Research
A Pillowcase that Tracks Sleep
A pillowcase monitoring system could offer a more comfortable and personalized way to understand what happens while we sleep.
An interdisciplinary team of researchers led by Columbia Engineering Associate Professor Xia Zhou has developed PillowSense, a fabric-based sleep monitoring system. The novel system uses soft sensors to measure heart rate, breathing rate, body movement, and sleep posture through natural contact with the sleeper’s neck.
Zhou, with the computer science department at Columbia Engineering, worked with experts in computing, engineering, and sleep health to develop PillowSense. The system is detailed in “My Pillow Knows My Sleep: Sleep Monitoring with Computational Fabrics in the Pillowcase,” presented at the Pervasive and Ubiquitous Computing Conference (UbiComp ’26) and received a Distinguished Paper Award at UbiComp in October, 2026. The project is supported by Columbia’s Seed Funding Program for Interdisciplinary Research (RISE).
Over 151 nights of at-home testing, PillowSense distinguished seven sleep postures with an F1 score of 88%, a measure of classification performance. The findings suggest a way to collect detailed sleep information without requiring people to wear a device overnight.
Why sleep position matters
Heart rate and breathing measurements provide useful information about sleep, but posture adds another part of the picture. Sleeping position can affect breathing and comfort, while changes in position reveal behavior that vital signs alone do not capture.
“It provides the physical context that makes physiological measurements such as heart rate or breathing more meaningful,” Zhou explained.
The project grew out of the group’s work on sensing physical activity and motion using everyday fabrics. The researchers wanted to explore whether fabric sensors could also measure physiological signals. Sleep offered a compelling application: monitoring needs to capture useful information without disturbing the person being observed.
Reading signals through fabric
Every heartbeat and muscle contraction produces electrical signals that can be detected at the skin’s surface. PillowSense captures these signals using conductive fabric sewn into a pillowcase. The fabric acts as an electrode, while sensitive electronics measure small voltage differences between sensors touching the sleeper’s neck.
The system separates signals from the heart and neck muscles, using them to estimate heart and breathing rates and identify patterns associated with different sleeping positions.
One unexpected discovery expanded what the researchers could detect: neck muscle activity could also reflect whether the sleeper’s legs were bent.
“This opened up the possibility of recognizing more detailed sleep postures using sensors confined to a pillowcase,” said the study’s lead author, Qijia Shao PhD’26, Zhou’s former doctoral student, now an assistant professor at the Hong Kong University of Science and Technology.
Testing during natural sleep
Making the system work throughout an ordinary night presented challenges. When people shift position, the contact between their neck and the pillow changes. Differences in skin condition and hair can also affect signal quality.
The team developed signal-processing and machine learning methods to handle those variations. Testing in people’s homes allowed the researchers to observe movements over full nights of sleep, beyond what shorter laboratory sessions could capture.
Among the healthy participants studied, movement was relatively infrequent. That finding also had practical implications: because movement triggers the more computationally demanding posture-classification process, fewer movements meant less unnecessary processing.
The results came with an important qualification. Poor sensor contact sometimes created gaps in the data, and those gaps were excluded from the reported posture-classification results.
A design for everyday use
Comfort and maintenance shaped the researchers’ approach to design. The sensing layer detaches for hand washing, while the rest of the pillowcase is machine washable.
The researchers envision a system that people could slip over their own pillow and pair with a smartphone app. In the morning, users could review how their posture, movement, heart rate, and breathing changed overnight. Local processing on the phone could help protect privacy, alongside secure storage and controls for keeping, deleting, or sharing data.
The approach could benefit people who find wearable sensors uncomfortable or difficult to tolerate. It could also help sleep researchers collect observations over many nights at home.
Next steps
PillowSense remains a research prototype rather than a diagnostic tool. Clinical applications would require comparisons with clinical reference measurements and studies involving larger, more diverse populations, including older adults and people with sleep disorders.
The team also sees opportunities to improve sensor contact, address interference from hair, and simplify the electronics and initial setup. A companion app would need to show when measurements are incomplete or uncertain.
“Moving beyond a research prototype means showing that the system remains accurate, comfortable, affordable, and easy to maintain over months of everyday use,” Zhou said.