computation

Review
Factors, Prediction, and Explainability of Vehicle Accident Risk
Due to Driving Behavior through Machine Learning:
A Systematic Literature Review, 2013–2023
Javier Lacherre 1, * , José Luis Castillo-Sequera 2 and David Mauricio 1

1 Faculty of Systems Engineering and Informatics, National University of San Marcos, Lima 15081, Peru;
[email protected]
2 Department of Computer Sciences, Polytechnic School, University of Alcala, 28871 Alcala de Henares, Spain;
[email protected]
* Correspondence: [email protected]

Abstract: Road accidents are on the rise worldwide, causing 1.35 million deaths per year, thus
encouraging the search for solutions. The promising proposal of autonomous vehicles stands out in
this regard, although fully automated driving is still far from being an achievable reality. Therefore,
efforts have focused on predicting and explaining the risk of accidents using real-time telematics data.
This study aims to analyze the factors, machine learning algorithms, and explainability methods most
used to assess the risk of vehicle accidents based on driving behavior. A systematic review of the
literature produced between 2013 and July 2023 on factors, prediction algorithms, and explainability
methods to predict the risk of traffic accidents was carried out. Factors were categorized into
five domains, and the most commonly used predictive algorithms and explainability methods were
determined. We selected 80 articles from journals indexed in the Web of Science and Scopus databases,
identifying 115 factors within the domains of environment, traffic, vehicle, driver, and management,
with speed and acceleration being the most extensively examined. Regarding machine learning
advancements in accident risk prediction, we identified 22 base algorithms, with convolutional neural
network and gradient boosting being the most commonly used. For explainability, we discovered
Citation: Lacherre, J.;
six methods, with random forest being the predominant choice, particularly for feature importance
Castillo-Sequera, J.L.; Mauricio, D.
analysis. This study categorizes the factors affecting road accident risk, presents key prediction
Factors, Prediction, and Explainability
algorithms, and outlines methods to explain the risk assessment based on driving behavior, taking
of Vehicle Accident Risk Due to
Driving Behavior through Machine
vehicle weight into consideration.
Learning: A Systematic Literature
Review, 2013–2023. Computation 2024, Keywords: machine learning; prediction algorithms; risk assessment; road accident
12, 131. https://doi.org/10.3390/
computation12070131

Academic Editor: Rafael Lahoz-Beltra

1. Introduction
Received: 25 April 2024 There are around 1.35 million deaths worldwide per year due to vehicle accidents [1];
Revised: 20 June 2024 in Europe, 60% of such deaths occur on two-lane roads [2]. In this regard, the United
Accepted: 26 June 2024 Nations Organization has proposed 17 sustainable development goals (SDGs) for the year
Published: 28 June 2024
2030, where SDG-3, “Good health and well-being”, aims to reduce deaths and injuries
resulting from traffic incidents by 50% worldwide [3]. One potential option is the imple-
mentation of autonomous vehicles. Nevertheless, complete automation in driving is still a
Copyright: © 2024 by the authors.
considerable distance away, making it unlikely in the foreseeable future [4]; furthermore,
Licensee MDPI, Basel, Switzerland.
extensive research is still needed, especially in the field of prediction.
This article is an open access article Since 1975, research has focused on predicting vehicle accident risk (VAR). Chipman
distributed under the terms and and Morgan [5] studied various factors such as demerit points, age, gender, license class,
conditions of the Creative Commons and accident history. Their findings highlighted demerit points as the key factor influencing
Attribution (CC BY) license (https:// future accident risk, offering a chance to prevent accidents when linked with driving
creativecommons.org/licenses/by/ behavior (DB). Extensive research over time has led to modifications and regulations in
4.0/). environmental, vehicular, traffic, driver, and management domains to reduce risks. These

Computation 2024, 12, 131. https://doi.org/10.3390/computation12070131 https://www.mdpi.com/journal/computation

Computation 2024, 12, 131 2 of 21

measures aim to reduce risks, such as using deceleration devices and central protection
barriers on roads for risk mitigation [2]. Additionally, mechanisms have been implemented
for collision prevention, pedestrian identification, lane change alerts, and detection of
driver distraction and drowsiness with feedback to the driver, among other capabilities [6].
These advancements prompted governments to implement safety manuals, such as the
Road Safety Manual, which includes a widely used VAR prediction model [7,8]; however,
this model does not consider DB in its statistical analysis [9].
The study problem in this article focuses on driving behavior (DB) and its impact on
traffic accident incidence. DB refers to the actions and responses of a driver during various
driving scenarios, encompassing the journey from an initial point to a final destination,
taking into account factors such as travel time [10]. DB can be categorized into distinct
groups with similar patterns, facilitating the estimation of driving risk levels [11]. These
groups include the following: normal, drowsy, and aggressive behaviors [12].
In the VAR context, DB holds utmost significance as it accounts for the highest in-
cidence of traffic accidents—surpassing 70% of accidents in certain countries, such as
Peru [13]. Therefore, the vehicle accident risk due to driving behavior (DBVAR) refers to
the probability of a traffic accident occurring due to actions taken by drivers behind the
wheel, which can increase the chances of suffering an accident and endanger road safety.
Identifying this risk is fundamental to protecting human lives, promoting road safety,
reducing the costs associated with traffic accidents, and developing effective safety policies.
Research has been conducted to determine factors for predicting traffic incidents using
machine learning (ML) methods. Xu et al. [14] found that there was a strong correlation
between aggressive DB and aspects of the driver, vehicle, and environment. In a similar vein,
Li et al. [15] included the environment, vehicle, driver, and traffic. Likewise, Niu et al. [16]
and Yang et al. [17] included the management domain. It is important to study each of
these factors, not only to find a better model but also to mitigate the risk of accidents and
their consequences, in addition to improving road safety [18].
Regarding prediction, different artificial intelligence algorithms have been used to
predict the risk of vehicle accidents. Geng et al. [2] presented an extensive modeling frame-
work for evaluating truck safety on two-lane rural roads using extreme gradient boosting
(XGboost), achieving an impressive accuracy of 96.67%. In the study by Peng et al. [19], it
was noted that long short-term memory (LSTM) is suitable for extracting significant and
continuous information from vehicles such as accelerations and decelerations, which they
applied for DBVAR prediction and achieved a 93.5% accuracy.
On the other hand, various algorithms have also been used for DBVAR explainability.
In the study by Masello et al. [20], Shapley additive explanations (SHAP) was applied, and it
was found that the speed limit was a very relevant factor for the riskiest events. In the same
sense, the study by Alfai et al. [21], based on the random forest (RF) feature importance
method, discovered that the most significant predictors for DBVAR were the mean speed
of the vehicle, the vehicle’s instantaneous speed, and its longitudinal acceleration.
The amount of research on DBVAR has motivated various researchers to perform state-
of-the-art studies. In the study by Bouhsissin et al. [22], 93 articles were reviewed between
2015 and 2022, from which it was highlighted that ML algorithms occupied the predominant
position with 60%, followed by deep learning (DL) algorithms and statistical methods (with
34.87% and 5.15%, respectively). The most-used algorithms were support vector machine
(SVM), logistic regression (LR), LSTM, artificial neural network (ANN), k-nearest neighbors
(KNN), RF, and convolutional neural network (CNN). In parallel, 39 relevant factors were
identified in this area. In the study by Paredes et al. [23], 27 articles were analyzed between
2015 and 2020, finding 17 ML algorithms in which Bayesian algorithms and decision
trees mainly stood out. In addition, 21 relevant factors were identified in this context,
coinciding with the results of Bouhsissin et al. [22], where the most used were acceleration,
deceleration, and speed. Likewise, in the research of Elassad et al. [24], 82 articles from the
period 2009–2019 were reviewed, and the factors and prediction aspects were analyzed. A
total of 14 general factors grouped into the dimensions of driver, vehicle, and environment
Computation 2024, 12, 131 3 of 21

were identified, and it was found that SVM, neural network (NN), Bayesian learners (BL),
and ensemble learners (EL) were the four most used algorithms, present in 72% of the
selected studies. On the other hand, in the study by Silva et al. [25], the prediction and
explainability aspects were studied in relation to the frequency of accidents and severity
classification, based on 26 articles from the period 2003–2020, and it was found that the
main techniques were KNN and decision tree (DT); however, ANN was found to be the
most suitable for predicting accident frequency. Furthermore, they highlighted the road
environment, human behaviors, accident characteristics, and vehicle-related elements as
the main contributors to the elucidation of accident causes.
Studies in the field have revealed that a wealth of knowledge exists that needs to be
inventoried, analyzed, and classified. However, in the context of ML, there is a tendency
to use algorithms that evaluate risk based on accident frequency and DBVAR, without
differentiating between light and heavy vehicles or associated factors related to vehicle
trip management. These factors include the estimated delay time to the destination or
whether a heavy vehicle is loaded or empty. Furthermore, current approaches focus
on contributing factors that explain the frequency or severity of accidents but do not
identify the factors contributing to DBVAR. This gap is crucial as regulations increasingly
mandate the incorporation of mechanisms for reading trajectory and security data. Through
analyzing these data, conducting prediction in real time, and explaining the causes, we can
significantly mitigate the number of accidents.
This study aims to systematically review all the important developed aspects related
to the factors, prediction, and explainability of DBVAR based on ML and aims to answer the
following research question: Which factors, ML advances for prediction, and explainability
methods have been investigated in relation to DBVAR?
The main contributions of this article are as follows:
• Providing a comprehensive catalog of traffic accident risk factors, classified into
five dimensions;
• Identifying the various prediction algorithms, data sets used, and performance metrics
employed in the analysis;
• Compiling the various studies utilizing multiple methods to explain factors contribut-
ing to DBVAR;
• Providing the reader with a wide range of bibliographic references that they can utilize
to delve deeper into understanding the models based on ML that facilitate prediction
and explanation of DBVAR.
This article is organized into five sections, as follows. In Section 2, the methodology
followed for the systematic review of the literature is presented. Section 3 presents the
results, focused on answering the research questions, the discussion of which is presented
in Section 4. Finally, the conclusions follow in Section 5.

2. Methodology
For this article, a systematic review of the literature was carried out based on the
model applied by Silva et al. [25] and Shiguihara et al. [26] to ensure scientific rigor, which
consisted of the following phases:
• Planning: Define the research questions to be addressed, establish the sequence of
steps to be carried out to search, and identify primary studies in indexed databases,
also including the inclusion/exclusion criteria used for the selection of articles.
• Development: The selection of primary studies is carried out in accordance with
planning, following which the quality is evaluated and the data are extracted
and synthesized.
Results: Statistics on publications are shown, and the research questions are answered
in Sections 2.3 and 3, respectively.
Computation 2024, 12, 131 4 of 21

2.1. Planning
Three research questions were proposed in order to determine the aspects developed
to understand the factors, prediction, and explainability of the DBVAR:
• RQ1: What are the factors considered in predicting DBVAR?
• RQ2: What are the advances of ML in DBVAR prediction?
• RQ3: What advances in explainable artificial intelligence (XAI) exist for DBVAR
prediction?
In order to address the research questions, we conducted a review of primary publica-
tions in journals indexed in the SCOPUS and Web of Science (WoS) databases, using the
following search string:
(“vehicle accident risk” OR “car accident risk” OR “car following” OR “driving be-
havior ” OR “driving style” OR “driver behavior ” OR “driving risk” OR “driver risk”
OR “road safety”) AND ((factors OR features OR causes) OR (predicti* OR forecast* OR
progno*) OR (explainability OR explainable OR interpretabl* OR xai)) AND (“machine
learning” OR “deep learning” OR lstm).
As shown in Table 1, the string was applied in “title-abs-key” format for Scopus and
“topic” format for WoS, considering the period from January 2013 to July 2023. Additionally,
the search was limited to publications with SCImago journal ranking impact factor. Finally,
the inclusion and exclusion criteria established in Table 2 were applied.

Table 1. Database search string.

Database Search String

TITLE-ABS-KEY ((“vehicle accident risk” OR “car accident risk” OR “car following”

OR “driving behavior” OR “driving style” OR “driver behavior” OR “driving risk”
Scopus OR “driver risk” OR “ road safety”) AND ((factors OR features OR causes) OR
(predicti* OR forecast* OR progno*) OR (explainability OR explainable OR
interpretabl* OR xai)) AND (“machine learning” OR “deep learning”))
Results for (“vehicle accident risk” OR “car accident risk” OR “car following” OR
“driving behavior” OR “driving style” OR “driver behavior” OR “driving risk” OR
WoS “driver risk” OR “road safety”) AND ((factors OR aspects OR causes) OR (predicti*
OR forecast*) OR (explainability OR explainable OR interpretable OR xai)) AND
(“machine learning” OR “deep learning”) (Topic)

Table 2. Inclusion and exclusion criteria.

Inclusion Criteria Exclusion Criteria

CI1: Studies that answer the research questions CE1: Studies aimed at cost reduction
(factors, prediction models, or explainability) CE2: Studies not related to
CI2: Primary type studies vehicular transportation
CI3: Studies that present metrics to evaluate CE3: Studies that do not present test results
the quality of predictive models CE4: Studies that are not of the “journal” type
CI4: Studies presented in English of article

2.2. Development
The possible original investigations found during the search were subjected to a
selection procedure based on the criteria detailed in Table 2, covering both inclusion and
exclusion criteria. To achieve this, it was necessary to carry out a prior review of the content,
in order to determine its relevance for the present study and find those studies related to the
factors, prediction, or explainability of DBVAR using ML. Most of the works were discarded
as they corresponded to unrelated topics such as driver identification, energy consumption,
autonomous vehicles, vehicles with fewer than four wheels, racing cars, pollution, level of
accident severity, traffic study, or time and cost optimization. Figure 1 explains the applied
process and identifies the activities carried out to select or reject studies.
 discarded as they corresponded to unrelated topics such as driver identification, energy
consumption, autonomous vehicles, vehicles with fewer than four wheels, racing cars,
pollution, level of accident severity, traffic study, or time and cost optimization. Figure 1
explains the applied process and identifies the activities carried out to select or reject stud-
Computation 2024, 12, 131 ies. 5 of 21

Figure 1. Systematic review process according to PRISMA [27].

Figure 1. Systematic review process according to PRISMA [27].
2.3. Results
2.3. Results
Potentially eligible studies and selected studies
Potentially eligible
The systematic studies
review and selected
search studies
conducted in Scopus and WoS resulted in 1674 articles,
The systematic
of which review search
80 were selected conducted
(see Table 3). in Scopus and WoS resulted in 1674 articles,
of which 80 were selected (see Table 3).
Table 3. Potentially eligible studies and selected studies.
Table 3. Potentially eligible studies and selected studies.
Source Potentially Eligible Studies Selected Studies
Source Potentially Eligible Studies Selected Studies
Scopus 1115 52
Scopus 1115 52
WoS 559 28
WoS 559 28
Total Total 1674 1674 80aa
80
a
a 26 26 studiesremoved
studies removed from
from WoS
WoSfor
forbeing duplicates
being in Scopus.
duplicates in Scopus.

Trend
Trendofofstudies
studiesbybyyear
year
TheThe number
number ofof publications
publications inin the
the aspects
aspects ofof factors,
factors, prediction,
prediction, oror explainability
explainability ofof
DBVARshowed
DBVAR showedexponential
exponential growth
growthboth
bothinin
potential articles
potential (see(see
articles Figure 2a) and
Figure 2a)inand
selected
in
articles (see Figure 2b). This could be explained by the increasing number of traffic accidents
and the introduction of ML technologies for accident prediction and explainability.
Computation 2024, 12, 131 6 of 22

selected articles (see Figure 2b). This could be explained by the increasing number of tra
Computation 2024, 12, 131 selected articles (see Figure 2b). This could be explained 6 of 22of traf-
Computation 2024, 12, 131 fic accidents and the introduction of ML technologies forby the increasing
accident number
prediction6 and
of 21 explai
fic accidents and the introduction of ML technologies for accident prediction and explain-
ability.
ability.
selected articles (see Figure 2b). This could be explained by the increasing number of traf-
fic accidents and the introduction of ML technologies for accident prediction and explain-
ability.

(a) (a) (b)

(b)
Figure (a)
2. Number of publications per year: (a) potentially
(b)eligible and (b) selected studies.
FigureFigure
2. Number of publications
2. Number of publications per year:(a)
per year: (a)potentially
potentially eligible
eligible and
and (b) (b) selected
selected studies. studies.
Figure 2. Number of publications per year: (a) potentially eligible and (b) selected studies.
Study trends across different countries
Study trends acrossdifferent
different countries
Study trends
Figure 3 across
illustrates the countries
distribution of studies based onauthors’
the authors country of affili-
Study trends across
Figure 3 illustrates thedifferent countries
distribution of studies based on the country of affilia-
Figure
ation, 3
with illustrates
China
Figure
tion, with and the
the
3 illustrates
China and the thedistribution
United States
distribution
United of studies
representingbased
of studies based
States representing 45%
45% ofon on
of
the
the the
the authors
country of affili- of affi
total
authors
total country
concentration.
concentration.
ation, ation,
with with
China and the United States representing 45% of the total concentration.
China and the United States representing 45% of the total concentration.

Figure Studies by
byauthors’
authors country of affiliation.
Figure 3.3.3.Studies
Figure Studiesby authors country
country ofof
affiliation.
affiliation.
Articles
Articlesselected
selectedby
byjournal
journalquality
qualityfactor
factor
Figure 3. Articles
Studies selected
Regarding
by
Regarding the
authors by country
the journal
journal
journal quality
quality
quality factor,
offactor, factor
60%
affiliation.
60% (48)
(48) of the
of the articles
articles werewere categorized
categorized in
in quar-
Regarding
quartile Q1 and the
35% journal
(28) in quality
quartile factor,
Q2, 60% (48)
indicating of
that the
95% articles
of the were
articles categorized
fell within
tile Q1 and 35% (28) in quartile Q2, indicating that 95% of the articles fell within the top in quar-
the
tile
top Q1
twoand 35%
quartiles
two quartiles (28)
(see in
(see Figure quartile
Figure 4). Q2,
This indicating
highlights
4). Thisquality
highlights thethat 95%
quality ofof the
the articles
studies.
the quality of the studies. fell within the top
Articles selected by journal factor
two quartiles (see Figure 4). This highlights the quality of the studies.
Regarding the journal quality factor, 60% (48) of the articles were categorized in qua
tile Q1 and 35% (28) in quartile Q2, indicating that 95% of the articles fell within the to
two quartiles (see Figure 4). This highlights the quality of the studies.

Figure4.4.Articles
Figure Articlesby
byquality
qualityfactor.
factor.

Figure 4. Articles by quality factor.

Computation 2024, 12, 131 7 of 22
Computation 2024, 12, 131 7 of 21

Articles selected by journal

Articles selected by journal
Figure 5 illustrates that the two most prominent journals—Accident Analysis and
Figure 5 illustrates that the two most prominent journals—Accident Analysis and
Prevention and IEEE Access—were situated in the Q1 quartile and collectively accounted
Prevention and IEEE Access—were situated in the Q1 quartile and collectively accounted for
for 25% of the publications. Notably, there were 27 other journals categorized under “Oth-
25% of the publications. Notably, there were 27 other journals categorized under “Others”,
ers”, each contributing a single article.
each contributing a single article.

Figure 5. Articles by journal.

3. Results
3. Results
This section addresses the research questions posed in Section 2.1 based on the se-
This section addresses the research questions posed in Section 2.1 based on the se-
lected studies.
lected studies.
A. RQ1:
A. RQ1: What
Whatare
arethe
thefactors
factors considered
considered in
in predicting
predicting DBVAR?
DBVAR?
DB encompasses
DB encompasses aa driver
driver’s actions, awareness,
s actions, awareness, and and adherence
adherence to to road
road regulations.
regulations.
These factors can directly impact a driver’s behavior or prompt changes,
These factors can directly impact a driver s behavior or prompt changes, and compre- and compre-
hending them aids in enhancing safety standards [28]. In this context,
hending them aids in enhancing safety standards [28]. In this context, 115 factors were 115 factors were
found in 48 studies, which were classified considering three of the four
found in 48 studies, which were classified considering three of the four categories from categories from
Silva et al. [25], separating the factors related to traffic from the environment
Silva et al. [25], separating the factors related to traffic from the environment category andcategory and
adding aa management
adding managementcategory,
category,then
thenexcluding
excluding thethe
accident
accidentcategory
category(characteristics of the
(characteristics of
occurred accident type) as this was a result and not a risk, and so, it did not correspond
the occurred accident type) as this was a result and not a risk, and so, it did not correspond to
a DBVAR. The resulting categories were as follows:
to a DBVAR. The resulting categories were as follows:
(1)
(1) Environment: environment
Environment: environment andand geographical
geographical distribution.
distribution.
(2)
(2) Traffic: related to vehicles surrounding to the one
Traffic: related to vehicles surrounding to the one beingbeing studied.
studied.
(3)
(3) Vehicle: static or moving mode features.
Vehicle: static or moving mode features.
(4) Driver: related to the human who drives the vehicle.
(4) Driver: related to the human who drives the vehicle.
(5) Management: efficient vehicle fleet and drivers control and coordination.
(5) Management: efficient vehicle fleet and drivers control and coordination.
Environment factors: A total of 20 factors were found from 23 articles, where the
Environment factors: A total of 20 factors were found from 23 articles, where the
weather was the most used (in 9), followed by date–time and slope (in 8 and 5, respectively;
weather was the most used (in 9), followed by date–time and slope (in 8 and 5, respec-
see Table 4).
tively; see Table 4).
Table 4. Environmental factors used in DBVAR.
Table 4. Environmental factors used in DBVAR.

ID Factor Description
ID Factor Description # # Studies
Studies
Atmospheric Atmospheric conditions affecting visibility
conditions affecting visibility and traction, increasing and
01 Weather 01 Weather 9 9 [14–16,20,29–33]
[14–16,20,29–33]
accident risk. traction, increasing accident risk.
Specific
Specific time and date of time and date
travel, influencing ofcongestion
traffic travel, influencing
and
02 Date–time 02 Date–time 8 8 [30–32,34–38]
[30–32,34–38]
driver fatigue. traffic congestion and driver fatigue.
03 Slope Terrain inclination impacting
Terrain vehicle speed and
inclination control. vehicle speed
impacting 5 [2,14,15,18,20]
04 Lane 03 Slope
Vehicle position on the road, influencing collision risk. 5 5 [20,34,39–41]
[2,14,15,18,20]
and control.
05 Road condition Pavement quality and obstacles
Vehicle compromising
position on the safety.
road, influencing 4 [15,16,18,20]
04 Lane 5 [20,34,39–41]
collision risk.
Computation 2024, 12, 131 8 of 21

Table 4. Cont.

ID Factor Description # Studies

Meteorological Atmospheric elements such as rain and snow, affecting visibility

06 4 [15,20,29,31]
conditions and vehicle adherence.
07 Light conditions Level of available light impacting visibility and driver reaction time. 4 [20,30–32]
08 Road type Road design affecting speed and maneuverability. 3 [12,20,42]
09 Road obstruction Roadside obstacles posing hazards. 3 [16,41,43]
10 Curve type Shape and degree of road curves that affect the driver’s driving. 3 [14,18,20]
11 Segment length Distance between road reference points. 3 [18,40,42]
12 Curve radius Measure of road curve curvature. 2 [2,40]
13 Road safety Presence of safety measures on the road. 1 [18]
14 Number of lanes Quantity of lanes available on the road. 1 [40]
15 Weekday Day of the week of travel. 1 [30]
16 Road Measurements Specific road data. 1 [40]
17 Crosswalk Designated pedestrian crossing areas. 1 [31]
18 Population density Number of individuals living in a specific area. 1 [17]
19 Employment density Concentration of workplaces in a given area. 1 [17]
20 Land use Utilization of land along the road. 1 [17]

Traffic factors: A total of 17 were identified, where the most studied were the distance
between two vehicles, the time to collision, and the traffic density, in 13, 10, and 9 studies
out of 25, respectively (see Table 5).

Table 5. Traffic factors used in DBVAR.

ID Factor Description # Studies

Distance between Space separating two vehicles on the road, influencing

01 13 [2,11,14,30,33–35,39,42,44–47]
two vehicles collision likelihood.
Estimated time before a collision between vehicles,
02 Time to collision assuming current speeds and trajectories 10 [11,14,29,33–35,42,46,48,49]
remain unchanged.
Volume of vehicles on the road, impacting
03 Traffic density 9 [2,14,17,18,20,30,31,34,35]
accident frequency.
Driving at a speed exceeding legal limits, elevating
04 Overspeeding 5 [14,15,20,31,40]
accident risk.
Speed difference Variation in velocity between two vehicles, affecting
05 4 [30,33,39,45]
between two vehicles collision potential.
Signs indicating traffic regulations or hazards,
06 Road signals 3 [15,17,20]
influencing driver actions and accident likelihood.
07 Time headway Time interval between vehicles, affecting collision risk. 3 [42,46,48]
Degree of vulnerability to vehicular accidents,
08 Accident risk level 3 [11,31,40]
influenced by various factors.
Mean velocity of vehicles, affecting
09 Average speed 2 [2,40]
accident probability.
Distribution of vehicle types on the road, impacting
10 Density by vehicle type 2 [2,17]
accident dynamics.
Non-compliance with
11 Failure to adhere to traffic laws, elevating accident risk. 2 [16,43]
regulations
Lateral distance Distance between vehicles and roadside objects,
12 1 [44]
with objects affecting collision probability.
Computation 2024, 12, 131 9 of 21

Table 5. Cont.

ID Factor Description # Studies

Acceleration difference Variation in acceleration rates between vehicles,

13 1 [39]
between two vehicles influencing collision potential.
There is a Presence of neighboring vehicles, affecting driving
14 1 [15]
surrounding vehicle dynamics and accident risk.
Absence or lax enforcement of traffic regulations,
15 Lack of laws 1 [16]
increasing accident likelihood.
State of traffic signals, influencing driver behavior and
16 traffic light status 1 [50]
accident risk.
Vehicle in front with Leading vehicle using high beam headlights, impacting
17 1 [16]
high beams visibility and accident risk.

Vehicle factors: A total of 44 factors were identified in 39 articles, where the most
used were speed, acceleration, and steering angle, in 27, 23, and 9 studies, respectively (see
Table 6).

Table 6. Vehicle factors used in DBVAR.

ID Factor Description # Studies

The rate at which a vehicle is traveling, measured in

distance per unit of time, directly impacting the vehicle’s [10,11,14,15,21,29,30,33,35–
01 Speed 27
ability to respond to hazards and increasing the severity of 37,39,42–45,47,48,50–58]
potential collisions.
The rate of change of velocity of a vehicle over time, either
[11,12,14,15,19,21,30,33,34,
increasing or decreasing, crucial for determining the
02 Acceleration 23 36,38,39,42,44,45,48,51,53,
vehicle’s ability to adjust its speed and navigate safely
55,56,58–60]
through traffic, influencing collision potential.
03 Steering angle Degree of wheel rotation, affecting vehicle trajectory. 9 [14,19,29,47,48,50,51,54,57]
Global position coordinates, crucial for navigation and
04 Vehicle GPS position 8 [11,12,19,31,32,35,37,41]
accident location determination.
Direction of vehicle travel, crucial for navigation and
05 Heading angle 7 [33,35,37,48,51,58,59]
collision avoidance.
06 Braking Deceleration of the vehicle, critical for collision avoidance. 6 [10,15,40,43,53,58]
Position of accelerator and brake pedals, impacting vehicle
07 Pedals position 6 [29,48,51,52,54,57]
speed control.
Assigned lane on the road, influencing collision risk
08 Lane number 5 [11,34,47,51,58]
during lane changes.
Angle of rotation around the vertical axis, affecting
09 Yaw angle 5 [12,29,35,44,55]
vehicle stability.
10 RPM Engine speed, influencing vehicle acceleration and control. 5 [21,29,51,52,57]
Temperature of the engine coolant, impacting
11 Coolant temperature 4 [21,29,36,52]
vehicle performance.
Classification of the vehicle, influencing handling
12 Vehicle type 3 [11,29,39]
characteristics and collision dynamics.
Change in lane position, increasing collision risk due to
13 Lane change 3 [33,51,53]
potential blind spots.
Length of the vehicle, influencing maneuverability and
14 Vehicle length 3 [2,11,19]
collision severity.
Angle of vehicle balance, affecting stability and risk
15 Balancing angle 3 [12,35,55]
of rollover.
Computation 2024, 12, 131 10 of 21

Table 6. Cont.

ID Factor Description # Studies

Demand placed on the engine, impacting vehicle

16 Engine load 3 [21,52,54]
performance and stability.
Remaining combustible in the tank, critical for propulsion
17 Fuel 3 [21,29,57]
and impacting vehicle range.
The specific maneuver a vehicle intends to execute, such as
18 Turn type a left turn, right turn, or U-turn, crucial for anticipating 3 [10,16,53]
traffic flow and collision avoidance
Rate of change of acceleration, affecting passenger comfort
19 Jerk 2 [11,30]
and vehicle control.
20 Pitch angle Angle of vehicle tilt, influencing stability and collision risk. 2 [12,55]
Temperature of the braking system, affecting braking
21 Brake temperature 2 [16,29]
efficiency and collision avoidance.
Elevation above sea level, influencing engine performance
22 Altitude 2 [21,37]
and vehicle handling.
Distance covered by the vehicle, impacting fatigue and
23 Traveled distance 2 [29,36]
collision risk.
Malfunction of the braking system, increasing
24 Brake failure 2 [16,19]
collision risk.
Vehicle direction of travel, crucial for collision avoidance
25 Directional 2 [10,57]
and navigation.
Height of the vehicle suspension, affecting stability and
26 Suspension height 2 [19,29]
collision risk.
Width of the vehicle, impacting maneuverability and
27 Vehicle Width 1 [19]
collision risk.
Abrupt changes in acceleration, impacting passenger
28 Harsh accelerations 1 [40]
comfort and vehicle control.
Mechanism for engaging and disengaging the engine from
29 Clutch 1 [15]
the transmission, crucial for vehicle control.
Angle of the vehicle wheels, influencing steering and
30 Wheel angle 1 [51]
collision risk.
Forces acting on the vehicle in three-dimensional space,
31 G force in all three axes 1 [29]
impacting vehicle stability and control.
Lubricant for the vehicle engine, crucial for engine
32 Oil 1 [29]
function and longevity.
Pressure in the vehicle cooling system, impacting engine
33 Water pressure 1 [29]
temperature regulation.
Pressure in the vehicle tires, crucial for tire performance
34 Air pressure 1 [21]
and vehicle stability.
Contact points between the vehicle and the road, crucial
35 Tires 1 [29]
for traction and vehicle control.
Damaged rear-view Impaired visibility to the rear of the vehicle, increasing
36 1 [16]
mirror collision risk.
Warning system for exceeding speed limits, crucial for
37 Overspeed alarm 1 [16]
collision avoidance.
Loaded with Transporting dangerous substances, increasing collision
38 1 [16]
hazardous material risk and potential for environmental damage.
Damaged Impaired visibility in adverse weather conditions,
39 1 [16]
windshield wiper increasing collision risk.
Transmission setting, impacting vehicle speed
40 Gear 1 [10]
and acceleration.
Mechanism for transferring engine power to the wheels,
41 Transmission 1 [19]
crucial for vehicle propulsion.
Computation 2024, 12, 131 11 of 21

Table 6. Cont.

ID Factor Description # Studies

Gear setting for backward vehicle movement, crucial for

42 Reverse 1 [10]
maneuvering and collision avoidance.
Audible warning device, crucial for communication and
43 Horn 1 [10]
collision avoidance.
Illumination for visibility in low-light conditions, crucial
44 Vehicle exterior light 1 [16]
for collision avoidance.

Driver factors: A total of 25 were identified, where the most used were heart rate and
eye, in 4 studies each, representing 39% of the 18 studies (see Table 7).

Table 7. Driver factors used in DBVAR.

ID Factor Description # Studies

Pulse rate indicating stress or fatigue levels affecting

01 Heart rate 4 [29,37,51,59]
driving performance.
Eye movements and tracking, influencing attention and
02 Eye 4 [50,51,61,62]
reaction times.
03 Head Head position and movement, indicating focus and awareness. 3 [50,61,62]
04 Age Driver’s age, impacting reflexes and driving abilities. 3 [14,16,42]
05 Distraction Level of attentional diversion from driving tasks. 3 [14,16,63]
06 Electrocardiogram (ECG) Heart activity measurement, indicating stress or health issues. 2 [18,64]
07 Electrodermal Activity Skin conductance reflecting stress or arousal levels. 2 [51,59]
08 Breathing frequency Rate of breathing, indicating stress or fatigue. 2 [32,59]
A driver attribute used to analyze differences in driving behavior and
09 Gender 2 [14,42]
accident risk.
10 Driving experience Duration of driving practice, affecting skill and accident risk. 2 [16,42]
11 Driver’s mood Emotional state impacting focus and decision-making. 2 [16,34]
Electroencephalogram
12 Brain activity measurement indicating alertness levels. 1 [56]
(EEG)
13 Temperature Body temperature, affecting comfort and concentration. 1 [51]
14 Sleep Amount of rest influencing alertness and reaction times. 1 [14]
15 Driver video Visual monitoring of driver behavior and attention. 1 [18]
16 Educational background Education level influencing knowledge and adherence to traffic rules. 1 [16]
Origin of driver, potentially affecting driving habits and
17 Birthplace 1 [16]
risk perception.
Classification of the license, indicating permitted vehicle types and
18 Driver’s license type 1 [16]
driver qualifications.
19 Extreme excitement High arousal levels impacting decision-making and control. 1 [16]
Unaware of road
20 Lack of awareness about current road status, increasing accident risk. 1 [16]
conditions
21 Perinasal perspiration Sweat around the nose indicating stress or discomfort. 1 [59]
22 Face Facial expressions reflecting emotions and attention levels. 1 [61]
23 Mouth Mouth movements indicating speech or stress levels. 1 [62]
24 Reaction time Speed of response to stimuli, crucial for accident avoidance. 1 [45]
25 Driving time Duration of driving, impacting fatigue and alertness. 1 [15]

Management factors: A total of nine were identified in two studies (see Table 8).
Computation 2024, 12, 131 12 of 21

Table 8. Management factors used in DBVAR.

ID Factor Description # Studies

Assessment of a driver’s performance and skills, impacting safety

01 Driver evaluation 1 [16]
and accident risk.
Extended work hours contributing to driver fatigue and increased
02 Overtime work 1 [16]
accident risk.
Oversight of vehicle operations, ensuring safety and reducing
03 Vehicle management 1 [16]
accident likelihood.
Programs aimed at improving driver safety and reducing
04 Safety training 1 [16]
risky behaviors.
Measures ensuring driver well-being, impacting alertness and
05 Drivers’ care 1 [16]
accident risk.
06 Workload Amount of work assigned to drivers, affecting fatigue and focus. 1 [16]
07 Units monitoring Surveillance of vehicles to ensure compliance with safety standards. 1 [16]
08 Distance to destination point Remaining distance influencing driver fatigue and decision-making. 1 [17]
The concentration of storage locations in an area, influencing traffic
Density of
09 patterns and accident risks through truck and delivery vehicle flow, 1 [17]
warehousing facilities
potentially increasing congestion and interactions with other traffic.

On the other hand, the DBVAR prediction studies considered four variables of interest
that described the accident, which were presented in two studies (see Table 9) and were
used as a prediction object.

Table 9. Variables that describe the accident used in DBVAR prediction.

Studies Factor

[31] Severity, number of accidents

[32] Accident type, accident causes

B. RQ2: What are the advances of ML in DBVAR prediction?

Prediction based on statistical or ML methods allows behavior to be predicted in
the case of an event, in order to predict probable future results such as DB or traffic
accidents [30]. These models use the factors as input to make predictions; however, once
the result is obtained, the reasoning behind the decision becomes unknown, and it is
not possible to determine which of the factors has contributed most significantly to the
generated effect [65]. For this reason, they are called “closed box” techniques, and to fully
understand them, the use of additional explainability techniques is necessary.
To address this question, we examined 76 studies, identifying 22 core algorithms.
Among these, CNN and GB emerged as the primary choices, which were featured in 19 and
15 studies, respectively. These algorithms were employed either independently or in hybrid
models, resulting in the highest accuracy with XGBoost at 100%, as detailed in Table 10.
Additionally, Table 11 reveals that out of all the conducted studies, four were centered on
heavy vehicles, while two focused on rural roads. Moreover, Table 12 underscores that
the primary aspect under scrutiny in the analysis of DB was driving style, comprising
45 studies.
Computation 2024, 12, 131 13 of 21

Table 10. Algorithms used in the DBVAR.

Studies Algorithm a Data Set Study Area Result

ANN:
[54] D.B.D. Turkey acc = 90.00%
Backpropagation Levenberg–Marquardt
[52] ANN Own India acc = 99.00%
[66] SdsAE D.B.D. Turkey acc = 98.33%
[67] CNN:4D MRL Eye Czech Republic acc = 97.53%
[6] CNN: Inception v3 Own Mexico acc = 92.80%
[68] CNN: EFFNet-CA SF3D USA acc = 99.58%
[69] CNN SF3D USA acc = 95.00%
[70] CNN: HCF SF3D USA acc = 96.74%
[58] CNN Own Japan acc = 83.00%
[71] CNN: BiFPN DMD - acc = 95.60%
[64] CNN: DCNN Own USA acc = 95.51%
[72] CNN: DenseNet + GA SF3D USA acc = 99.80%
[57] CNN: GoogleNet Own - acc = 89.00%
[73] CNN: LWANet (VGG16) SF3D USA acc = 99.37%
[74] CNN: MobileNet COCO USA acc = 90.00%
[75] CNN: MobileNetV2 SF3D USA acc = 99.68%
[76] CNN: MobileNetv3 3D KITTI Germany acc = 99.95%
[77] CNN: MSFLD HNUFDD China acc = 99.13%
[78] CNN: VGG-19 AUCD2 Greece acc = 95.77%
[61] CNN + LSTM Own - acc = 94.00%
[60] CNN-GRU + XGBoost SHRP 2 USA acc = 97.50%
[15] DMNM NavInfo - acc = 99.00%
[47] GB BEBO The Netherlands acc = 81.00%
[10] GB Own Pakistan acc = 97.00%
[34] GB UAH- DriveSet Spain acc = 67.00%
[16] GB: GBDT Own China acc = 80.00%
[79] GB: LightGBM HighD Germany acc = 97.58%
[80] GRU Own China acc = 93.94%
[11] GB: XGboost NGSIM USA acc = 89.00%
[2] GB: XGboost Own China acc = 96.66%
[59] GB: XGboost SIM 1 USA acc = 89.24%
[38] GB:XGBoost D.B.D. Turkey acc = 100.00%
[33] GMM: HC + FA SH-NDS China acc = 87.00%
[50] AIO-HMM RoadLab Canada acc = 86.40%
[39] LSTM HighD Germany acc = 97.00%
[19] LSTM Own - acc = 93.50%
[44] LSTM:Ensemble Classifier Own China acc = 90.50%
[14] LSTM + HMM SH-NDS China acc = 84.00%
[32] LSTM: BCDU-Net Own China acc = 98.48%
[81] ConvLSTM: UMMFN Own China acc = 97.79%
ResNet + HRNN +
[63] SF3D USA acc = 96.23%
Inception
[82] EV-CNN + LSTM SF3D USA acc = 93.68%
[35] LSTM: Stacked-LSTM UAH-DriveSet Spain acc = 99.47%
[55] LSTM: Stacked-LSTM UAH-DriveSet Spain acc = 94.00%
[45] LSTM-NN SHRP 2 USA acc = 88.00%
[83] MediaPipe face mesh Own - acc = 95.80%
Computation 2024, 12, 131 14 of 21

Table 10. Cont.

Studies Algorithm a Data Set Study Area Result

[84] MLP Own Germany acc = 87.00%

[56] MLP Own China acc = 88.00%
[42] MLP Own China acc = 69.60%
[30] MLP SH-NDS China acc = 89.20%
[85] MLP + CNN + Tranformer SF3D USA acc = 99.91%
[86] ResNet: TSD-DLN AUCD2 Greece acc = 89.50%
[62] RF NTHUDDD Taiwan acc = 99.00%
[40] RF Own Greece acc = 89.30%
[51] RF Own Canada acc = 91.78%
[46] RF Own China acc = 93.00%
[37] RF Own India acc = 98.00%
[43] RF SHRP 2 USA acc = 90.00%
Traffic, Driving Style and
[21] RF Italy acc = 95.00%
Road Surface Condition
[12] RF UAH- DriveSet Spain acc = 91.60%
[87] RF SPMD USA acc = 92.77%
[31] RF UK Car Accident 2015 United Kingdom acc = 99.00%
[65] Sequential Covering D.B.D. Turkey acc = 96.25%
[88] SVM Own Pakistan acc = 87.00%
[41] DDPG SPMD USA RMSE = 0.4254
[18] GB: LightGBM Own China RMSE = 0.004
[89] GB: LightGBM HighD Germany RMSE = 0.0447
[20] GB: XGboost Own Germany RMSE = 0.0463
[17] GB: XGboost SWITRS USA RMSE = 4.058
[36] LSTM Own Taiwan RMSE = 0.733
[90] NB Freeway-USA USA RMSD = 0.7
[29] MCS: BL + KNN + SVM + MLP Own Morocco F1 = 93.56%
[91] DTW Own India dr = 100
[53] NMF Own South Korea drs = 72.9
[49] K-Means NGSIM USA TTCi = 3.1602
[48] sHDP-HMM/NPYLM-K-Means NUDrive corpus USA ROC = 0.953
Accuracy = acc; root mean squared deviation = RMSD; F1 score = F1; root mean square error = RMSE; detection
rate = dr; mean absolute percentage error = MAPE; time to collision = TTC; driving risk score = drs; area under the
curve = AUC; receiver operating characteristic = ROC. a Acronyms used for the algorithms used in DBVAR; multi-
classifier system (MCS), Bayesian learning (BL), multi-layer perceptron (MLP), EfficientNet with channel attention
(EFFNet-CA), stacked denoising sparse AutoEncoders (SdsAE), hybrid CNN framework (HCF), bi-directional
feature pyramid network (BiFPN), deep CNN (DCNN), genetic algorithm (GA), lightweight attention-based
network (LWANet), multi-scale facial landmark detector (MSFLD), gated recurrent unit (GRU), deep deterministic
policy gradient (DDPG), deep multichannel network model (DMNM), dynamic time warping (DTW), gradient
boosting (GB), gradient boosting decision trees (GBDT), Gaussian mixture model (GMM), hierarchical clustering
(HC), factor-analysis (FA), hidden Markov model (HMM), auto-regressive input output HMM (AIO-HMM),
convolutional LSTM (ConvLSTM), bi-directional ConvLSTM U-Net (BCDU-Net), unsupervised multi-modal
fusion network (UMMFN), hierarchical recurrent neural network (HRNN), ensemble variant CNN (EV-CNN),
non-negative matrix factorization (NMF), temporal–spatial double-line DL network (TSD-DLN), negative binomial
(NB), and sticky hierarchical Dirichlet process hidden Markov model (sHDP-HMM).

Table 11. Studies applied in heavy vehicles and rural roads.

Vehicle/Road Studies

Heavy vehicles [2,16,17,53]

Type of rural road [2,40]
Computation 2024, 12, 131 15 of 21

Table 12. Studies applied by type of driving risk.

DB # Studies

Lane change 4 [39,41,48,89]

Distraction 22 [6,45,51,57,58,61,63,64,68,70–77,80–82,85,86]
[2,10–12,14–17,19–21,29–38,40,42–44,46,47,49,50,52–
Driving style 45
56,59,60,65,66,69,79,80,84,87,90,91]
Stress 1 [18]
Drowsiness 4 [62,67,83,88]

C. RQ3: What advances in XAI exist for DBVAR prediction?

XAI allows for adequate interpretation of the prediction process [17], for which models
are used to analyze the importance and dependence of the factors that contribute to
explaining the result [40]; in this way, confidence and transparency in the predictions can be
ensured, such that they can reasonably be applied in the field of transportation safety [14].
To answer this question, 18 studies were found that used six methods to explain the
factors with the greatest contribution to the DBVAR. In this context, RF and GB feature
importance were the most used (in 50%), as well as SHAP (in 33%). They mainly focused on
explaining DB in accident risk, where China and the United States were the main countries
where the studies were applied (see Table 13).

Table 13. Methods used in the explainability of the DBVAR.

Studies Method Explanation #Factors Country

[15] DB 20 China
[89] Lane change 6 Germany
[18] Driving stress 22 China
SHAP
[40] Accident risk 10 Greece
[17] Injuries in accident 16 USA
[20] Driving risk 26 Germany
[90] Driving risk 6 USA
[21] Driving style 7 Italy
[31] Accident risk 16 United Kingdom
RF features importance
[37] DB 5 India
[16] DB 15 China
Aggressive/risk DB on
[87] 23 USA
horizontal curves
[39] Lane change 7 Germany
GB features importance Driving under the
[47] influence of different 36 The Netherlands
substances
[11] DB 3 USA
[10] Laplacian punctuation DB 14 Pakistan
[51] ExtraTrees Driver distraction 19 Canada
[14] Average attention weight Aggressive DB 8 China

4. Discussion
The result of this systematic literature review is a catalog of factors, prediction algo-
rithms, and methods used to explain the importance of the factors. Researchers can use
the different results to understand progress in the field and provide new approaches to
reduce the risk of accidents to protect human lives, promote road safety, reduce the costs
associated with traffic accidents, and develop effective safety policies. The relevance of this
Computation 2024, 12, 131 16 of 21

information is validated as 95% of studies were within the first two quartiles, such that the
quality of the results is guaranteed. The research questions are discussed below.

4.1. About Factors

In this study, it was observed that the factors were classified into five dimensions
(vehicle, environment, traffic, driver, and management), where vehicle was the most
studied. Speed, acceleration, and distance between two vehicles stood out as the most-
used factors due to their direct influence on the driver’s ability to control the vehicle in
various risk situations. In addition, they also determine the level of severity of an accident.
Additional crucial factors include the geographical location, determined through the Global
Positioning System (GPS), as it enables us to comprehend other related elements, such as the
geographical environment. The increasing prevalence of cost-effective sensors and cameras
in vehicles is driving a trend toward greater data acquisition in real time, consequently
enhancing the precision of models. At present, China leads research on DBVAR factors,
probably due to the growth, leadership, and expansion of its automotive industry.
Some studies have considered the accident domain; however, this refers to the results
and not the causes. Furthermore, they tend to focus on accident characterization, and so,
they have not been considered as factors; however, they could be considered as an object to
predict. Likewise, it is important to highlight that there were no factors associated with trip
management, such as delay in delivery, the driver’s experience on the route, or whether
the vehicle was loaded. Therefore, it is important to consider management-related factors
(i.e., those in the management dimension) to evaluate commercial vehicles and improve
the understanding of vehicle accidents as a whole.

4.2. About Prediction

In this study, the algorithms, vehicles, and roads types used in accident risk prediction
research were identified. The novelty lies in the growing use of algorithms that combine
convolutional and recurrent neural networks, taking advantage of the diversity of sensors
integrated into vehicles, which generate data and images. The most commonly used
algorithm is CNN or other combinations with CNN as it allows one to take advantage of
the individual strengths of each model through sequential data processing, such as text
and image analysis. Furthermore, it was identified that the most important algorithm
in terms of accuracy for detecting distracted driving was Mobilenetv3 as it showed high
accuracy in terms of real-time pattern recognition. However, for driving style recognition,
the algorithm with the best accuracy was XGBoost. Therefore, there is a trend toward
the use of deep learning, possibly due to the availability of larger volumes of data and
advances in hardware and software, as well as the ability to achieve better performance
overall. Regarding the metrics used to evaluate the algorithms, a consensus has been
found in the use of accuracy as one of the main indicators in most studies. However, it is
important to note, as observed in Table 10, that eight different metrics were identified, and
not all studies considered the same metrics. Additionally, it is worth mentioning that only
2.6% of the studies focused on rural roads and 5% on heavy vehicles. To improve precision
in this topic, it is suggested to explore the incorporation of transformer neural networks
and dynamic Bayesian networks, which can capture long-term relationships in time series
data. Additionally, alerts can be implemented for drivers and fleet managers regarding risk
level, such that they can take preventive actions based on the provided information.
Moreover, based on Table 10, it can be determined that the most commonly used
models (present in 80% of the studies) are MLP, CNN, GB, LSTM, and RF. This could
be explained by the fact that MLP is one of the pioneering models in ML, while the
other four stand out for their high accuracy, averaging 92.94%. Furthermore, an analysis of
Tables 4–9 on factors and Table 10 on models reveals that the common factors influencing the
performance of the most commonly used models are speed, acceleration, and heading angle.
Computation 2024, 12, 131 17 of 21

4.3. About Explainability

In this study, six explainability methods were identified in 18 studies, where the most
studied was “RF feature importance,” with influencing factors related to the environment,
such as road shape, road network, and weather. The increasing adoption of deep learning
algorithms has highlighted the importance of understanding and trusting model decisions,
driving the use of explainability methods to identify influential risk factors that might not
be obvious to humans. Although the reviewed research barely addressed management
factors, it is relevant to study their importance in explainability. Furthermore, there exist
very successful methods, such as local interpretable model-agnostic explanations (LIME),
which could provide good results in this context.

5. Conclusions
For this study, we conducted a systematic literature review related to DBVAR through
ML. Out of the 1674 articles identified, 80 research papers were meticulously chosen
through analysis, enabling the discovery of advancements in the field with respect to
factors, prediction, and explainability. Within this review, we identified 115 factors across
48 studies, 22 prediction algorithms within 76 studies, and 6 explainability algorithms
across 18 studies, all of which elucidated the influence of certain factors on prediction
outcomes. Unlike other state-of-the-art studies on DBVAR, this work considered three
crucial aspects: the influencing factors, accident prediction, and explainability. In relation
to factors, we identified five dimensions: environment (20 factors), traffic (17 factors),
vehicle (44 factors), driver (25 factors), and management (9 factors). In particular, speed,
acceleration, and distance between two vehicles were the most-studied factors. In the
realm of ML advancements, CNN and GB emerged as the most commonly employed
algorithms. Moreover, there is a growing trend in leveraging deep learning and hybrid
models for enhanced precision. Notably, XGboost achieved the highest accuracy at 100%
on a DBD data set of Turkish origin. It is worth noting that the majority of studies focused
on light vehicles, with limited research conducted on heavy vehicles and rural roads. In
reference to advances in explainability, it was found that the most-used method was the
RF algorithm with feature importance. Additionally, the most studied models were MLP,
CNN, GB, LSTM, and RF, and the common factors influencing their performance were
speed, acceleration, and heading angle.
This study had some limitations that should be considered. Only studies in English
were included, and only the WoS and Scopus databases were used as sources of information.
Based on our findings, future research should focus on developing practices and strategies
to address DBVAR factors in order to reduce the occurrence of traffic accidents, as well as
extending this study to include other languages and additional databases.

Author Contributions: Conceptualization, J.L.; methodology, J.L.; formal analysis, J.L.; investigation,
J.L.; resources, J.L.; writing—original draft preparation, J.L.; writing—review and editing, J.L. and
D.M.; supervision, D.M. and J.L.C.-S. All authors have read and agreed to the published version of
the manuscript.
Funding: This research received no external funding.
Conflicts of Interest: The authors declare no conflicts of interest.

References
1. World Health Organization. Global Status Report on Road Safety—Time for Action; World Health Organization: Geneva, Switzer-
land, 2018.
2. Geng, Z.; Ji, X.; Cao, R.; Lu, M.; Qin, W. A Conflict Measures-Based Extreme Value Theory Approach to Predicting Truck Collisions
and Identifying High-Risk Scenes on Two-Lane Rural Highways. Sustainability 2022, 14, 11212. [CrossRef]
3. Naciones Unidas. La Agenda 2030 y los Objetivos de Desarrollo Sostenible: Una Oportunidad para América Latina y el Caribe; Comisión
Económica para América Latina y el Caribe (CEPAL): Santiago de Chile, Chile, 2018; ISBN 978-92-1-058643-6.
4. Kashevnik, A.; Shchedrin, R.; Kaiser, C.; Stocker, A. Driver Distraction Detection Methods: A Literature Review and Framework.
IEEE Access 2021, 9, 60063–60076. [CrossRef]
Computation 2024, 12, 131 18 of 21

5. Chipman, M.L.; Morgan, P. The Role of Driver Demerit Points and Age in the Prediction of Motor Vehicle Collisions. J. Epidemiol.
Community Health 1975, 29, 190–195. [CrossRef]
6. Celaya-Padilla, J.M.; Galván-Tejada, C.E.; Lozano-Aguilar, J.S.A.; Zanella-Calzada, L.A.; Luna-García, H.; Galván-Tejada, J.I.;
Gamboa-Rosales, N.K.; Rodriguez, A.V.; Gamboa-Rosales, H. “Texting & Driving” Detection Using Deep Convolutional Neural
Networks. Appl. Sci. 2019, 9, 2962. [CrossRef]
7. AASHTO. Highway Safety Manual; American Association of State Highway and Transportation Officials: Washington, DC,
USA, 2010.
8. Das, S.; Tsapakis, I.; Khodadadi, A. Safety Performance Functions for Low-Volume Rural Minor Collector Two-Lane Roadways.
IATSS Res. 2021, 45, 347–356. [CrossRef]
9. Tola, A.M.; Demissie, T.A.; Saathoff, F.; Gebissa, A. Crash Distribution Dataset: Development and Validation for the Undivided
Rural Roads in Oromia, Ethiopia. Transp. Telecommun. J. 2022, 23, 11–24. [CrossRef]
10. Halim, Z.; Sulaiman, M.; Waqas, M.; Aydın, D. Deep Neural Network-Based Identification of Driving Risk Utilizing Driver
Dependent Vehicle Driving Features: A Scheme for Critical Infrastructure Protection. J. Ambient Intell. Humaniz. Comput. 2023, 14,
11747–11765. [CrossRef]
11. Shi, X.; Wong, Y.D.; Li, M.Z.-F.; Palanisamy, C.; Chai, C. A Feature Learning Approach Based on XGBoost for Driving Assessment
and Risk Prediction. Accid. Anal. Prev. 2019, 129, 170–179. [CrossRef]
12. Yi, D.W.; Su, J.Y.; Liu, C.J.; Quddus, M.; Chen, W.H. A Machine Learning Based Personalized System for Driving State Recognition.
Transp. Res. Part C Emerg. Technol. 2019, 105, 241–261. [CrossRef]
13. Observatorio Nacional de Seguridad Vial Boletín Estadístico de Siniestralidad Vial, 2021. Available online: https://www.onsv.
gob.pe/post/boletin-estadistico-de-siniestralidad-vial-2021/ (accessed on 20 May 2022).
14. Xu, W.; Wang, J.; Fu, T.; Gong, H.; Sobhani, A. Aggressive Driving Behavior Prediction Considering Driver’s Intention Based on
Multivariate-Temporal Feature Data. Accid. Anal. Prev. 2022, 164, 106477. [CrossRef] [PubMed]
15. Li, D.; Wang, Y.; Xu, W. A Deep Multichannel Network Model for Driving Behavior Risk Classification. IEEE Trans. Intell. Transp.
Syst. 2023, 24, 1204–1219. [CrossRef]
16. Niu, Y.; Li, Z.M.; Fan, Y.X. Analysis of Truck Drivers’ Unsafe Driving Behaviors Using Four Machine Learning Methods. Int. J.
Ind. Ergon. 2021, 86, 103192. [CrossRef]
17. Yang, C.; Chen, M.Y.; Yuan, Q. The Application of XGBoost and SHAP to Examining the Factors in Freight Truck-Related Crashes:
An Exploratory Analysis. Accid. Anal. Prev. 2021, 158, 106153. [CrossRef]
18. Zhong, S.; Fu, X.; Lu, W.; Tang, F.; Lu, Y. An Expressway Driving Stress Prediction Model Based on Vehicle, Road and Environment
Features. IEEE Access 2022, 10, 57212–57226. [CrossRef]
19. Peng, L.; Wang, Y.; Zhang, F.; Zhang, J.; Li, Z. Evaluation of Emergency Driving Behaviour and Vehicle Collision Risk in Connected
Vehicle Environment: A Deep Learning Approach. IET Intell. Transp. Syst. 2021, 15, 584–594. [CrossRef]
20. Masello, L.; Castignani, G.; Sheehan, B.; Guillen, M.; Murphy, F. Using Contextual Data to Predict Risky Driving Events: A Novel
Methodology from Explainable Artificial Intelligence. Accid. Anal. Prev. 2023, 184, 106997. [CrossRef]
21. Al-refai, G.; Elmoaqet, H.; Ryalat, M. In-Vehicle Data for Predicting Road Conditions and Driving Style Using Machine Learning.
Appl. Sci. 2022, 12, 8928. [CrossRef]
22. Bouhsissin, S.; Sael, N.; Benabbou, F. Driver Behavior Classification: A Systematic Literature Review. IEEE Access 2023, 11,
14128–14153. [CrossRef]
23. Paredes, J.J.; Yepes, S.F.; Salazar-Cabrera, R.; de la Cruz, Á.P.; Molina, J.M.M. Intelligent Collision Risk Detection in Medium-Sized
Cities of Developing Countries, Using Naturalistic Driving: A Review. J. Traffic Transp. Eng. (Engl. Ed.) 2022, 9, 912–929.
[CrossRef]
24. Elassad, Z.E.A.; Mousannif, H.; Moatassime, H.A.; Karkouch, A. The Application of Machine Learning Techniques for Driving
Behavior Analysis: A Conceptual Framework and a Systematic Literature Review. Eng. Appl. Artif. Intell. 2020, 87, 103312.
[CrossRef]
25. Silva, P.B.; Andrade, M.; Ferreira, S. Machine Learning Applied to Road Safety Modeling: A Systematic Literature Review. J. Traffic
Transp. Eng. (Engl. Ed.) 2020, 7, 775–790. [CrossRef]
26. Shiguihara, P.; De Andrade Lopes, A.; Mauricio, D. Dynamic Bayesian Network Modeling, Learning, and Inference: A Survey.
IEEE Access 2021, 9, 117639–117648. [CrossRef]
27. Page, M.J.; McKenzie, J.E.; Bossuyt, P.M.; Boutron, I.; Hoffmann, T.C.; Mulrow, C.D.; Shamseer, L.; Tetzlaff, J.M.; Akl, E.A.;
Brennan, S.E.; et al. The PRISMA 2020 Statement: An Updated Guideline for Reporting Systematic Reviews. Syst. Rev. 2021,
10, 89. [CrossRef]
28. Alkinani, M.H.; Khan, W.Z.; Arshad, Q. Detecting Human Driver Inattentive and Aggressive Driving Behavior Using Deep
Learning: Recent Advances, Requirements and Open Challenges. IEEE Access 2020, 8, 105008–105030. [CrossRef]
29. Elassad, Z.E.A.; Mousannif, H.; Moatassime, H.A. A Real-Time Crash Prediction Fusion Framework: An Imbalance-Aware
Strategy for Collision Avoidance Systems. Transp. Res. Part C Emerg. Technol. 2020, 118, 102708. [CrossRef]
30. Shangguan, Q.; Fu, T.; Wang, J.; Luo, T.; Fang, S. An Integrated Methodology for Real-Time Driving Risk Status Prediction Using
Naturalistic Driving Data. Accid. Anal. Prev. 2021, 156, 106122. [CrossRef]
31. Zhao, H.; Li, X.; Cheng, H.; Zhang, J.; Wang, Q.; Zhu, H. Deep Learning-Based Prediction of Traffic Accidents Risk for Internet of
Vehicles. China Commun. 2022, 19, 214–224. [CrossRef]
Computation 2024, 12, 131 19 of 21

32. Hu, Z.; Zhou, J.; Zhang, E. Improving Traffic Safety through Traffic Accident Risk Assessment. Sustainability 2023, 15, 3748.
[CrossRef]
33. Wang, J.; Xu, W.; Fu, T.; Jiang, R. Recognition of Trip-Based Aggressive Driving: A System Integrated With Gaussian Mixture
Model Structured of Factor-Analysis, and Hierarchical Clustering. IEEE Trans. Intell. Transp. Syst. 2022, 23, 20442–20451.
[CrossRef]
34. Ghandour, R.; Potams, A.J.; Boulkaibet, I.; Neji, B.; Barakeh, Z.A. Driver Behavior Classification System Analysis Using Machine
Learning Methods. Appl. Sci. 2021, 11, 10562. [CrossRef]
35. Khodairy, M.A.; Abosamra, G. Driving Behavior Classification Based on Oversampled Signals of Smartphone Embedded Sensors
Using an Optimized Stacked-LSTM Neural Networks. IEEE Access 2021, 9, 4957–4972. [CrossRef]
36. Li, D.C.; Lin, M.Y.-C.; Chou, L.-D. Macroscopic Big Data Analysis and Prediction of Driving Behavior with an Adaptive Fuzzy
Recurrent Neural Network on the Internet of Vehicles. IEEE Access 2022, 10, 47881–47895. [CrossRef]
37. Arumugam, S.; Bhargavi, R. Road Rage and Aggressive Driving Behaviour Detection in Usage-Based Insurance Using Machine
Learning. Int. J. Softw. Innov. 2023, 11, 1–29. [CrossRef]
38. Kanwal, K.; Rustam, F.; Chaganti, R.; Jurcut, A.D.; Ashraf, I. Smartphone Inertial Measurement Unit Data Features for Analyzing
Driver Driving Behavior. IEEE Sens. J. 2023, 23, 11308–11323. [CrossRef]
39. Shangguan, Q.; Fu, T.; Wang, J.; Fang, S.; Fu, L. A Proactive Lane-Changing Risk Prediction Framework Considering Driving
Intention Recognition and Different Lane-Changing Patterns. Accid. Anal. Prev. 2022, 164, 106500. [CrossRef]
40. Nikolaou, D.; Ziakopoulos, A.; Dragomanovits, A.; Roussou, J.; Yannis, G. Comparing Machine Learning Techniques for
Predictions of Motorway Segment Crash Risk Level. Safety 2023, 9, 32. [CrossRef]
41. Guo, H.; Xie, K.; Keyvan-Ekbatani, M. Modeling Driver’s Evasive Behavior during Safety–Critical Lane Changes: Two-
Dimensional Time-to-Collision and Deep Reinforcement Learning. Accid. Anal. Prev. 2023, 186, 107063. [CrossRef]
42. Lyu, N.; Wang, Y.; Wu, C.; Peng, L.; Thomas, A.F. Using Naturalistic Driving Data to Identify Driving Style Based on Longitudinal
Driving Operation Conditions. J. Intell. Connect. Veh. 2022, 5, 17–35. [CrossRef]
43. Abdelrahman, A.E.; Hassanein, H.S.; Abu-Ali, N. Robust Data-Driven Framework for Driver Behavior Profiling Using Supervised
Machine Learning. IEEE Trans. Intell. Transp. Syst. 2022, 23, 3336–3350. [CrossRef]
44. Wang, H.; Wang, X.; Han, J.; Xiang, H.; Li, H.; Zhang, Y.; Li, S. A Recognition Method of Aggressive Driving Behavior Based on
Ensemble Learning. Sensors 2022, 22, 644. [CrossRef]
45. Liu, Z.; Ren, S.; Peng, M. Identification of Driver Distraction Based on SHRP2 Naturalistic Driving Study. Math. Probl. Eng. 2021,
2021, 6699327. [CrossRef]
46. Zhao, L.; Xu, T.; Zhang, Z.; Hao, Y. Lane-Changing Recognition of Urban Expressway Exit Using Natural Driving Data. Appl. Sci.
2022, 12, 9762. [CrossRef]
47. van der Wall, H.E.C.; Doll, R.J.; van Westen, G.J.P.; Koopmans, I.; Zuiker, R.G.; Burggraaf, J.; Cohen, A.F. The Use of Machine
Learning Improves the Assessment of Drug-Induced Driving Behaviour. Accid. Anal. Prev. 2020, 148, 105822. [CrossRef]
48. Yurtsever, E.; Yamazaki, S.; Miyajima, C.; Takeda, K.; Mori, M.; Hitomi, K.; Egawa, M. Integrating Driving Behavior and Traffic
Context Through Signal Symbolization for Data Reduction and Risky Lane Change Detection. IEEE Trans. Intell. Veh. 2018, 3,
242–253. [CrossRef]
49. Wang, K.; Yang, Y.; Wang, S.; Shi, Z. Research on Car-Following Model Considering Driving Style. Math. Probl. Eng. 2022,
2022, 7215697. [CrossRef]
50. Rahman, M.J.; Beauchemin, S.S.; Bauer, M.A. Predicting Driver Behaviour at Intersections Based on Driver Gaze and Traffic Light
Recognition. IET Intell. Transp. Syst. 2020, 14, 2083–2091. [CrossRef]
51. Misra, A.; Samuel, S.; Cao, S.; Shariatmadari, K. Detection of Driver Cognitive Distraction Using Machine Learning Methods.
IEEE Access 2023, 11, 18000–18012. [CrossRef]
52. Malik, M.; Nandal, R.; Dalal, S.; Jalglan, V.; Le, D.N. Driving Pattern Profiling and Classification Using Deep Learning. Intel.
Autom. Soft Comput. 2021, 28, 887–906. [CrossRef]
53. Seo, H.; Shin, J.; Kim, K.-H.; Lim, C.; Bae, J. Driving Risk Assessment Using Non-Negative Matrix Factorization with Driving
Behavior Records. IEEE Trans. Intell. Transp. Syst. 2022, 23, 20398–20412. [CrossRef]
54. Lattanzi, E.; Freschi, V. Machine Learning Techniques to Identify Unsafe Driving Behavior by Means of In-Vehicle Sensor Data.
Expert Syst. Appl. 2021, 176, 114818. [CrossRef]
55. Kadri, N.; Ellouze, A.; Ksantini, M.; Turki, S.H. New LSTM Deep Learning Algorithm for Driving Behavior Classification. Cybern.
Syst. 2023, 54, 387–405. [CrossRef]
56. Zhang, X.; Yan, X. Predicting Collision Cases at Unsignalized Intersections Using EEG Metrics and Driving Simulator Platform.
Accid. Anal. Prev. 2023, 180, 106910. [CrossRef]
57. Tran, D.; Do, H.M.; Sheng, W.H.; Bai, H.; Chowdhary, G. Real-Time Detection of Distracted Driving Based on Deep Learning. IET
Intell. Transp. Syst. 2018, 12, 1210–1219. [CrossRef]
58. Nakano, K.; Chakraborty, B. Real-Time Distraction Detection from Driving Data Based Personal Driving Model Using Deep
Learning. Int. J. Intell. Transp. Syst. Res. 2022, 20, 238–251. [CrossRef]
59. Panagopoulos, G.; Pavlidis, I. Forecasting Markers of Habitual Driving Behaviors Associated with Crash Risk. IEEE Trans. Intell.
Transp. Syst. 2020, 21, 841–851. [CrossRef]
Computation 2024, 12, 131 20 of 21

60. Shi, L.; Qian, C.; Guo, F. Real-Time Driving Risk Assessment Using Deep Learning with XGBoost. Accid. Anal. Prev. 2022,
178, 106836. [CrossRef]
61. Fan, X.; Wang, F.; Song, D.; Lu, Y.; Liu, J. GazMon: Eye Gazing Enabled Driving Behavior Monitoring and Prediction. IEEE Trans.
Mob. Comput. 2021, 20, 1420–1433. [CrossRef]
62. Albadawi, Y.; AlRedhaei, A.; Takruri, M. Real-Time Machine Learning-Based Driver Drowsiness Detection Using Visual Features.
J. Imaging 2023, 9, 91. [CrossRef]
63. Alotaibi, M.; Alotaibi, B. Distracted Driver Classification Using Deep Learning. Signal Image Video Process. 2020, 14, 617–624.
[CrossRef]
64. Taherisadr, M.; Asnani, P.; Galster, S.; Dehzangi, O. ECG-Based Driver Inattention Identification during Naturalistic Driving
Using Mel-Frequency Cepstrum 2-D Transform and Convolutional Neural Networks. Smart Health 2018, 9–10, 50–61. [CrossRef]
65. Haque, M.M.; Sarker, S.; Dewan, M.A.A. Driving Maneuver Classification from Time Series Data: A Rule Based Machine Learning
Approach. Appl. Intell. 2022, 52, 16900–16915. [CrossRef]
66. Hu, J.; Zhang, X.; Maybank, S. Abnormal Driving Detection with Normalized Driving Behavior Data: A Deep Learning Approach.
IEEE Trans. Veh. Technol. 2020, 69, 6943–6951. [CrossRef]
67. Jahan, I.; Uddin, K.M.A.; Murad, S.A.; Miah, M.S.U.; Khan, T.Z.; Masud, M.; Aljahdali, S.; Bairagi, A.K. 4D: A Real-Time Driver
Drowsiness Detector Using Deep Learning. Electronics 2023, 12, 235. [CrossRef]
68. Khan, T.; Choi, G.; Lee, S. EFFNet-CA: An Efficient Driver Distraction Detection Based on Multiscale Features Extractions and
Channel Attention Mechanism. Sensors 2023, 23, 3835. [CrossRef]
69. Abosaq, H.A.; Ramzan, M.; Althobiani, F.; Abid, A.; Aamir, K.M.; Abdushkour, H.; Irfan, M.; Gommosani, M.E.; Ghonaim, S.M.;
Shamji, V.R.; et al. Unusual Driver Behavior Detection in Videos Using Deep Learning Models. Sensors 2023, 23, 311. [CrossRef]
70. Huang, C.; Huang, C.; Wang, X.; Cao, J.; Wang, S.; Wang, S.; Zhang, Y.; Zhang, Y. HCF: A Hybrid CNN Framework for Behavior
Detection of Distracted Drivers. IEEE Access 2020, 8, 109335–109349. [CrossRef]
71. Li, T.; Zhang, Y.; Li, Q.; Zhang, T. AB-DLM: An Improved Deep Learning Model Based on Attention Mechanism and BiFPN for
Driver Distraction Behavior Detection. IEEE Access 2022, 10, 83138–83151. [CrossRef]
72. Aljohani, A.A. Real-Time Driver Distraction Recognition: A Hybrid Genetic Deep Network Based Approach. Alex. Eng. J. 2023,
66, 377–389. [CrossRef]
73. Lin, Y.C.; Cao, D.X.; Fu, Z.H.; Huang, Y.M.; Song, Y.Y. A Lightweight Attention-Based Network towards Distracted Driving
Behavior Recognition. Appl. Sci. 2022, 12, 4191. [CrossRef]
74. Kabir, M.F.; Roy, S. Real-Time Vehicular Accident Prevention System Using Deep Learning Architecture. Expert Syst. Appl. 2022,
206, 117837. [CrossRef]
75. Hossain, M.U.; Rahman, M.A.; Islam, M.M.; Akhter, A.; Uddin, M.A.; Paul, B.K. Automatic Driver Distraction Detection Using
Deep Convolutional Neural Networks. Intell. Syst. Appl. 2022, 14, 200075. [CrossRef]
76. Lin, P.-W.; Hsu, C.-M. Innovative Framework for Distracted-Driving Alert System Based on Deep Learning. IEEE Access 2022, 10,
77523–77536. [CrossRef]
77. Xiao, W.; Liu, H.; Ma, Z.; Chen, W.; Sun, C.; Shi, B. Fatigue Driving Recognition Method Based on Multi-Scale Facial Landmark
Detector. Electronics 2022, 11, 4103. [CrossRef]
78. Ezzouhri, A.; Charouh, Z.; Ghogho, M.; Guennoun, Z. Robust Deep Learning-Based Driver Distraction Detection and Classifica-
tion. IEEE Access 2021, 9, 168080–168092. [CrossRef]
79. Xue, Q.; Gao, K.; Xing, Y.; Lu, J.; Qu, X. A Context-Aware Framework for Risky Driving Behavior Evaluation Based on Trajectory
Data. IEEE Intell. Transp. Syst. Mag. 2023, 15, 70–83. [CrossRef]
80. Fan, Y.; Gu, F.; Wang, J.; Wang, J.; Lu, K.; Niu, J. SafeDriving: An Effective Abnormal Driving Behavior Detection System Based
on EMG Signals. IEEE Internet Things J. 2022, 9, 12338–12350. [CrossRef]
81. Zhang, Y.; Chen, Y.; Gao, C. Deep Unsupervised Multi-Modal Fusion Network for Detecting Driver Distraction. Neurocomputing
2021, 421, 26–38. [CrossRef]
82. Boucetta, Z.; Fazziki, A.E.; Adnani, M.E. Integration of Ensemble Variant CNN with Architecture Modified LSTM for Distracted
Driver Detection. Int. J. Adv. Comput. Sci. Appl. 2022, 13, 440–458. [CrossRef]
83. Safarov, F.; Akhmedov, F.; Abdusalomov, A.B.; Nasimov, R.; Cho, Y.I. Real-Time Deep Learning-Based Drowsiness Detection:
Leveraging Computer-Vision and Eye-Blink Analyses for Enhanced Road Safety. Sensors 2023, 23, 6459. [CrossRef]
84. Jardin, P.; Moisidis, I.; Kartal, K.; Rinderknecht, S. Adaptive Driving Style Classification through Transfer Learning with Synthetic
Oversampling. Vehicles 2022, 4, 1314–1331. [CrossRef]
85. Wang, H.; Chen, J.; Huang, Z.; Li, B.; Lv, J.; Xi, J.; Wu, B.; Zhang, J.; Wu, Z. FPT: Fine-Grained Detection of Driver Distraction
Based on the Feature Pyramid Vision Transformer. IEEE Trans. Intell. Transp. Syst. 2023, 24, 1594–1608. [CrossRef]
86. Ping, P.; Huang, C.; Ding, W.; Liu, Y.; Chiyomi, M.; Kazuya, T. Distracted Driving Detection Based on the Fusion of Deep Learning
and Causal Reasoning. Inf. Fusion 2023, 89, 121–142. [CrossRef]
87. Jahangiri, A.; Berardi, V.J.; MacHiani, S.G. Application of Real Field Connected Vehicle Data for Aggressive Driving Identification
on Horizontal Curves. IEEE Trans. Intell. Transp. Syst. 2018, 19, 2316–2324. [CrossRef]
88. Siddiqui, H.U.R.; Saleem, A.A.; Brown, R.; Bademci, B.; Lee, E.; Rustam, F.; Dudley, S. Non-Invasive Driver Drowsiness Detection
System. Sensors 2021, 21, 4833. [CrossRef]
Computation 2024, 12, 131 21 of 21

89. Zhang, Y.; Chen, Y.; Gu, X.; Sze, N.N.; Huang, J. A Proactive Crash Risk Prediction Framework for Lane-Changing Behavior
Incorporating Individual Driving Styles. Accid. Anal. Prev. 2023, 188, 107072. [CrossRef]
90. Cai, B.; Di, Q. Different Forecasting Model Comparison for Near Future Crash Prediction. Appl. Sci. 2023, 13, 759. [CrossRef]
91. Singh, G.; Bansal, D.; Sofat, S. A Smartphone Based Technique to Monitor Driving Behavior Using DTW and Crowdsensing.
Pervasive Mob. Comput. 2017, 40, 56–70. [CrossRef]

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual
author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to
people or property resulting from any ideas, methods, instructions or products referred to in the content.