Video Analysis of Sound-Induced Pain in Mice

A new machine learning framework can now objectively measure pain in mice caused by loud sound, providing a direct behavioral model for studying a human condition known as pain hyperacusis. Researchers at Johns Hopkins University have developed an automated video analysis method that quantifies pain-related behaviors, a critical step for understanding how normally tolerable sounds can become physically painful. The work, led by Benjamin J. Seicol and colleagues, was published in the journal *eNeuro* (Seicol et al., 2026, PMID: 42595472).

How Machine Learning Decodes Pain from a Mouse’s Face and Posture

The research team’s method relies on two well-established indicators of pain in rodents: the Mouse Grimace Scale and changes in body movement or posture. To measure these objectively, they used a deep neural network—a type of machine learning model—trained to extract specific facial features (like eye squinting or cheek bulging) and body positions from video recordings taken with a single camera. This automated approach removes human bias and allows for the high-speed analysis of large amounts of behavioral data.

To ensure the model was accurately detecting pain, they first tested it on a controlled painful condition. They induced migraine-like pain in the mice by injecting calcitonin gene-related peptide (CGRP), a neuropeptide strongly linked to migraine in humans. The machine learning system successfully quantified a distinct pain response from this injection, differentiating it from the animals’ normal, baseline behavior. This step was essential, as it allowed the researchers to define a concrete, data-driven “pain threshold.”

High-Intensity Sound Crosses the Migraine Pain Threshold

With a validated pain detection tool, the team then exposed freely moving mice to different sound stimuli. The results were clear. When sounds reached high intensities, the mice exhibited significant changes in both facial grimace and body position. Critically, the magnitude of these behavioral changes was not just noticeable; it surpassed the pain threshold established by the CGRP-induced migraine model.

This finding provides strong behavioral evidence that intensely loud sound is not just startling or aversive to mice—it is genuinely painful. “These data revealed that exposure to intense sound triggers significant pain behavioral responses,” the authors state. The study offers a direct animal model for human pain hyperacusis, a condition where everyday sounds cause physical discomfort or pain, often linked to migraine and related disorders.

Deaf Mice Confirm the Sound Must Be Heard to Cause Pain

A vital control experiment confirmed that the observed pain behavior was specifically caused by the auditory experience of the sound, not by other factors like vibration. The researchers tested Tmie-knockout mice, which are functionally deaf due to a lack of sound transduction in the cochlea. When exposed to the same high-intensity sounds, these deaf mice showed no changes in facial grimace or body position. Their behavior did not cross the pain threshold.

This result confirms that the pain response depends on the neural signals generated by the ear. It rules out the possibility that the pain behaviors were a general stress response to the experimental environment or a physical reaction to air pressure changes. The pain is evoked by the sound signal being processed by the auditory system.

A New Foundation for Research on Painful Sound

The practical implication of this work is the creation of a powerful new research tool. This automated, quantitative framework allows scientists to systematically study the biological mechanisms of sound-evoked pain. Future studies can use this model to ask specific questions: Which neural pathways are involved? How does chronic exposure to noise alter pain sensitivity in the brain? Are there genetic factors that increase vulnerability?

Answering these questions could inform the development of treatments for conditions like hyperacusis and severe phonophobia in misophonia. Understanding the shared mechanisms between sound-evoked pain and migraine, as suggested by the CGRP validation, could also open new therapeutic avenues. Furthermore, this model could be used to test the ototoxic potential of very loud sounds, complementing traditional measures of hidden hearing loss that focus on nerve damage rather than pain perception.

The study, “Machine Learning Guided Video Analysis Identifies Sound-Evoked Pain-Related Behaviors from Facial Grimace and Body Cues in Mice”, shifts the study of auditory pain from subjective reporting to objective measurement in an animal model. It provides a clear behavioral benchmark from which to explore the peripheral and central nervous system processes that turn sound into suffering.