Collision Frequency Analysis Using Tree-Based Stratification

Although several factors are known to contribute to collisions at highway-railway grade crossings, the mixed effects of the control factors such as a highway class and other countermeasures on collision occurrence are less well explored. This study evaluates the relationship between countermeasures and collision occurrence by using a sequential analytic strategy that combines the tree-based data stratification method with the generalized linear regression technique. After the control factor effects were controlled with the use of the tree-based data partitioning method, stratified collision prediction models were developed, and the adjusted effect of the selected countermeasures was identified. The conventional statistical models were also provided for comparison with stratified models. The study used Canadian inventory and collision data to demonstrate the expected collision reductions resulting from changes in selected countermeasures, such as warning devices and posted speed limit.

Various countermeasures can be introduced to reduce collisions at highway–railway grade crossings. These countermeasures may take different forms, such as passive and (or) active driver warning devices, supplementary traffic controls (four quadrant barriers, wayside horn, closed circuit television (CCTV) monitoring, etc.), illumination, signage and highway speed limit, etc. In this research, we present a structured model that makes use of data mining techniques to estimate the effect of changes in countermeasures on the expected number of collisions at a given crossing. This model serves as a decision-support tool for the evaluation and development of cost-effective and practicable safety program at highway–railway grade crossings. The use of data mining techniques helps to resolve many of the problems associated with conventional statistical models used to predict the expected number of collisions for a given type of crossing. Statistical models introduce biases that limit their ability to fully represent the relationship between selected countermeasures and resultant collisions for a mix of crossing attributes. This paper makes use of Canadian inventory and collision data to illustrate the potential merits of the proposed model to provide decision support.Key words: highway–railway grade crossing, collision prediction model, countermeasures, Poisson regression.

Current collision prediction models fail to account for the full spectrum of relevant factors affecting the number of collisions at specific highway– railway grade crossings. A number of reasons contribute to this failure, including biases in model parameters resulting from collinearity in the model inputs, absence of important variables in the prediction model caused by lack of statistical significance, the inability of models to consider higher-order interactions, and the presence of unexplained variation in the prediction estimates. These problems have compromised the use of collision prediction models in decisions concerning the development and evaluation of cost-effective safety treatments or countermeasures for application at specific crossings. This paper introduces a stratified collision prediction model for highway–railway grade crossings. The development of this model involves three steps: ( a) crossing inventory variables are expressed in terms of a limited number of orthogonal (nonlinear) underlying attributes or factors; ( b) factor scores are estimated for each crossing and factor, and these scores are used as “seed points” in a subsequent clustering exercise to yield groups or clusters of crossings with similar underlying attributes; and ( c) for each cluster, separate collision prediction models are developed and include important treatment input variables of interest to decision makers and planners. The paper describes an application of a stratified collision prediction model to Canadian highway–railway grade crossing inventory and collision occurrence data for the period 1993 to 2001. The usefulness of the model in estimating collision reduction benefits of selected treatments is illustrated with reference to two countermeasure strategies: upgrades in the type of warning device and the removal of whistle prohibition.

The overall goal of this study was to develop warrants or guidelines for the installation of active warning devices without speed selection (fixed-distance systems) and active warning devices with speed selection (constant warning time systems) at railroad highway grade crossings. This objective was achieved by the use of the FRA grade crossing inventory and accident data bases to develop accident rates, effectiveness values, statistical models to determine which variables are significantly related to accidents, and a benefit cost analysis, all of which were used in warrant development. Overall results of this study tended to confirm the hypothesis that constant warning time systems have greater credibility with motorists than do fixed-distance systems. The effectiveness values for changes from fixed-distance to constant warning time systems confirmed this. Accident rates for both fixed-distance and constant warning time systems were developed and showed that on the average, constant warning time systems tended to have lower rates. A separate regression technique showed that the only two variables that were significantly related to accident rates were warning device types and exposure. In general, constant warning time systems were warranted at a higher exposure level than fixed-distance systems. Keeping the cost ratio fixed, flashing lights of fixed distance are warranted at the least exposure level for crossings formerly equipped with passive devices, followed by flashing lights of constant warning time, gates of fixed distance and last, gates of constant warning time. For crossings formerly equipped with gates of fixed distance, gates of constant warning time will rarely be warranted.

Comparing conventional statistical models and machine learning in a small cohort of South African cardiac patients

Introduction: Congestive heart failure (CHF) in acute phase, many clinical fundamental parameters including vital sign, laboratory data, or initial treatment were investigated, and we can roughly estimate the prognosis with the parameters. However, machine learning method for prediction of the prognosis is still challenging. Hypothesis: Machine learning methods have higher performance for predicting cardiac prognosis in patients with CHF than conventional statistical model. Methods: We enrolled consecutive 300 patients with CHF (79.5±12.1 years, 158 Males). Patients with acute coronary syndrome, mechanical circulatory support cases, and cardio-pulmonary arrest cases were excluded. Objective variable was set as cardiac death in one year. First, multivariate stepwise logistic regression analysis with Akaike’s information criterion (AIC) was performed, and extracted parameters which had the most versatile predictive value. Propensity score (PS) was calculated with the extracted parameters, and the prediction model constructed by cut-off value of ROC curve analysis of PS. Next, machine learning (random forest method and deep learning) to build predictive model was performed with the predictors. Finally, accuracy of each predictive model was compared. Results: Thirty cases showed cardiac death in one year. Logistic regression with AIC extracted 8 predictors, and the cut off-value of propensity score with the 8 parameters was 0.110. The accuracy was 0.714 and area under ROC (AUROC) was 0.836. Conversely, random forest method demonstrated the accuracy as 0.927, AUROC 0.860. On deep learning, the accuracy was 0.937 and AUROC 0.901.The top 4 high feature importance of random forest were Cl/red blood cell count/pH/Anion Gap. However, accuracy of those predictors was lower than that of machine learning. Conclusions: Machine learning was a powerful tool to predict cardiac prognosis of CHF, comparing with conventional statistical model.

The purpose of this research investigation was to analyze the effects of environment, tonography, geometry, and highway and rail traffic patterns with respect to rail-highway grade crossing accidents in rural areas.

A latent class modeling approach for identifying vehicle driver injury severity factors at highway-railway crossings

A design methodology for gate delay and gate interval time for at-grade crossings using four-quadrant gates is developed. The design approach is based on the concept of dilemma zones related to signal change intervals at signalized intersections. The design approach is validated based on data from six sites in Illinois on a proposed high-speed rail corridor. Gate delay and gate interval times are determined that provide an optimal safe decision point to allow a driver to stop before the crossing or to proceed through the crossing without becoming trapped by the exit gates.

Recent experiments (1) on % MathType!MTEF!2!1!+-% feaafiart1ev1aaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn% hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr% 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq-Jc9% vqaqpepm0xbba9pwe9Q8fs0-yqaqpepae9pg0FirpepeKkFr0xfr-x% fr-xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaWexLMBb50ujb% qeguuDJXwAKbacfaGaf8hCaaNbaebacqWFTaqlcqWFWbaCaaa!3E94! $$\bar p - p$$ annihilation have indicated an excess of π- over π+ in % MathType!MTEF!2!1!+-% feaafiart1ev1aaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn% hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr% 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq-Jc9% vqaqpepm0xbba9pwe9Q8fs0-yqaqpepae9pg0FirpepeKkFr0xfr-x% fr-xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaWexLMBb50ujb% qeguuDJXwAKbacfaGaf8hCaaNbaebaaaa!3C50! $$\bar p$$ (forward) and an equal excess of π+ in p-direction (backward). Also, the experimental results (2) on the distribution of angle between like pion-pairs and unlike pion-pairs differ significantly from that in conventional statistical model. To accommodate these effects we are naturally led to consider the emission of some free mesons from the clouds of proton and antiproton and a statistical model for the annihilation of cores with angular momentum conservation and pion-pion resonance. The model reproduces fairly the experimental results.

In biomarker discovery studies, uncertainty associated with case and control labels is often overlooked. By omitting to take into account label uncertainty, model parameters and the predictive risk can become biased, sometimes severely. The most common situation is when the control set contains an unknown number of undiagnosed, or future, cases. This has a marked impact in situations where the model needs to be well-calibrated, e.g., when the prediction performance of a biomarker panel is evaluated. Failing to account for class label uncertainty may lead to underestimation of classification performance and bias in parameter estimates. This can further impact on meta-analysis for combining evidence from multiple studies. Using a simulation study, we outline how conventional statistical models can be modified to address class label uncertainty leading to well-calibrated prediction performance estimates and reduced bias in meta-analysis. We focus on the problem of mislabeled control subjects in case-control studies, i.e., when some of the control subjects are undiagnosed cases, although the procedures we report are generic. The uncertainty in control status is a particular situation common in biomarker discovery studies in the context of genomic and molecular epidemiology, where control subjects are commonly sampled from the general population with an established expected disease incidence rate.

Application of Bayesian model averaging in modeling long-term wind speed distributions

Law and social science scholars have long elucidated ways of governing built around state governance of populations and subjects. Yet many are now grappling with the growing prevalence of practices of governance that depart, to varying degrees, from received models. The profusion of digital data, and the deployment of machine learning in its analysis, are redirecting states’ and international organizations’ attention away from the governance of populations as such and toward the amassing, analysis, and mobilization of hybrid data repositories and real-time data flows for governance. Much of this work does not depend on state data sources or on conventional statistical models. The subjectivities nurtured by these techniques of governance are frequently not those of choosing individuals. Digital objects and mediators are increasingly prevalent at all scales. This article surveys how scholars are beginning to understand the nascent political technologies associated with this shift toward governance by data.

This review delves into the potential of artificial intelligence (AI), particularly machine learning (ML), in enhancing graft-versus-host disease (GVHD) risk assessment, diagnosis, and personalized treatment. Recent studies have demonstrated the superiority of ML algorithms over traditional multivariate statistical models in donor selection for allogeneic hematopoietic stem cell transplantation. ML has recently enabled dynamic risk assessment by modeling time-series data, an upgrade from the static, "snapshot" assessment of patients that conventional statistical models and older ML algorithms offer. Regarding diagnosis, a deep learning model, a subset of ML, can accurately identify skin segments affected with chronic GVHD with satisfactory results. ML methods such as Q-learning and deep reinforcement learning have been utilized to develop adaptive treatment strategies (ATS) for the personalized prevention and treatment of acute and chronic GVHD. To capitalize on these promising advancements, there is a need for large-scale, multicenter collaborations to develop generalizable ML models. Furthermore, addressing pertinent issues such as the implementation of stringent ethical guidelines is crucial before the widespread introduction of AI into GVHD care.

The market for cryptocurrency has thrived for more than 10 years and has experienced a drastic change. The success of cryptocurrencies was concerned and analyzed worldwide. This research discusses the way to build machine learning and statistical models to predict the future price of the cryptocurre