If it's as accurate as the team says it is, this is pretty astounding IMO. They'd have to be capturing some sort of baseline in the beginning of the conversation, and recording relative shifts of variables (ie; smile, eye shift/location) to come up with how "depressed" the person is. Just writing some software that says "oh this person is looking down" probably wouldn't cut it-- because shy people look down too. I noticed in the demo it seemed as if it started monitoring these things after certain questions were asked, which would be the correct way to measure a shift from baseline. Elicit a predictable period of what could be change and record.
Comments
If it's as accurate as the team says it is, this is pretty astounding IMO. They'd have to be capturing some sort of baseline in the beginning of the conversation, and recording relative shifts of variables (ie; smile, eye shift/location) to come up with how "depressed" the person is. Just writing some software that says "oh this person is looking down" probably wouldn't cut it-- because shy people look down too. I noticed in the demo it seemed as if it started monitoring these things after certain questions were asked, which would be the correct way to measure a shift from baseline. Elicit a predictable period of what could be change and record.