In this paper, we present a new method for public health
research combining GIS with social media. Compared with traditional
data collection method, our automated method for detecting
MDD users is faster and cheaper for analysis and diagnosis. The
system can be applied to some online forum for detecting depression
topic and forwarding related questions to psychiatrists. Our
GIS results also provide novel knowledge about this disorder by
examining the geographic clustering of MDD users and relationship
with SES.
In this research, we didn't say that our method for MDD diagnosis
can replace the work of a clinical psychologist. Our method
can improve diagnosis techniques for depression. Further detailedclinical contexts are needed to make a formal diagnosis. Future
study should probe into the difference between depression detected
online and self-reported depression reported by a professional
clinical scale table.
Secondly, the Twitter APIs only allows free access to a one
percent convenience sampling of tweets. Data acquired are
restricted to users with public profiles. These may bring some bias
to our result.
Additionally, we only include tweets written in English and
users who identified themselves living in the U.S. Changing any of
those may affect our results.