Thursday, March 10, 2016

Study to develop means of mining social media data to detect signs of mental illness

Every minute of every day there are about 347,000 tweets on Twitter. On Facebook there are 293,000 statuses. Four hundred hours of YouTube videos are uploaded.

 1 of 2 
Facebook users Liked more than 4.1 million posts per minute in the most recent year. Twitter users tweeted more than 347,000 times, a considerable increase from 277,000 last year. Apple uses downloaded 51,000 apps, a slight increase from 48,000 a year before.
This data is being mined by marketers in order to produce target ads, but it is also being used by governments, scientists, and law enforcement agencies to help with their needs.The data has been used for purposes as diverse as predicting epidemics and foiling cyber-terrorists. Soon the data will be mined as a means to identify and monitor people who show signs of mental illness on line.
Minister of Science in the Liberal government, Kirsty Duncan announced that as part of the Natural Sciences and Engineering Council of Canada's (NSERC) Strategic Partnership Grants that $464,100 had been granted to Diana Inkpen of the University of Ottawa. The grant was for a three-year-long project callled "Social web mining and sentiment analysis for mental illness detection".
The researchers are not just at the University of Ottawa but at the University of Alberta and also at Montpellier University in France. The team will used social media data to screen for individuals who are at risk of mental health issues. Inkpen said to CBC news: "We want to look at what kind of emotions people express, and then we will focus in particular on negative emotions that might show some early signs of possible mental disorders, It could be depression, it could be anorexia, it could be other kinds of early mental illness signs."
The researchers intend to develop tools that could be used by doctors, psychologists, school counselors and other researchers that would enable them to flag certain patterns found in posts by those using social media.
Inkpen said that data would be collected from public sites such as Twitter, public medical forums, and different interest groups on Facebook.
Ottawa-based company Advanced Symbolics is partnering with scientists to collect data from social media in both French and English. The Inkpen team will use algorithms in order to discover patterns within the data and to predict what these patterns mean. The algorithms can pick up such patterns as strong very negative emotions that appear frequently or over a long period. Changes in online activities can be picked up as well. If a doctor's patient agreed to be monitored in this way, the doctor could receive automatic alerts should the patterns indicate a risk for the patient.
Tools for detecting possible cyber-bullying could be used to notify parents or school counselors if a child begins to post very angry or strange messages on line. Inkpen admitted that the tools developed could be harmful if in the wrong hands. However, she was nevertheless confident that there would be many positive potential uses for the tools the group hopes to unveil in 2018.


No comments: