Researchers from MIT have created an algorithm that can read a person's face and gauge the level of pain they are in. The technology is intended to help doctors get a better understanding of the amount of pain their patients are going through.
People experience pain differently and doctors estimate this pain in different ways, says Dianbo Liu, who headed the team that developed this program. That is one of the reasons why a person's self-reported level of pain might be different from what the doctor gauges, he added.
This system can reportedly take into account a person's age, sex, skin tone, and personalise each patient for higher levels of accuracy. Researchers have said that this system is not a one-size- fits- all approach to pain recognition.
While the program is still in its initial stages, Liu says that it will eventually be possible to port it to smartphones in the form of an app for doctors to use with ease.
Researchers created this technology by "training" the algorithm with videos of people experiencing pain. Each of the videos used people with varying degrees of shoulder pain as the camera captured their pain levels expressed in their faces, reports New Scientist. Participants were told to make different movements that give them varying levels of pain and then rate what they feel.
The algorithm, it is reported, can read subtle differences in people's facial expressions and make estimations based on them.
Liu pointed out that certain movements in the face of the patient like around the nose and mouth, "tended to suggest higher self-reported pain scores," according to the publication.
A study by the University of California, San Diego, was mentioned in the report that found that a computer based algorithm can point out people who are faking pain much better than even trained humans. It said that humans were right only 55% of the time when compared to the 85% accuracy rate that computers have.
Liu, however, insists that it is not a replacement for real doctors as the algorithm was trained using videos captured in ideal lighting and camera conditions, so it might not be as accurate when applied to real patients just yet.
Jeffrey Cohn, professor of psychology at the University of Pittsburgh said that these metrics might be useful in "determining real pain from faked pain," and that it could be the difference between spotting a faker and a patient who really needs pain medication.
Liu noted that objectively measuring pain levels is a tricky task and doctors have to make such estimates on a daily basis. Prescription medication and their overuse is a major issue in the American healthcare system with the government actually telling their doctors to stop writing flurry of prescriptions for addictive, deadly pain pills.
Last year, a study made by the National Safety Council found that 99% of doctors prescribe highly addictive opioid medication for periods longer than what the Centers for Disease Control and Prevention recommends. The report also mentioned how 23% of doctors write up one-month doses and that only 30-day usages can cause changes in the brain.
So, a more objective, computer based facial recognition program could potentially reduce instances where fakers are prescribed painkillers by doctors.