How the natural environment in downtown neighborhood affects physical activity and sentiment: Using social media data and machine learning

Background: Natural environment might encourage physical exercise, hence enhancing human health and wellbeing. Social media offers an extensive repository of spatiotemporal data, containing details on the feelings and behaviors of individuals. However, investigations on physical activity and public sentiment in the natural environment of the downtown neighborhood are lacking in the existing literature. Methods: To extract environmental and behavioral information from social media data and other multi-source data, natural language processing, semantic segmentation, instance segmentation, and fully convolutional neural networks are employed. The research examines how neighborhood blue-green spaces and other health-promoting facilities affect physical activity and public sentiment. Results: The results reveal that blue space visibility, activity facilities, street furniture, and safety all have a favorable influence on physical activity with a social gradient. Amenities, perceived street safety and beauty positively correlated to public sentiment. The findings from social media about the environment and physical activity are consistent with traditional surveys from the same time period with a 0.588 kappa value. Conclusion: According to our findings, social media data might be utilized to learn more about how urban en-vironments influence people's physical activity patterns. Also, the health-promoting effects of blue space require more investigation.