Using machine learning to examine street green space types at a high spatial resolution: Application in Los Angeles County on socioeconomic disparities in exposure.

Background: Compared to commonly-used green space indicators from downward-facing satellite imagery, street view-based green space may capture different types of green space and represent how environments are perceived and experienced by people on the ground, which is important to elucidate the underlying mechanisms linking green space and health. Objectives: This study aimed to evaluate machine learning models that can classify the type of vegetation (i.e., tree, low-lying vegetation, grass) from street view images; and to investigate the associations between street green space and socioeconomic (SES) factors, in Los Angeles County, California. Methods: SES variables were obtained from the CalEnviroScreen3.0 dataset. Microsoft Bing Maps images in conjunction with deep learning were used to measure total and types of street view green space, which were compared to normalized difference vegetation index (NDVI) as commonly-used satellite-based green space measure. Generalized linear mixed model was used to examine associations between green space and census tract SES, adjusting for population density and rural/urban status. Results: The accuracy of the deep learning model was high with 92.5% mean intersection over union. NDVI were moderately correlated with total street view-based green space and tree, and weakly correlated with low-lying vegetation and grass. Total and three types of green space showed significant negative associations with neighborhood SES. The percentage of total green space decreased by 2.62 [95% confidence interval (CI): -3.02, -2.21, p < 0.001] with each interquartile range increase in CalEnviroScreen3.0 score. Disadvantaged communities contained approximately 5% less average street green space than other communities. Conclusion: Street view imagery coupled with deep learning approach can accurately and efficiently measure eye-level street green space and distinguish vegetation types. In Los Angeles County, disadvantaged communities had substantively less street green space. Governments and urban planners need to consider the type and visibility of street green space from pedestrian's perspective.

Associations between green space and preterm birth: Windows of susceptibility and interaction with air pollution.

BACKGROUND: Recent studies have reported inconsistent associations between maternal residential green space and preterm birth (PTB, born < 37 completed gestational weeks). In addition, windows of susceptibility during pregnancy have not been explored and potential interactions of green space with air pollution exposures during pregnancy are still unclear. OBJECTIVES: To evaluate the relationships between green space and PTB, identify windows of susceptibility, and explore potential interactions between green space and air pollution. METHODS: Birth certificate records for all births in California (2001-2008) were obtained. The Normalized Difference Vegetation Index (NDVI) was used to characterized green space exposure. Gestational age was treated as a time-to-event outcome; Cox proportional hazard models were applied to estimate the association between green space exposure and PTB, moderately PTB (MPTB, gestational age < 35 weeks), and very PTB (VPTB, gestational age < 30 weeks), after controlling for maternal age, race/ethnicity, education, and median household income. Month-specific green space exposure was used to identify potential windows of susceptibility. Potential interactions between green space and air pollution [fine particulate matter < 2.5 µm (PM2.5), nitrogen dioxide (NO2), and ozone (O3)] were examined on both additive and multiplicative scales. RESULTS: In total, 3,753,799 eligible births were identified, including 341,123 (9.09%) PTBs, 124,631 (3.32%) MPTBs, and 22,313 (0.59%) VPTBs. A reduced risk of PTB was associated with increases in residential NDVI exposure in 250 m, 500 m, 1000 m, and 2000 m buffers. In the 2000 m buffer, the association was strongest for VPTB [adjusted hazard ratio (HR) per interquartile range increase in NDVI: 0.959, 95% confidence interval (CI): 0.942-0.976)], followed by MPTB (HR = 0.970, 95% CI: 0.962-0.978) and overall PTB (HR = 0.972, 95% CI: 0.966-0.978). For PTB, green space during the 3rd - 5th gestational months had stronger associations than those in the other time periods, especially during the 4th gestational month (NDVI 2000 m: HR = 0.970, 95% CI: 0.965-0.975). We identified consistent positive additive and multiplicative interactions between decreasing green space and higher air pollution. CONCLUSION: This large study found that maternal exposure to residential green space was associated with decreased risk of PTB, MPTB, and VPTB, especially in the second trimester. There is a synergistic effect between low green space and high air pollution levels on PTB, indicating that increasing exposure to green space may be more beneficial for women with higher air pollution exposures during pregnancy.