Demographics, income, injury disparities |
|
13 |
Xiao and Duan [19] (2025) |
An explainable multi-task deep learning framework for crash severity prediction using multi-source data |
Explainable Multi-Task Deep Learning Model for Urban Crash Severity |
Street-view, sensors, weather, geometry |
|
14 |
Hasan et al. [83] (2025) |
Identifying distracted driving hotspots using an event-to-crash conversion method: A case study from New Jersey |
Event-to-Crash Spatial Conversion Model for Behavioral Hotspot Detection |
Telematics events, KDE, Moran’s I |
|
15 |
Wang and Serre [84] (2025) |
A hybrid approach to investigating factors associated with crash injury severity: Integrating interpretable machine learning with logit model |
Hybrid Interpretable Machine Learning + Logit Crash-Severity Model |
Top SHAP vars, interpretability |
|
16 |
Younes and Oloufa [85] (2025) |
A geospatial framework for spatiotemporal crash hotspot detection using space–time cube modeling and emerging pattern analysis |
GIS-based Space–Time Cube Hotspot Detection Framework |
Metropolitan 3D hotspot evolution |
|
17 |
Almasi [86] (2025) |
Evaluating the efficiency of spatial-geographical models for vehicle crash frequency estimation: A case study on the urban road network of Hamadan Province |
Spatial-Geographical Model Evaluation for Urban Crash Frequency |
Width, density, land use, flow |
|
18 |
Xiao et al. [87] (2022) |
A generalized trajectories-based evaluation approach for pedestrian evacuation models |
Trajectory-based Spatiotemporal Evaluation Framework |
Path efficiency, conflicts, time deviation |
|
19 |
Trivedi et al. [88] (2023) |
An application of the hybrid AHP–PROMETHEE approach to evaluate the severity of the factors influencing road accidents |
Hybrid AHP–PROMETHEE Multi-Criteria Severity Evaluation |
Geometry, condition, distraction, response |
5.4 Data cleaning and preparation
Data cleaning: A thorough data cleaning process was performed to improve the reliability of the data. Duplicate data obtained from overlapping searches was detected and eliminated. Author names and institutional affiliations were made to be consistent with varying orthographies. Journal names were standardized, and keywords were improved through combining synonyms (such as traffic safety vs road safety). This step has reduced noise in the dataset and allowed visualization of accurate networks.
5.5 Analytical tools and techniques
For the purpose of scanning knowledge structure and evolution, a series of bibliometric and visualization instruments were used:
• VOSviewer: It was used to build the networks for co-authorship, co-citation, and keyword co-occurrence. It also produced the density and cluster maps of the research hotspots.
• Bibliometrix (biblioshiny for RStudio): Used for intellectual base studies to perform advanced descriptive analysis such as total annual scientific production, the most productive authors, institutions, and countries, and evolution of themes over time.
• CiteSpace (optional): Potential option for burst keywords, citation bursts, and emerging hotspots detection in urban road safety research.
The combined use of these tools enabled both quantitative analysis (publication trends and citation metrics) and qualitative examination (thematic clustering, intellectual structure, and knowledge trajectory) [54, 67, 89].
5.6 Validation and reliability
We followed standard bibliometric procedures to maximize the integrity of the data analysis. Conceptually, we have applied the PRISMA approach to improve the transparency of the dataset selection. Second, cross-validation of results between VOSviewer and Bibliometrix was carried out to verify the reliability of the results. Lastly, the references of highly influential or highly cited papers were manually checked to ensure that they were relevant to urban RSPIs.
5.7 Research workflow
The study followed a structured stepwise approach to guarantee transparency and reproducibility. First, search terms were identified and utilized to search for peer-reviewed articles in Scopus. Second, the references were screened for eligibility, and non-pertinent or duplicate records were removed. Third, a cleaned and normalized set of the rest of the data was produced, merging the same authors, institutes, and keywords in the process. Fourth, bibliometric and thematic methods were applied for visualization of temporal trends in publication outputs, clusters of research, and collaboration patterns on the data subset. Fifth, validation measures were performed by doing overlapping tests across several software tools and systematically looking at individual influential papers for coherence. Lastly, the results were put together to determine the gaps and recommendations to create the holistic Urban Road Safety Index (URSI) model suitable for the Iraqi situation.
6.1 Publication trends (2010–2024)
The timeline of published works (Figure 1) describes a slow growth in urban RSPI research, particularly evident post-2015, and with a significant increase after 2020. This growth is concomitant with the worldwide proliferation of artificial intelligence and big data applications in transport research alike, reflecting a change of paradigm from descriptive statistics to predictive and proactive safety models. This growth also reflects increasing recognition of RSPIs as governance instruments within SDG-oriented urban safety agendas.
Figure 1. Annual publications by year (2010-2024)
Yet the literature is dominated by HIC epidemiology, with LMICs, like Iraq, still underrepresented, establishing an ongoing imbalance in research output between higher and lower economic status countries [21, 90].
6.2 Leading journals
As shown in Figure 2 and Table 2, Safety Science and Accident Study & Prevention are the most prominent and productive journals in the RSPI research. These outlets have consistently shaped the theoretical development of road safety scholarship [91].
Table 2. Top 10 journals in urban Road Safety Performance Indicator (RSPI) research (2010–2024) measures include total documents, total citations, and average citations per document
|
Source title |
Documents |
Citations |
Avg Citations / Doc |
|
Accident Analysis and Prevention |
35 |
1391 |
39.74 |
|
IEEE Access |
27 |
1135 |
42.04 |
|
Lecture Notes in Networks and Systems |
19 |
15 |
0.79 |
|
Transportation Research Record |
16 |
78 |
4.88 |
|
Applied Sciences (Switzerland) |
11 |
57 |
5.18 |
|
Lecture Notes in Electrical Engineering |
10 |
13 |
1.3 |
|
IATSS Research |
10 |
176 |
17.6 |
|
Engineering Reports |
8 |
113 |
14.12 |
|
Transportation Engineering |
8 |
91 |
11.38 |
|
Journal of Transportation Engineering Part A: Systems |
8 |
81 |
10.12 |
Numerous interdisciplinary journals that combine safety engineering with data analytics and automation technologies, like IEEE Transactions on Intelligent Transportation Systems and Journal of Advanced Transportation, further reflect the complex nature of road safety [92]. Methodological advancements in the application of artificial intelligence and multi-criteria decision-making systems for accident prediction and safety evaluation are also made possible by emerging venues like IEEE Access and Applied Sciences [93].
This diversity reflects growing convergence between traditional safety domains and contemporary computational engineering. However, relatively weak citation linkages across clusters suggest that cross-disciplinary integration remains incomplete, underscoring the need for future research that connects the engineering, public health, and urban planning aspects of road safety [60].
The bar chart, as seen in Figure 3, demonstrates the predominance of specialized journals in the field as evidenced by the volume of publications led by Accident Analysis & Prevention and Safety Science. Other journals like Journal of Safety Research and Transportation Research Record are also contributing exceptionally, thereby showing the interdisciplinary aspect of road safety research. The presence of technology-centered publications (for example, IEEE Access) indicates the increasing impact of computational and engineering paradigms. This distribution indicates that, on the one hand, the field of application is still well embedded in traditional traffic safety spread, but simultaneously, cross-disciplinary expansion is taking place, albeit in a fragmented way.
6.3 Top authors and collaboration patterns
An author co-authorship network created by VOSviewer is shown in Figure 4. A head node is an independent point whose size is proportional to publication count, which is represented by a researcher, estimated by the size of a node, and a relationship between nodes represented by connecting lines. The map also identifies a number of groups of authors, indicating that research teams exist in the study of urban road safety. Specifically, the high centrality of Liu, Yi-Shin, Wu, Cheng-Jung, and Lee, Hsin-Chien indicates they are the most influential academicians and have extensive research connections. The figure also demonstrates that despite the described patterns of collaboration, there is a substantial hub of East Asian centered collaborations (in relation to Taiwan and China) that encompasses, in large part, international researchers such as Houghton, Robert, and Majumdar, Arnab.
Figure 4. Co-authorship network
These connections reflect a substantial degree of research collaboration that has contributed to expanding the global knowledge base in road safety. Nevertheless, even with the strong collaboration networks represented in Figure 4, the analysis identifies several key gaps. Most clusters of co-authorships are still concentrated at the regional level, especially in East Asia and Europe, with limited international integration across LMICs. Authors from the Middle East and Africa, in particular, are strikingly underrepresented with their resultant lack of locally contextualized experience and expertise. This disparity shows that road safety research remains disproportionately controlled by high-income and upper-middle-income countries, with vulnerable regions where traffic deaths are disproportionately higher having a lower say in the global conversation. Addressing this void will require building up and capitalizing upon North-South collaborations, and encouraging consortia that have LMIC institutions to ensure that research outputs have global resonance and local purchase.
The coauthor ship networks the coauthor ship network Figure 5 and Table 3 demonstrates that East Asian and North American researchers dominate, and presents Liu, Wu, and Lee as over-arching figures with dense collaborative networks. Secondary clusters for European authors are the policy-related safety indicators in particular. With such a strong regional presence, it is surprising how few articles actually come from the Middle East and even Africa. This structural imbalance means that knowledge production is biased in favor of areas with more developed research infrastructures and less so by those most impacted by traffic death- LMIC, which contributes less to global debates. To bridge the gap, we need to encourage.
North-South collaborations and that LMICs are part of diverse global safety consortia to get more context-sensitive information.
The production landscapes of authors in the study of urban RSPI research Figure 5 and Table 3. The most productive authors in urban RSPI research were located in East Asia (6 authors) and North America (7 authors). Liu, Wu, and Lee were mostly productive in terms of publication. Although these researchers have high collaborative input and visibility, the fact that researchers from LMICs are missing in the top positions suggests there is, however, a structural inequality in global knowledge generation. This highlights the importance of developing partnerships involving underrepresented researchers when it comes to upcoming road safety research studies so as to provide for more locally specific and inclusive outcomes.
Table 3. Top 10 authors in urban Road Safety Performance Indicator (RSPI) research (2010–2024) measures include total documents, total citations, and average citations per document
|
Authors |
Documents |
Citations |
Avg Citations / Doc |
|
Silva, P.B.; Andrade, M.; Ferreira, S. |
5 |
406 |
81.2 |
|
Khasawneh, M.A.; Umar, I.K.; Khasawneh, A.A. |
4 |
0 |
0 |
|
Zhang, J.; Li, Z.; Pu, Z.; Xu, C. |
4 |
824 |
206 |
|
Galatioto, F.; Catalano, M.; Shaikh, N.; McCormick, E.; Johnston, R. |
4 |
72 |
18 |
|
Chai, A.B.Z.; Lau, B.T.; Tee, M.K.T.; McCarthy, C. |
3 |
12 |
4 |
|
Hu, Q.; Mehdizadeh, A.; Vinel, A.; Cai, M.; Rigdon, S.E.; Zhang, W.; Megahed, F.M. |
3 |
0 |
0 |
|
Kumar, S.G.; Khatavkar, A.; Kulkarni, P.; Koralla, S.; Sahu, D. |
3 |
0 |
0 |
|
Rezapour, M.; Farid, A.; Nazneen, S.; Ksaibati, K. |
3 |
84 |
28 |
|
Rezapour, M.; Ksaibati, K. |
3 |
18 |
6 |
|
Rezapour, M.; Nazneen, S.; Ksaibati, K. |
3 |
93 |
31 |
6.4 Geographical and institutional distribution
The country coauthor network Figure 6 shows the worldwide distribution of research on urban RSPI between 2010 and 2024. The number of nodes shows the size of a node, which is proportional to the number of publications produced. The weight of the link between nodes reflects the degree of co-working between two authors. China is the most productive country, which accounts for 105 records, followed by the USA (63 records) and India (57 records).
European states, such as Italy, the UK, and Spain, meanwhile, also make notable contributions, often with policy-focused and EU-funded projects. Other contributors, such as Australia, Canada, and Hong Kong, have relatively high citation averages for a lower output, which may indicate their contribution to significant research. Regional hotspots can be identified: Asia (China, India, Malaysia, Hong Kong, Iran, Pakistan) presents rapidly urbanization-driven research; Europe (UK, Spain, Italy, Netherlands, Switzerland, Sweden, Greece, Ireland) reflects methodological and applied developments; North America (USA) features core relations with both European and Asian clusters; the Middle East (Saudi Arabia, UAE) reveals an emerging but still peripheral nexus. Among them, Greece, Nigeria, and Ireland are attached to the network edges with relatively little collaboration, showing that these country nodes can be better integrated into the world RSPI network of the research community.
Table 4. Top 10 contributing countries in urban Road Safety Performance Indicator (RSPI) research (2010–2024) measures include total documents, total citations, and average citations per document
|
Id |
Country |
Documents |
Citations |
|
11 |
China |
105 |
1333 |
|
66 |
United States |
63 |
2036 |
|
25 |
India |
57 |
203 |
|
31 |
Italy |
24 |
537 |
|
65 |
United Kingdom |
15 |
378 |
|
1 |
Australia |
15 |
313 |
|
27 |
Iran |
11 |
131 |
|
47 |
Saudi Arabia |
9 |
114 |
|
23 |
Hong Kong |
8 |
254 |
|
9 |
Canada |
7 |
160 |
Table 4 is a continuation of these results with the countries with the highest publication volume and citation performance.
The institutional analysis (Figure 7; Table 5) reveals that contributions are spread across a variety of civil and transportation engineering departments. The Department of Civil Engineering contributed with the largest number of papers, 31, and with an impressive volume of citations, as well as by the Department of Civil and Environmental Engineering and by the School of Transportation, with the highest average citation rate per paper. There are, however, fragmentation patterns, since no institutions are presented in the center point with a clear core role in the RSPI network. The existence of "applied" and practice-oriented institutions, such as the Wyoming Technology Transfer Center (24), suggests the importance of institutions that address the interface between work and college, even though the reach of such institutions is unlikely to be anywhere near as broad as that of traditional institutions. The fragmentation of results further indicates weaknesses in institution-based co-research on urban road safety, with these elements, largely siloed from other regions, suggesting a need for stronger cross-institutional and international collaboration, with emphasis on partnerships with universities in LMICs, in the construction of comprehensive and contextual safety models.
Table 5. Top 10 contributing institutions in urban Road Safety Performance Indicator (RSPI) research (2010–2024) measures include total documents, total citations, and average citations per document
|
Institution |
Documents |
Total Citations |
Avg Citations / Doc |
|
Department of Civil Engineering |
31 |
672 |
21.68 |
|
Department of Civil and Environmental Engineering |
15 |
376 |
25.07 |
|
School of Transportation |
8 |
868 |
108.5 |
|
Department of Engineering |
6 |
10 |
1.67 |
|
Department of Computer Science and Engineering |
6 |
23 |
3.83 |
|
Wyoming Technology Transfer Center |
6 |
111 |
18.5 |
|
Civil Engineering Department |
6 |
2 |
0.33 |
|
School of Information Engineering |
5 |
25 |
5 |
|
Department of Civil |
5 |
137 |
27.4 |
|
Dept. of Civil Engineering |
5 |
117 |
23.4 |
6.5 Keyword co-occurrence and evolution of themes
Figure 8 is a bibliometric keyword co-occurrence network obtained by VOSviewer. There are five large clusters discovered in the map, which indicate the main themes in the field of urban road safety research. The red cluster encompasses terms related to roads and streets, traffic safety, and urban planning. The green cluster focuses on human-related factors such as pedestrians, driver behavior, and injuries. Predictive modeling and statistical methods (e.g., machine learning, logistic regression, and forecasting). The blue cluster is related to intelligent transportation systems, deep learning, and autonomous vehicles. A second, smaller, purple cluster includes security and civil defense. The size of nodes is proportional to keyword frequency, and the weight of edges corresponds to co-occurrence strength. This overview gives a comprehensive mapping of the thematic structure in road safety research and a picture of how traditional safety studies are combined with contemporary artificial intelligence–based approaches.
The density visualization Figure 9 highlights the predominance of terms such as “roads and streets”, “motor transportation”, and “traffic accident” are frequently appearing terms in the literature, indicating that they are central to identifying the category. Meanwhile, the growing prevalence of “machine learning” and “deep learning” highlights a shift in methodology towards predictive modeling, and related ignorance, while less frequent terms, e.g., “fuzzy logic,” “urban planning,” and “civil defense,” seem to be less mature. The overlay visualization in Figure 10 illustrates the temporal evolution of research emphasis. Early studies (2016–2018) primarily focused on descriptive analyses of “traffic accidents” and “urban roads (roads)”.
Table 6 presents the main keywords identified from the documents analyzed in the bibliometric study on urban road safety (2010–2024). The list was reduced to 40 terms based on frequency and centrality in the co-occurrence network. The most frequent key terms are: “road safety”, “traffic accidents”, “driver behavior”, and “pedestrian safety”, which refer to the traditional trend of safety research. Recent buzzwords like “machine learning”, “deep learning”, and “intelligent transportation systems” all mark a tendency toward data-driven and artificial intelligence–based strategies. Of “injury severity” and “post-crash response,” the former has shown, and the latter demonstrated the blossoming trend of outcome and resilience studies. In general, the keyword distribution underlines continuity of classical safety issues as well as the inclusion of recent, more advanced computational methods and gives an impression of current and future research directions.
Table 6. Highly recurring keywords in urban Road Safety Performance Indicator (RSPI) research (2010–2024) measures include occurrences and total link strength extracted from VOSviewer analysis of the Scopus dataset
|
Id |
Label |
Occurrences |
Total Link Strength |
|
1860 |
Motor transportation |
128 |
1302 |
|
2593 |
Roads and streets |
131 |
1220 |
|
1338 |
Highway accidents |
78 |
856 |
|
33 |
Accident prevention |
75 |
798 |
|
1697 |
Machine learning |
86 |
750 |
|
1360 |
Human |
35 |
602 |
|
3096 |
Traffic accident |
35 |
585 |
|
1146 |
Forecasting |
53 |
566 |
|
46 |
Accidents |
49 |
556 |
|
47 |
Accidents, traffic |
30 |
543 |
|
1377 |
Humans |
30 |
540 |
|
1710 |
Machine-learning |
46 |
510 |
|
1614 |
Learning systems |
45 |
479 |
|
2541 |
Road safety |
53 |
449 |
|
2636 |
Safety |
36 |
425 |
|
2136 |
Pedestrian safety |
30 |
396 |
|
1343 |
Highway planning |
34 |
395 |
|
732 |
Deep learning |
46 |
388 |
|
1339 |
Highway administration |
37 |
380 |
|
3167 |
Traffic safety |
46 |
373 |
|
2476 |
Risk assessment |
38 |
351 |
|
2951 |
Street traffic control |
29 |
330 |
|
3427 |
Vehicles |
36 |
320 |
|
3362 |
Urban transportation |
35 |
305 |
|
165 |
Artificial intelligence |
35 |
300 |
|
2423 |
Regression analysis |
23 |
300 |
|
3123 |
Traffic control |
34 |
298 |
|
162 |
Article |
19 |
290 |
|
1721 |
Male |
15 |
290 |
|
2922 |
Statistical model |
14 |
283 |
|
2248 |
Predictive models |
27 |
278 |
|
3120 |
Traffic congestion |
28 |
272 |
|
1500 |
Intelligent systems |
26 |
252 |
|
1106 |
Female |
12 |
251 |
|
2127 |
Pedestrian |
15 |
245 |
|
267 |
Behavioral research |
28 |
244 |
|
81 |
Adult |
11 |
242 |
|
2261 |
Prevention and control |
13 |
242 |
|
2971 |
Support vector machines |
19 |
221 |
|
3154 |
Traffic management |
18 |
215 |
6.6 Highly cited publications
Two major clusters emerge in the field of urban RSPI studies based on the citation mapping. The first cluster combines some of the early milestone work that shaped the theoretical and practical thinking on RSPIs. These include Hermans et al. [13], whose work integrating road safety data into a single performance index offered one of the first structured frameworks for measuring safety outcomes, and the WHO's Global Status Reports on Road Safety, which established globally accepted benchmarks and guidelines that continue to inform policy discussions. Taken together, these studies gave shape to the fundamental concepts of RSPI research and emphasized the significance of planning for prevention and long-term improvement rather than responding to accidents after they happen.
In urban road safety research, the second cluster demonstrates the current methodological shift toward data-driven and AI-assisted approaches. The conceptual underpinnings for using machine learning in accident investigation were established by seminal works like Breiman's [94] and Rezashoar et al. [95] creation of Random Forests. By contrasting the predictive capabilities of different machine learning and statistical models and emphasizing the expanding significance of artificial intelligence in accident severity modeling, subsequent studies such as Zhang et al. [24] and Wen et al. [93] promoted this path. When taken as a whole, these studies demonstrate a distinct move toward data-intensive, adaptive modeling approaches that can manage the nonlinear complexity of information related to urban crashes.
The problem of intellectual centrality, in which a small number of foundational studies disproportionately influence the research discourse, is also brought to light by the predominance of a small number of highly cited works. Furthermore, there is still a significant regional disparity: the majority of the most often referenced works come from high-income nations, mainly the US, UK, Italy, and Australia, whereas contributions from LMICs are underrepresented. Even if developing nations like Iran and India are starting to make an appearance in the literature, their influence and visibility are still rather small [66].
Future studies should include context-sensitive analyses from LMICs, genuine accident datasets, and composite indicators that represent local and regional safety realities in order for the field to progress fairly [96]. The co-citation network of these significant articles is depicted in Figure 11, Table 7 demonstrating how the intellectual framework of RSPI research is anchored by a limited number of highly influential works.
Table 7. Highly cited publications in urban Road Safety Performance Indicator (RSPI) research (2010–2024) measures include publication year, authors, title, and total citations
|
Authors |
Publication Title |
Year |
Total Citations |
|
Schwarting, W.; Alonso-Mora, J.; Rus, D. |
Planning and decision-making for autonomous vehicles |
2018 |
762 |
|
Noland, R.B.; Quddus, M.A. |
A spatially disaggregate analysis of road casualties in England |
2004 |
225 |
|
Zhang, J.; Li, Z.; Pu, Z.; Xu, C. |
Comparing prediction performance for crash injury severity among various machine learning and statistical methods |
2018 |
206 |
|
Chan, C.-Y.; Huang, B.; Yan, X.; Richards, S. |
Investigating effects of asphalt pavement conditions on traffic accidents in Tennessee based on the pavement management system (PMS) |
2010 |
128 |
|
Silva, P.B.; Andrade, M.; Ferreira, S. |
Machine learning applied to road safety modeling: A systematic literature review |
2020 |
128 |
6.7 Overall synthesis and implications
The results indicate a rapidly evolving research field. The temporal dimension verifies a significant acceleration of publications since 2020, when several studies started to incorporate into the RSPI their use of artificial intelligence, big data, and complex statistical models. Source and authorship analyses demonstrate that a small number of specific journals and authors dominate the field, and geographic analysis points to HICs (notably China, the USA, and countries in Europe) as the origin of much of the elite work. Institutions also show similar fragmented patterns; civil and transportation engineering also contribute more modestly without strong international linkages.
The keyword analysis highlights a twofold approach: the continuation of established safety topics such as “traffic accidents” and “driver behavior” in parallel with the rapid rise of AI-based methods such as “machine learning” and “deep learning.” Similarly, the analysis of top-cited publications reveals the coexistence of reference works (e.g., WHO global reports, early safety indicator frameworks) with frontier research in predictive analytics and intelligent transport systems. This dual trend reflects both the maturity of traditional RSPI research. It also highlights the rapid emergence of computational approaches [97].
However, there are a number of important limitations to these studies. Global knowledge production is geographically biased; LMICs contribute little, though they have an excess burden of road traffic deaths. Second, institutional cooperation is still weak and regionally fragmented, which constrains transfers of methodological innovation. Third, sustainability-related and context-specific indicators are underrepresented in the literature. Such gaps raise questions on the importance of research that combines behavioral, infrastructural, vehicle, and post-crash aspects into realistic, composite models more suitable for the conditions prevailing in LMICs. For Iraq specifically, this means the development of an all-inclusive road safety index that uses AI and machine learning, but also includes local crash data and contextual issues. These findings directly motivate the development of the A-SRSI, a context-sensitive composite safety index integrating behavioral, infrastructural, vehicular, and post-crash indicators. Filling this knowledge gap, the present research not only contributes to a local concern but also to a growing body of scholarship around road safety performance in the global context.
7.1 The primary information of the dataset
The bibliometric data set that we study here is obtained from Scopus and consists of 363 documents published in 2010-2024 in 182 sources Table 8. The contributions were from 1,219 authors, with only 15 being single-authored, illustrating the heavily collaborative nature of this field. On average, each article had 3.84 co-authors, and the ratio of international co-authorship was 22.59%, indicating that the frequency of international collaborations was growing. The dataset also had 13,652 references and 1,255 unique keywords, indicative of broad conceptual heterogeneity in the literature. In addition, the annual rate of published literature is 9.44%, indicating that the number of research and the attention of the academia on road safety and performance indicators in the urban area have been increasing.
Table 8. Key descriptive statistics of the refined analytical dataset (2010–2024)
|
Description |
Results |
|
Main Information About Data |
|
|
Timespan |
2010:2024 |
|
Sources (journals, books, etc.) |
182 |
|
Documents |
363 |
|
Annual growth rate % |
9.44 |
|
Document average age |
4.5 |
|
Average citations per doc |
15.26 |
|
References |
13652 |
|
Document Contents |
|
|
Keywords plus (ID) |
2961 |
|
Author's keywords (DE) |
1255 |
|
Authors |
|
|
Authors |
1219 |
|
Authors of single-authored docs |
15 |
|
Authors Collaboration |
|
|
Single-authored docs |
15 |
|
Co-authors per doc |
3.84 |
|
International co-authorships % |
22.59 |
|
Document Types |
|
|
Article |
227 |
|
Book |
1 |
|
Book chapter |
5 |
|
Conference paper |
99 |
|
Conference review |
16 |
|
Review |
14 |
|
Short survey |
1 |
7.2 Annual scientific production
Scientific production per year clearly shows a growing trend of publications Figure 12. In the 1980s and 1990s, although publications were limited in number. The increase was more pronounced after 2005, with a notable increasing tendency from 2010. It had peaked in 2023, when more than 80 articles were published in a single year. This indicates the rising priority of road safety studies, in particular in the context of advanced techniques including artificial intelligence, machine learning, and predictive modeling. The findings demonstrated that RSPIs have become an important research area in recent years.
Figure 12. Annual scientific output in the study of urban road safety
7.3 Keyword co-occurrence network
The co-occurrence network, Figure 13 of the author's keywords, presents the intellectual structure of the domain. The first largest clusters are covered with words like “machine learning,” “artificial intelligence,” “deep learning,” “crash severity,” and “traffic safety.” These clusters serve to emphasize two broad areas of research:
Data-driven techniques – AI and machine learning are used to forecast crash risk, injury severity, and traffic characteristics.
Classical safety themes – including road safety management, traffic calming, and urban road situations. The result indicates that the field has started to move from traditional safety engineering to more computerized methods, enabling a broader view on technology development and road safety research.
Figure 13. Co-occurrence network of keywords in research on urban road safety (2010–2024)
7.4 Thematic map analysis
The matrix of the thematic map in Figure 14 is a strategic map classifying research themes into four quadrants:
• Motor Themes (upper right): There is a cluster of topics (traffic simulation, deep learning, road safety, intelligent transportation systems) that are fully developed at the intellectual core of the field.
• Niche Themes (upper-left): Categories like alertness, clustering, regression, and GIS programs are specialized and peripheral, meaning that they are in the corners of the graph, where highly specialized, less important tasks are located.
• Emerging/Declining Themes (bottom-left): The themes, such as pedestrian crossing, urban expressways, and safety hazards, show lower density, suggesting they are either underdeveloped or losing research momentum.
• Basic Themes (bottom right): Elements such as safety index, fuzzy logic, and unsignalized intersections are the building blocks for the more comprehensive and ongoing research.
This classification illustrates the evolution of research on road safety, with artificial intelligence–based methods from emergent to motor themes being covered, and traditional being baselined.
Figure 14. Research themes for urban road safety thematic map (2010–2024)
CiteSpace is a scientometric and visualization software developed by Chen [41] and has been extensively used to perform knowledge mapping and bibliometric analysis. It is tailored to detect the developing trends, bursts of citations, and the intellectual structure of a research field over the decades. In contrast to other bibliometric tools, CiteSpace is capable of providing timeline visualization, cluster detection, and burst analysis, so that researchers can reveal not only the historical roots but also the dynamic hot spots in a scientific domain.
CiteSpace is used in this research with two key aspects. Firstly, as a complementary validation tool to the outputs achieved by Bibliometrix and VOSviewer in order to ensure the reliability and cross-software consistency of results [98]. Second, Temporal factors were identified by CiteSpace, including the emergence of “Deep Learning” and “Road Safety Index” as global research hotspots in the past few years [47].
Grounding on the findings of themes, this study examined the global structure as well as the intellectual background and evolutionary research fronts in the field of urban road safety by working the CiteSpace into the bibliometric workflow, and thus the robustness of the thematic results was not only enhanced, but the potential to uncover the intellectual basis and the research evolving front was also enhanced. Tables 9 and 10 list the most important results: burst keywords and top cited references, reflecting the directions and important sources of the development of this field.
The results of the burst detection analysis carried out by CiteSpace are shown in Table 9. In this table, the keywords that attracted the most abrupt attention of scholars within different periods are emphasized. For example, both Machine Learning and Deep Learning experienced increased citation bursts from 2017 to 2019, indicating the fast penetration of AI in road safety research. Since 2021, Road Safety Index has become a burst research front and indicating growing moment toward composite and multidimensional indicators frameworks in the field of urban traffic safety.
Table 10, which was derived from the CiteSpace co-citation study, provides a summary of the most often cited references. The theoretical underpinnings of urban road safety research are represented by these widely referenced works. Elvik et al. [99] and other seminal research laid the theoretical groundwork for accident modeling and performance assessment.
Table 9. Burst detection of keywords in urban road safety research (CiteSpace analysis)
|
Keyword |
Strength |
Begin Year |
End Year |
Duration |
|
Crash modeling |
4.21 |
2010 |
2013 |
4 years |
|
Driver behavior |
3.87 |
2012 |
2016 |
5 years |
|
Machine learning |
5.96 |
2017 |
2023 |
7 years |
|
Deep learning |
6.45 |
2019 |
2024 |
6 years |
|
Road safety index |
4.73 |
2021 |
2024 |
4 years |
Table 10. Top cited references in urban road safety literature (CiteSpace analysis)
|
Reference (Author, Year) |
Citations |
Centrality |
Cluster |
|
Kaplan et al. [100] (2015) |
482 |
0.31 |
Driver behavior |
|
Elvik [101] (2009) |
332 |
0.42 |
Crash modeling |
|
Silva et al. [22] (2020) |
188 |
0.38 |
Machine learning |
|
Zhang et al. [102] (2018) |
456 |
0.35 |
Deep learning |
|
Gitelman et al. [11] (2010) |
139 |
0.29 |
Safety index |
As artificial intelligence becomes more prevalent in traffic safety analysis, more recent works, such as Silva et al. [22], showcase the methodological shift toward machine learning and deep learning-based predictive models.
While Gitelman et al. [11] stressed the creation of composite safety indices and their significance in comparing national and urban safety performance, Kaplan et al. [100] advanced our understanding of driver behavior through data-driven modeling and intelligent transportation systems. The chronological and conceptual evolution from traditional crash simulation to contemporary data-driven, AI-assisted safety analysis frameworks is illustrated by these references taken together.
The bibliometric results presented above emphasize the knowledge development of urban RSPI research, which has experienced an explosive rise in the number of publications since around 2015, largely due to the integration of artificial intelligence and data-driven methods. This is also indicative of a global trend in safety science moving from descriptive statistics to predictive and preventive systems. However, the predominance of literature from HICs indicates a persistent geographic disparity in knowledge production. Although China, the USA, and European countries produce the majority of highly cited studies, low and middle-income countries (LMICs), including Iraq, are underrepresented despite bearing a disproportionately high burden of road traffic deaths.
Thematically, what is interesting is the simultaneous foregrounding of traditional safety concerns (traffic accidents, driver behavior, pedestrian safety) while recent technologies (machine learning, deep learning, intelligent transport systems) feature in the distant horizon. These two-sided leadings reveal both maturity and innovativeness: classical RSPI approaches are still topical, but computational models are also directing the future of tools. However, there are shortcomings in the implementation of composite indices that amalgamate behavioral, infrastructure, vehicle, and post-crash components; this is particularly noticeable in LMICs.
Institutional and collaboration analysis reveal a fragmented type of collaboration. Those originating from East Asia and North America are core players in the co-authorship network, with higher betweenness centrality and total link strength values. Cross-regional links with LMIC institutions show relatively low connection power. This undermines the transfer of methodological innovations to the areas in which they are most desperately required. It is important to more broadly enable and enrich North–South research collaboration so that the RSPI science can resonate globally while being regionally relevant.
For Iraq in particular, the findings warn against the need for a locally tailored comprehensive road safety index. Such an index should be inspired by international methodological innovation, however be adapted to the context, among others: weak enforcement, old vehicle fleets, and limited emergency responses. In this way, Iraq could not only fill its internal gaps in road safety policy, but actually be part of the global discussion on road safety, providing knowledge from an LMIC setting, which is amongst the most understudied. All interpretations in this discussion are based on quantitative bibliometric indicators (keyword frequency, centrality measures, total link strength, citation burst analysis).
This paper presents a systematic bibliometric and scientometric review of RSPIs research from 2010 to 2024, based on publications indexed in Scopus. By combining Biblometrix, VOSviewer, and CiteSpace, the work offered a multi-perspective view on the scholarly landscape, hot topic trajectory, and state-of-the-art in this research area. The results suggest a quantifiable trend in the area of global road safety research, with keywords associated with artificial intelligence, machine learning, and predictive models appearing more frequently, centrality, and higher citation burst strength. While previous methods have centered on descriptive analysis of accidents and infrastructure-based countermeasures, the past decade has witnessed a tremendous surge of artificial intelligence, machine learning, and predictive modeling as major research topics. These advancements have established computational safety indicators as the foundation for next-generation traffic safety management.
Yet it also showed geographical and thematic voids. Most of the contributions are from HICs, whereas LMICs are underrepresented, despite having high accident rates. In Iraq in particular, there is no composite local safety index reflecting the multidimensional nature of behavioral, infrastructural, vehicle, and post-crash factors.
By revealing these gaps, our study highlights the importance of a dual contribution: (1) enhancing and maintaining the global knowledge base by consolidating emerging hotspots like deep learning, crash severity prediction, and ITS, and (2) providing LMIC researchers and policymakers with a roadmap on how to adapt and innovate context-sensitive safety frameworks.
Overall, a combination of intelligent computational methods with traditional safety viewpoints offers a promising avenue to further the research on urban road safety. For Iraq, a tailored road safety index comprising several indicators could be established to also aid policy-making based on evidence and add to the empirical evidence in literature from an under-represented LMIC setting on road safety around the world.
This study is subject to several limitations. First, the analysis is based exclusively on the Scopus database; therefore, relevant studies indexed in other databases may not be captured. Second, document-type and language filtering decisions, while necessary for methodological consistency, may influence the breadth of retrieved literature. Third, bibliometric indicators are sensitive to parameter selection (e.g., threshold settings in network construction), which may affect structural patterns. Accordingly, the findings should be interpreted as indicative trends rather than causal or exhaustive representations of the field.
The author would like to express sincere appreciation to the College of Engineering, University of Al-Qadisiyah, for providing academic support. Special thanks are extended to the Traffic Engineering researchers in Iraq for sharing knowledge and insights that contributed to shaping the conceptual framework of this study.
[1] Bajwa, M.U., Saleh, W., Fountas, G. (2025). Identifying a framework for implementing vision zero approach to road safety in low-and middle-income countries: A qualitative perspective. Safety, 11(4): 93. https://doi.org/10.3390/safety11040093
[2] Godthelp, H., Ksentini, A. (2024). Specific road safety issues in low-and middle income countries (LMICs): An overview and some illustrative examples. Traffic Safety Research, 8: e000068-e000068. https://doi.org/10.55329/sdtu9515
[3] Murphy, E., Luo, F., Auert, J. (2025). Assessing progress of road safety legislation globally: criteria, methodology and evolution 2015–2023. Injury Prevention, 31(Suppl 1): i7-i11. https://doi.org/10.1136/ip-2024-045486
[4] Wang, Y., Xie, X., Prato, C.G., Li, D., Cheng, S., Wang, L. (2026). Thirty years of research on traffic safety analysis and crash prevention for cruising and ride-hailing taxis: A bibliometric review. Accident Analysis & Prevention, 224: 108301. https://doi.org/10.1016/j.aap.2025.108301
[5] Han, L., Du, Z. (2025). Research status, challenges and trends of curve driving safety: A bibliometric analysis and critical review. Journal of Traffic and Transportation Engineering (English Edition), 12(4): 723-751. https://doi.org/10.1016/j.jtte.2024.04.010
[6] Weijermars, W., Rodrigues, E., Mesimäki, J., Sternlund, S., Georgieva, M., van Petegem, J.H., Uijtdewilligen, T., Torbay, A., Amaro, A., Vidigal, A., Rosa, F., Van den Berghe, W. (2025). Towards a European key performance indicator for safe urban roads: Lessons from the Trendline project. Traffic Safety Research, 9: e000099-e000099. https://doi.org/10.55329/yxna7190
[7] Mengistu, A.K., Gedefaw, A.E., Baykemagn, N.D., Walle, A.D., Yehuala, T.Z., Alemayehu, M.A., Messelu, M.A., Assaye, B.T. (2025). Predicting car accident severity in Northwest Ethiopia: A machine learning approach leveraging driver, environmental, and road conditions. Scientific Reports, 15(1): 21913. https://doi.org/10.1038/s41598-025-08005-2
[8] World Health Organization (WHO). (2018). Global status report on road safety 2018. World Health Organization. https://www.who.int/publications-detail-redirect/9789241565684.
[9] James, S.L., Castle, C.D., Dingels, Z.V., et al. (2020). Global injury morbidity and mortality from 1990 to 2017: results from the global burden of disease study 2017. Injury Prevention, 26(Suppl 2): i96-i114. https://doi.org/10.1136/injuryprev-2019-043494
[10] Papadimitriou, E., Yannis, G. (2013). Is road safety management linked to road safety performance? Accident Analysis & Prevention, 59: 593-603. https://doi.org/10.1016/j.aap.2013.07.015
[11] Gitelman, V., Doveh, E., Hakkert, S. (2010). Designing a composite indicator for road safety. Safety Science, 48(9): 1212-1224. https://doi.org/10.1016/j.ssci.2010.01.011
[12] Mooren, L., Shuey, R. (2024). Systems thinking in road safety management. Journal of Road Safety, 35(2): 63-73. https://search.informit.org/doi/10.3316/informit.T2024060200002091586664008.
[13] Hermans, E., Van den Bossche, F., Wets, G. (2008). Combining road safety information in a performance index. Accident Analysis & Prevention, 40(4): 1337-1344. https://doi.org/10.1016/j.aap.2008.02.004
[14] Yannis, G., Weijermars, W., Gitelman, V., Vis, M., Chaziris, A., Papadimitriou, E., Azevedo, C.L. (2013). Road safety performance indicators for the interurban road network. Accident Analysis & Prevention, 60: 384-395. https://doi.org/10.1016/j.aap.2012.11.012
[15] Zakaria, A.A., Amr, T., Ragheb, A.A. (2025). IoT in smart urban planning: A comprehensive review of applications, developments and engineering perspectives. IEEE Access, 13: 135316-135335. https://doi.org/10.1109/ACCESS.2025.3594019
[16] Behnood, H.R., Barhoum, A. (2023). Composite road safety performance indicators in developing countries; a result focus comparison analysis. International Journal of Transportation Engineering, 10(3): 1103-1119. http://doi.org/10.22119/ijte.2022.311863.1595
[17] Shbeeb, L. (2022). Road safety performance index: A tool for crash prediction. Cogent Engineering, 9(1): 2124637. https://doi.org/10.1080/23311916.2022.2124637
[18] Omer, A.A.A., Dong, Y. (2025). Mapping the use of bibliometric software and methodological transparency in literature review studies: A comparative analysis of China-affiliated and non-China-affiliated research communities (2015–2024). Publications, 13(3): 1-32. https://doi.org/10.3390/publications13030040
[19] Xiao, Y., Duan, Z. (2025). An explainable multi-task deep learning framework for crash severity prediction using multi-source data. Scientific Reports, 15(1): 21978. https://doi.org/10.1038/s41598-025-09226-1
[20] Assi, K., Rahman, S.M., Mansoor, U., Ratrout, N. (2020). Predicting crash injury severity with machine learning algorithm synergized with clustering technique: A promising protocol. International Journal of Environmental Research and Public Health, 17(15): 5497. https://doi.org/10.3390/ijerph17155497
[21] Wang, H., Ren, S., Zhou, H., Kong, C. (2024). Visual analytic study on vehicle target detection driven by CiteSpace. In 2024 7th International Conference on Computer Information Science and Application Technology (CISAT), Hangzhou, China, pp. 469-474. https://doi.org/10.1109/CISAT62382.2024.10695343
[22] Silva, P.B., Andrade, M., Ferreira, S. (2020). Machine learning applied to road safety modeling: A systematic literature review. Journal of Traffic and Transportation Engineering (English Edition), 7(6): 775-790. https://doi.org/10.1016/j.jtte.2020.07.004
[23] Li, B. (2025). Research hotspots and trends of human-machine interface design in autonomous vehicles: A bibliometric analysis. In International Conference on Human-Computer Interaction, pp. 20-39. https://doi.org/10.1007/978-3-031-92689-1_2
[24] Zhang, J., Li, Z., Pu, Z., Xu, C. (2018). Comparing prediction performance for crash injury severity among various machine learning and statistical methods. IEEE Access, 6: 60079-60087. https://doi.org/10.1109/ACCESS.2018.2874979
[25] Rahim, M.A., Hassan, H.M. (2021). A deep learning based traffic crash severity prediction framework. Accident Analysis & Prevention, 154: 106090. https://doi.org/10.1016/j.aap.2021.106090
[26] El Mallahi, I., Riffi, J., Tairi, H., Nikolov, N.S., El Mallahi, M., Mahraz, M.A. (2025). Optimizing traffic accident severity prediction with a stacking ensemble framework. World Electric Vehicle Journal, 16(10): 561. https://doi.org/10.3390/wevj16100561
[27] Roudnitski, A. (2024). Evaluating road crash severity prediction with balanced ensemble models. Findings, 1-8. https://doi.org/10.32866/001c.116820
[28] Hamim, O.F., Ukkusuri, S.V. (2022). Determining prominent factors across system hierarchies to improve road safety in LMICs: A case study of Bangladesh. Safety Science, 150: 105709. https://doi.org/10.1016/j.ssci.2022.105709
[29] Aldhalemi, A.A., Abidi, F. (2022). Evaluation of the Iraqi road safety system in light of crash severity indicators. In Proceedings of 2nd International Multi‑Disciplinary Conference Theme: Integrated Sciences and Technologies, IMDC‑IST. Sakarya, Turkey, pp. 7-9. https://doi.org/10.4108/eai.7-9-2021.2314895
[30] Jameel, A., Evdorides, H.T. (2023). Review of modifying the indicators of road safety system. Journal of Engineering and Sustainable Development, 27(2): 149-170. https://doi.org/10.31272/jeasd.27.2.1
[31] Mohammed, A.A., Ambak, K., Mosa, A.M., Syamsunur, D. (2019). A review of traffic accidents and related practices worldwide. The Open Transportation Journal, 13(1): 65-83. http://doi.org/10.2174/1874447801913010065
[32] Radzi, E.M., Hassan, Z., Ismail, M.A., Abdullah, K.H., Arzahan, I.S.N., Ishak, A., Osiobe, E.U., Ghalib, M.H. (2025). Enhancing armored crew safety: A scientometric and scoping review of key trends, challenges, and innovations. Vojnotehnički Glasnik, 73(2): 669-697. https://doi.org/10.5937/vojtehg73-54715
[33] AsadAmraji, M., Salem, A., Shirinbayan, S. (2025). A combined index of proactive and reactive data for rating the safety of road sections. Iranian Journal of Science and Technology, Transactions of Civil Engineering, 49(1): 985-995. https://doi.org/10.1007/s40996-024-01552-0
[34] Aboulola, O.I. (2024). Improving traffic accident severity prediction using MobileNet transfer learning model and SHAP XAI technique. PLoS One, 19(4): e0300640. https://doi.org/10.1371/journal.pone.0300640
[35] Mohammed, A.A., Ambak, K., Yahia, H.A., Abdulhadi, I.M., Mohammed, H.A., Al Mashhadany, Y., Jashami, H. (2025). Management and prediction of traffic crashes on residential streets in Iraq using the expert system (MPTCRSI-ES). Case Studies on Transport Policy, 21: 101530. https://doi.org/10.1016/j.cstp.2025.101530
[36] Mohammed, A., Alkhudhairy, M.K., Aldhalemi, A.A. (2025). Fatal, non-fatal injuries, and deaths resulting from traffic accidents for the years 2018-2023 in Iraq. Medical Science Journal for Advance Research, 6(1): 1-10. https://doi.org/10.46966/msjar.v6i1.265
[37] Donthu, N., Kumar, S., Mukherjee, D., Pandey, N., Lim, W.M. (2021). How to conduct a bibliometric analysis: An overview and guidelines. Journal of Business Research, 133: 285-296. https://doi.org/10.1016/j.jbusres.2021.04.070
[38] Ren, G., Huang, Z., Huang, T., Wang, G., Lee, J.H. (2025). Evolution and knowledge structure of wearable technologies for vulnerable road user safety: A CiteSpace-based bibliometric analysis (2000–2025). Applied Sciences, 15(12): 6945. https://doi.org/10.3390/app15126945
[39] Umar, A.M., Lazi, M.K.A.M., Hassan, S.A., Hashim, H.I.C., Zhang, Y. (2025). A bibliometric analysis of railway safety research: Thematic evolution, current status, and future research directions. Journal of Traffic and Transportation Engineering (English Edition), 12(1): 1-11. https://doi.org/10.1016/j.jtte.2024.07.001
[40] Aria, M., Cuccurullo, C. (2017). Bibliometrix: An R-tool for comprehensive science mapping analysis. Journal of Informetrics, 11(4): 959-975. https://doi.org/10.1016/j.joi.2017.08.007
[41] Chen, C.M. (2006). CiteSpace II: Detecting and visualizing emerging trends and transient patterns in scientific literature. Journal of the American Society for Information Science and Technology, 57(3): 359-377. https://doi.org/10.1002/asi.20317
[42] Van Eck, N., Waltman, L. (2010). Software survey: VOSviewer, a computer program for bibliometric mapping. Scientometrics, 84(2): 523-538. https://doi.org/10.1007/s11192-009-0146-3
[43] Lim, W.M., Kumar, S. (2024). Guidelines for interpreting the results of bibliometric analysis: A sensemaking approach. Global Business and Organizational Excellence, 43(2): 17-26. https://doi.org/10.1002/joe.22229
[44] Liang, Y., You, J., Wang, R., Qin, B., Han, S. (2024). Urban transportation data research overview: A bibliometric analysis based on CiteSpace. Sustainability, 16(22): 9615. https://doi.org/10.3390/su16229615
[45] Lee, M.B., Lee, C.T., Abas, M.A., Chong, W.W.F. (2025). Advancing pedestrian safety in the era of autonomous vehicles: A bibliometric analysis and pathway to effective regulations. Journal of Traffic and Transportation Engineering (English Edition), 12(4): 772-794. https://doi.org/10.1016/j.jtte.2024.05.004
[46] Haghani, M., Behnood, A., Dixit, V., Oviedo-Trespalacios, O. (2022). Road safety research in the context of low-and middle-income countries: Macro-scale literature analyses, trends, knowledge gaps and challenges. Safety Science, 146: 105513. https://doi.org/10.1016/j.ssci.2021.105513
[47] Ma, W., Alimo, P.K., Wang, L., Abdel-Aty, M. (2022). Mapping pedestrian safety studies between 2010 and 2021: A scientometric analysis. Accident Analysis & Prevention, 174: 106744. https://doi.org/10.1016/j.aap.2022.106744
[48] Anastasiadou, K., Kehagia, F. (2025). Road safety improvement and sustainable urban mobility: Identification and prioritization of factors and policies through a multi-criteria approach. Urban Science, 9(4): 93. https://doi.org/10.3390/urbansci9040093
[49] Cuthbertson, J., Drummond, G. (2025). Prehospital care Post-Road-Crash: A systematic review of the literature. Prehospital and Disaster Medicine, 40(2): 94-100. https://doi.org/10.1017/S1049023X25000202
[50] Pinna, M., Serra, P., Porcu, E., Ponti, M., Fancello, G. (2025). An integrated approach for urban road safety analysis. Transportation Research Procedia, 90: 702-709. https://doi.org/10.1016/j.trpro.2025.06.089
[51] Khan, M.N., Das, S. (2024). Advancing traffic safety through the safe system approach: A systematic review. Accident Analysis & Prevention, 199: 107518. https://doi.org/10.1016/j.aap.2024.107518
[52] Cobo, M.J., López‐Herrera, A.G., Herrera‐Viedma, E., Herrera, F. (2012). SciMAT: A new science mapping analysis software tool. Journal of the American Society for information Science and Technology, 63(8): 1609-1630. https://doi.org/10.1002/asi.22688
[53] Raina, S., Mandal, S.K. (2024). Perception of safety in public space: A bibliometric and network analysis from 1978–2023. International Journal of Sustainable Development and Planning, 19(11): 4331-4348. https://doi.org/10.18280/ijsdp.191122
[54] Wu, Q., Su, Y., Tan, W., Zhan, R., Liu, J., Jiang, L. (2025). UAV Path planning trends from 2000 to 2024: A bibliometric analysis and visualization. Drones, 9(2): 128. https://doi.org/10.3390/drones9020128
[55] Zhi, H., Zolotova, M. (2025). Wearable devices & elderly: A bibliometric analysis of 2014–2024. Healthcare, 13(16): 2066. https://doi.org/10.3390/healthcare13162066
[56] Chen, X., Tian, W., Fang, H. (2025). Bibliometric analysis of natural language processing using CiteSpace and VOSviewer. Natural Language Processing Journal, 10: 100123. https://doi.org/10.1016/j.nlp.2024.100123
[57] Zhang, C., Yang, X., Ye, J., Cai, Y., Zhang, H., Fang, Y., Zhang, L., Cai, S. (2025). Mapping the research landscape of traditional Chinese medicine in insomnia management: A bibliometric study (2005–2024). Frontiers in Neurology, 16: 1614948. https://doi.org/10.3389/fneur.2025.1614948
[58] Alesaily, Z., Albialy, A. (2025). Future cities: A bibliometric review, 1875 to 2024. Sustainable Futures, 9: 100801. https://doi.org/10.1016/j.sftr.2025.100801
[59] Hassan, M., Mahin, H.D., Al Nafees, A., Paul, A., Shraban, S.S. (2025). Big data applications in intelligent transport systems: A bibliometric analysis and review. Discover Civil Engineering, 2(1): 49. https://doi.org/10.1007/s44290-025-00205-z
[60] İnce, E.C. (2025). Mapping the path to sustainable urban mobility: A bibliometric analysis of global trends and innovations in transportation research. Sustainability, 17(4): 1480. https://doi.org/10.3390/su17041480
[61] Du, Z., Deng, M., Lyu, N., Wang, Y. (2023). A review of road safety evaluation methods based on driving behavior. Journal of Traffic and Transportation Engineering (English Edition), 10(5): 743-761. https://doi.org/10.1016/j.jtte.2023.07.005
[62] Guo, F., Zhou, Y., Wang, X., Li, W., Cai, J. (2024). Literature review of driving fatigue research based on bibliometric analysis. Journal of Traffic and Transportation Engineering (English Edition), 11(6): 1401-1419. https://doi.org/10.1016/j.jtte.2024.03.005
[63] Angarita-Zapata, J.S., Maestre-Gongora, G., Calderín, J.F. (2021). A bibliometric analysis and benchmark of machine learning and automl in crash severity prediction: The case study of three colombian cities. Sensors, 21(24): 8401. https://doi.org/10.3390/s21248401
[64] Al-Mahbashi, M., Li, G., Peng, Y., Al-Soswa, M., Debsi, A. (2025). Real-time distracted driving detection based on GM-YOLOv8 on embedded systems. Journal of Transportation Engineering, Part A: Systems, 151(3): 04024126. https://doi.org/10.1061/JTEPBS.TEENG-8681
[65] Wang, Y., Li, Z., Liu, P., Xu, C., Chen, K. (2024). Surrogate safety measures for traffic oscillations based on empirical vehicle trajectories prior to crashes. Transportation Research Part C: Emerging Technologies, 161: 104543. https://doi.org/10.1016/j.trc.2024.104543
[66] World Health Organization (WHO). (2023). Global status report on road safety 2023. World Health Organization. https://www.who.int/publications/i/item/9789240086517.
[67] Mei, T., Liu, H., Tong, B., Tong, C., Zhu, J., Wang, Y., Kou, M. (2025). Exploring knowledge domain of intelligent safety and security studies by bibliometric analysis. Sustainability, 17(4): 1475. https://doi.org/10.3390/su17041475
[68] Chen, B., Zhao, X., Li, Y., Liu, X. (2025). Mapping the knowledge domain of crash risk in older drivers studies: A scientometric analysis. Journal of Traffic And Transportation Engineering (English Edition), 12(3): 587-602. https://doi.org/10.1016/j.jtte.2024.05.002
[69] Habibi, R.F., Prasetyo, C., Ivan, M.T., Rijaluddin, A., Prasetijo, J. (2025). A systematic literature review on the influence of road curves on urban mobility performance and safety. RIGGS: Journal of Artificial Intelligence and Digital Business, 4(2): 5578-5589. https://doi.org/10.31004/riggs.v4i2.1474
[70] Hartati, N.D., Puspitasari, N.R., Muizz, F.A., Rijaluddin, A., Taufik, M. (2025). The impact of local road infrastructure on driver safety: A systematic literature review. RIGGS: Journal of Artificial Intelligence and Digital Business, 4(2): 5207-5218. https://doi.org/10.31004/riggs.v4i2.1389
[71] Mei, Z., Gong, J., Que, Z., Pan, J. (2025). Impacts of external factors on crash injury severity in urbanised areas: An exploratory analysis. IET Intelligent Transport Systems, 19(1): e70040. https://doi.org/10.1049/itr2.70040
[72] Marzouk, M., Bin Mahmoud, A.A., Al-Gahtani, K.S., Adel, K. (2025). Automation in construction (2000–2023): Science mapping and visualization of journal publications. Buildings, 15(15): 2789. https://doi.org/10.3390/buildings15152789
[73] Permana, I., Permana, A., Awaliyah, A., Herdiyana, A.D., Prasetijo, J. (2025). Bibliometric analysis of road performance and level of service (LoS) using VOSviewer. Leader Civil Engineering and Architecture Journal, 3(4): 221-238. https://doi.org/10.37253/leader.v3i4.10920
[74] Giri, O.P., Shahi, P.B., Bhaumik, A., Poddar, S. (2022). A bibliometric analysis on road traffic safety, 2009 to 2022. SSRN Electronic Journal, 31(2): 230-252. https://doi.org/10.2139/ssrn.4148189
[75] Feizizadeh, B., Omarzadeh, D., Sharifi, A., Rahmani, A., Lakes, T., Blaschke, T. (2022). A GIS-based spatiotemporal modelling of urban traffic accidents in Tabriz City during the COVID-19 pandemic. Sustainability, 14(12): 7468. https://doi.org/10.3390/su14127468
[76] Mohammed, H., Jaff, D., Schrock, S. (2019). The challenges impeding traffic safety improvements in the Kurdistan Region of Iraq. Transportation Research Interdisciplinary Perspectives, 2: 100029. https://doi.org/10.1016/j.trip.2019.100029
[77] Elvik, R. (2024). The development of a road safety policy index and its application in evaluating the effects of road safety policy. Accident Analysis & Prevention, 202: 107612. https://doi.org/10.1016/j.aap.2024.107612
[78] Darkhaneh, M.E., Effati, M., Arabani, M. (2025). Factors affecting the injury severity of head-on crashes on undivided rural roads under different weather conditions. International Journal of Transportation Science and Technology, 21: 384-398. https://doi.org/10.1016/j.ijtst.2025.02.007
[79] Inada, H., Ichikawa, M. (2025). Association between automatic emergency braking and pedestrian and cyclist injury severity in Japan. Accident Analysis & Prevention, 218: 108091. https://doi.org/10.1016/j.aap.2025.108091
[80] Elfahim, O., El Midaoui, M., Youssfi, M., Bouattane, O. (2023). Traffic violations analysis: Identifying risky areas and common violations. Heliyon, 9(9): e19058. https://doi.org/10.1016/j.heliyon.2023.e19058
[81] Abdalazeem, M., Oke, J. (2025). Roadway crash typology of census tracts enables targeted interventions via interpretable machine learning. Data Science for Transportation, 7(2): 14. https://doi.org/10.1007/s42421-025-00128-2
[82] Calder, R.S., Summa, C., Clark, R. (2025). Trends and disparities in motor vehicle collision injuries in Washington, DC. Accident Analysis and Prevention, 223: 108243. https://doi.org/10.1016/j.aap.2025.108243
[83] Hasan, A.S., Patel, D., Islam, M.S., Al-Sheikh, O., Jalayer, M. (2025). Identifying distracted driving hotspots using an event-to-crash conversion method: A case study from New Jersey. Case Studies on Transport Policy, 22: 101604. https://doi.org/10.1016/j.cstp.2025.101604
[84] Wang, C., Serre, T. (2025). A hybrid approach to investigating factors associated with crash injury severity: Integrating interpretable machine learning with logit model. Applied Sciences, 15(19): 10417. https://doi.org/10.3390/app151910417
[85] Younes, S., Oloufa, A. (2025). A geospatial framework for spatiotemporal crash hotspot detection using space–time cube modeling and emerging pattern analysis. Urban Science, 9(10): 411. https://doi.org/10.3390/urbansci9100411
[86] Almasi, S.A. (2025). Evaluating the efficiency of spatial-geographical models for vehicle crash frequency estimation: A case study on the urban road network of Hamadan Province. Transportation Engineering, 21: 100362. https://doi.org/10.1016/j.treng.2025.100362
[87] Xiao, Y., Xu, J., Chraibi, M., Zhang, J., Gou, C. (2022). A generalized trajectories-based evaluation approach for pedestrian evacuation models. Safety Science, 147: 105574. https://doi.org/10.1016/j.ssci.2021.105574
[88] Trivedi, P., Shah, J., Moslem, S., Pilla, F. (2023). An application of the hybrid AHP-PROMETHEE approach to evaluate the severity of the factors influencing road accidents. Heliyon, 9(11): e21187. https://doi.org/10.1016/j.heliyon.2023.e21187
[89] Meng, L., Tao, J., Zheng, G., Ren, J., Shang, L., Li, D., Liu, H., Bao, Y., Hua, B. (2025). Bone metastasis and pain research through the dual lens of Bibliometrics and bioinformatics: knowledge structure, frontiers, and core pathway analysis (2015–2024). Frontiers in Medicine, 12: 1619607. https://doi.org/10.3389/fmed.2025.1619607
[90] Das, S., Kong, X., Wei, Z., Liu, J. (2023). Scientometric and bibliographic analysis of pedestrian safety research. Transportation Research Record, 2677(12): 65-82. https://doi.org/10.1177/03611981231167158
[91] Wei, T., Zhu, T., Lin, M., Liu, H. (2024). Predicting and factor analysis of rider injury severity in two-wheeled motorcycle and vehicle crash accidents based on an interpretable machine learning framework. Traffic Injury Prevention, 25(2): 194-201. https://doi.org/10.1080/15389588.2023.2284111
[92] Fu, Y., Li, C., Yu, F.R., Luan, T.H., Zhang, Y. (2021). A survey of driving safety with sensing, vehicular communications, and artificial intelligence-based collision avoidance. IEEE Transactions on Intelligent Transportation Systems, 23(7): 6142-6163. https://doi.org/10.1109/TITS.2021.3083927
[93] Wen, X., Xie, Y., Jiang, L., Pu, Z., Ge, T. (2021). Applications of machine learning methods in traffic crash severity modelling: Current status and future directions. Transport Reviews, 41(6): 855-879. https://doi.org/10.1080/01441647.2021.1954108
[94] Breiman, L. (2001). Random forests. Machine Learning, 45(1): 5-32. https://doi.org/10.1023/A:1010933404324
[95] Rezashoar, S., Kashi, E., Saeidi, S. (2026). Comparison of machine learning algorithms for predicting traffic accident severity (case study: United Kingdom from 2010 to 2014). International Journal of Crashworthiness, 31(2): 124-133. https://doi.org/10.1080/13588265.2025.2507474
[96] Dhibi, M. (2019). Road safety determinants in low and middle income countries. International Journal of Injury Control and Safety Promotion, 26(1): 99-107. https://doi.org/10.1080/17457300.2018.1482926
[97] Malashenko, M., Gutman, S. (2025). Modelling of road transport safety indicators in russian regions. Sustainability, 17(14): 6584. https://doi.org/10.3390/su17146584
[98] Cobo, M.J., López‐Herrera, A.G., Herrera‐Viedma, E., Herrera, F. (2011). Science mapping software tools: Review, analysis, and cooperative study among tools. Journal of the American Society for information Science and Technology, 62(7): 1382-1402. https://doi.org/10.1002/asi.21525
[99] Elvik, R., Høye, A., Vaa, T., Sørensen, M. (2009). The Handbook of Road Safety Measures. Emerald Group Publishing Limited. https://doi.org/10.1108/9781848552517
[100] Kaplan, S., Guvensan, M.A., Yavuz, A.G., Karalurt, Y. (2015). Driver behavior analysis for safe driving: A survey. IEEE Transactions on Intelligent Transportation Systems, 16(6): 3017-3032. https://doi.org/10.1109/TITS.2015.2462084
[101] Elvik, R. (2009). The Power Model of the Relationship between Speed and Road Safety: Update and New Analyses (No. 1034/2009). Transportøkonomisk institutt. http://worldcat.org/isbn/9788248010012.
[102] Zhang, Z., He, Q., Gao, J., Ni, M. (2018). A deep learning approach for detecting traffic accidents from social media data. Transportation Research Part C: Emerging Technologies, 86: 580-596. https://doi.org/10.1016/j.trc.2017.11.027