Study on the mechanisms associating community outdoor public spaces with elderly behavior.
Journal:
Scientific reports
Published Date:
Aug 22, 2025
Abstract
As the global population ages, enhancing community outdoor public spaces to accommodate the needs of senior citizens has emerged as a critical challenge. This research delves into the intricate relationship between community outdoor public spaces and the behavioral patterns of the elderly, seeking to inform strategies for optimizing these spaces. The complexity and diversity of the mechanisms linking elderly behaviors with the characteristics of their outdoor environments pose challenges in identifying clear guidelines for improvement. Traditional methods of collecting behavioral data, such as questionnaires and manual observations, are time-consuming and limit the scope and detail of data captured. In contrast, computer vision technologies offer an efficient alternative for gathering behavioral data. However, the application of computer vision to specifically identify various behaviors of the elderly population presents certain challenges. This study addresses two key issues: improving the use of computer vision to recognize diverse behaviors of the elderly; and elucidating how community outdoor public spaces shape the outdoor activities of seniors and identifying crucial influencing factors. The research proceeds by initially categorizing elderly behavior characteristics and typologies of outdoor public spaces based on the physiological and psychological needs of seniors. The spatial elements are classified into four metrics: spatial, greenness, functional facilities, and accessibility. A computer vision-based behavior detection algorithm is then constructed to effectively identify six typical activities of the elderly: exercising, jogging, sitting, standing, walking, and playing chess or cards. Subsequently, a set of quantifiable indicators for community outdoor public spaces is established, and nonlinear machine learning models (Random Forest, Gradient Boosting Decision Tree, and eXtreme Gradient Boosting) are employed to reveal the association mechanisms between these six behaviors and the four categories of spatial metrics. The findings highlight 16 major characteristics that have a significant impact on elderly behavior, such as area size, form, green enclosure, and types of workout equipment.
