MIT (Massachusetts Institute of Technology) has been working on the ability to identify depression or anxiety within somebody’s speech patterns using deep learning. They are referring to it as “talk diagnosis”, and are hoping that it could help to diagnose mental health issues, allowing these situations to be treated quickly.
By doing this, it may be less necessary for patients to have to visit a doctor and undergo the series of questions doctors use to draw conclusions on various illnesses. Instead, your own computer might be able to draw this conclusion based on your everyday conversations instead.
The study is made possible thanks to a computer farm packed full of NVIDIA (Titan X) GPUs, which work together to analyze given data and develop a model that can be used for detecting the various changes in a person’s conversation.
“There is significant signal in the data that will cue you to whether people have depression,” – Tuka Alhanai, (MIT research assistant and Ph.D. candidate in computer science). “You listen to overall conversation and absorb the trajectory of the conversation and speech, and the larger context in which things are said.”
The computer monitors conversations looking for these cues, and thus far they have seen a 70% success rate in detecting depression during their trials, using their finished model. They currently seek faster solutions as technology progresses in the near future, to better fine tune these results with, as well as improving the technology to also detect such signs within body language or writing.
At this rate, it won’t be long before they have a working solution, hopefully in the upper 90’s percentage-wise, allowing computers to help predict warning signs well ahead of time. The technology could also be used to monitor current patients to follow them throughout their treatment.
This could be a great technology to help combat depression and prevent it from having long-term effects on a person’s body and mind. As long as it doesn’t lead to a future similar to the movie Idiocracy, where a Carl’s Jr vending machine decides that you are an unfit parent based on your interaction with ordering your meal, or an excuse to simply medicare as many people as possible for the purposes of profit (but that’s a completely new topic in itself we won’t get into the politics of).
