Aging populations around the world pose unique challenges in urban planning and design. To address these challenges, it is crucial to create public spaces that cater to the needs and preferences of older adults. This study proposes the use of deep learning algorithms to assess and improve the design features of public spaces to enhance their suitability for aging individuals. This study presents a multi-scale attention convolutional neural network (MSACNN) to integrate deep learning models with the evaluation of intelligent design for aging-friendly public environments. In this study, we suggest using an attention-based multi-scale network to learn the connection consciously. To better study the link between the quality aspects of intelligent design for aging adaptive public spaces, it mixes multiple scale features at the same spatial location utilizing parallel networks with different convolution kernels. Extensive tests are conducted, and the findings show that the multi-scale network based on the attention mechanism performs better than other networks. In comparison to other machine learning techniques, MSACNN has higher precision and recall. The outcomes of this research contribute to the field of urban planning and provide valuable insights for policymakers and designers in creating inclusive environments for all age groups.