Unraveling the travel patterns of ride-hailing trips for medical mobility: comparative analysis across hospital tiers

With the rapid growth of urban healthcare demand, understanding hospital-related trips is crucial for improving equity in service accessibility. While ride-hailing offers flexible mobility, most existing studies focus on general travel and rarely quantify hospitals’ spatial attractiveness using large-scale origin–destination (OD) data. This study investigates the spatiotemporal patterns and determinants of hospital-related ride-hailing trips using over 150,000 hospital-related records in Tianjin, China. Hospital-related trips are identified from text-inclusive OD data, and a Shannon entropy index is constructed at the traffic analysis zone (TAZ) scale to quantify hospitals’ spatial attractiveness. We integrate multi-sourced data, including hospital attributes, built environment characteristics and TAZ-side socioeconomic and infrastructure variables, to examine the determinants of hospital-related ride-hailing trips. Gradient Boosting Decision Tree (GBDT) models with Shapley Additive exPlanations (SHAP) interpretation are employed to reveal the nonlinear determinants under temporal variations (weekday versus weekend) and directional differences (origin versus destination). The comparative analysis indicates that hospital tier and area are decisive determinants. Tertiary hospitals display the highest entropy values, while primary hospitals mainly serve localized demand. Hospital-related trips are more frequently generated on weekdays than on weekends. Compared with hospital-buffer variables, TAZ-side factors exert stronger influence. Population density and proximity to the city center have positive effects, and housing prices, hotel density, automobile service density, and transit station density exhibit significant nonlinear effects. Finally, we propose policy implications aimed at improving hospital-related ride-hailing service efficiency and equity.