About the Project
Researchers are collaborating across species lines to investigate the use of facial expression for pain detection in animals. The underlying musculature that produces facial expression in mammals is relatively well preserved, and Facial Action Coding System (FACS) offers a new perspective for detecting and measuring pain in animals. Thankfully, humans can communicate where they feel pain, although referred pain is a relatively common phenomenon. Facial expression of true pain is involuntary, in that it cannot be masked. “Fake” pain expression involves voluntary movement of the face in humans, which is discernable to the trained eye. In the 1970s, the human psychologist Paul Ekman, PhD began to study facial expression in humans and developed the Facial Action Coding System, or FACS. This system scored the face based on the muscles engaged in each expression and was performed from still images or short videos of faces by human coding experts. This type of objective research paved the way for more advanced understanding of facial expression of pain and emotion in humans.
In pediatric medicine, this understanding of voluntary vs involuntary use of facial musculature to depict real vs fake pain, became useful clinically. Researchers at UCSD pioneered some of the early work to train a computer to detect facial expression in children, and measure expression changes in pixilated detail. In studies that ensued, the computer was more sensitive than most nurses or doctors in detecting subtle changes in expression indicative of pain. Digitized evaluation of images served as microscope which exceeded the speed and capability of the human eye.
Veterinary medicine is much like pediatric medicine in that animals cannot verbally express the source or degree of pain they are experiencing. To complicate matters, prey animals such as livestock and horses hide pain, as pain signals vulnerability to would be predators. Researchers across species lines have observed that facial expression of pain often emerges before other symptoms, such as elevated HR, or changes in feeding behavior. Thus earlier recognition and treatment of pain are possible if knowledge of species specific facial expression of pain patterns are elucidated. As with human medicine, earlier intervention in disease is often associated with improved outcomes.
Animals, like people are living longer as veterinary medicine advances. The Human Animal Bond has been shown to benefit the well-being of humans in several different ways and an estimated 195 million Americans own pets. The pet industry contributes billions of dollars to the annual economy and there is a growing need to keep geriatric pets, and working animals such as horses comfortable and ambulatory. In comes citizen science. In this day and age, all one has to do is visit Facebook to visualize the global impact of animals on human emotion and compassion.
In order to properly “train a computer” to recognize pain or no pain on an animal image, one first has to train the computer to recognize the animal in a frame. This is difficult, as there is tremendous individual variation within and across the mammalian species. Consider ear, eye and nostril shape within the dog species for example. Maheen Rashid has successfully trained a computer to recognize the equine face. Now the task at hand is to run thousands of images past the computer that are accurately annotated as pain or no pain. Computer recognition of pain will need to be validated, and citizen science is likely the road to success.
A validated scoring system for Facial expression of pain is likely to significantly enhance current pain research. A recent retrospective of existing pain scale and scoring systems reveals significant inadequacies in differentiating acute and chronic pain, and pain coming from different body systems. In horses for example, scales developed to assess orthopedic pain are not applicable to abdominal pain. Physically obtaining vitals such as HR or blood pressure affects pain expression in horses and stall side assessment often significantly underestimates true pain in prey species. Brit Coles and her team have provided evidence that remote monitoring of post -surgical equine patients results in significantly higher pain scores and different treatment decisions than “in stall” assessment.
Accurate Pain scoring procedures for livestock are truly lacking, which stalls the development and approval of pain medications in those species. The burden of proof that a medication is working is difficult to achieve when measurements are inefficient and unrepresentative of grades of pain. Earlier detection of infection or injury in food producing animals, might lead to less medication use in those species. There is currently an initiative in the United States to dramatically reduce the amount of antibiotic use in food producing animals, to minimize the development of antibiotic resistance in humans.
It is clear that pain detection and management across species lines is due for a major overhaul. This new meld of technology and observation may hold the key to improving welfare on a grand scale.
Every day, 115 US citizens die of opiod associated overdose.
We need better options for treating pain in humans and animals. In order to research diagnostic and therapeutic options for pain, we need to be able to objectively assess pain, score it, and monitor for improvement. Recently, Dr. Jamie Peyton DVM was awarded a Big Idea Award for her work with human pain specialists on the concept of an Integrated Pain Center; with the focus of uniting human and veterinary medical teams to research new options for the alleviation of pain. This project is in its initial stages at UC Davis, and we expect the Center to elevate options for the management of pain across species lines.