Built environment, Mental health, Machine learning, Road ratio, COVID-19, Urbanization. City and country, HT361-384, Political institutions and public administration (General), JF20-2112

Road ratio, representing the proportion of roads in the street view, exerts varying degrees of visual influence on the mental well-being of residents. In our study, we surveyed the psychological conditions of 2,636 Hong Kong residents across four periods: before, during, and after the pandemic. Utilizing machine learning algorithms, we analyzed street view images within a 100-m radius of the residents' locations to determine the proportion of roads within the street views. This served as a representation of the visual impact of roads on residents. Subsequently, we employed Ordinary Least Squares (OLS) models and Multinomial Logit (MNL) models to investigate the relationship between the proportion of road presence in street views and the frequency of various forms of stress among residents across the four identified periods. Our findings indicate that an increase in road ratio correlates with a higher incidence of diverse stress forms. This effect was particularly pronounced during the pandemic, where the influence of road ratio on the frequency of depressive episodes intensified and persisted even after the pandemic had ended. The significance of our research lies in its implications for future urban planning, specifically in how road ratio near residential areas can be reduced and offset with more natural elements to mitigate the adverse effects of road ratio on residents' mental health.