Microsoft study finds Twitter posts can predict post-natal depression
A Microsoft study investigating the linguistic quirks of Twitter users has found that subtle shifts in language used by expectant mothers can predict their subsequent likelihood of developing post-natal depression.
A computer program devised for the tech firm was used to distillate musings from the micro blog; identifying general signs of increased negativity through use of words such as hate, miserable and disappointed.
The computer code shrink was also tasked with picking out increased usage of the first person pronoun ‘I’ and expletives on the basis that such vocabulary has been long-associated with the condition by psychiatrists.
Researchers first searched for patterns in the language used by women suffering from post-natal depression and then re-wound the clock to posts made in the weeks prior to birth to see if there were any signs of changing mood.
Eric Horvits, co-director of Microsoft research, said: “We found that two to three weeks before the birth, the same clues were there in about 80 per cent of cases. It’s very subtle. It’s things like an increase in the word ‘frustrated’ and, for example, they curse more often.”
It is hoped that an app with such software built-in could enable new mothers to seek help earlier.