During a conversation humans tends to hide their depression in their words but if listened carefully this grasp may interfere in their voice thus making it possible to be to be depressed, some humans are good at hiding their depression in their voice thus making it hard to be noticed by their acquaintances.
Now, due to this MIT researchers have come up with an algorithm that can detect if the friend is depressed or not, depression is one of the most widely suffered and undiagnosed conditions in the U.S. The reason why it matters for depression to be detect it’s because, according to research about 1 in 15 adults of 37 million American experience major depressive episodes but often go untreated.
The background of Al can diagnose dozens of diseases as accurately as a human expert as from images of tissue samples, also Al can detect breast cancer better than human expert can do, and this is achieved when a 3D eye scan is given.
Lately ,a Ph.D. candidate at MIT focused on language understanding trained an AI system using more than 140 recorded conversations to assess whether a person is depressed and if so how severely and this was the work of Tuka Alhanai.
In order to show how it works Alhanai used a neural network to find characteristics of speech like variation of voice and the breathing pace, then she employed another algorithm to surface patterns that mostly likely to depression.
This system was more accurate when it considered responses in context as Tuka Alhanai says. Thus, due to this she ensured to use a neural network that can find patterns across answers rather than to analyze each in isolation. The best performing system was classified at 83% of the test cases were correct.
The advantages of this AI system are beneficial;
Some people who are depressed may find it difficult to visit professionals but with this AI system one can notice the depression and advise them to see a professional in this depressive matter and one being treated. Also the AI can detect depression that would be definitely missed by human.