Title:

Computational indicators of urban mental health

Abstract:

More than 1/7th of the world’s population is now active on social media sharing the most intimate details of their lives in real-time on a public forum. Using Natural Language Processing techniques and geolocation methods we can observe the well-being of individuals in different location over time at high levels of temporal resolution. In previous work we demonstrated the ability to detect mood disorders, and possible early warning indicators of mental health transitions, in longitudinal Twitter data. Here I discuss our recent work on studying the degree to which urban and rural areas differ in terms of the well-being of their denizens, and whether we can derive indicators of their evolving mental health status from their online posts. In the case of natural disasters we can track the time it takes an urban community’s well-being to revert back to normal which may indicate their emotional resilience to withstand exogenous shocks.