Researchers at Penn State say they have developed a stretchable, rechargeable sticker designed to detect genuine emotions by combining facial movement data with physiological signals such as skin temperature and heart rate.
The team argues the approach could help clinicians understand what patients feel in real time, especially when facial expressions alone are misleading or emotions are intentionally concealed.
In a study published in Nano Letters, the researchers describe a BandAid-sized patch that measures several body signals linked to emotional states, including temperature, humidity, heart rate and blood oxygen levels.
The device is built from thin, flexible layers of metals such as platinum and gold, shaped to remain sensitive even when bent, pulled or twisted during natural facial movement.
How the emotion-tracking patch works
To reduce measurement errors, the sensors are arranged so they operate independently, with protective layers intended to prevent stretching or moisture from distorting readings from neighboring components.
Alongside the biosignals, facial strain sensors capture subtle changes in expression, and the system fuses those inputs to separate acted emotions from those tied to physiological responses.
AI training and early accuracy results
The researchers trained an AI model using repeated facial-expression performances across six categories: happiness, surprise, fear, sadness, anger and disgust.
In the reported tests, the model classified performed facial expressions with 96.28% accuracy, based on data collected while participants repeatedly displayed each expression.
To probe real emotions, participants watched video clips intended to elicit feelings while the patch tracked physiological changes associated with emotional arousal.
The system identified emotions with 88.83% accuracy in those tests, with sensor readings aligning with known links between emotions and changes in metrics such as skin temperature and heart rate.
Potential uses in telemedicine care
The patch wirelessly transmits measurements to mobile devices and cloud systems, which the researchers say could support remote monitoring in telemedicine settings.
The team also says the device is designed to avoid collecting personal information beyond sensor signals, aiming to reduce privacy risks while still enabling clinical interpretation.
While the work remains at a research stage, the authors suggest the platform could eventually support broader health applications, including monitoring non-verbal patients and tracking conditions where behavioral signals are difficult to assess.
The project was supported by funding from the U.S. National Institutes of Health and the U.S. National Science Foundation, according to the researchers.
Leave a Reply