Cross-centerline crashes occur rarely in the United States but are especially severe. This type of crash is characterized by one vehicle departing over a centerline and encountering a vehicle traveling in the opposite direction. In recent years, automakers have started developing and implementing lane departure warning (LDW) on newer vehicles. This system provides the potential to reduce or significantly impact the frequency of cross-centerline crashes. The objective of this thesis was to estimate the potential crash and injury benefits of a LDW system if installed on every vehicle in the US fleet. This research includes the following 1) a characterization of cross-centerline crashes in the United States today with current and future prevention methods, 2) a reconstruction methodology used for all crashes including rollovers and heavy vehicles, and 3) a simulation model and approach method used to estimate potential benefits of LDW systems on cross-centerline crashes. Cross over to left crashes account for only 4% of non-junction non-interchange crashes but account for 44% of serious injury crashes of the same type. As part of this research, 42 cross-centerline crashes were reconstructed and simulated as if they had a LDW system installed. Accounting for driver capability to react to a LDW alert, crash reduction benefits ranged from 22 – 30%.Using injury risk curves, the probability of experiencing a MAIS2+ injury in a cross-centerline crash was reduced by 29% when using a LDW system. / Master of Science

Vehicular crashes are the leading cause of death for young adult drivers, however, very little life course research focuses on drivers in their 20s. Moreover, most data analyses of crash data are limited to simple correlation and regression analysis. This thesis proposes a data-driven approach and usage of machine-learning techniques to further enhance the quality of analysis. We examine over 10 years of data from the Iowa Department of Transportation by transforming all the data into a format suitable for data analysis. From there, the ages of drivers present in the crash are discretized depending on the ages of drivers present for better analysis. In doing this, we hope to better discover the relationship between driver age and factors present in a given crash. We use machine learning algorithms to determine important attributes for each age group with the goal of improving predictivity of individual methods. The general format of this thesis follows a Knowledge Discovery workflow, preprocessing and transforming the data into a usable state, from which we perform data mining to discover results and produce knowledge. We hope to use this knowledge to improve the predictivity of different age groups of drivers with around 60 variables for most sets as well as 10 variables for some. We also explore future directions this data could be analyzed in.

BACKGROUND- Motor vehicle crashes pose a serious public health problem. Many serious crashes are due to faulty driving by unfit operators, including several categories of neurological patients. Unfortunately, there seems to be little agreement among health professionals, driving experts, and state government on how to advise these individuals. REVIEW SUMMARY- This article reviews the question of driving in neurological patients. Decisions on driver fitness should be based on empirical observations of performance and not on criteria of age or medical diagnosis, which alone are unreliable predictors. Relevant data can be collected either on a road test or off-road, using different probes of vision and cognition, in the setting of a Department of Motor Vehicles office or medical clinic. The use of a driving simulator is also feasible. The predictive value of these performance assessments is a topic of active research. CONCLUSION- Understanding how performance data from off-road and on-road observations correlate with real-life crash risk is a key step toward developing safe, fair, and accurate means of predicting driver fitness. One potential benefit is the prevention of injury, and another is the preservation of mobility and independence of individuals whose licenses are being unduly revoked because of old age or illness.

Time series involving count data are present in a wide variety of applications. In many applications, the observed counts are usually small and dependent. Failure to take these facts into account can lead to misleading inferences and may detect false relationships. To tackle such issues, a Poisson parameter-driven model is assumed for the time series at hand. This model can account for the time dependence between observations through introducing an autoregressive latent process. In this thesis, we consider Bayesian approaches for estimating the Poisson parameter-driven model. The main challenge is that the likelihood function for the observed counts involves a high dimensional integral after integrating out the latent variables. The main contributions of this thesis are threefold. First, I develop a new single-move (SM) Markov chain Monte Carlo (MCMC) method to sample the latent variables one by one. Second, I adopt the idea of the particle Gibbs sampler (PGS) method \citep{andrieu} into our model setting and compare its performance with the SM method. Third, I consider Bayesian composite likelihood methods and compare three different adjustment methods with the unadjusted method and the SM method. The comparisons provide a practical guide to what method to use. We conduct simulation studies to compare the latter two methods with the SM method. We conclude that the SM method outperforms the PGS method for small sample size, while they perform almost the same for large sample size. However, the SM method is much faster than the PGS method. The adjusted Bayesian composite methods provide closer results to the SM than the unadjusted one. The PGS and the selected adjustment method from simulation studies are compared with the SM method via a real data example. Similar results are obtained: first, the PGS method provides results very close to those of the SM method. Second, the adjusted composite likelihood methods provide closer results to the SM than the unadjusted one.