Validating medical digital twins for clinical decision support: beyond predictive accuracy.

Population aging has emerged as one of the most significant demographic transformations of the 21st century, creating substantial challenges for public health systems worldwide (). Promoting healthy aging has therefore become a central priority in global health agendas (), with regular physical activity widely recognized as a key behavioral strategy for maintaining functional capacity, preventing chronic diseases, and enhancing overall quality of life among older adults (). Nevertheless, despite the well-documented benefits of physical activity, participation levels among older populations remain insufficient in many settings (), highlighting the need to better understand the determinants of physical activity engagement in later life. Early research on physical activity behavior primarily emphasized individual-level determinants, including health status, motivation, knowledge, and behavioral intentions (). While these factors remain important, an increasing body of public health research has shifted toward ecological perspectives (), which emphasize the role of environmental and contextual conditions in shaping behavioral outcomes (). Ecological models suggest that individuals’ opportunities and motivations for engaging in physical activity are strongly influenced by the environments in which they live (), including neighborhood infrastructure, accessibility of recreational resources, environmental safety, and community-level social norms (). Such environmental influences may be particularly salient for older adults, who often rely more heavily on local community environments due to mobility limitations, safety concerns, and reduced spatial activity ranges. Empirical studies conducted across diverse geographic contexts have consistently demonstrated positive associations between supportive neighborhood environments and physical activity participation among older adults (). Features such as accessible green spaces, pedestrian-friendly neighborhoods, availability of recreational facilities, and supportive social environments have been shown to facilitate physical activity engagement (). However, two issues remain insufficiently addressed. First, much of the existing literature has examined environmental indicators either in isolation or through aggregated environmental indices (), which limits the ability to identify the relative contribution of specific environmental dimensions to physical activity participation (). Second, comparatively limited research has incorporated lifestyle characteristics into the same analytical framework, even though these factors may confound the relationship between environmental conditions and physical activity outcomes (). These limitations make it necessary to examine community environmental dimensions and lifestyle covariates within an integrated modeling framework.