Factors, Prediction, and Explainability of Vehicle Accident Risk Due to Driving Behavior through Machine Learning: A Systematic Literature Review, 2013–2023

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This report reviews published studies on vehicle accident risk due to driving behavior (DBVAR). It presents several findings, including the factors used in DBVAR studies, and highlights significant concerns and potential challenges that require further investigation.

1. As the title suggests, this paper aims to review studies investigating vehicle accident risk due to driving behavior. However, the main text covers a variety of topics and deviates from the specified scope.

In the database search string, driving behavior and its variations are mixed with other research topics, as shown in the search string: ("vehicle accident risk" OR "car accident risk" OR "car following" OR "driving behavior" OR "driving style" OR "driver behavior" OR "driving risk" OR "driver risk" OR "road safety"). This approach is problematic because: i) it includes many out-of-scope variables, and more importantly, ii) it compromises the relationship between the variables of interest. This results in search outcomes that include studies where either driving behavior or other variables are present, rather than focusing on vehicle accident risk due to driving behavior.

The authors need to redesign the search logic to explicitly reveal the relationship between vehicle accident risk and driving behavior by reviewing relevant publications. The Results section presents different factors used in DBVAR. However, without addressing the critical question of how driving behavior causes vehicle accident risk, these analyses fall short of providing useful information to support the study.

2. The authors have included a variety of studies in the review. However, a more in-depth analysis is necessary to provide insight into the development of research on vehicle accident risk due to driving behavior.

For instance, the authors list different factors used in DBVAR and sort the results by their frequency in studies. However, it is necessary to synthesize these factors, explaining how they are defined and used in the published literature, as well as exploring possible relationships between them. Simply listing the factors does not help to understand how these variables relate to the research topic, compromising the quality of this work.

Another example is that the authors list ML/DL algorithms used in DBVAR and provide brief results about their performance. However, there are many inconsistencies between and within the algorithms. The authors need to provide detailed evidence on model development and evaluation for each algorithm, summarize the performance across publications, and explore potential reasons for the inconsistencies.

3. The authors should revise the manuscript to improve the language. For instance, the first sentence of the second paragraph on page 3 states, "Art studies reveal that a lot of knowledge exists and needs to be inventoried, analyzed, and classified." This sentence is unclear and lacks a reference to support the claim. Additionally, the term "art studies" is ambiguous in this context. The authors need to clarify the meaning and provide appropriate references for such statements.

Comments on the Quality of English Language

  This report needs moderate revision to enhance the language.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

Comments and Suggestions for Authors

The manuscript analyzes the factors influencing vehicle accident risks due to driving behavior and the methods for their assessment.

1. Some of the images, such as Figures 4 and 5, are not clear. Please provide higher resolution versions.

2. The Introduction section requires revision to improve its coherence. Moreover, it is necessary to highlight the contributions of this review.

3. Sections 2 and 4.4 are overly brief, with only a few lines describing the methodology and limitations. These sections should be expanded to provide a more comprehensive explanation.

4. The units of measurement in Table 10 are inconsistent. Please ensure that all results are presented with uniform units for clarity.

Comments on the Quality of English Language

Moderate editing of English language is required.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

I appreciate the authors' responses and revisions. The inclusion of additional information enhances the integrity and transparency of this paper, addressing most of my concerns and suggestions. However, I have a follow-up suggestion. More details should be provided regarding the advances in ML for DBVAR prediction. As noted in my previous comments, there are substantial discrepancies in performance even with the same or similar ML/DL models. The authors should at least summarize the key factors influencing model performance for commonly used models. Simply listing the model performance metrics is less informative and may cause further confusion. If the authors incorporate this feedback, I do not need to review the revised version again.

Comments on the Quality of English Language

Minor revision of the language is suggested.

Author Response

Please see the attachment.

Author Response File: Author Response.docx