Identification of high risk depression patient via SNA method

             Xin Li-MBA-2016

Problem/Challenge

Depression is a very common mental disorder.  According to WHO, there are approximate 350 million people that are diagnosed as depression covering all age groups.  Depression is also one of the major reason that people get disability. Furthermore, depression can even cause patient to suicide. 

Even though there are some effective therapy for depression patients, less than 50% of depression patients get effective treatment globally.  In many countries, only less than 10% of patients get effective treatment.  One of main reasons is that identifying people in high risk of depression is very difficult.  Therefore, finding a method to identify high risk depression patient has significant meaning for public health and human wellness.

Social network analysis method and development of social media with big data provides us a way to resolve this issue. 

What data do I need? Easy/hard to get?

Data needed:	Easy /Hard to get?	Reason why I need this data:	How will I deal with data? Coding.
Name	Easy	Name is important biometry information to identify people	Recorded as biometry information.
Gender	Easy	Woman has more possibility to get depression.	Weigh more for female
Job title	Easy	Some high pressure industry may has high risk that lead to depression.	High risk:  3 Mid  risk: 2 Low risk:  1 No risk:   0
Living country	Easy	People from some country in cold area has high risk like ones in Russia, North Europe.  And also there some political and nationality reason that cause high risk of depression.	Weight for high risk countries.
Historical post	Easy	High risk people like to post their own angry or over-excited feeling in social media.	Filter key word and rate them as weight.
On-line interval	Easy	The possibility to get depression is higher for people who stay long on-line than the one for people who stay shorter on-line. On the other hand, longer on-line time means less sleeping time that is potential symptom of depression.	> 12h    5 8-12h    4 6-8h    3 4-6h    2 2-4h    1 0-2h    0
Response time and frequency for reply and like/unlike	Easy	The faster and the more frequently response, like and unlike,  the more possibility to get depression	Frequency=FR Response time=RT Higher FR/RT means higher risks. FR/RT percentile:   99%        3 97%        2 68%        1
Family connections	Easy and hard	Genetic factors: If subject has family connections who are identified as depression patients, the subject has higher risk.  The connections information are easy to get, but it’s hard to get depression patients’ information.	SNA method
Employment status	Easy	Unemployment subject has higher possibilities of depression.	Yes 0                 No 1
If Heart disease?	Hard	Heart disease patient has higher possibilities of depression.	Yes    1 No    0
History of losing family or horrible accident	Hard	The event of losing family or horrible accident can cause depression disease.	Yes  1 No  0

What will be the most important network measures? What will the SNA help me do?

Once I calculated all the weight, I will combine the weight information for each subject I got for first round analysis. I can label high weighted subject, for example bigger size, and analyze the social network via SNA method.

If I can find some subjects with bigger weight who have family connections with some confirmed depression patients, the subjects might have high risk.    

Next steps are to further investigate each risk factors of the selected subjects.  Based on the attributes of subjects, check his/her historical posts to further evaluate.

With healthcare professional’s help, we can identify the high risk subject more scientifically. And then provide interception treatment proactively.

Let’s make the world happier!