Spatial temporal surveillance of post disaster mental health

Mass traumatic events, such as terrorist attacks and natural disasters, have increased globally in recent years and can have substantial consequences for population mental health. Early emotional reactions after traumatic... [ view full abstract ]

Mass traumatic events, such as terrorist attacks and natural disasters, have increased globally in recent years and can have substantial consequences for population mental health. Early emotional reactions after traumatic events are predictive of later psychiatric symptoms. Targeting geographic areas wherein citizens are exhibiting emotional responses in the immediate aftermath of mass trauma could therefore mitigate long-term mental health consequences.
However, no systematic efforts have investigated the possibility of a space-time syndromic surveillance system for the early emotional consequences of mass trauma. Here we show that a near real-time examination of social media data combining sentiment analysis and a space-time scan statistic has the potential to identify geographic areas with early emotional responses in the aftermath of a disaster. We extracted seven basic emotions from Twitter data in the wider Paris arrondissements in the aftermath of the recent terrorist attacks. We then identified geographic areas in which emotional responses were concentrated. We found geographic concentrations of fear and sadness in the tweets of users in areas around which the attacks took place. A syndromic surveillance system as outlined here may help with the early detection of geographic areas with emerging mental health needs.
We anticipate our study to be a starting point for more sophisticated models in the early detection of mental health risk after mass traumatic events. For example, further efforts could be prospective incorporating real-time social media data also in other languages over larger geographic areas with more fine-grained temporal resolutions. In countries or areas without efficient emergency infrastructure, this approach may also have potential for the early detection of mass trauma and to guide emergency care and rescue efforts.