At Rutgers University-New Brunswick, Elizabeth Torres is teaching computers to read pain in the flicker of an eyelid. In a new study published in Frontiers in Neuroscience, the computational neuroscientist and her doctoral researcher Mona Elsayed have discovered that tiny facial micromovements—movements so subtle the human eye cannot detect them—reveal exactly how much pain someone is experiencing, often more reliably than asking them directly.
The research arrives as a quiet revolution in how medicine measures one of the most difficult symptoms to quantify. For decades, clinicians have relied on patients' answers to a simple, one-size-fits-all question: "On a scale of 1 to 10, how bad is your pain?" The problem is that pain tolerance varies dramatically from person to person, and some patients cannot answer at all—young children, stroke survivors, individuals with dementia, or anyone whose condition makes speech difficult. Torres's team wanted to move beyond this limitation by letting the body speak for itself.
The study involved 45 adults who were subjected to controlled, brief pressure pain while researchers filmed them at rest and during various tasks—drawing, touching objects, and performing memory exercises. Using video analysis and artificial intelligence, Torres and Elsayed tracked the tiniest fluctuations in facial muscle activity and paired this data with measurements of heart rate variability, the subtle timing between heartbeats. What they found was striking: as pain intensified, the heart's rhythm became increasingly irregular, and these physiological changes were reflected in rapid micromovement spikes around the eyes and face—changes visible only to algorithmic analysis, not to the human observer.
"Within seconds, we could see the body's pain response reflected in tiny facial movements," Torres explained. "The more dysregulated the heart became, the more clearly it showed up in the face."
The researchers also discovered that context matters. Pain registered most clearly during tactile tasks like drawing or object manipulation, when the link between facial movements and heart rhythm was strongest. But when participants engaged in memory or attention-demanding tasks, that connection weakened. "A higher cognitive load essentially crowds out the pain," Torres noted, suggesting that focused mental engagement may naturally redirect attention away from discomfort—a finding that could inform therapeutic approaches to chronic pain management.
Torres's work emerges from her broader Sensory Motor Integration Lab, which has spent years studying micromovements in people with autism, Parkinson's disease, and other neurological conditions. By decoding internal states through subtle body language, she has identified patterns that clinicians and caregivers might otherwise miss, particularly when it comes to physical distress in nonverbal individuals.
The practical implications are significant. Right now, pain assessment in patients who cannot speak relies on caregiver interpretation—valuable but incomplete. An objective measurement tool could transform care in nursing homes, pediatric clinics, and remote settings. Torres notes that while current work requires specialized heart monitors paired with facial video, advances in AI and smartphone cameras may eventually enable widespread deployment of this technology.
The study represents an early stage in what Torres and collaborators are developing into a smartphone-accessible tool. The next phase will test the approach in larger, more diverse populations, particularly those living with chronic pain. For patients whose pain has gone unheard or misunderstood, this small window into the body's signals offers something rare: the possibility of being truly seen.