Assessing health-regulating ecosystem services in rural landscapes: An integrated approach using physiological and...

To support health-oriented rural revitalization, effective management of rural green infrastructure requires robust tools for quantifying restorative ecosystem services and informing evidence-based design. This study develops and tests a multimodal evaluation framework that integrates subjective mood assessment (POMS), physiological stress indicators (GSR and HRV), and deep-learning-based visual semantic analysis (SegFormer) to examine how landscape structure relates to health-relevant restorative responses. The framework was applied in four rural landscape typologies-flat-farmland, hilly-forest, lowland-water-network, and plain-development-through a standardized walking experiment that collected continuous physiological and psychological data from human participants. The results show that specific visual-semantic features were associated with physiological recovery: greater sky openness and lower dominance of built elements were linked to improved heart rate variability, while higher vegetation coverage was associated with reduced skin conductance. The analysis also revealed a dynamic temporal effect, with the quality and continuity of visual landscapes becoming more important for physiological regulation as walking exposure progressed. In particular, transitions from built to more natural environments emerged as critical transitional phases of positive physiological change, suggesting that the sequencing of landscape elements is an essential consideration in restorative route design. Overall, the study extends restorative-environment research into real rural walking settings and translates its findings into an operational multimodal framework for the evaluation and optimization of rural greenways and ecological spaces. The proposed approach offers an empirical basis for integrating health-related outcomes into rural environmental management and planning.