Evaluation models of environmental cognitive performance in the indoor visual environments of care facilities for older adults

The visual environment, as a non-pharmacological intervention, plays a crucial role in supporting cognitive compensation and promoting functional independence among older adults with cognitive impairment. The absence of performance-driven evaluation methods for environmental cognitive compensation effectiveness hinders the development of cognitive-friendly environments. Therefore, we propose the concept of environmental cognitive performance (ECP) to measure the compensatory effects of the environment on cognition. First, an innovative ECP evaluation system was developed by integrating interface, spatial, and semantic dimensions with the cognitive needs of older adults with mild cognitive impairment. Subsequently, five machine learning models were employed to construct and optimizes a series of evaluation models across commonly used functional spaces. Finally, the importance and compensatory effects of ECP indicators in these functional spaces for supporting different cognitive abilities were analyzed, and critical thresholds for key indicators were identified. The findings emphasize that (1) the ECP evaluation models developed using the RF model exhibit superior explanatory power, with most models achieving R² values exceeding 0.6; (2) the cognitive compensation effects of ECP indicators are highly dependent on specific functional spaces and particular ECP goals; (3) the semantic dimension exhibits the most pronounced cognitive compensation effects, particularly the window proportion within the field of view, followed by the spatial and interface dimensions; (4) key ECP indicators demonstrate nonlinear cognitive compensation effects, with their critical thresholds providing decision-making guidance for designing cognitive-friendly environments. This study offers designers and managers a quantitative ECP evaluation system and models, establishing a reproducible performance- driven evaluation paradigm.