This solution is provided for the Cure Depression challenge.
The problem description is;
There is no blood test or imaging that can be done to diagnose depression. The diagnosis is based on self-reported symptoms, and there is a lot of overlap between different psychiatric issues in terms of the symptoms people experience.
Once a diagnosis has been made, doctors have the same problem in monitoring effects of treatment -- they see the patient at one point in time and have to judge whether they are better or worse.
Design a way to collect data from existing patients. Design a tool to help physicians make better diagnosis. Help non-academic (community) physicians have access to newest methodology.
Because, the fact that 300 million prescriptions for anti-depressants every year and suicide is 2nd highest cause of death for under 25 age group is a very series problem that requires an immediate intervation
We developed a conceptual frameowrk that work as a solution.
- We develop a tool that monitor user social web activity with aim of extracting depression related twittes or posts.
For this task, first we studied basic depression related activities like worthlessness, and/or helplessness , thoughts of suicide, suicide attempts and ect.
We also studied symptoms of depression like talking about suicide (killing one's self),saying things like "It would be better if I wasn't here" or "I want out" and making comments about being hopeless, helpless, or worthless and etc.
Then we employed opinion mining techniques to classify twittes and posts as depression symptoms.
We analyzed reply twittes and posts with aim of detecting anomalies in social conversation: For example when some relied "I m a worthless dude !" for A happy BD post.
We developed a working mockups to show the processes in solving the problem
Next we plan to develop a protype and carry out an evaluation