P3.O287: Comparing green space type and area to adolescent obesity rates in metropolitan areas using geospatial analysis and machine learning.

Background: Insufficient physical activity, influenced by limited access to exercise-friendly greenspaces, contributes to increasing childhood obesity rates. However, rigorous evaluation of greenspace quality on a citywide scale is lacking. This study aims to characterize greenspace vegetation types and correlate them with adolescent obesity outcomes, identifying optimal characteristics for promoting health. This data could aid city planners in optimizing greenspace allocation and improving vegetation choices.

Methods: Satellite imagery and municipal data were used to map ten cities, classifying urban, water, grass, and tree areas while excluding areas unlikely to be used by children. A Random Forest model was trained for each city and assessed using validation data. Adolescent obesity rates were analyzed in relation to grass, tree, and greenspace percent cover using a Bayesian regression model. Slope and 95% Credible Interval (CI) were calculated. Percent confidence in slope direction was obtained by sampling from the posterior distribution.

Results: Classified maps are provided for ten major US cities. For grass, a positive correlation is suggested (slope 0.21, CI -0.2 to 0.63, 85% confidence slope>0), while greenspace indicates a negative correlation (slope -0.13, CI -0.4 to 0.13, 85.5% confidence slope < 0) with adolescent obesity rates. Similarly, tree area exhibits a negative correlation (slope -0.16, CI -0.37 to 0.06, 93.5% confidence slope < 0). Income’s effect on the study was assessed; findings suggest it is a weak confounding variable.

Conclusion: Given the rising trend in childhood obesity, alterations to children’s environments are essential for promoting health. Our study suggests increasing tree cover may reduce pediatric obesity rates, with negative associations found between percent greenspace and tree area and adolescent obesity rates (85.5 and 93.5% confidence, respectively). Cities should prioritize using Geographic Information Systems to identify areas needing access to safe green spaces with shade-providing vegetation like trees, potentially lowering adolescent obesity rates and improving citywide health outcomes.