Can tweets reveal where fear and sadness cluster before and after a hurricane?

Details

Disasters have substantial consequences for population mental health. Social media data offer an opportunity to monitor mental health after disasters and help identify areas with mental health needs.

This observational study aimed to identify specific basic emotions expressed on Twitter in the greater New York City area during Hurricane Sandy, which made landfall on October 29, 2012, and to detect and map space-time clusters representing excess risk of these emotions. Researchers applied advanced sentiment analysis to 344,957 tweets from the study area over eleven days, October 22 to November 1, 2012, to extract basic emotions. They used a space-time scan statistic (SaTScan) and a geographic information system (QGIS) to detect and map excess risk of these emotions.

Sadness and disgust were among the most prominent emotions identified. The analysis identified 24 spatial clusters of excess risk of basic emotions over time: four for anger, one for confusion, three for disgust, five for fear, five for sadness, and six for surprise. Anger, confusion, disgust, and fear clusters appeared before the disaster; a surprise cluster was found around the time of the disaster; and a sadness cluster emerged afterward.

The researchers proposed a novel social-media-based syndromic surveillance approach for mental health that may support conventional approaches by adding useful information during disasters. They showed that excess risk of multiple basic emotions could be mapped across space and time, as a step toward anticipating acute stress in the population and rapidly and efficiently identifying community mental health needs after a disaster.

Further studies are needed to better control for bias, identify associations with reliable and valid mental health measurement instruments, and explore computational methods for continued model-fitting, causal relationships, and ongoing evaluation. This study may also provide a starting point for more fully developed models that prospectively detect mental health risk using real-time social media data, or detect excess risk of emotional reactions in areas lacking efficient infrastructure during and after disasters. Social media data may therefore help monitor mental health after large-scale disasters, identify areas with mental health needs, and guide where interventions might most effectively reduce disasters’ mental health consequences.