AI from MIT Diagnoses Depression
This is how a therapy consultation typically goes: a patient comes in, the doctor asks them invasive questions over an hour or so, and then they decide if they'd be a good fit for each other, what goals they'd like to achieve with therapy, and other things that are important to the treatment of mental illness.
But with the AI from MIT, there might be a chance that they could avoid some, if not most, of the invasive questioning- questions that ask about family history, personal history, and questions about other topics that might overwhelm a patient on their first time attending.
You, sir, are a depressed bitch
The AI scans natural speech and tone in a casual and non-intimidating setting, so that the AI could carry a conversation about something that has nothing to do with mental illness and still detect that there's something wrong with the patient based on the audio input that the AI receives.
Apparently, depressed individuals have flatter, more monotone voices, and there are certain words that keep naturally popping up on their everyday vocabulary, such as 'down' and 'sad'.
This could be the stepping stone for other AI's to help and diagnose patients who are unable to attend therapy due to financial, personal, or locational reasons. For young people with parents who don't understand a lot about mental illness or those who simply don't know there's a problem, this could be a great aid because it'll tell the clinician about the red flags they've detected that the clinician might have missed.
Better diagnosing might mean a less tumultuous treatment and recovery for patients with depression and other mental disorders.
Medication time, hoooooman
They say that their next focus might be on identifying patients with dementia and similar disorders, which although different, is a similar concept in which they will analyze patterns in speech and communication to properly say if someone is cognitively impaired or not.
WRITTEN BY: RAE DIZON
No comments: