Psychiatry is the field of medicine concerned with mental health and encompasses everything from anxiety to paranoid schizophrenia. It is often criticised for lacking in objective measures with which to diagnose these conditions; you can’t just do a blood test and diagnose someone with OCD. For this reason, psychiatry has the largest sections in the diagnostic manuals and psychiatrists have to spend a lot longer with each of their patients in order to get the best understanding of what is going on.
And diagnosis isn’t the only problem. Predicting the course of some of these conditions which are notoriously tumultuous can be nigh on impossible - leaving both the doctors and the people with these conditions to deal with problems after they arise.
What if there were a way to predict the course of, say, schizophrenia? Surely that would be useful. Any means of predicting problems before they arise would certainly stand you in good stead for dealing with them. Well that’s what Mariano Sigman, a neuroscientist, and his partner Guillermo Cecchi set out to do, seeing if the words people use could be an indicator of future mental health.
But how do you objectively analyse words to meaningfully get a prediction? Sigman and Cecchi had a cunning solution that had been used in other fields before. They used what Sigman describes as a ‘Space of Words’. This Space of Words is a hypothetical space in which words are organised according to their relationship with other words. Words that are known to be used together often, and thus are likely to be more connected to each other, are grouped closer together and words infrequently used together are separated by a distance. When developing the Space of Words he noticed it was very easy to delineate areas within this space into their own ‘semantic neighbourhoods’. Words such as ‘sun’, ‘star’ and ‘planet’ would be in their own semantic neighbourhood relating to astronomy, but other words like ‘proton’ and ‘molecule’ would be close by but in their own semantic neighbourhood relating more to physics.
Sigman and Cecchi then took 34 young people at a high risk of developing schizophrenia and recorded them speaking. They looked at the words that were being used by these young people, the topics they were talking about, and looked at whether this could be predictive of developing psychosis within the next three years. Alas, it could not.
The things people say are only part of the picture though. How people put words together and how people link topics can give great insight into how a person’s mind is working. In fact, this is a technique psychiatrists already use to gauge someone’s current mental state.