This Algorithm Can Tell How Much Pain You’re In
Doctors may soon measure pain with an app
/https://tf-cmsv2-smithsonianmag-media.s3.amazonaws.com/filer/af/36/af36c84a-2b95-49e2-8044-b8e36af87683/istock-622444078.jpg)
MIT scientists have developed an algorithm that can learn to recognize miniscule tics and expressions on the human face to quantify how much pain that person is experiencing, reports Matt Reynolds for New Scientist. The algorithm could help with what's often a tricky task for doctors: gauging how a person is feeling, and whether they're exaggerating or minimizing that pain.
Currently, pain is reported by ranking scales using numbers or pictures, reports Luke Dormehl for Digital Trends. Yet these "visual analog scales" for pain can be imprecise, and difficult to use for people who have trouble translating symbols or faces and matching that meaning to their own experiences. One particular group that struggles using these systems are people with autism, Rose Eveleth reported for The Atlantic in 2015. And no self-reported scale can get around somebody trying to fake pain to get a drug prescription.
Computers could be one answer to these problems. Researchers have previously trained artificial neural networks to learn to pick up cues that people often make when in pain far more accurately than other humans. But these programs work on a single scale for all humans, even though pain is different in different people and depends on lots of factors, Christianna Reedy and Karla Lant write for Futurism.
Researchers are trying to fix this problem. A new program, described last month in the Journal of Machine Learning Research, moves beyond this one-size-fits-all approach with a pain algorithm that can be fine-tuned based on a person's demographics, facial features and other factors that affect how pain appears on their face. The program, called DeepFaceLIFT, was trained using videos of people with shoulder injuries who were asked to move their injured arm, writes Reynolds.
DeepFaceLIFT isn't intended to replace traditional pain reporting, but rather augment it, reports Reynolds. Its creators hope to develop it into a mobile app that could be used by doctors with smartphones while interviewing patients. Perhaps in the future, the doctor will know even if you are trying to put on a brave face.