How does the built environment in high-density cities affect subway travel for the older adults: insights from travel chain and explainable machine learning

Introduction Understanding how the built environment influences older adults’ mobility is vital for inclusive transport planning in aging, high-density cities. This study explores the determinants of metro travel patterns among older adults in Shanghai. Methods This study employed large-scale metro smart card data from Shanghai and applied a Geographically Weighted Random Forest model with SHAP interpretability. Older adults’ travel chains were reconstructed to identify destination stations. Nonlinear and interaction effects were analyzed, and clustering was used to classify stations by population density and ridership. Results Functional facilities, particularly shopping, living services, and government institutions, were the strongest drivers of metro use. In contrast, traditional density measures such as intersection density and floor area ratio showed limited impact due to saturation effects in highly developed urban areas. Nonlinear analysis revealed threshold effects, with moderate provision of facilities and green spaces preferred over excessive density. Interaction analysis demonstrated synergistic or suppressive effects between POI diversity and intersection density. Clustering further identified distinct station types, highlighting spatial heterogeneity in built environment influences. Conclusion Integrating GWRF with SHAP provides robust insights into spatially varying and nonlinear effects on older adults’ metro travel. Findings highlight the primacy of functional facilities over density measures and the importance of threshold and interaction effects. These results offer evidence to guide age-friendly and targeted transport policies in rapidly urbanizing contexts.