Safety Prediction Model Research Articles

Quantitative read-across structure-activity relationship (q-RASAR): A novel approach to estimate the subchronic oral safety (NOAEL) of diverse organic chemicals in rats

We have developed a quantitative safety prediction model for subchronic repeated doses of diverse organic chemicals on rats using the novel quantitative read-across structure-activity relationship (q-RASAR) approach, which uses similarity-based descriptors for predictive model generation. The experimental -Log (NOAEL) values have been used here as a potential indicator of oral subchronic safety on rats as it determines the maximum dose level for which no observed adverse effects of chemicals are found. A total of 186 data points of diverse organic chemicals have been used for the model generation using structural and physicochemical (0D–2D) descriptors. The read-across-derived similarity, error, and concordance measures (RASAR descriptors) have been extracted from the preliminary 0D–2D descriptors. Then, the combined pool of RASAR and the identified 0D–2D descriptors of the training set were employed to develop the final models by using the partial least squares (PLS) algorithm. The developed PLS model was rigorously validated by various internal and external validation metrics as suggested by the Organization for Economic Co-operation and Development (OECD). The final q-RASAR model is proven to be statistically sound, robust and externally predictive (R2 = 0.85, Q2LOO = 0.82 and Q2F1 = 0.94), superseding the internal as well as external predictivity of the corresponding quantitative structure-activity relationship (QSAR) model as well as previously reported subchronic repeated dose toxicity model found in the literature. In a nutshell, the q-RASAR is an effective approach that has the potential to be used as a good alternative way to improve external predictivity, interpretability, and transferability for subchronic oral safety prediction as well as ecotoxicity risk identification.

Mechanical safety prediction of a battery-pack system under low speed frontal impact via machine learning

In response to nonrenewable energy consumption and environmental pollution, the development of electric vehicles has accelerated. The safety of electric vehicles has garnered considerable attention. There are numerous incalculable foreign objects on the road that may collide with or scratch the battery-pack’s frontal, resulting in battery-pack system damage or even explosion. This poses a significant risk to the safety of passengers and drivers. In this paper, a diversity of mechanical safety prediction models for battery-pack systems are proposed. These models support data-driven structural optimization of the battery-pack system by utilizing the numerical results of the bottom shell deformation. These simulation-based prediction models combine response surface method and machine learning algorithms. First, a nonlinear finite element model of the battery-pack system is established. The efficacy of the model is verified using constrained modal analysis in a variety of commercial software packages. Second, the collision simulations are executed and the data in various collision conditions are collected. Different sample sizes are used to develop various response surface models and machine learning models. The machine learning algorithms adopt support vector machine, Gaussian process regression, and neural network models. Third, the prediction accuracy of multiple prediction models is investigated according to error functions. The results show that the neural network model can predict the most accurate deformation under the condition of low speed frontal impact. The prediction average absolute percentage error of the neural network model is only 0.34% within the design domain, and only 2.54% outside the design domain. The proposed prediction model can be used for reliable design of the battery-pack system in electric vehicles. It also can be employed to design the early warning system of the battery-packs.

Construction safety predictions with multi-head attention graph and sparse accident networks

The reliability of risk assessment is crucial for designing effective construction safety management strategies. Construction safety prediction using machine learning models is still suboptimal, considering representativeness, interpretability, and efficiency challenges. This paper enhances the representativeness, interpretability, and accuracy of construction safety prediction models using a multi-head graph attention network (GAT) and a novel sparse construction safety network. Through its accommodation of connectivity information between accident records, the proposed approach enhances the interpretability and representativeness of non-graph-based machine learning models. In addition, it improves the message aggregation of the embedded accident information and adaptability to hyperparameter variations of the state-of-the-art graph convolutional network. The evaluation of multi-head GAT on three sparse construction safety networks achieved accuracies of 86.2%, 87.1%, and 86.9%, which implies the benefit of the self-attention mechanism in learning the importance of connected accident records. This substantiates the reliability of the proposed approach for integration within a risk assessment procedure, the outcome of which is instrumental in designing effective management strategies.

Cycle-level traffic conflict prediction at signalized intersections with LiDAR data and Bayesian deep learning

Real-time safety prediction models are vital in proactive road safety management strategies. This study develops models to predict traffic conflicts at signalized intersections at the signal cycle level, using advanced Bayesian deep learning techniques and efficient LiDAR points. The modeling framework contains three phases, which are data preprocessing, base deep learning model development, and Bayesian deep learning model development. The core of the framework is the long short-term memory (LSTM) employed to predict the conflict frequency of a cycle by using traffic features of the previous five cycles (e.g., dynamic traffic parameters, traffic conflict frequency). Four Bayesian deep learning models were developed, including Bayesian-Standard LSTM, Bayesian-Hybrid-LSTM, Bayesian-Stacked-LSTM Encoder-Decoder, and Bayesian-Multi-head Stacked-LSTM Encoder-Decoder. The developed models were applied to traffic conflicts extracted from LiDAR points that were collected from a signalized intersection in Harbin, China with a total duration of seven days. Traffic conflicts, measured by the modified time-to-collision conflict indicator, were identified using the peak over threshold approach. The models were thoroughly evaluated from the aspects of reliability, transferability, sensitivity, and robustness. The results show that the developed four models can predict traffic conflict frequency per cycle per lane simultaneously with its uncertainty. Moreover, the two Bayesian encoder-decoder models perform better than Bayesian-Standard LSTM and Bayesian-Hybrid-LSTM in the four tests. Bayesian-Multi-head Stacked-LSTM Encoder-Decoder is suggested as the optimal model for its high reliability under uncertainty, good transferability in three scenarios, low sensitivity to different parameters, and sound robustness against small noise. The proposed framework could benefit studies on the state-of-the-art data-driven approach for real-time safety prediction.

Construction Worker Safety Prediction and Active Warning Based on Computer Vision and the Gray Absolute Decision Analysis Method

Although a great deal of worker safety risks analysis has been conducted, safety accidents continue to occur and recur at regular intervals. On construction sites, various activities have different levels of accident probability and severity, and existing methods are limited because they involve the use of the same weights to assess accident probability and accident severity. In addition, uncertainties that occur over time result in rapidly changing risks. The majority of existing approaches provide postevent warnings, so workers may not have sufficient time to prevent accidents once they are warned about the possibility. Thus, there is a need for mechanisms that can be used to assess safety risk regularity, predict worker risk levels, and provide proactive warnings based on the comprehensive consideration of the impacts of worker behaviors and environments. To address these issues, a safety prediction model was proposed for use, and an active warning mechanism was constructed for construction workers. The prediction model performs accident potential regularity analysis based on attribute-based safety risk analysis and precursor analysis. Further, it quantifies worker risk levels using decision matrix risk assessment (DMRA) and the gray absolute decision analysis (GADA) method. The model overcomes the limitation of using the same weights in DMRA to assess accident probability and severity. An active warning mechanism for construction workers was created to validate the efficacy of the safety prediction model, and the proposed safety prediction model is embedded in the mechanism. The mechanism mainly consists of three modules: (1) a data collection module that mainly includes expert knowledge and dynamic safety information from surveillance cameras; (2) a data analysis module that mainly uses the proposed safety prediction model to predict individual worker risk levels; and (3) an early warning module that displays the predicted risk levels, dynamically ranks risk indicators, and provides corresponding early warning measures. Finally, the feasibility and operability of the proposed active warning mechanism are demonstrated through a practical case study.

Physics-informed multi-step real-time conflict-based vehicle safety prediction

Real-time vehicle safety prediction is critical in roadway safety management as drivers or vehicles can be altered beforehand to take corresponding evasive actions and avoid possible collisions. This study proposes a physics-informed multi-step real-time conflict-based vehicle safety prediction model to enhance roadway safety. Physics insights (i.e., traffic shockwave properties) are combined with data-driven features extracted from deep-learning techniques to improve prediction accuracy. A time series of future vehicle safety indicators are predicted such that vehicles/drivers have enough time to take precautions. The safety indicator at each time stamp is a continuous value that the sign reflects the presence of conflict risks, and the absolute value indicates the conflict risk level to advise different magnitudes of evasive actions. A customized loss function is developed for the proposed prediction model to give more attention to risky events, which are the focus of safety management. The prediction superiority of the proposed model is proven through numerical experiments by comparing it with three benchmarks constructed based on the literature. Further, sensitivity analysis on key model parameters is carried out to advise parameter selections in developing real-world conflict-based vehicle safety prediction applications.

Analysis of Roadway Grade Estimations From Global Positioning Systems and Barometer Measurements

Roadway grade is used as an input variable for several highway analysis tasks, including evaluating curve margin-of-safety, applying safety prediction models, assessing the adequacy of sight distance, and maintaining state roadlog databases. However, collecting accurate grade data is often expensive owing to the need to obtain field measurements or review as-built plan sets. Agencies would benefit from the development of a method to compute roadway grade from an automated data collection system. This paper documents a comparison of roadway grade estimations computed from global positioning system (GPS) and barometric altimeter data streams obtained during test drives on highway segments of interest. To calibrate the estimation method, the authors obtained ground-truth grade measurements collected from the field and then modeled those measures as responses in a time series model with the two data stream types as explanatory variables. The authors found that for several applications, the elevation data obtained from GPS were adequate to obtain reasonable estimates, but such estimations could be improved with a supplemental data stream from a barometer.

Did Operating Speeds During COVID-19 Result in More Fatal and Injury Crashes on Urban Freeways?

Impacts of the COVID-19 pandemic in the transportation arena included less traffic, higher speeds, and higher fatal and injury crash frequencies. Many news media reported on speeding and its impact. However, the majority of these reporting are based on partial or incomplete information. The current study aims to understand the association between speed and crash on the freeways of Dallas (Texas) by collecting data from the National Performance Management Research Dataset, the Texas Department of Transportation’s (TxDOT’s) roadway inventory, and TxDOT’s crash database for 2018–2020. The results show decreasing traffic volume, increasing average operating speed, and increasing fatal and severe crash frequencies per 100 million vehicle miles traveled during 2020 (April–November). This study developed 8-month- and daily-level safety prediction models for fatal and injury crashes. The 8-month-level dataset contains speed measures as an aggregate for the 8-month period. The daily-level database includes operating speeds and fatal and injury crashes at the daily level where segments experiencing fatal and injury crashes were temporally matched with the same segment with the same day of the week and with no fatal and injury crash occurrences. For the 8-month models, average operating speed and speed variability are positively associated with fatal and injury crash frequencies during the COVID period. This association was also found for daily-level models. The findings of this study can help transportation agencies in developing strategies (for example, posted speed limit reconsideration, additional enforcement at specific locations) for crash reduction.

County-level crash prediction models for Pennsylvania accounting for income characteristics

The majority of current safety prediction models utilize roadway and traffic data as independent variables to describe safety performance at a microscopic level. Recent work moves toward predicting these measures within some region as a function of roadway and traffic data, as well as non-traditional variables, such as socioeconomic measures. This paper aims to provide a holistic view of the intersection of socioeconomics and safety in Pennsylvania by investigating possible relationships between wealth and various aspects of safety performance, including crash frequency, severity, and cost. The analyses presented in this paper serve as case studies with intentions to promote the development of more robust, wealth-inclusive safety analyses in the future. The study reveals relationships between socioeconomic-related measures and crash frequency, severity, and cost estimations. These relationships indicate counties with increased levels of socioeconomic distress (quantified by multiple socioeconomic-related variables) are estimated to experience more crashes – particularlyrelated to alcohol usage – and higher total crash costs, and crashes that occur in counties with increased levels of socioeconomic distress are estimated to be more likely to result in an increased injury severity level compared to crashes that occur elsewhere. These resultssupport previous work andexpand on that work by considering multiple socioeconomic-related variables and their impacts on three unique safety-related measures. The existence of a relationship between crash frequency, severity, and cost and wealth-related variables opens the door to further exploration of including wealth in traditional safety analyses. This paper discusses these relationships and offers recommendations for future work.

Research on artificial intelligence safety prediction and intervention model based on ship driving habits

Based on the analysis of the causes of ship accidents, the development prospect and development direction of ship intelligent safe driving, the artificial intelligence safety prediction and intervention model is put forward. This model solves the problem of ship intelligent safety prediction by using intelligent analysis technology and network technology, and promotes the development of ship intelligence and ship safety navigation technology. Additionally, it expands the channels of obtaining information, connects the ship's mechanical and electrical equipment, collects, stores and analyzes the data reasonably, and constructs the intelligent analysis and processing platform of ship small-world data processing to implement intelligent intervention. What is impressive is that it makes ship navigation safer, more economical, more reasonable and optimized, and accelerates the development of ship artificial intelligence safe navigation.

Safety prediction model of fire detection and alarm system based on optimized neural network algorithm

The function of the fire detector is to capture specific fire signals. The sensitivity of the detector determines the sensitivity of the response to the characteristics of the fire severity. The sensitivity and reliability of the detector become the key parameters that the detector needs to consider in balance. Aiming at the problem of the most reliable detector types, we first use MATLAB to calculate the reliability and failure rate of the detectors according to the index weights. Then, the SPSSPRO evaluation TOPSIS method is used to score and rank the detectors, and in descending order, the detector with the highest ranking is selected as the detector with the highest reliability. In response to the problem of alarm accuracy, we first preprocess the data, and find suitable indicators according to the requirements of the topic to establish an intelligent judgment model for the type of regional alarm components. The indicators we find here include the name of the component, the name of the project, the affiliation of the fire department, and the probability of fire occurrence, which can be considered, six indicators of the probability of real alarm, and then uses the principal component analysis method to determine the weight through dimension reduction, and further screens the data, and then predicts the indicators through the method of neural network analysis to determine the alarm accuracy rate.

A data-driven, kinematic feature-based, near real-time algorithm for injury severity prediction of vehicle occupants

Accurate real-time prediction of occupant injury severity in unavoidable collision scenarios is a prerequisite for enhancing road traffic safety with the development of highly automated vehicles. Specifically, a safety prediction model provides a decision reference for the trajectory planning system in the pre-crash phase and the adaptive restraint system in the in-crash phase. The main goal of the current study is to construct a data-driven, vehicle kinematic feature-based model to realize accurate and near real-time prediction of in-vehicle occupant injury severity. A large-scale numerical database was established focusing on occupant kinetics. A first-step deep-learning model was established to predict occupant kinetics and injury severity using a convolutional neural network (CNN). To reduce the computational time for real-time application, the second step was to extract simplified kinematic features from vehicle crash pulses via a feature extraction method, which was inspired by a visualization approach applied to the CNN-based model. The features were incorporated with a low-complexity machine-learning algorithm and achieved satisfactory accuracy (85.4 % on the numerical database, 78.7 % on a 192-case real-world dataset) and decreased computational time (1.2 ± 0.4 ms) on the prediction tasks. This study demonstrated the feasibility of using data-driven and feature-based approaches to achieve accurate injury risk estimation prior to collision. The proposed model is expected to provide a decision reference for integrated safety systems in the next generation of automated vehicles.

Driving Safety Analysis Using Grid‐Based Water‐Filled Rut Depth Distribution

The water accumulated in the rutted road sections poses a threat to the safety of vehicles. Water‐filled ruts will cause partial or complete loss of the friction between tires and the road surface, leading to driving safety hazards such as hydroplaning and sliding. At present, the maximum water depth of left and right ruts is mostly adopted to analyze the safety of water‐filled ruts, ignoring the uneven change of ruts in the driving direction and the cross‐section direction, which cannot fully reflect the actual impact of asymmetric or uneven longitudinal ruts on the vehicle. In order to explore the impact of water‐filled ruts on driving safety, a three‐dimensional (3D) tire‐road finite element model is established in this paper to calculate the adhesion coefficient between the tire and the road surface. Moreover, a model of the 3D water‐filled rut‐adhesion coefficient vehicle is established and simulated by the dynamics software CarSim. In addition, the influence of the water depth difference between the left and right ruts on the driving safety is quantitatively analyzed, and a safety prediction model for the water‐filled rut is established. The results of the case study show that (1) the length of dangerous road sections based on vehicle skidding is longer than that based on hydroplaning, and the length of dangerous road sections based on hydroplaning is underestimated by 9.4%–100%; (2) as the vehicle speed drops from 120 km/h to 80 km/h, the length of dangerous road sections obtained based on vehicle sliding analysis is reduced by 93.8%. Therefore, in order to ensure driving safety, the speed limit is controlled within 80 km/h to ensure that the vehicle will not skid. The proposed method provides a good foundation for the vehicles to actively respond to the situation of the water‐filled road section.

Persation: an IoT Based Personal Safety Prediction Model Aided Solution

The number of attacks on innocent victims in moving vehicles, and abduction of individuals in their vehicles has risen alarmingly in the past few years. One common scenario evident from the modus operandi of this kind of attack is the random motion of these vehicles, due to the driver's unpredictable behaviours. To save the victims in such kinds of assault, it is essential to offer help promptly. An effective strategy to save victims is to predict the future location of the vehicles so that the rescue mission can be actioned at the earliest possibility. We have done a comprehensive survey of the state-of-the-art personal safety solutions and location prediction technologies and proposes an Internet of Things (IoT) based personal safety model, encompassing a prediction framework to anticipate the future vehicle locations by exploiting complex analytics of current and past data variables including the speed, direction and geolocation of the vehicles. Experiments conducted based on real-world datasets demonstrate the feasibility of our proposed framework in accurately predicting future vehicle locations. In this paper, we have a risk assessment of our safety solution model based on the OCTAVE ALLEGRO model and the implementation of our prediction model.

Deep learning model of concrete dam deformation prediction based on CNN

The concrete dam deformation prediction model is a key measure to predict the evolution of structural behavior and evaluate the safe service status. This paper uses open-source deep learning framework TensorFlow as the platform and uses the mature convolutional neural network technology in deep learning theory to establish the concrete dam deformation safety prediction model based on a deep learning. The application of engineering examples shows that the residual map, mean square error, and average percentage error are used as the model fitting and prediction accuracy evaluation standards. Compared with the shallow neural network model and the traditional Statistical model, the concrete dam deformation prediction model based on deep learning has higher prediction accuracy and more stable performance, providing a new method for concrete dam deformation monitoring.

Rule-based safety prediction models for rural two-lane run-off-road crashes

• Analyzed rural two-lane run-off-road crashes in Louisiana. • Applied both statistical and machine learning models to determine the best-fit model. • Cubist provides the highest accuracies with rules-based regression models. • Rules-based models are capable in addressing localized issues by improving prediction accuracies. During 2015–2017, the yearly average of fatalities caused by roadway departure (RwD) crashes was 19,233 in the U.S. Roadway departure crashes, or crashes in which a vehicle crosses an edge line, centerline, or otherwise leaves the traveled way, account for 52 percent of all traffic fatalities in the U.S. during this time. A majority of RwD crashes are run-off-road (ROR) crashes; these are crashes that result in a vehicle crossing the edge line on either side of the roadway. In this study, the research team analyzed seven years (2010–2016) of rural two-lane ROR crash data from Louisiana to better comprehend ROR crashes in refining safety predictions for rural two-lane highways. Statistical model (negative binomial model) and three separate machine learning models (random forest, support vector machine, and Cubist) were applied to determine the best fit models. Overall, Cubist is characterized by a better performance in estimating ROR crashes on rural two-lane roadways. The Cubist approach introduced rules-based safety performance functions (SPFs) for total and fatal and injury crashes. This approach will be beneficial for the safety practitioners in tackling localized issues in crash data with an emphasis on prediction accuracy.

Dairy Safety Prediction Based on Machine Learning Combined with Chemicals.

Dairy safety has caused widespread concern in society. Unsafe dairy products have threatened people's health and lives. In order to improve the safety of dairy products and effectively prevent the occurrence of dairy insecurity, countries have established different prevention and control measures and safety warnings. The purpose of this study is to establish a dairy safety prediction model based on machine learning to determine whether the dairy products are qualified. The 34 common items in the dairy sampling inspection were used as features in this study. Feature selection was performed on the data to obtain a better subset of features, and different algorithms were applied to construct the classification model. The results show that the prediction model constructed by using a subset of features including "total plate", "water" and "nitrate" is superior. The SN, SP and ACC of the model were 62.50%, 91.67% and 72.22%, respectively. It was found that the accuracy of the model established by the integrated algorithm is higher than that by the non-integrated algorithm. This study provides a new method for assessing dairy safety. It helps to improve the quality of dairy products, ensure the safety of dairy products, and reduce the risk of dairy safety.

Safety Performance of One-Way Arterials

Safety analysts are generally interested in understanding the differences in the safety performance when a two-way street is converted to a one-way operation or vice-versa. Literature exists to understand and predict the safety of two-way streets. However, safety prediction procedures are currently not available for assessing the safety performance of one-way arterials. This research was undertaken to develop safety prediction models for one-way arterials. To accomplish this objective, data collected in California, Illinois, Michigan, Oregon, and Texas were assembled that included a wide range of geometric design features, traffic control features, traffic characteristics, and crash records. The data were used to calibrate predictive models, each of which included a safety performance function (SPF) and several crash modification factors (CMFs). Separate SPFs were developed for fatal and injury crashes (i.e., fatal, incapacitating injury, non-incapacitating injury, and possible injury crash) and property-damage-only crashes. The SPFs were estimated using the negative binomial modeling structure. Severity distribution functions (SDFs) were also calibrated using the fatal and injury data. These functions can be used with the predictive models to estimate the expected crash frequency for each of four injury severity levels.

Determining Skid Resistance Needs on Horizontal Curves for Different Levels of Precipitation

Horizontal curves are a major cause of crashes that lead to fatal and serious injuries. Much research has been conducted on the safety implications of geometric and traffic characteristics of curves. Variables describing curve geometry and speed have been incorporated into safety prediction methodologies. However, relatively less research has been conducted on the effects of pavement friction and weather data on safety. The objective of this study is to develop a methodology for determining the pavement friction needs for different levels of precipitation. To accomplish the study objective, rural two-lane, four-lane undivided, and four-lane divided horizontal curve data from Texas were used. Safety prediction models were developed that included traffic and geometric characteristics, skid number, and annual precipitation rate. These models were then used to develop the guidelines for assessing the safety performance of a curve of interest by accounting for curve geometry, pavement skid resistance, and exposure to the wet-weather conditions that are most relevant for considerations of skid resistance. For conducting a planning-level analysis to identify candidate sites for pavement friction treatments, researchers developed thresholds based on the combined effect of skid number and annual precipitation variables. Researchers also provided skid number thresholds for high-priority sites for two example locations that experience significantly different levels of annual precipitation.

Safety Prediction Model for Reinforced Highway Slope using a Machine Learning Method

Recycled plastic pin (RPP) has been proved to be an effective and inexpensive solution for shallow slope stabilization. Current practice suggests conducting numerical modeling to find out the desired factor of safety (FS) using RPP in the design of landslide repair. While the slope stability is heavily dependent on soil strength parameters and slope geometry, RPP length and spacing can also play a significant role in reaching the target factor of safety for the highway slope. During this study, a safety prediction model was developed using both statistical and machine learning (ML) approaches to use RPP in slope stabilization. Initially, parametric study was conducted using five different soil strengths, six slope heights, three slope ratios, three RPP lengths, and five RPP spacing configurations. Using the strength reduction techniques of Finite Element Modeling Software PLAXIS 2D, FS was determined for more than 1,000 combinations. Afterwards, a statistical approach was undertaken to determine a safety prediction model containing all possible parameters. Finally, an ML approach was conducted for safety model. The ML approach was found to be more accurate than the classical statistical approach with 85% accuracy of predicting the FS for an RPP reinforced highway slope. The developed model was validated against the values obtained from the numerical modeling, which indicated that the SF obtained from the developed model was in good agreement with those from finite element method (FEM) analysis.