Computer Vision System Better than Humans at Detecting Real Pain


There are no accurate scales to determine whether or not a patient is actually feeling pain or how strong their pain is. This is in part due to each person’s varying tolerance to pain and partially because the science on the subject has been very limited. However, researchers at the University of California in San Diego have tried to make determining whether or not a patient is truly in pain more simple by developing a computer vision system. The system has algorithms which determine the slight differences in facial muscle contractions during real and fake pain.

To test the system’s accuracy compared to a trained professional’s ability at determining real pain, subjects were recorded dipping their hand into icy water (real pain), then dipping their hand into a bowl of warm water and feigning an expression of pain. Afterward, the images were fed to the computer vision system as well as shown to trained medical professionals. The trained professionals could only determine true expressions of pain with 55 percent accuracy, while the computer system was accurate 85 percent of the time. The researchers hope that this technology could various uses, including helping ER doctors perform pain assessments.

(Source: ieeeSpectrum)

Posted in 1. Robotics. Comments Off on Computer Vision System Better than Humans at Detecting Real Pain