Exploring the associations between street-view green space quantity and quality,and influenza in Guangzhou,China through machine learning and spatial regression: A socio-economic equity perspective

Abstract

Existing studies have highlighted that green space is associated with non-communicable diseases. However, scant attention has been paid to the association between green space quantity and quality with communicable diseases. Here, we explore the relationships between green space and influenza cases in Guangzhou, China, using street-view green (SVG) space quantity and SVG-quality indicators, which offer a better assessment of urban green space than traditional remote sensing metrics. Influenza cases were collected from hospitalization records, while street-level green space was measured by street-view data and deep neural networks. The neighbourhood deprivation index (NDI) was also used as a proxy for neighbourhood-level socio-economic status. We employed the Random Effects-Eigenvector Spatial Filtering (RE-ESF) regression model because of its usefulness in handling spatial dependence. Findings showed that higher levels of SVG-quantity and quality are associated with a lower number of influenza cases, implying a negative relationship. Specifically, the marginal effects for SVG indicate that influenza may decrease by 145 cases for every unit increase in SVG-quantity, and by 11 cases for every unit increase in SVG-quality. In terms of planning, this could mean that though green quality is essential for the aesthetic part of urban life, quantity is much more critical concerning the containment of influenza. In addition, SVG-quantity and quality moderated the positive association between NDI and influenza cases. In other words, people in more deprived neighbourhoods were more influenced by SVG-quantity and quality compared to people living in less deprived areas. This means that more green space should be added to such neighbourhoods. We also observed that the association between SVG-quality and influenza cases was weaker for females, people aged between 18 and 45, and employed people. Because influenza is the most common pandemic worldwide, green space at the street level should be considered when promoting equitable public health and this study provides quantifiable evidence for the negative effect of green space quantity and quality over influenza cases.