Supporting equitable and responsible highway safety improvement funding allocation strategies - Why AI prediction biases matter.

The existing methodologies for allocating highway safety improvement funding closely rely on the utilization of crash prediction models. Specifically, these models produce predictions that estimate future crash hazard levels in different geographical areas, which subsequently support the future funding allocation strategies. In recent years, there is a burgeoning interest in applying artificial intelligence (AI)-based models to perform crash prediction tasks. Despite the remarkable accuracy of these AI-based crash prediction models, they have been observed to yield biased prediction outcomes across areas of different socioeconomic statuses. These biases are primarily attributed to the inherent measurement and representation biases of AI-based prediction models. More precisely, measurement bias arises from the selection of target variables to reflect crash hazard levels, while representation bias results from the issue of imbalanced number of samples representing areas with different socioeconomic statuses within the dataset. Consequently, these biased prediction outcomes have the potential to perpetuate an unfair allocation of funding resources, contributing to worsen social inequality over time. Drawing upon a real-world case study in North Carolina, this study designs an AI-based crash prediction model that utilizes previous sociodemographic and crash-related variables to predict future severe crash rate of each area to reflect the crash hazardous level. By incorporating a fair regression framework, this study endeavors to transform the crash prediction model to become both fair and accurate, aiming to support equitable and responsible safety improvement funding allocation strategies.

Modeling the lagged impacts of hourly weather and speed variation factors on the segment crash risk of rural interstate freeways: Applying a space-time-stratified case-crossover design.

In the realm of traditional roadway crash studies, cross-sectional modeling methods have been commonly employed to investigate the intricate relationship between the crash risk of roadway segments and variables including roadway geometrics, weather conditions, and speed distribution. However, these methodologies assume that the explanatory variables and target variable are only associated within the same time period. Although this assumption is well-founded for static factors like roadway geometrics, it proves inadequate when dealing with highly time-varying variables related to weather conditions and speed variation. Recent investigations have unveiled that these time-varying variables may exhibit lagged impacts on segment crash risk, necessitating the adoption of more comprehensive time-series modeling methods. This study employs two interpretable statistical methods, namely the distributed lag model (DLM) and the distributed lag nonlinear model (DLNM), to elucidate meaningful and interpretable patterns of the lagged impacts of weather and speed variation factors on segment crash risk. Empirical evidence based on crash data collected from rural interstate freeways in the state of Texas demonstrates coherent and interpretable lagged impact patterns of these variables. This study's results serve as strong support for the existence of lagged impacts on roadway segment-level crash risk, emphasizing the need for considering time-series effects in future crash modeling research. Furthermore, these findings could offer practical implications for the design of real-time crash warning systems and the effective implementation of variable speed limits to enhance road safety.

Applying an interpretable machine learning framework to study mobility inequity in the recovery phase of COVID-19 pandemic.

The COVID-19 pandemic is a public health crisis that also fuels the pervasive social inequity in the United States. Existing studies have extensively analyzed the inequity issues on mobility across different demographic groups during the lockdown phase. However, it is unclear whether the mobility inequity is perennial and will continue into the mobility recovery phase. This study utilizes ride-hailing data from Jan 1st, 2019, to Mar 31st, 2022, in Chicago to analyze the impact of various factors, such as demographic, land use, and transit connectivity, on mobility inequity in the different recovery phases. Instead of commonly used statistical methods, this study leverages advanced time-series clustering and an interpretable machine learning algorithm. The result demonstrates that inequity still exists in the mobility recovery phase of the COVID-19 pandemic, and the degree of mobility inequity in different recovery phases is varied. Furthermore, mobility inequity is more likely to exist in the census tract with more families without children, lower health insurance coverage, inflexible workstyle, more African Americans, higher poverty rate, fewer commercial land use, and higher Gini index. This study aims to further the understanding of the social inequity issue during the mobility recovery phase of the COVID-19 pandemic and help governments propose proper policies to tackle the unequal impact of the pandemic.

Finite-time state feedback stabilization of strict-feedback switched systems with asymmetric output-constraints.

Finite-time stabilization of strict-feedback switched systems with asymmetric output constraints (AOCs) via state feedback is investigated. First, an elaborately constructed fraction-type barrier Lyapunov function (BLF) is presented. Then, with a mild assumption on strict-feedback switched systems, state feedback laws are constructed by revamping the adding a power integrator approach (AAPIA) and meanwhile a common Lyapunov function (CLF) is also obtained. The resultant closed-loop systems are finite-time stable (FTS) and the asymmetric output constraint is satisfied too. The approach, proposed in this note, can make strict-feedback switched systems with/without AOCs finite-time stable in a unified frame.

Stabilization of planar switched systems with an output constraint via output feedback.

In this paper, smooth output feedback stabilization (OFS) problem of planar output-constrained switched nonlinear systems (SNS) is investigated. Adding a power integrator technique (APIT) is employed to design state feedback controllers in a systematic manner combining with a constructed common logarithm-type barrier Lyapunov function (BLF). Then, by incorporating the variable-gain reduced-order observers designed constructively, switched systems can be stabilized via smooth output feedback and the output constraint is guaranteed simultaneously. Finally, simulations results are shown to exemplify the validity of the method proposed in this paper.