Researchers at Columbia University have developed a smart pillowcase that monitors heart rate, breathing, body movement and sleeping position without requiring users to wear a tracking device. The experimental system, called PillowSense, uses conductive fabric to collect physiological signals through contact with the neck. Initial home testing suggests that the technology could offer a less intrusive alternative to conventional sleep trackers, although further research is needed before it can be used for medical purposes.
Sleep monitoring has become increasingly accessible through smartwatches, fitness trackers and other consumer devices. These technologies can provide information about sleep duration, movement and physiological activity, but they also have limitations. Some people find wearable devices uncomfortable during sleep, while the accuracy of individual measurements can vary.
An interdisciplinary research team led by Xia Zhou, associate professor at Columbia Engineering, investigated whether everyday textiles could provide an alternative. Their solution, PillowSense, integrates soft electronic sensors into a pillowcase. The system was presented at the 2026 ACM International Joint Conference on Pervasive and Ubiquitous Computing (UbiComp), where the research received a Distinguished Paper Award.
Measuring sleep through conductive fabric
PillowSense uses conductive fabric sewn into a pillowcase to detect electrical signals generated by the body. These include signals associated with cardiac activity and muscle contractions. The fabric functions as an electrode, while sensitive electronics measure small voltage differences between sensors that come into contact with the sleeper's neck.
Signal-processing algorithms separate cardiac and muscular activity, allowing the system to estimate heart rate and breathing frequency. Machine learning is subsequently used to recognise body movements and different sleeping positions. Unlike many conventional sleep trackers, which primarily collect measurements from the wrist or finger, PillowSense also provides information about posture.
This is relevant because sleeping position can influence breathing and comfort. Combining positional information with physiological measurements could therefore provide researchers with a more detailed picture of what happens during sleep. The researchers also discovered that electrical activity in the neck muscles could provide information about the position of other body parts, including whether a person's legs were bent. This made it possible to distinguish more detailed sleeping postures using sensors located exclusively in the pillowcase.
Home testing demonstrates potential
The researchers evaluated PillowSense during 151 nights of sleep in participants' homes. Testing in a natural environment allowed them to investigate how the system performed during ordinary overnight movements and changes in sleeping position. The technology distinguished seven sleeping postures with an F1 score of 88%. This statistical measure combines precision and recall to evaluate classification performance. The findings suggest that fabric-based sensors can identify sleeping positions with relatively high accuracy under the conditions studied.
However, the researchers also identified several technical limitations. Sensor contact can change when someone moves, while hair and differences in skin condition can interfere with signal quality. Poor contact occasionally resulted in missing measurements. Importantly, periods with insufficient sensor contact were excluded from the reported posture-classification results. The 88% performance figure therefore does not represent uninterrupted accuracy throughout every night. The participants were also healthy individuals, meaning the findings cannot automatically be applied to older adults or patients with sleep disorders.
What can sleep trackers actually measure?
The development comes amid growing interest in consumer sleep monitoring and questions about the clinical value of the information these devices provide. As we previously reported, the increasing popularity of sleep trackers raises an important distinction between measuring physiological signals and accurately determining sleep quality. Consumer devices typically estimate sleep-related parameters using indirect measurements such as movement, heart rate and variations in cardiac activity. Algorithms then translate these signals into estimates of sleep duration, sleep stages or other indicators.
These estimates should not automatically be regarded as equivalent to measurements obtained through clinical sleep studies. PillowSense approaches monitoring differently by collecting electrical signals through fabric and combining them with information about body position. Nevertheless, it faces a similar challenge: demonstrating that its measurements reliably reflect clinically relevant aspects of sleep.
The ability to identify posture or estimate breathing and heart rate does not, by itself, establish whether someone has a sleep disorder or is receiving sufficient restorative sleep. For that reason, the researchers describe PillowSense as a monitoring prototype rather than a diagnostic system.
Practical application
The development team designed the pillowcase with everyday use in mind. Its sensing layer can be detached for hand washing, while the remaining fabric can be machine washed. The researchers envisage a future version that could be placed over an ordinary pillow and connected to a smartphone application. Users could then review overnight changes in posture, movement, heart rate and breathing. Processing measurements locally on a smartphone could potentially reduce the need to transmit sensitive physiological information to external servers. Appropriate data storage and user controls would nevertheless remain important.
Before wider deployment, PillowSense will require validation against established clinical reference measurements, as well as testing in larger and more diverse populations. Further improvements are also needed to maintain reliable sensor contact, reduce interference and simplify the electronics.
References
Add ICT&health on Google
Show more content from ICT&health in Google Search.
