How Much Does It Hurt?: Artificial Intelligence and Pain Detection
Photo Source: Public Domain Pictures
Everyone experiences pain. The presence or absence of pain, in part, defines the human condition. It is essential because pain signals to the mind that something bad is happening to the body. Pain can indicate an injury or that damage may occur unless you stop what you are doing. In medicine, pain management constitutes an integral part of treating patients. Optimal pain management plays a key role in reducing suffering, speeding the healing process, and improving rehabilitation. Studies like the one in the Journal of Anesthesia & Analgesia titled “Changes in mortality after total hip and knee arthroplasty over a ten-year period” showed post-operative outcomes in knee and hip replacement significantly improve with better pain management. (JAA) However, physicians do not have a simple, objective instrument to determine a patient’s pain level unlike blood pressure, body temperature, or weight. Doctors must ask people to describe and assess their pain to get a sense of its severity. Thus, pain evaluation is subjective and likely to vary from person to person.
To bring a semblance of order to pain assessment, doctors frequently use the Numerical Pain Scale or NPS. NPS scores pain on a scale from zero to ten, with 0 as “no pain” to 5 being moderate up to 10 as “worst pain possible.” Although the scale has clear numerical markers for pain states, it depends on the patient’s understanding of all the pain levels and judging their pain against the worst pain possible. Additionally, different people perceive pain differently. For example, a burnt finger feels tolerable to one person (a 3 or 4), but to another, it could feel severe (a 7 or 8). Moreover, when a parent or a doctor has a young child who cannot understand the scale, the caregivers must determine the pain level of their infant or small child—the same holds for those people caring for the impaired, such as dementia patients.
Over the past ten years, the development of artificial intelligence has opened the door to new tools that can score pain levels based on subtle physical changes in the patient. For example, human faces make very brief changes called micro-expressions that reflect emotional states and the sensation of pain. Unfortunately, facial micro-expressions happen as fast a 1/30th of a second—too fast for the human eye. Still, artificial intelligence can catch these expressions and learn to connect them to different levels of pain in the patient. Additionally, research has shown that changes in sweat production correlate with varying thresholds of pain. Therefore, a technique called electrodermal activity measurement can monitor the sweat production in the skin.
Changes in micro-expressions and perspiration coincide with different pain levels, and cameras and sensors can easily detect these changes. Artificial intelligence can rapidly analyze these information streams to provide doctors and caregivers objective tools to assess pain states. For example, research published in IEEE Transactions in Biomedical Engineering titled “Automated Pain Assessment in Children using Electrodermal Activity and Video Data Fusion via Machine Learning” demonstrates an artificial intelligence approach to measuring pain. In the paper, the authors show that with artificial intelligence, they can tell the difference between clinical and non-clinical pain in children with over 90% accuracy. (IEEE) Such a capability cannot be understated. Studies in clinical settings have shown that doctors often underestimate pain in children and infants. A recent study involving emergency room doctors titled “A comparison of pain assessment by physicians, parents, and children in an outpatient setting” published in the Emergency Medicine Journal demonstrated that physicians significantly underestimated pain in children under three. In another category, the doctors provided pain medication to only 42% of children in severe pain. (EMJ)
Pain affects people throughout their lives, from acute injuries to chronic inflammation and nerve injury. From a doctor’s, caregiver’s, and parent’s perspective, understanding the severity of pain someone feels remains a subjective evaluation on a scale from 0-10. Such imprecision makes it difficult to control pain accurately, and for those too young or incapacitated, the estimation of pain becomes even more difficult. Studies have shown that doctors consistently underestimate pain in children, especially those below three years old. Research in artificial intelligence and pain perception has opened the door to new, more objective ways to measure pain that use facial micro-expressions and perspiration. With the help of video and sensors, computers can see indications of pain that a person cannot, opening up a promising horizon for better pain management for reducing suffering, and helping people heal faster.
Dr. Smith’s career in scientific and information research spans the areas of bioinformatics, artificial intelligence, toxicology, and chemistry. He has published a number of peer-reviewed scientific papers. He has worked over the past seventeen years developing advanced analytics, machine learning, and knowledge management tools to enable research and support high level decision making. Tim completed his Ph.D. in Toxicology at Cornell University and a Bachelor of Science in chemistry from the University of Washington.
You can buy his book on Amazon in paperback and in kindle format here.