The dream programme for psychologists: detects depressive bloggers

Four clinical psychologists have pored through the results given by an automatic depression detection programme that draws on analysis of entries across 300,000 blogs. The machine was 78% right.

Researchers at Ben-Gurion University in Israel have developed a depression detection programme based on the analysis of texts published online. 300,000 blogs served as the basis for the trials. The programme came up with a TOP 100 of the most depressive authors, then a second TOP 100 of the least depressive (decidedly online depression is all the rage today on Digital Versus).

The reults were submitted to four clinical psychologists to test the machine analysis. Verdict: the computer was right in 78% of cases.

Machines no replacement for humans (yet)!

Professor Yair Neuman will present the study in detail in Toronto at the beginning of September. He has already given an initial explanation in a paper entitled "Are Your Texts Depressed? The Computer Knows, Maybe" saying that the program is designed to detect depressive content contained in language that isn't obviously depressive (ie. not words such as "depression", "suicide", "black"). He does add however that the programme won't be a substitute for human judgement.

The programme has been designed to provide a screening process and flag up those vulnerable to suicide, who don't perhaps realise themselves that they are depressed. Possible applications include warnings for parents of abnormal behaviour in their children through social networking sites, blogs and so on.


> Buyer's Guides: Our Pick of The Best Products