The aggravating trend of population aging has contributed to the growing demand for ageing-friendly housing design, and traditional housing environments are difficult to meet the physiological and psychological needs of the elderly. Existing senior housing has significant deficiencies in heat and humidity environment, sound and light environment, and air quality, and lacks scientific prediction models to guide design decisions. The rapid development of intelligent technology provides new ideas to solve this problem, and building smart senior living communities by combining artificial intelligence with age-appropriate design has become an important way to meet the challenges of aging. This study analyzes the influencing factors of aging-friendly residential design comfort through the field measurements of senior living buildings in five cities, namely Dalian, Dezhou, Yulin, Xinyang, and Wuhan, and constructs an aging-friendly residential design demand prediction model based on BP neural network. The study adopts the neural network algorithm with 25 evaluation indexes as the input layer neurons and the comprehensive expectation value as the output layer neurons, and the model is trained and validated by 328 valid questionnaire data. The results show that the average indoor temperatures in northern cities with heating range from 22.98°C to 25.12°C, which are significantly higher than those in southern cities without heating, which range from 11.49°C to 11.62°C. The average indoor PM2.5 concentration in rural dwellings is 156.9 μg/m³, which is well above the design limit of 75 μg/m³; the relative error of the prediction model is controlled within 5%, and the absolute error is maximum 0.333 It is concluded that the BP neural network model can effectively predict the design needs of ageing homes, provide a scientific basis for the design of homes in smart elderly communities, and promote the optimization and improvement of the environment of ageing homes.
