The spatio-temporal residual network (ST-ResNet) leverages the power of deep learning (DL) for predicting the volume of citywide spatio-temporal flows. However, this model, neglects the dynamic dependency of the input flows in the temporal dimension, which affects what spatio-temporal features may be captured in the result. This study introduces a long short-term memory (LSTM) neural network into the ST-ResNet to form a hybrid integrated-DL model to predict the volumes of citywide spatio-temporal flows (called HIDLST). The new model can dynamically learn the temporal dependency among flows via the feedback connection in the LSTM to improve accurate captures of spatio-temporal features in the flows. We test the HIDLST model by predicting the volumes of citywide taxi flows in Beijing, China. We tune the hyperparameters of the HIDLST model to optimize the prediction accuracy. A comparative study shows that the proposed model consistently outperforms ST-ResNet and several other typical DL-based models on prediction accuracy. Furthermore, we discuss the distribution of prediction errors and the contributions of the different spatio-temporal patterns.

Traffic crashes present substantial challenges to human safety and socio-economic development in urban areas. Developing a reliable and responsible traffic crash prediction model is crucial to address growing public safety concerns and improve the safety of urban mobility systems. Traditional methods face limitations at fine spatiotemporal scales due to the sporadic nature of high-risk crashes and the predominance of non-crash characteristics. Furthermore, while most current models show promising occurrence prediction, they overlook the uncertainties arising from the inherent nature of crashes, and then fail to adequately map the hierarchical ranking of crash risk values for more precise insights. To address these issues, we introduce the Spatiotemporal Zero-Inflated Tweedie Graph Neural Networks (STZITD-GNN), the first uncertainty-aware probabilistic graph deep learning model in road-level daily-basis traffic crash prediction for multi-steps. Our model combines the interpretability of the statistical Tweedie family with the predictive power of graph neural networks, excelling in predicting a comprehensive range of crash risks. The decoder employs a compound Tweedie model, handling the non-Gaussian distribution inherent in crash data, with a zero-inflated component for accurately identifying non-crash cases and low-risk roads. The model accurately predicts and differentiates between high-risk, low-risk, and no-risk scenarios, providing a holistic view of road safety that accounts for the full spectrum of probability and severity of crashes. Empirical tests using real-world traffic data from London, UK, demonstrate that the STZITD-GNN surpasses other baseline models across multiple benchmarks, including a reduction in regression error of up to 34.60% in point estimation metrics and an improvement of above 47% in interval-based uncertainty metrics.

A cooperative routing strategy for daily operations is necessary to maintain the effects of hotspot policing and to reduce crime and disorder. Existing robot patrol routing strategies are not suitable, as they omit the peculiarities and challenges of daily police patrol including minimising the average time lag between two consecutive visits to hotspots, as well as coordinating multiple patrollers and imparting unpredictability to patrol routes. In this research, we propose a set of guidelines for patrol routing strategies to meet the challenges of police patrol. Following these guidelines, we develop an innovative heuristic-based and Bayesian-inspired real-time strategy for cooperative routing police patrols. Using two real-world cases and a benchmark patrol strategy, an online agent-based simulation has been implemented to testify the efficiency, flexibility, scalability, unpredictability, and robustness of the proposed strategy and the usability of the proposed guidelines

Predicting trip purpose from comprehensive and continuous smart card data is beneficial for transport and city planners in investigating travel behaviors and urban mobility. Here, we propose a framework, ActivityNET, using Machine Learning (ML) algorithms to predict passengers’ trip purpose from Smart Card (SC) data and Points-of-Interest (POIs) data. The feasibility of the framework is demonstrated in two phases. Phase I focuses on extracting activities from individuals’ daily travel patterns from smart card data and combining them with POIs using the proposed “activity-POIs consolidation algorithm”. Phase II feeds the extracted features into an Artificial Neural Network (ANN) with multiple scenarios and predicts trip purpose under primary activities (home and work) and secondary activities (entertainment, eating, shopping, child drop-offs/pick-ups and part-time work) with high accuracy. As a case study, the proposed ActivityNET framework is applied in Greater London and illustrates a robust competence to predict trip purpose. The promising outcomes demonstrate that the cost-effective framework offers high predictive accuracy and valuable insights into transport planning.

Travel mode choice prediction is critical for travel demand prediction, which influences transport resource allocation and transport policies. Travel modes are often characterised by severe class imbalance and inequality, which leads to the inferior predictive performance of minority modes and bias in travel demand prediction. In existing studies, the class imbalance in travel mode prediction has not been addressed with a general approach. Basic resampling methods were adopted without much investigation, and the performance was assessed by commonly used metrics (e.g., accuracy), which is not suitable for predicting highly imbalanced modes. To this end, this paper proposes an evaluation framework to systematically investigate the combination of six over/undersampling techniques and three prediction methods. In a case study using the London Passenger Mode Choice dataset, results show that applying over/undersampling techniques on travel mode substantially improves the F1 score (i.e., the harmonic mean of precision and recall) of minority classes, without considerably downgrading the overall prediction performance or model interpretation. These findings suggest that combining over/undersampling techniques and statistical/machine-learning methods is appropriate for predicting travel mode, which effectively mitigates the influence of class imbalance while achieving high predictive accuracy and model interpretation. In addition, the combination of over/undersampling techniques and prediction methods enriches the model options for predicting mode choice, which would better support transport planning.

Background: The COVID-19 pandemic has revealed significant disparities in global vaccine accessibility, particularly affecting low and middle-income countries (LMICs). However, current research on COVID-19 vaccine accessibility primarily focuses on individual countries or high-income countries (HIC). We aimed to evaluate geographic accessibility to COVID-19 vaccination on a multicountry scale, covering comparisons across LMICs and HICs. Additionally, we explored the potential economic factors related to accessibility and their impacts on health outcomes. Methods: We collected population data at a 1 km resolution and geocoded all vaccination sites across the selected countries/regions. Four measures were used to evaluate vaccine accessibility from different perspectives: population coverage with varying travel time thresholds, driving time to vaccination sites, the number of sites within a 30-min threshold and a geographic accessibility index using enhanced two-step floating catchment area method. Finally, we explored the relationships between geographic accessibility and several factors: gross domestic product per capita, vaccination uptake and mortality. Findings: We found substantial disparities in vaccine accessibility across the selected countries/regions. In 24.07% of these countries/regions, over 95% of the population can access the nearest vaccination services within 15 min. In contrast, in countries/regions such as Manitoba (Canada), Zimbabwe and Bhutan, less than 30% of the population can reach these sites within 60 min. Underserved areas, termed ‘vaccine deserts’, were identified in both HICs and LMICs. We found that countries/regions with higher vaccine accessibility tend to achieve higher vaccination rates, whereas those with lower vaccine accessibility are likely to experience substantial increases in COVID-19 mortality rates. Conclusion: LMICs require enhanced attention to improve geographic accessibility to vaccination. Additionally, there are internal disparities in accessibility within both HICs and LMICs. National public health officials and global health initiatives are suggested to prioritize ‘vaccine deserts’ and to ensure equitable vaccine access in future pandemics.