Computers help doctors better treat patients with mental illness

To assess a patient’s condition, doctors usually look at vital signs such as heartbeat, blood pressure and rate of breath to establish a baseline.

Dr. Justin Baker explained that when a patient is bi-polar or has schizophrenia, those types of vitals pointing to a baseline don’t exist.

“We have expert judgment of these folks who have trained for, you know, years and years,” Baker said

Subtle body movements and changes in voice patterns are key when treating mental health patients, but Baker said even highly skilled psychiatrists may not pick up on everything.


That’s why for the last two years, his team at McLean Hospital in Massachusetts, along with a group from Carnegie Mellon, have been training computers to help treat mental health patients.

Using a camera and microphone, a computer is able to analyze facial expressions and voice patterns.

“We feel like it’s going to be really important to establish some of these ground truth type measures,” Baker said. “How fast is someone speaking? How much is the tone of their voice changing? How much are they shifting around in their chair? These are just things we take for granted as kind of giving us insight into whether someone is behaving abnormally.”

So far, they’ve looked at the data of a few dozen participants as well as patients tracked over time in the hospital.

Baker told Channel 2 Action News that they are shifting from data-gathering mode to beginning to try to apply what they’ve learned.

“We’re able to pick up on certain sort of social behaviors, like smiling and speaking, that are more consistent with someone having, perhaps, a psychotic disorder like schizophrenia.” Baker said. “Having these kinds of objective metrics to really ground the treatments we’re providing, I think it’s a really key way to work toward cost savings, but also toward providing more reliable care.”