Man vs. Computer: Which Can Best Spot Pain Fakers?
(SAN DIEGO) -- Researchers at the University of California, San Diego, are working on a breakthrough that could change how doctors treat patients and their pain.
Many doctors’ offices have started displaying charts with faces showing various levels of pain, but what if a person is faking it?
The UCSD team conducted a simple test, pitting man versus machine.
They had human observers watch videos of students experiencing pain and faking it and then asked them to guess which person was faking and who wasn’t. Then they ran the videos through an experimental computer vision system.
“The system did pattern recognition and we tried to figure out if these computer and the pattern recognition system could pick up on those differences between real and faked pain and could they do it differently or better than the human observers could,” said Marian Bartlett, a lead author on the study and from UCSD’s Institute for Neural Computation.
The result: The computer system performed a lot better. While humans were right only half the time -- that number reached 55 percent with a little training -- the computer had a success rate of 85 percent.
“Pretty much the humans were much like guessing. And that’s been known for a long time that human observers are not very good, for the most part, at telling many kinds of deception, whether it’s verbal or nonverbal kinds of deception,” Bartlett said. “Human judges really have trouble telling the difference between the two.”
According to Bartlett, when it comes to real pain, muscle movement is much more random, variable and fleeting. She said the fakers in the experiment opened and closed their mouths too consistently.
The computer is much better at detecting and deciphering the patterns of pain than humans are.
Bartlett said her team was evaluating whether a system like theirs could be used in a clinical setting.
“We’re working with a local children’s hospital in San Diego to develop this and evaluate that, in order to see if, for example, with kids can we do better at estimating their pain intensity levels so that their pain can be better managed,” she said.
Researchers from Ohio State University have computers looking at faces and recognizing as many as 21 different expressions.
The goal is that one day this technology might one day replace the polygraph as a lie detector and perhaps be used at airport security.
Copyright 2014 ABC News Radio