(Nanowerk Spotlight) Each exhaled breath carries distinct signatures of human activity – quick and shallow during exercise, deep and regular in sleep, irregular while speaking. These breathing patterns provide a window into behavior, but capturing and interpreting these subtle respiratory variations has challenged scientists and engineers.
Camera systems can track physical movements but miss physiological nuances. Wearable devices require multiple sensors that increase bulk and complexity. Traditional humidity sensors, while promising for breath detection, lack the sensitivity and response speed (around 2.2 seconds for recovery in this sensor) required to capture rapid moisture changes in exhaled air. The technical challenge lies in detecting minute humidity variations quickly enough to map them to specific behaviors.
Chinese researchers have engineered a solution: a humidity sensor that uses microscopic “nanoforests” to detect subtle changes in breath moisture. Published in Microsystems & Nanoengineering (“An intelligent humidity sensing system for human behavior recognition”), their system combines these specialized structures with precise temperature control to achieve 96.2% accuracy in distinguishing between nine different human behaviors.
The sensor’s nanoforests create an extensive surface area covered in hydrophilic groups – molecular structures that readily interact with water molecules. Elevated operating temperatures further increase the activity and diffusion speed of these molecules, significantly boosting sensor sensitivity.
When a person exhales, water vapor from their breath adheres to these surfaces through hydrogen bonding, forming initial chemical bonds. As humidity increases, additional water molecules stack onto this first layer through weaker physical bonds, creating multiple layers that the sensor can detect.
a Fabrication process for the nanoforest (NF)-based humidity sensor. b An optical image of the humidity sensor. c, d SEM images of NFs on the device. (Image: Reprinted from DOI:10.1038/s41378-024-00863-6, CC BY) (click on image to enlarge)
A built-in micro-heater maintains the sensor at 57.1°C, increasing its sensitivity by nearly six times compared to room temperature operation. This enhancement allows detection of even slight variations in breath moisture. An integrated thermistor continuously monitors temperature, providing additional data about breathing patterns.
The system processes this combined humidity and temperature data through a machine learning algorithm that converts the measurements into two-dimensional maps. These maps serve as input for a neural network trained to recognize specific behaviors. The researchers tested the system’s ability to identify nine distinct states: working, speaking, walking, playing electronic games, sleeping, sighing, breath holding, jumping, and exercising.
The results demonstrated perfect recognition of five behaviors – working, walking, sleeping, sighing, and breath holding. The system occasionally confused similar activities, such as misclassifying jumping as exercise due to comparable breathing patterns. Speaking was sometimes mistaken for gaming activity, likely due to overlapping respiratory characteristics.
The sensor maintained consistent performance through more than 1,000 consecutive breathing cycles, demonstrating robust stability. It also showed high selectivity for water vapor compared to other breath components like oxygen, carbon dioxide, and nitrogen, ensuring accurate humidity measurements even in complex respiratory environments.
The researchers integrated their sensor into a face mask that wirelessly transmits breathing data to smartphones or computers for real-time analysis. This implementation enables continuous behavior monitoring without requiring multiple devices or complex setups.
The technology offers particular utility in healthcare settings, where automated behavior tracking could help monitor patient activity levels and sleep patterns. In smart homes, the system could adjust environmental controls based on detected behaviors. The non-invasive nature of humidity sensing preserves privacy while providing detailed insights into physical and physiological states.
The sensor’s ability to extract behavioral information from breath moisture represents a shift in human activity monitoring. By focusing on this single, information-rich parameter, the system achieves sophisticated behavior recognition without the complexity of multiple sensor types or the privacy concerns of video monitoring.
The research demonstrates how precise measurement of a fundamental physiological process – breathing – can reveal complex patterns of human behavior. This technical advancement brings automated behavior recognition closer to practical implementation in healthcare and daily life applications.
Get our Nanotechnology Spotlight updates to your inbox!
Thank you!
You have successfully joined our subscriber list.
Become a Spotlight guest author! Join our large and growing group of guest contributors. Have you just published a scientific paper or have other exciting developments to share with the nanotechnology community? Here is how to publish on nanowerk.com.
