The technology could potentially be used to help more people get treatment for depression.
Researchers at MIT have developed an artificial intelligence system that can identify depression simply from listening to people talk or by monitoring their texts.
The technology, which uses a neural-network model, can listen or read natural conversations in order to identify speech and communication patterns that indicate depression.
“The first hints we have that a person is happy, excited, sad, or has some serious cognitive condition, such as depression, is through their speech,” Tuka Alhanai, first author of the paper outlining the technology, told MIT News.
Doctors diagnose depression by asking their patients questions and listening to their responses. Machines have been hailed as a way to improve diagnostics in recent years.
However, many of the existing systems require a person to answer specific questions and then make a diagnosis based on the answers that a person provides. “But that’s not how natural conversations work,” said Alhanai, a researcher at the Computer Science and Artificial Intelligence Laboratory (CSAIL).
The new system can be used in more situations because it monitors natural conversations.
“We call it ‘context-free’ because you’re not putting any constraints into the types of questions you’re looking for and the type of responses to those questions,” Alhanai says. “If you want to deploy [depression-detection] models in a scalable way… you want to minimize the amount of constraints you have on the data you’re using. You want to deploy it in any regular conversation and have the model pick up, from the natural interaction, the state of the individual.”
The new model works by analyzing speech and text from people who were depressed and those who were not. It then identified patterns in each group. For example, people with depression might speak more slowly or take longer pauses between words. In text messages they might use words like “low,” “sad” or “down” more commonly.
“The model sees sequences of words or speaking style, and determines that these patterns are more likely to be seen in people who are depressed or not depressed,” Alhanai said. “Then, if it sees the same sequences in new subjects, it can predict if they’re depressed too.”
The technology could potentially be used to help more people get treatment for depression. Although the condition is very common, 37% of people with depression do not receive any treatment.
Alhanai’s team said their technology could be used to develop apps that monitor a person’s conversations and send alerts when their mental health might be deteriorating. It could also be used in a traditional counseling or medical setting to assist medical professionals.
