Rutgers University uses AI to measure pain through facial micromovements and heart rhythms

Tracking Facial Micromovements to Measure Pain - Neuroscience News

Summary: A pioneering computational neuroscience study identified a precise, data-driven method to quantify human pain by tracking invisible, high-speed facial micromovement spikes. The research challenges the traditional, highly subjective 1-to-10 scale by using artificial intelligence (AI) and video analysis to decode tiny motor fluctuations that escape the human eye. The team demonstrated a direct neurological link between these facial micro-spikes and heart rate variability during episodes of physical distress. This physiological "window" offers an objective diagnostic tool for patients who cannot verbally articulate their suffering, including young children, stroke survivors, and individuals living with dementia. Researchers at Rutgers University-New Brunswick are working to measure pain more accurately beyond the single, subjective question patients are often asked: "On a scale of 1 to 10, how bad is your pain?" In their new study, published in Frontiers in Neuroscience, the researchers suggest a more precise way to quantify this discomfort by tracking tiny facial micromovement spikes. These rapid, high-speed motor fluctuations -- too subtle for the human eye to notice -- offer objective clues to what an individual is experiencing, particularly when they cannot articulate their level of distress. "The motivation was to move beyond a one-size-fits-all pain scale," said Elizabeth Torres, a psychology professor with the Rutgers School of Arts and Sciences who conducted the study with doctoral researcher Mona Elsayed. "Every individual has a different threshold for pain tolerance. By measuring that response directly from the body's own signals, we can begin to tailor care in a much more individualized way." To test whether facial movements could reveal pain-related signals, Torres and Elsayed recorded 45 adults before and during episodes of controlled, brief pressure pain. Participants were observed at rest and while performing tasks involving movement, touch and memory. Using video analysis and artificial intelligence (AI), the team tracked facial muscle activity alongside heart rate variability -- a measure of the timing between heartbeats. This revealed a direct link between micromovement spikes and the body's physiological response: As pain intensified, heart rhythms became increasingly irregular, with the most pronounced changes appearing around the eyes. "Within seconds, we could see the body's pain response reflected in tiny facial movements," Torres said. "The more dysregulated the heart became, the more clearly it showed up in the face." The researchers also found that different activities changed how pain appeared in the data. Pain registered most clearly during tactile tasks, such as drawing or manipulating objects, when the link between facial movements and heart rhythm was strongest. In contrast, tasks requiring memory or attention weakened that connection. "A higher cognitive load essentially crowds out the pain," Torres said. "This kind of engagement may act as a natural distractor, offering a potential therapeutic tool for redirecting attention." The pain study emerged from a broader line of research in Torres' Sensory Motor Integration Lab, which has long studied micromovements in people with autism, Parkinson's disease and other neurological conditions. Torres, a computational neuroscientist, uses mathematical modeling to decode internal states through subtle body language. In her studies of nonverbal autism, these patterns provided vital clues to physical distress that clinicians and caregivers might otherwise miss. By applying this approach to facial movements and heart rhythms, Torres suggests clinicians can objectively evaluate pain in patients who are not able to describe their symptoms including young children, stroke survivors, and individuals with dementia. "Right now, we rely on caregivers' interpretations, which are valuable but incomplete," Torres said. "This gives us a window into the physiology itself." Monitoring these signals requires pairing facial videos with specialized heart monitors. But Torres said widely available tools, such as smartphones, eventually could capture this data. Advances in video analysis and AI now enable the detection of physical markers that previously required specialized equipment, which could make pain assessment easier to scale in clinics, nursing homes or remote settings. The research remains in its early stages. Torres said the study's size was modest but with significant statistical power, given the high sensitivity of the personalized micro-movements' metrics. The next step is to test the approach in larger, more diverse populations, including patients with chronic pain. Torres and her collaborators are also translating the technology into a smartphone tool through Neuroinversa LLC, a Rutgers-New Brunswick spinoff startup company that licensed the technology from Rutgers. Although the app is still in development, Torres said it eventually could help clinicians and individuals monitor treatment response. "You can see whether a medication is working, how quickly it's taking effect, and whether adjustments are needed," Torres said. "It's a much more precise way to monitor outcomes." Torres said the simplicity of a short facial scan is what could eventually make the approach useful beyond specialized research settings. "Instead of a piece of paper with emojis, you have a digital dashboard where you can measure yourself day to day," she said. "It gives people a sense of control over their own biorhythms." Author: Megan Schumann Source: Rutgers University Contact: Megan Schumann - Rutgers University Image: The image is credited to Neuroscience News Original Research: Open access. "Facial micro-movements as a proxy of increasingly erratic heart rate variability while experiencing pressure pain" by Elizabeth B. Torres and Mona Elsayed. Frontiers in Neuroscience DOI:10.3389/fnins.2026.1702124 Abstract Facial micro-movements as a proxy of increasingly erratic heart rate variability while experiencing pressure pain Introduction: The sensation of pain varies from person to person. These patterns of individual variation are difficult to capture using coarse subjective self-reports. However, they are important when prescribing therapies and tailoring them to each person's own sensations. Pain can be experienced differently by the same person and can fluctuate based on context; yet, most analyses treat the problem with a one-size-fits-all model. Methods: In this work, we introduce a series of assays to assess pressure pain across tasks with different motoric and cognitive demands, in relation to a resting state. In a cohort of healthy individuals, we examine pain-free vs. pain states at rest, during drawing with heavy cognitive demands, during pointing to a visual target, and during a grooved peg task, such as inserting a grooved key into a matching keyhole. We adopt a standardized data type called micro-movement spikes (MMS) to characterize the biorhythmic activities of facial micro-expressions and the micro-fluctuations in the heart's inter-beat interval timings. Results: Using the MMS peaks, we find that the continuous Gamma family of probability distribution functions best fits the frequency histograms of both the facial and heart data. Furthermore, we find that the Gamma shape and scale parameters in both signals span a scaling power law whereby, as the noise-to-signal ratio (Gamma scale parameter) increases, so does the randomness of the stochastic process. We find that as the heart IBI becomes more erratic (noisier and more random), the facial ophthalmic region also increases in noise and randomness, with higher linear correlation for tasks requiring haptic feedback (R 0.84) and lower correlation for tasks requiring greater cognitive and memory loads (R 0.77). Conclusion: Increases in transfer entropy show that recent past activity (~167 ms back) of the heart IBI and facial data combined lower the uncertainty in predicting the present ophthalmic facial activity, suggesting that this facial region may serve as a proxy for the increasingly dysregulated heart. These results have implications for the detection and monitoring of pressure pain.

Facial micromovements may help doctors measure pain more accurately

Rutgers UniversityMay 19 2026 Researchers at Rutgers University-New Brunswick are working to measure pain more accurately beyond the single, subjective question patients are often asked: "On a scale of 1 to 10, how bad is your pain?" In their new study, published in Frontiers in Neuroscience, the researchers suggest a more precise way to quantify this discomfort by tracking tiny facial micromovement spikes. These rapid, high-speed motor fluctuations-too subtle for the human eye to notice-offer objective clues to what an individual is experiencing, particularly when they cannot articulate their level of distress. "The motivation was to move beyond a one-size-fits-all pain scale," said Elizabeth Torres, a psychology professor with the Rutgers School of Arts and Sciences who conducted the study with doctoral researcher Mona Elsayed. "Every individual has a different threshold for pain tolerance. By measuring that response directly from the body's own signals, we can begin to tailor care in a much more individualized way." To test whether facial movements could reveal pain-related signals, Torres and Elsayed recorded 45 adults before and during episodes of controlled, brief pressure pain. Participants were observed at rest and while performing tasks involving movement, touch and memory. Using video analysis and artificial intelligence (AI), the team tracked facial muscle activity alongside heart rate variability-a measure of the timing between heartbeats. This revealed a direct link between micromovement spikes and the body's physiological response: As pain intensified, heart rhythms became increasingly irregular, with the most pronounced changes appearing around the eyes. Within seconds, we could see the body's pain response reflected in tiny facial movements. The more dysregulated the heart became, the more clearly it showed up in the face." Elizabeth Torres, psychology professor, Rutgers School of Arts and Sciences The researchers also found that different activities changed how pain appeared in the data. Pain registered most clearly during tactile tasks, such as drawing or manipulating objects, when the link between facial movements and heart rhythm was strongest. In contrast, tasks requiring memory or attention weakened that connection. "A higher cognitive load essentially crowds out the pain," Torres said. "This kind of engagement may act as a natural distractor, offering a potential therapeutic tool for redirecting attention." The pain study emerged from a broader line of research in Torres' Sensory Motor Integration Lab, which has long studied micromovements in people with autism, Parkinson's disease and other neurological conditions. Torres, a computational neuroscientist, uses mathematical modeling to decode internal states through subtle body language. In her studies of nonverbal autism, these patterns provided vital clues to physical distress that clinicians and caregivers might otherwise miss. By applying this approach to facial movements and heart rhythms, Torres suggests clinicians can objectively evaluate pain in patients who are not able to describe their symptoms including young children, stroke survivors, and individuals with dementia. "Right now, we rely on caregivers' interpretations, which are valuable but incomplete," Torres said. "This gives us a window into the physiology itself." Monitoring these signals requires pairing facial videos with specialized heart monitors. But Torres said widely available tools, such as smartphones, eventually could capture this data. Advances in video analysis and AI now enable the detection of physical markers that previously required specialized equipment, which could make pain assessment easier to scale in clinics, nursing homes or remote settings. The research remains in its early stages. Torres said the study's size was modest but with significant statistical power, given the high sensitivity of the personalized micro-movements' metrics. The next step is to test the approach in larger, more diverse populations, including patients with chronic pain. Torres and her collaborators are also translating the technology into a smartphone tool through Neuroinversa LLC, a Rutgers-New Brunswick spinoff startup company that licensed the technology from Rutgers. Although the app is still in development, Torres said it eventually could help clinicians and individuals monitor treatment response. "You can see whether a medication is working, how quickly it's taking effect, and whether adjustments are needed," Torres said. "It's a much more precise way to monitor outcomes." Torres said the simplicity of a short facial scan is what could eventually make the approach useful beyond specialized research settings. "Instead of a piece of paper with emojis, you have a digital dashboard where you can measure yourself day to day," she said. "It gives people a sense of control over their own biorhythms." Rutgers University Journal reference: Torres, E. B., & Elsayed, M. (2026). Facial micro-movements as a proxy of increasingly erratic heart rate variability while experiencing pressure pain. Frontiers in Neuroscience. DOI: 10.3389/fnins.2026.1702124. https://www.frontiersin.org/journals/neuroscience/articles/10.3389/fnins.2026.1702124/full