Incorporating uncertainty quantification into deep-learning-based travel mode choice modeling: A Bayesian Neural Network approach and an uncertainty-guided active survey framework

Determinants of green travel modes in old towns: Evidence from Beijing, China

Gendered transition in travel mode patterns following childbirth: A longitudinal analysis of pre-childbirth working parents in the Netherlands

Geographic Impact of Surgical Conferences on Travel Burden and Carbon Emissions: Multisociety Study.

ABSTRACT Electric micromobility (e-micromobility) has gained significant popularity in cities worldwide, yet urban infrastructure often fails to support adequate access to and use of these modes for a wide range of users. This systematic literature review applies the Capabilities Approach to identify key physical features of the built environment necessary to foster e-micromobility as an equitable and inclusive travel option. We analyse how personal, social, political, and economic conversion factors interact with physical features of the built environment to impact the translation of e-micromobility resources into three capabilities: e-micromobility access, mobility and accessibility. Drawing from the review of 31 studies, we discuss the interactions between conversion factors, propose a definition of equitable e-micromobility, and identify infrastructure elements to be prioritised by planners to foster the three capabilities. Key priorities include (i) dockless parking; (ii) dense and connected dedicated lanes, physically segregated from motorised vehicles and pedestrians; (iii) affordable fares of shared systems; and (iv) regulations to reduce conflicts between modes. The findings suggest the need for research that approaches e-micromobility equity from a systems perspective: focusing on the relationships between the conversion factors and on the interactions of e-micromobility with other travel modes. The explicit comparison of how differently e-bikes and e-scooters serve vulnerable groups and the direct elicitation of perceptions and preferences within these groups are also important gaps that should be filled to guide the design and implementation of more equitable e-micromobility systems.

Evolution of urban travel mode choice modelling from discrete choice models to machine learning

ABSTRACT This study explores how psychological anxiety shaped individual transport mode choices during the early stages of the COVID-19 outbreak, prior to the implementation of nationwide restriction policies. Focusing on the intersection of public health and urban mobility, we investigate how factors such as awareness, fear, and coronaphobia influenced modal shift decisions in the absence of formal restrictions. A web-based survey was conducted in Daegu, South Korea – one of the first metropolitan areas to experience a major outbreak. Data were collected from 417 residents, covering psychological perceptions, transport usage before and after the outbreak, and preferred safety measures. We applied an ensemble classification model using the AdaBoost algorithm to predict modal shifts across various trip purposes, leveraging both sociodemographic and psychological indicators. Results show that heightened psychological anxiety led to significant reductions in public transit use, with a corresponding increase in the use of private cars, bicycles, and walking. These shifts occurred even without policy-enforced restrictions, suggesting that behavioural response to health risk perception plays a critical role in shaping urban transport demand. The study highlights the importance of integrating psychological dimensions into transport planning, particularly during crisis conditions. Findings support the development of adaptive, health-sensitive mobility strategies capable of responding not only to government interventions but also to individual psychological reactions in times of public health emergencies. Highlights Psychological anxiety shaped urban travel choices during the coronavirus outbreak. Fear of infection reduced public transport use and encouraged safer travel modes. A two-stage framework assessed whether and how travelers changed modes. Machine learning predicted pandemic transport mode shifts with high overall accuracy. Findings support health-sensitive planning for resilient urban transport systems.

The barrier effect refers to the difficulties experienced by pedestrians in their movements due to transport infrastructures or high traffic volumes and speeds ( physical barriers ), and the consequent creation of unpleasant environments surrounding these infrastructures due to factors such as noise, air pollution and fear of crime ( environmental barriers ). Few studies have addressed environmental barriers or employed indices to measure both physical and environmental barriers. In this cross-sectional study, we used a questionnaire to collect data in Santiago Province, Chile. The main components associated with residents’ perceptions of the barrier effect were determined using factor analysis. Cluster analysis was then applied to develop barrier effect indices based on the factor scores. Finally, we used logistic regression models to estimate relationships between these indices, the choice of active modes, and the frequency of interactions with neighbours. The sample comprised 1812 valid responses. Our findings indicated that perceptions of the barrier effect can be explained by two components: perceptions of vehicular traffic and crossing points (physical barriers) and perceptions of the surroundings of transport infrastructures (environmental barriers). Travelling to local destinations by active modes was more likely among participants who were carers than among those who were employed. For supermarket trips in particular, participants with basic education were more likely to use active modes than those with a university education. Additionally, perceptions of physical barriers were significantly associated with the use of active modes, whereas neither type of perceived barrier was associated with the frequency of interactions with neighbours. This study shows the importance of integrating the perceptions of pedestrians and cyclists into the planning and design of both new and existing transport infrastructures to improve accessibility. Such an approach is relevant for promoting safe and comfortable environments and encouraging the use of active modes for local activities. • The study examined the barrier effect, travel mode, and interactions with neighbours. • Associations were assessed using multivariate logistic regression. • Carers were more likely than employed people to travel locally by active modes. • Interactions with neighbours were not associated with perceptions of the barrier effect.

Inferring crowd crush accidents in typical high-density pedestrian movement zones via the vision-trajectory fusion neural network.

Trips for educational purposes represent a significant portion of morning and evening peak hour trips. These trips, if carried out by private transport, can lead to several negative consequences including increased traffic congestion, air and noise pollution, and driver discomfort. This study aimed at predicting the mode choices of school-going students in two medium-sized South Asian cities, Kandy, Sri Lanka, and Sahiwal, Pakistan. City-specific classification models were developed for each city, followed by cross-city evaluations using a subset of common features. SHapley Additive exPlanations (SHAP) were employed to interpret model behavior and assess the stability of learned decision logic across contexts. Ensemble models, particularly CatBoost and Gradient Boosting, consistently outperformed linear and single-tree classifiers in both cities, with substantially stronger predictive performance observed in Sahiwal due to richer household and contextual information. SHAP analyses reveal a shared behavioral foundation across cities in which cost-related variables dominate mode choice decisions. Higher costs for both private and sustainable modes are associated with continued reliance on the corresponding mode, indicating necessity-driven, mode-aligned behavior rather than cost-induced switching. Distance, income, and school type exert secondary but context-dependent effects within cities. Cross-city transferability analysis demonstrates limited and asymmetric generalizability. Models trained in one city experience pronounced performance degradation and systematic classification biases when applied to the other. SHAP-based diagnostics show that transferred models undergo marked reconfiguration of decision logic, including reduced spatial sensitivity and disproportionate reliance on cost signals, with evidence of decision-structure collapse under certain transfer directions. These results highlight the strong context dependence of school travel behavior and the need for locally calibrated, explainable modeling approaches.

• Survey examines commuting patterns and attitudes toward emission reduction. • Commuting patterns differ significantly between students and university staff. • Single-occupancy cars account for nearly 90 % of commuting-related GHG emissions. • Commuting data enhances Scope 3 emissions reporting and mitigation planning. • Universities can influence commuting emissions through targeted strategies. Commuting is one of the 15 categories of Scope 3 emissions and is recognized as a significant contributor to universities' carbon footprints. This study investigates the commuting patterns of students and staff at LUT University, Finland, estimates the associated greenhouse gas (GHG) emissions in accordance with the GHG Protocol, and explores potential mitigation measures. Data were collected through an online survey capturing travel modes, commuting distances and frequencies, as well as attitudes toward emission reduction initiatives. The analysis revealed marked differences between the commuting behaviors of students and staff. Students primarily used walking, buses, and single-occupancy cars in similar proportions, whereas nearly half of the staff relied on single-occupancy cars. As a result, car use accounted for approximately 90% of total commuting-related GHG emissions. Recommended mitigation strategies include reducing single-occupancy car use and promoting public transportation and active mobility through incentives, improved infrastructure, expanded mobility services, and engagement strategies. Although commuting emissions fall outside universities’ direct control, institutions play a critical role in shaping travel behavior. Comprehensive and accurate emissions accounting is essential for identifying effective interventions and tracking progress toward climate goals.

Active commuting to and from school among Spanish adolescents: social inequalities and environmental characteristics.

ABSTRACT Synergistic governance of air pollution control (AP) and carbon mitigation (CM) is a pivotal strategy for China. This study evaluates provincial-level AP-CM synergistic governance performance from 2013 to 2020, using an indicator evaluation system and a coupling coordination degree model. To understand the spatially varying relationships underlying these dynamics, it compares three models’ performance, including OLS (Ordinary Least Squares), GWR (Geographically Weighted Regression), and MGWR (Multi-scale Geographically Weighted Regression). The results indicate that China’s synergistic governance performance improved significantly over the study period, primarily driven by stringent environmental policies and enhanced coordination mechanisms. Spatially, a distinct gradient was observed, where the eastern, central, and southern regions demonstrated higher synergy levels, while western regions exhibited lower levels. This evolution was shaped by the varying performances of the individual AP and CM subsystems across provinces and time periods. Moreover, driving factors such as economic development, energy utilization, the industrial production structure, and public travel modes significantly influenced the synergistic governance performance. The MGWR analysis revealed substantial spatial heterogeneity in these effects across China. The findings of this paper underscore the importance of developing locally tailored strategies that address specific contexts and spatially varying driver impacts, rather than applying a “one-size-fits-all” strategy.

Population health interventions may change physical, social or fiscal environments to shift health behaviours and often operate within a complex system of interacting factors. Systems methods are frequently suggested as a way to provide an overview of the interactions that determine behaviours or actions and may help to understand how or why interventions might (or might not) change health. However, there are few examples of how these methods have been used to guide empirical evaluations of public health interventions. We describe our application of a qualitative complex systems approach within an ongoing evaluation of a population health intervention: traffic restriction schemes, often referred to as School Streets, where motor vehicle access outside schools is restricted at pick-up and drop-off times. This paper presents a methodological case study illustrating how these methods can be applied in practice. We intended to use systems methods across our evaluation; here we describe how we use these methods to build theory. Following the five-step approach outlined by Alvarado et al., (2023), we produced research propositions, systems archetypes and a causal loop diagram to understand the underlying system in which traffic restriction schemes are implemented and their potential impact on school-based active travel. In bringing together diverse and multi-disciplinary evidence from different stakeholders we identified unanticipated systems interactions such as increased initial tensions and conflicts between users of different active travel modes (e.g. cyclists vs pedestrians). Using causal loop diagrams and systems archetypes, we developed research propositions focused on funding and implementation, safety, habit formation and the potential for unintended consequences. Taking a complex systems approach has deepened our understanding of how traffic restriction schemes interact with their broader context and helped us develop a guiding theoretical framework for our ongoing evaluation. We intend this work to stimulate discussion, offer insights for future evaluations of population-level health interventions, and encourage public health researchers to adopt systems methods in their own evaluations.

BackgroundVarious interventions have been developed with the intention to increase the amount of physical activity among children, including those promoting active school travel (AST). However, no gold standard currently exists for measuring different travel modes in AST. This study evaluates the criterion validity of a web-based data collection tool designed for children to self-report their school travel.PurposeTo assess the criterion validity of a web-based data collection tool for daily self-registration of commuting mode, time, and distance among middle school children in Sweden.MethodsThirty children (10–12 years old) from six schools in Falun, Sweden, were recruited using snowball sampling. The children self-reported their school travel data for one day, including travel mode, commuting time, and distance. Their reports were compared to a criterion standard based on direct observations. Spearman correlation and the Wilcoxon signed-rank test were used to analyze the accuracy of the self-reported data.ResultThe web-based data collection tool demonstrated 100% agreement between self-reported and observed travel modes, with high correlations for commuting time and distance (rs=0.953–0.989, p < 0.001). The Wilcoxon signed rank test showed no significant differences between self-reported data and criterion standard (p = 0.243–0.903).ConclusionThe strong agreement between self-reported and observed travel data indicates high criterion validity, suggesting that the web-based data collection tool is a reliable method for middle school children to self-report their daily school travel.

Optimizing mobile health clinic placement via geospatial modeling

Timely access to public healthcare is fundamental human rights and a key measure of social equity. In Thailand, transportation barriers especially in rural and underserved areas continue to restrict equitable access to medical services, reinforcing existing social disparities. This study investigates the determinants of hospital transport service utilization, focusing on the differences in travel behavior between urban and rural populations. A dataset of 1,200 respondents was analyzed using Categorical Boosting (CatBoost), a gradient-boosting machine learning algorithm known for high predictive accuracy and interpretability. The results indicate that The CatBoost model outperformed traditional statistical approaches, namely the Binary Logit Model, in identifying behavioral and contextual determinants of transport use. Key influencing factors included travel time, waiting time, travel cost, and parking fees, alongside demographic attributes such as age, income, and travel frequency. Findings reveal persistent inequities in healthcare accessibility shaped by transportation infrastructure and socioeconomic status. By integrating interpretable machine learning with a social equity perspective, this study demonstrates how data driven insights can inform inclusive and context sensitive health transport policies. The results contribute to global discussions on mobility justice and equitable healthcare access, emphasizing the need for socially responsive interventions to enhance accessibility, efficiency, and well-being across urban and rural communities.

Spatial patterns and determinants of children's school travel mode choice in México City: The link between escorting and active travel

• We examine travellers’ choices between autonomous vehicles (AVs) and human-driven vehicles (HVs) under different information conditions. • A repeated mode-choice game was implemented where participants made decisions based on varying levels of information about travel outcomes. • Mixed multinomial logit (MMNL) models were estimated to analyse the factors influencing travellers’ choices. • Learning patterns, heterogeneous preferences, and fairness concerns significantly influence mode choice in mixed AV-HV traffic environments. With rapid advancements in automation, autonomous vehicles (AVs) will soon coexist with human-driven vehicles (HVs) in mixed traffic. This study investigates the coordination behaviour in mode choice between AVs and HVs with positive and negative externalities, respectively. Participants play a mode choice game repeatedly under partial information (PI) or full information (FI) feedback on payoffs of the alternatives. The descriptive analysis indicates that choices under PI are close to the outcome at the Nash equilibrium; however, choices under FI are consistently below this equilibrium outcome. This discrepancy leads to a decline in social welfare, despite the provision of additional information. To test the factors influencing choice behaviour, we built a series of mixed multinomial logit (MMNL) models. The first model reveals substantial heterogeneity in individual preferences for AVs and indicates that full information feedback weakens the influence of inherent preferences while reinforcing the role of dynamic experience learning. In the second model, the memory effect diminishes further when payoff fairness is introduced. Fairness considerations significantly shape mode choices: while some participants prioritise equitable payoffs, others pursue individual payoff maximisation, producing aggregate outcomes between the distributions induced by the payoff equity and by the Nash equilibrium. By understanding the factors influencing choice behaviour and payoff preferences, policymakers and transport planners can develop targeted interventions to promote AV adoption, ultimately facilitating a smoother transition towards autonomous vehicle usage.

Smartphone-based travel surveys are gaining popularity for collecting trip diaries, where much of the travel information (i.e. travel modes, purposes) is inferred and the participants are asked to verify or correct the inferred information. While this allows researchers to collect high-resolution mobility data from larger samples with minimal user input, in many cases, the participants do not verify all of their trips, resulting in a considerable amount of untagged or missing data. To better impute missing trip purposes for individuals with partially tagged trips, we propose using posterior analysis within a mixed multinomial logit (MMNL) modelling framework. Posterior analysis leverages Bayes' rule to predict missing data by conditioning on the individual's previously observed trip purposes. Using a two-week-long dataset collected in the UK, we find this approach provides a better predictive performance on a testing dataset in terms of predicting market shares and the probability of choosing the correct alternative when compared to not using posteriors in MMNL or using a standard multinomial logit model. We further investigate the impact of using the imputed trip purposes in a mode choice model. Our findings demonstrate that imputing trip purposes using MMNL with posterior analysis achieves a better model fit and values of travel time compared to mode choice models that use (a) other methods of imputing trip purposes, and (b) ignore trip purpose as an explanatory variable. This highlights the advantages of our imputation strategy, as well as the benefits of incorporating imputed trip purpose data in developing travel behaviour models.