Can Facebook Predict Suicide Risk?

Investors are using social media to predict securities fluctuations. Economists are using them to predict changes in demand. Now  new project will attempt to use algorithmic analysis of social media posts to identify individuals who may attempt to harm themselves.

Just like the internet itself, seed funding will come indirectly from the military as the initial target research group will be US veterans who have had an unusually high propensity to commit suicide since the advent of the Iraq and Afghanistan conflicts, much higher than that of the general population. 

The basis for the selection criterion appears based on solid grounding: 65 percent of veterans who committed suicide repeatedly used key words and phrases. It is hoped that,  eventually, such applications may have broader utility, but given the urgency of the need and the possibility of successful identification and poential intervention, this may be a crucial app in every sense of the word. JL

Marshall Honorof comments in Tech New Daily via Mashable:

If you've been thinking about killing yourself, your social media might give you away. An initiative called the Durkheim Project will use artificial intelligence (AI) algorithms to identify common words and phrases among those who might be contemplating suicide.
The program, which launched on July 2, currently targets only veterans, who have disproportionately high suicide rates. Veterans opt into the Durkheim Project, which installs an app on computers, iOS and Android devices. These apps keep track of what users post and upload it to a medical database. A medical AI monitors the data in real time, picking out patterns that might lead to self-harm.
The Durkheim Project app monitors content from Facebook, Twitter and LinkedIn. In addition, it stores information from a user's mobile device. A database at the Geisel School of Medicine at Dartmouth University will keep track of users' locations and text messages, and will not share any information with third parties. Additionally, the system will be guarded by a firewall to ward off would-be hackers. "The study we've begun with our research partners will build a rich knowledge base that eventually could enable timely interventions by mental health professionals," said Chris Poulin, principal investigator on the project, in a statement. "Facebook's capability for outreach is unparalleled."
This project has a dark side, however: While future versions of the app may notify professionals if an individual appears to be at risk for harmful behavior, its current version is completely noninterventional. Since veterans commit suicide far more often than the general populace, the Durkheim Project may gain some of its most valuable data by tracking active social media users who go on to kill themselves.

That said, the research rests on solid ground. Poulin and a team of investigators ran the program's first phase in 2011, which examined social media from veterans who were active online. The findings were telling: more than 65 percent of users who went on to commit suicide employed key words or phrases on a regular basis on their social media accounts.
The Durkheim Project may not achieve its long-term goal: The program requires users to opt-in, and those who feel suicidal may not feel inclined to reach out for help. Additionally, the original study only tracked correlation: There's no indication that veterans who post negative statuses necessarily go on to kill themselves.
Even so, a project hoping to reduce suicide among veterans is a noble goal, and the Durkheim Project welcomes anyone who wants to help. Through cooperation among mental health professionals and technology experts, a veteran's social media page could be much more than a collection of sad statuses leading to his or her untimely death.