The researchers at MIT have now come up with an AI that can tell whether a person is depressed or not, it detects the words and intonations that offer clues about a person being depressed or not.
How it works?
According to a paper published by MIT says, "A neural network model has been developed by the scientists that can be applied to raw text and audio from interviews of patients to discover speech patterns indicative of depression.
Questions about lifestyle, mood, past mental illness and a few methods that help professionals tackle depression in patients. They identify the patient's condition based on his response to these methods. Tuka Alhanai (a researcher in the Computer Science and Artificial Intelligence Laboratory says, "the first hints we have that a person is happy, sad, excited or has some serious cognitive condition such as depression is through their speech." The model is able to tell if a person is depressed or not without needing any other information about the questions and answers and that too accurately. Alhanai continued by saying, "if you want to deploy depression detection models in scalable way, you want to minimise the amount of constraints you have on the data you are using. You want to deploy it in any regular conversation and have the model pick up from the natural interaction."
The model is able to detect patterns typical of depression; it then maps those patterns to a new person without adding any additional information.
How was it applied?
The model was applied to a data set of 142 interactions from the Distress Analysis Interview Corpus that contains audio, text and video interviews of the people with mental health issues and virtual agents controlled by humans. Each patient is rated in terms of depression in a scale between 0-27. A score of 10-14 is considered a moderate and 15-19 are considered moderate severe case of depression. Anyone below that rating is a non depressed patient. According to the database around 20 percent of the patients were depressed.
The researchers hope this method can be used to develop tools that are able to detect signs of depression during a natural conversation. For future instances the model can be used in mobile apps that could monitor a user's text and voice for mental distress and send alerts. This method can be of great help for the people who cannot get in touch with the professionals for the diagnosis because of distance, money or lack of knowledge about something being wrong.