Accident Prediction Models Research Articles

Development of an Optimized Accident Prediction Model for Construction Safety Using Big Data Analysis

Background and objective: The objective of this study was to develop an intuitive accident prediction model that can be readily applied at construction sites to effectively reduce the incidence of accidents through the analysis of construction accident data.Methods: These accidents significantly contribute to the construction industry's overall accident rate. Accident types were categorized into fatalities, injuries, and material damages to construct the accident prediction model. A total of 24 factors were considered across eight major variables, which were identified during the first and second phases of data preprocessing to analyze construction accident big data. Machine learning techniques were employed, specifically supervised learning and ensemble learning, to identify the optimal predictive model.Results: Among the models tested, XGBoost emerged as the most effective due to its highly balanced accuracy, even in the presence of class imbalance.Conclusion: The implementation of the XGBoost accident prediction model, along with the feature importance codes developed in this study, enables the prediction of accident types for specific tasks performed at construction sites. This predictive capability is expected to inform the implementation of targeted accident prevention measures, such as enhancing safety protocols or adjusting work procedures based on the prediction outcomes.

Research on fire accident prediction and risk assessment algorithm based on data mining and machine learning

Forest fire is a kind of natural disaster that is destructive, easy to spread, and difficult to extinguish. It greatly harms the balance of ecosystem and human life and property. The prediction and risk assessment of forest fire accidents can find forest fires as early as possible and then take corresponding remedial measures; the loss of forest fires will be minimized. This paper first collected the relevant data information in the Lesser Hinggan Mountain region, established the fire driving factor database, and then analyzed the impact and distribution characteristics of forest fire driving factors such as temperature and rainfall. Finally, a fire accident prediction model was built based on a deep neural network, and the model’s performance was compared with the SVM and RF models. The analysis results show that this paper’s fire accident prediction model is more accurate.

Analysis of causes of traffic accidents for taxi drivers based on MIMIC and machine learning

Through questionnaire survey, information on individual attributes, physical fatigue perception, work pressure, risky driving behavior and traffic accident experience of 2391 taxi drivers was collected. The multi-indicator multi-cause (MIMIC) model was used for path analysis to explore the inducing effect of physical fatigue perception and risky driving behavior on traffic accidents, and to verify the influence of gender, age and work pressure on physical fatigue perception and distracted driving behavior. Four machine learning algorithms, logistic regression, naive Bayes, support vector machine and random forest, were selected to predict taxi accidents. The results show that the increase of physical fatigue perception, negligent driving behavior, aggressive driving behavior and distracted driving behavior can lead to an increase in accident rate, and gender, age and work pressure will affect the frequency of physical fatigue perception and distracted driving behavior. The accident prediction model based on machine learning has excellent effect, among which the random forest has the best prediction effect. The prediction accuracy of using single feature variables "risky driving behavior", "distracted driving behavior" and "physical fatigue perception" is acceptable. When the personal attribute indicators of "work pressure", "age" and "gender" are introduced, the prediction accuracy is further improved.

Development of Motorway Horizontal Alignment Databases for Accurate Accident Prediction Models

The safe and efficient operation of highways minimizes the environmental impact, reduces accidents, and promotes the reliability of the transportation infrastructure, all in support of sustainable transportation. The horizontal alignment of highways holds particular importance as it directly impacts driver behavior, vehicle stability, and overall road safety. In many cases, highway inventory data held by infrastructure operators may contain inaccurate or outdated information. The accuracy of the variables used in crash prediction models eliminates possible bias in the variable estimators. This research proposes a methodology to obtain accurate horizontal geometric features from digital imagery based on the analysis of the planimetry, feature geolocation and centerline azimuth sequence. The reliability of the method is verified by means of numerical and statistical procedures. This methodology is applied to 150 km of motorway segments in Portugal. Although it is found that the geometric characteristics of most of the inventory segments closely matched the extracted alignments, very significant differences are found in some sections. The results of the proposed procedure are illustrated with several examples. Finally, the propagation of error in the determination of the geometric design independent variables in the fitting of the statistical models is discussed based on the results.

Development of Accident Prediction Model under Mixed Traffic Conditions: A Case Study

Development of Accident Prediction Model under Mixed Traffic Conditions: A Case Study

Prediction model of factors causing traffic accidents on rural arterial roads: A binary logistic regression approach

In rural areas, land use activities around primary arterial roads influence the road section’s traffic characteristics. Regulations dictate the design of primary arterial roads to accommodate high speeds. Hence, there is a mix of traffic between high-speed vehicles and vulnerable road users (pedestrians, bicycles, and motorcycles) around the land. As a result, researchers have identified several arterial roads in Indonesia as accident-prone areas. Therefore, to improve the road user’s safety on primary arterial roads, it is necessary to develop models of the influence of various factors on road traffic accidents. This research uses binary logistic regression analysis. The independent variables are carelessness, disorderliness, high speed, horizontal alignment, road width, clear zone, road shoulder width, signs, markings, and land use. Meanwhile, the dependent variable is the frequency of accidents, where the frequency of accidents consists of multi-accident vehicles (MAV) and single-accident vehicles (SAV). This study collects data for a traffic accident prediction model based on collision frequency in accident-prone areas. The results, road shoulder width, and road sign factor all have an impact on the frequency of traffic accidents. According to a realistic risk analysis, MAV and SAV have no risk difference. After validation, this model shows a confidence level of 92%. This demonstrates that the model generates estimations that accurately reflect reality and are applicable to a wider population. This research has the potential to assist engineers in improving road safety on primary arterial roads. In addition, the model can help the government measure the impact of implemented policies and engage the public in traffic accident prevention efforts.

A risk prediction model for Maritime accidents

This study proposes analytical tools to predict maritime accidents involving dangerous goods to help improve maritime safety and preserve maritime and coastal heritage. Maritime accidents of dangerous goods can have devastating consequences, causing loss of life, damage to the environment and economic losses. There have been numerous studies attempting to predict maritime accidents, but most have focused on a single type of accident (e.g. oil spills) or a single region (e.g. Baltic Sea, Maritime Silk Road, etc.). This study takes a different approach, using a global dataset on maritime accidents of dangerous goods from 2010 to 2019 (that includes information on the type of casualty, the location, the amount of material released, the type of material released, the cause of the accident, and the outcome), it applies both a machine learning technique and a statistical approach based on the Fourier distribution of rare events as a dual approach to address the problem. Using the Tyrrhenian area as a case study, the results show that the proposed tools can predict the probability of an accident occurring with an acceptable level of accuracy. The paper can provide a valuable tool for decision makers and stakeholders, who can use the findings to identify regions at risk of maritime accidents and take measures to prevent their occurrence.

A deep neural network algorithm-based approach for predicting recovery period of accidents according to construction scale

Despite ongoing safety efforts, construction sites experience a concerningly high accident rate. Notwithstanding that policies and research to reduce the risk of accidents in the construction industry have been active for a long time, the accident rate in the construction industry is considerably higher than in other industries. This trend may likely be further exacerbated by the rapid growth of large-scale construction projects driven by urban population expansion. Consequently, accurately predicting recovery periods of accidents at construction sites in advance and proactively investing in measures to mitigate them is critical for efficiently managing construction projects. Therefore, the purpose of this study is to propose a framework for developing accident prediction models based on the Deep Neural Network (DNN) algorithm according to the scale of the construction site. This study suggests DNN models and applies the DNN for each construction site scale to predict accident recovery periods. The model performance and accuracy were evaluated using mean absolute error (MAE) and root-mean-square error (RMSE) and compared with the widely used multiple regression analysis model. As a result of model comparison, the DNN models showed a lower prediction error rate than the regression analysis models for both small-to-medium and large construction sites. The findings and framework of this study can be applied as the opening stage of accident risk assessment using deep learning techniques, and the introduction of deep learning technology to safety management according to the scale of the construction site is provided as a guideline.

Improving traffic accident severity prediction using MobileNet transfer learning model and SHAP XAI technique.

Traffic accidents remain a leading cause of fatalities, injuries, and significant disruptions on highways. Comprehending the contributing factors to these occurrences is paramount in enhancing safety on road networks. Recent studies have demonstrated the utility of predictive modeling in gaining insights into the factors that precipitate accidents. However, there has been a dearth of focus on explaining the inner workings of complex machine learning and deep learning models and the manner in which various features influence accident prediction models. As a result, there is a risk that these models may be seen as black boxes, and their findings may not be fully trusted by stakeholders. The main objective of this study is to create predictive models using various transfer learning techniques and to provide insights into the most impactful factors using Shapley values. To predict the severity of injuries in accidents, Multilayer Perceptron (MLP), Convolutional Neural Network (CNN), Long Short-Term Memory (LSTM), Residual Networks (ResNet), EfficientNetB4, InceptionV3, Extreme Inception (Xception), and MobileNet are employed. Among the models, the MobileNet showed the highest results with 98.17% accuracy. Additionally, by understanding how different features affect accident prediction models, researchers can gain a deeper understanding of the factors that contribute to accidents and develop more effective interventions to prevent them.

An Accident Prediction Model Based on ARIMA in Kuala Lumpur, Malaysia, Using Time Series of Actual Accidents and Related Data

Recently, there has been an emerging trend to analyse time series data and utilise sophisticated tools for optimally fitting time series models. To date, Malaysian industrial accident data is underutilised and lacks informative records. Thus, this paper aims to investigate the Malaysian accident database and further evaluate the optimal forecasting models in accident prediction. The model’s input was based on available data from the Department of Occupational Safety and Health, Malaysia (DOSH), from 2018 until 2021, with 80% of the dataset to train the models and the remaining 20% for validation. The negative binomial and Poisson distribution prediction showed a mean absolute percentage error (MAPE) of 33% and 51%, respectively. It indicated that the negative binomial performed better than the Poisson distribution in accident frequency prediction. The available time series accident data were gathered for four years, and stationarity was checked in R Studio software for the Augmented Dickey-Fuller test. The lowest Akaike Information Criterion (AIC), Bayesian Information Criterion (BIC) and other error values were used to justify the best model, which was the ARIMA(2,0,2)(2,0,0)(12) model. The ARIMA models were considered after the data showed autocorrelation. The MAPE for both ARIMA in R and manual time series were 40% and 49%, respectively. Therefore, the accident prediction by using R Studio would outperform the manually negative binomial and Poisson distribution. Based on the findings, industrial safety practitioners should report accidents to DOSH truthfully in the era of digitalisation. It could enable future data-driven accident predictions to be carried out.

Causative factors and risk prediction model of hydrogen leakage accidents: Machine learning based on case evidence

This study innovatively derives a risk prediction model for hydrogen leakage accidents based on existing case evidence by combining MORT (Management Oversight and Risk Tree) with machine learning algorithms. 30 hydrogen leakage accidents causative factors are derived by analyzing 23 accident cases through MORT, and constructed a dataset of hydrogen leakage accident causative factors through data quantification. Using feature selection to optimize the risk indicators, 14 salient feature indicators are selected from the 30 full feature indicators. Two cross-validation strategies are introduced and combined with six machine learning models to establish 24 working conditions for numerical comparison experiments. The results show that the support vector machine model based on stratified K-fold cross-validation strategy has the best prediction performance, with the model Accuracy of 0.939, Precision of 0.848, Recall of 0.848, F1-score of 0.848, and AUC score of 0.98. This will effectively solve the problem of hydrogen leakage risk prediction and accident prevention in industrial systems.

Decision analysis with IDOCRIW-QUALIFLEX approach in the 2TLq-ROF environment: An application of accident prediction models in Pakistan

In Pakistan, the assessment of road safety measures within road safety management systems is commonly seen as the most deficient part. Accident prediction models are essential for road authorities, road designers, and road safety specialists. These models facilitate the examination of safety concerns, the identification of safety improvements, and the projection of the potential impact of these modifications in terms of collision reduction. In the context described above, the goal of this paper is to utilize the 2-tuple linguistic q-rung orthopair fuzzy set (2TLq-ROFS), a new and useful decision tool with a strong ability to address uncertain or imprecise information in practical decision-making processes. In addition, for dealing with the multi-attribute group decision-making problems in road safety management, this paper proposes a new 2TLq-ROF integrated determination of objective criteria weights (IDOCRIW)-the qualitative flexible multiple criteria (QUALIFLEX) decision analysis method with a weighted power average (WPA) operator based on the 2TLq-ROF numbers. The IDOCRIW method is used to calculate the weight of attributes and the QUALIFLEX method is used to rank the options. To show the viability and superiority of the proposed approach, we also perform a case study on the evaluation of accident prediction models in road safety management. Finally, the results of the experiments and comparisons with existing methods are used to explain the benefits and superiority of the suggested approach. The findings of this study show that the proposed approach is more practical and compatible with other existing approaches.

Harnessing Generative Pre-Trained Transformers for Construction Accident Prediction with Saliency Visualization

Leveraging natural language processing models using a large volume of text data in the construction safety domain offers a unique opportunity to improve understanding of safety accidents and the ability to learn from them. However, little effort has been made to date in regard to utilizing large language models for the prediction of accident types that can help to prevent and manage potential accidents. This research aims to develop a model for predicting the six types of accidents (caught-in-between, cuts, falls, struck-by, trips, and others) by employing transfer learning with a fine-tuned generative pre-trained transformer (GPT). Additionally, to enhance the interpretability of the fine-tuned GPT model, a method for saliency visualization of input text was developed to identify words that significantly impact prediction results. The models were evaluated using a comprehensive dataset comprising 15,000 actual accident records. The results indicate that the suggested model for detecting the six accident types achieves 82% accuracy. Furthermore, it was observed that the proposed saliency visualization method can identify accident precursors from unstructured free-text data of construction accident reports. These results highlight the advancement of the generalization performance of large language processing-based accident prediction models, thereby proactively preventing construction accidents.

Development and validation of prediction model for fall accidents among chronic kidney disease in the community.

The population with chronic kidney disease (CKD) has significantly heightened risk of fall accidents. The aim of this study was to develop a validated risk prediction model for fall accidents among CKD in the community. Participants with CKD from the China Health and Retirement Longitudinal Study (CHARLS) were included. The study cohort underwent a random split into a training set and a validation set at a ratio of 70 to 30%. Logistic regression and LASSO regression analyses were applied to screen variables for optimal predictors in the model. A predictive model was then constructed and visually represented in a nomogram. Subsequently, the predictive performance was assessed through ROC curves, calibration curves, and decision curve analysis. A total of 911 participants were included, and the prevalence of fall accidents was 30.0% (242/911). Fall down experience, BMI, mobility, dominant handgrip, and depression were chosen as predictor factors to formulate the predictive model, visually represented in a nomogram. The AUC value of the predictive model was 0.724 (95% CI 0.679-0.769). Calibration curves and DCA indicated that the model exhibited good predictive performance. In this study, we constructed a predictive model to assess the risk of falls among individuals with CKD in the community, demonstrating good predictive capability.

A comparative analysis of accident modification functions for traffic law enforcement

Traffic law enforcement is a road safety measure whose effects on accidents or injuries is best described by means of a function rather than a point estimate. An informative function should comprise both increases and decreases in enforcement. Currently available accident modification functions cannot serve this need. A fruitful approach to developing accident modification functions covering both increases and decreases in enforcement is differences-in-differences estimates based on multivariate accident prediction models. The paper explains how to develop such estimates and illustrates them. The interpretation of the results of empirical studies can be informed by a game-theoretic model of the effects of enforcement, previously published in Accident Analysis and Prevention (Bjørnskau and Elvik 1992, 507–520).

Design and construction of age-friendly smart buildings based on BIM technology

Abstract This paper explores the entry point of BIM technology in promoting intelligent construction and proposes an engineering safety accident prediction model for the construction of age-friendly intelligent buildings based on the gray prediction GM(1,1) method. Based on the physical condition of the elderly, the needs of the settlement, at the same time, the construction of an intelligent old-age design system, according to the needs of the corresponding response to the old-age building design, to facilitate the residence. The intelligent floor plate is identified as an example of the intelligent design results of age-friendly buildings by empirically analyzing their willingness to construct age-friendly intelligent buildings. In the willingness analysis, the survey results were analyzed for reliability validity and correlation; the reliability α coefficients were all greater than 0.7, the AVEs of the reliability were all higher than 0.5, the CRs were all higher than 0.7, and the reliability validity of the survey was high. The correlation coefficients were all >0, and there was a significant positive correlation between the survey variables. The vibration length was used as a criterion, and the accuracy, precision and recall were all 1. The experiments showed that the elderly have a high degree of intention to build age-appropriate intelligent buildings, and the buildings have a high accuracy in interpreting the state of the elderly, and they can detect problems in time and report them to the relevant personnel.

Investigating Highway–Rail Grade Crossing Inventory Data Quality’s Role in Crash Model Estimation and Crash Prediction

The highway–rail grade crossings (HRGCs) crash frequency models used in the US are based on the Federal Railroad Administration’s (FRA) database for highway–rail crossing inventory. Inaccuracies or missing values within this database directly impact the estimated parameters of the crash models and subsequent crash predictions. Utilizing a set of 560 HRGCs in Nebraska, this research demonstrates variations in crash predictions estimated by the FRA’s 2020 Accident Prediction (AP) model under two scenarios: firstly, employing the unchanged, original FRA HRGCs inventory dataset as the input, and secondly, utilizing a field-validated inventory dataset for the same 560 HRGCs as input to the FRA’s 2020 Accident Prediction (AP) model. The findings indicated a significant statistical disparity in the predictions made with the two input datasets. Furthermore, two new Zero-inflated Negative Binomial (ZINB) models were estimated by employing 5-year reported HRGCs crashes and the two inventory datasets for the 560 HRGCs. These models facilitated the comparison of model parameter estimates and estimated marginal values. The results indicated that errors and missing values in the original FRA HRGCs inventory dataset resulted in crash predictions that statistically differed from those made using the more accurate and complete (validated in the field) HRGCs inventory dataset. Furthermore, the crash prediction model estimated upon the corrected inventory data demonstrated enhanced prediction performance, as measured by the statistical fitness criteria. The findings emphasize the importance of collecting complete and accurate inventory data when developing HRGCs crash frequency models. This will enhance models’ precision, improve their predictive capabilities to aid in better resource allocation, and ultimately reduce HRGCs crashes.

Dynamic Multi-Period Maritime Accident Susceptibility Assessment Based on AIS Data and Random Forest Model

Maritime accidents, such as ship collisions and oil spills, directly affect maritime transportation, pollute the water environment, and indirectly threaten life and property safety. Predicting the maritime accident susceptibility and taking measures in advance can effectively avoid the accident probability and reduce the risk. Therefore, this study established dynamic multi-period (monthly, yearly, and five-yearly) maritime accident prediction models based on the random forest (RF) algorithm and Automatic Identification System (AIS) data for susceptibility assessment. First, based on historical maritime accidents and influencing factor data, we generated the feature matrixes and selected the conditioning factors using the Pearson correlation coefficient. Then, we constructed the accident susceptibility models using the RF method and evaluated the model performances based on the accuracy, recall, precision, F1-measure, ROC, and AUC values. Finally, we developed accident susceptibility maps for different period scales. The results show that the monthly, yearly, and five-yearly models performed well according to the validation values. And the three-period susceptibility maps show similar patterns. The high-susceptibility areas are close to the shore, especially from the Shanghai shore to the Guangxi shore. In addition, the ship density and bathymetry are the most critical factors among the ten influencing factors in the three models, contributing around 25% and 20% of the total information. These models and maps can provide technological support for maritime accident susceptibility assessment on a multi-period scale, which can be helpful for route planning and resource allocation in marine management.

Determinants of bicycle crashes at urban signalized intersections

Problem: Bicycle volumes are increasing in many regions worldwide leading to higher relevance of an in-depth understanding of bicyclist safety mechanisms. Detailed studies on bicyclist safety that consider exposure and distinguish by intersection category and crash types are missing for urban signalized intersections, which are of particular relevance for bicyclist safety. Method: Based on a comprehensive dataset of motorist and bicyclist volumes and infrastructure characteristics for a sample of 269 signalized intersections in two German cities, we utilize a top-down approach to analyze firstly, bicycle crashes of all types and secondly, bicycle crashes by type including turning, right-of-way and loss-of-control. A combination of descriptive statistics and Accident Prediction Models (APM) are applied as analysis methods. Results: Bicycle volumes are relevant for all types of intersections and crashes, whereas the effect of motor vehicle volumes differ between these different applications. The separation of bicyclists from motor vehicles in time and space increases their safety but also leads to behavioral adaption and risk compensation. The likelihood of right-of-way crashes even increases with more separation in the signaling scheme. The main predictor for loss-of-control crashes in terms of infrastructure are tram tracks. Summary: This study provides insights on relevant determinants of bicycle crashes at urban signalized intersections at several levels of detail. Exposure variables as well as the physical separation of bicyclists from motor vehicles show consistent effects on bicycle crash numbers whereas the effects of signaling differ between crash types. Practical Applications: The different types of intersections and crashes follow each specific mechanism of bicyclist safety. The separation of bicyclists and motorists in time and space are paramount at intersections with high bicycle volumes. Risk compensation such as red light running becomes more important as intersections get smaller and motor vehicle volumes decrease.

Accident Prediction Modeling for Indian Metro Cities

Road accidents are one of the biggest concerns to the road safety of developing nations. In India, around 150,000 fatal accidents occur annually. Road accident prediction models help in accessing the factors responsible for and those that contribute more to accidents. Most of the prediction models focus on the parameters like road characteristics, traffic characteristics, driver characteristics, and road geometrics. In this study, we considered socio-economic and land-use parameters as input data for accident prediction modeling. The socio-economic and land-use variables data of 20 Indian metro cities were collected. The data were collected for a period of 5 years ranging from 2016 to 2020. A multiple linear regression model was developed between the total number of accidents that happened in the 20 metro cities and the socio-economic and land-use variables. ANN model was also developed to check its applicability to this study and the results obtained are satisfactory.